/src/libavif/ext/libyuv/source/convert_argb.cc
Line | Count | Source |
1 | | /* |
2 | | * Copyright 2011 The LibYuv Project Authors. All rights reserved. |
3 | | * |
4 | | * Use of this source code is governed by a BSD-style license |
5 | | * that can be found in the LICENSE file in the root of the source |
6 | | * tree. An additional intellectual property rights grant can be found |
7 | | * in the file PATENTS. All contributing project authors may |
8 | | * be found in the AUTHORS file in the root of the source tree. |
9 | | */ |
10 | | |
11 | | #include "libyuv/convert_argb.h" |
12 | | |
13 | | #include <assert.h> |
14 | | |
15 | | #include "libyuv/convert_from_argb.h" |
16 | | #include "libyuv/cpu_id.h" |
17 | | #include "libyuv/planar_functions.h" // For CopyPlane and ARGBShuffle. |
18 | | #include "libyuv/rotate_argb.h" |
19 | | #include "libyuv/row.h" |
20 | | #include "libyuv/scale_row.h" // For ScaleRowUp2_Linear and ScaleRowUp2_Bilinear |
21 | | #include "libyuv/video_common.h" |
22 | | |
23 | | #ifdef __cplusplus |
24 | | namespace libyuv { |
25 | | extern "C" { |
26 | | #endif |
27 | | |
28 | | // Copy ARGB with optional flipping |
29 | | LIBYUV_API |
30 | | int ARGBCopy(const uint8_t* src_argb, |
31 | | int src_stride_argb, |
32 | | uint8_t* dst_argb, |
33 | | int dst_stride_argb, |
34 | | int width, |
35 | 0 | int height) { |
36 | 0 | if (!src_argb || !dst_argb || width <= 0 || height == 0) { |
37 | 0 | return -1; |
38 | 0 | } |
39 | | // Negative height means invert the image. |
40 | 0 | if (height < 0) { |
41 | 0 | height = -height; |
42 | 0 | src_argb = src_argb + (height - 1) * src_stride_argb; |
43 | 0 | src_stride_argb = -src_stride_argb; |
44 | 0 | } |
45 | |
|
46 | 0 | CopyPlane(src_argb, src_stride_argb, dst_argb, dst_stride_argb, width * 4, |
47 | 0 | height); |
48 | 0 | return 0; |
49 | 0 | } |
50 | | |
51 | | // Convert I420 to ARGB with matrix. |
52 | | LIBYUV_API |
53 | | int I420ToARGBMatrix(const uint8_t* src_y, |
54 | | int src_stride_y, |
55 | | const uint8_t* src_u, |
56 | | int src_stride_u, |
57 | | const uint8_t* src_v, |
58 | | int src_stride_v, |
59 | | uint8_t* dst_argb, |
60 | | int dst_stride_argb, |
61 | | const struct YuvConstants* yuvconstants, |
62 | | int width, |
63 | 152 | int height) { |
64 | 152 | int y; |
65 | 152 | void (*I422ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
66 | 152 | const uint8_t* v_buf, uint8_t* rgb_buf, |
67 | 152 | const struct YuvConstants* yuvconstants, int width) = |
68 | 152 | I422ToARGBRow_C; |
69 | 152 | assert(yuvconstants); |
70 | 152 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
71 | 0 | return -1; |
72 | 0 | } |
73 | | // Negative height means invert the image. |
74 | 152 | if (height < 0) { |
75 | 0 | height = -height; |
76 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
77 | 0 | dst_stride_argb = -dst_stride_argb; |
78 | 0 | } |
79 | 152 | #if defined(HAS_I422TOARGBROW_SSSE3) |
80 | 152 | if (TestCpuFlag(kCpuHasSSSE3)) { |
81 | 152 | I422ToARGBRow = I422ToARGBRow_Any_SSSE3; |
82 | 152 | if (IS_ALIGNED(width, 8)) { |
83 | 35 | I422ToARGBRow = I422ToARGBRow_SSSE3; |
84 | 35 | } |
85 | 152 | } |
86 | 152 | #endif |
87 | 152 | #if defined(HAS_I422TOARGBROW_AVX2) |
88 | 152 | if (TestCpuFlag(kCpuHasAVX2)) { |
89 | 152 | I422ToARGBRow = I422ToARGBRow_Any_AVX2; |
90 | 152 | if (IS_ALIGNED(width, 16)) { |
91 | 17 | I422ToARGBRow = I422ToARGBRow_AVX2; |
92 | 17 | } |
93 | 152 | } |
94 | 152 | #endif |
95 | 152 | #if defined(HAS_I422TOARGBROW_AVX512BW) |
96 | 152 | if (TestCpuFlag(kCpuHasAVX512BW | kCpuHasAVX512VL) == |
97 | 152 | (kCpuHasAVX512BW | kCpuHasAVX512VL)) { |
98 | 0 | I422ToARGBRow = I422ToARGBRow_Any_AVX512BW; |
99 | 0 | if (IS_ALIGNED(width, 32)) { |
100 | 0 | I422ToARGBRow = I422ToARGBRow_AVX512BW; |
101 | 0 | } |
102 | 0 | } |
103 | 152 | #endif |
104 | | #if defined(HAS_I422TOARGBROW_NEON) |
105 | | if (TestCpuFlag(kCpuHasNEON)) { |
106 | | I422ToARGBRow = I422ToARGBRow_Any_NEON; |
107 | | if (IS_ALIGNED(width, 8)) { |
108 | | I422ToARGBRow = I422ToARGBRow_NEON; |
109 | | } |
110 | | } |
111 | | #endif |
112 | | #if defined(HAS_I422TOARGBROW_SVE2) |
113 | | if (TestCpuFlag(kCpuHasSVE2)) { |
114 | | I422ToARGBRow = I422ToARGBRow_SVE2; |
115 | | } |
116 | | #endif |
117 | | #if defined(HAS_I422TOARGBROW_SME) |
118 | | if (TestCpuFlag(kCpuHasSME)) { |
119 | | I422ToARGBRow = I422ToARGBRow_SME; |
120 | | } |
121 | | #endif |
122 | | #if defined(HAS_I422TOARGBROW_MSA) |
123 | | if (TestCpuFlag(kCpuHasMSA)) { |
124 | | I422ToARGBRow = I422ToARGBRow_Any_MSA; |
125 | | if (IS_ALIGNED(width, 8)) { |
126 | | I422ToARGBRow = I422ToARGBRow_MSA; |
127 | | } |
128 | | } |
129 | | #endif |
130 | | #if defined(HAS_I422TOARGBROW_LSX) |
131 | | if (TestCpuFlag(kCpuHasLSX)) { |
132 | | I422ToARGBRow = I422ToARGBRow_Any_LSX; |
133 | | if (IS_ALIGNED(width, 16)) { |
134 | | I422ToARGBRow = I422ToARGBRow_LSX; |
135 | | } |
136 | | } |
137 | | #endif |
138 | | #if defined(HAS_I422TOARGBROW_LASX) |
139 | | if (TestCpuFlag(kCpuHasLASX)) { |
140 | | I422ToARGBRow = I422ToARGBRow_Any_LASX; |
141 | | if (IS_ALIGNED(width, 32)) { |
142 | | I422ToARGBRow = I422ToARGBRow_LASX; |
143 | | } |
144 | | } |
145 | | #endif |
146 | | #if defined(HAS_I422TOARGBROW_RVV) |
147 | | if (TestCpuFlag(kCpuHasRVV)) { |
148 | | I422ToARGBRow = I422ToARGBRow_RVV; |
149 | | } |
150 | | #endif |
151 | | |
152 | 2.52k | for (y = 0; y < height; ++y) { |
153 | 2.37k | I422ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
154 | 2.37k | dst_argb += dst_stride_argb; |
155 | 2.37k | src_y += src_stride_y; |
156 | 2.37k | if (y & 1) { |
157 | 1.16k | src_u += src_stride_u; |
158 | 1.16k | src_v += src_stride_v; |
159 | 1.16k | } |
160 | 2.37k | } |
161 | 152 | return 0; |
162 | 152 | } |
163 | | |
164 | | // Convert I420 to ARGB. |
165 | | LIBYUV_API |
166 | | int I420ToARGB(const uint8_t* src_y, |
167 | | int src_stride_y, |
168 | | const uint8_t* src_u, |
169 | | int src_stride_u, |
170 | | const uint8_t* src_v, |
171 | | int src_stride_v, |
172 | | uint8_t* dst_argb, |
173 | | int dst_stride_argb, |
174 | | int width, |
175 | 0 | int height) { |
176 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
177 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
178 | 0 | &kYuvI601Constants, width, height); |
179 | 0 | } |
180 | | |
181 | | // Convert I420 to ABGR. |
182 | | LIBYUV_API |
183 | | int I420ToABGR(const uint8_t* src_y, |
184 | | int src_stride_y, |
185 | | const uint8_t* src_u, |
186 | | int src_stride_u, |
187 | | const uint8_t* src_v, |
188 | | int src_stride_v, |
189 | | uint8_t* dst_abgr, |
190 | | int dst_stride_abgr, |
191 | | int width, |
192 | 0 | int height) { |
193 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
194 | 0 | src_stride_v, // Swap U and V |
195 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
196 | 0 | &kYvuI601Constants, // Use Yvu matrix |
197 | 0 | width, height); |
198 | 0 | } |
199 | | |
200 | | // Convert J420 to ARGB. |
201 | | LIBYUV_API |
202 | | int J420ToARGB(const uint8_t* src_y, |
203 | | int src_stride_y, |
204 | | const uint8_t* src_u, |
205 | | int src_stride_u, |
206 | | const uint8_t* src_v, |
207 | | int src_stride_v, |
208 | | uint8_t* dst_argb, |
209 | | int dst_stride_argb, |
210 | | int width, |
211 | 0 | int height) { |
212 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
213 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
214 | 0 | &kYuvJPEGConstants, width, height); |
215 | 0 | } |
216 | | |
217 | | // Convert J420 to ABGR. |
218 | | LIBYUV_API |
219 | | int J420ToABGR(const uint8_t* src_y, |
220 | | int src_stride_y, |
221 | | const uint8_t* src_u, |
222 | | int src_stride_u, |
223 | | const uint8_t* src_v, |
224 | | int src_stride_v, |
225 | | uint8_t* dst_abgr, |
226 | | int dst_stride_abgr, |
227 | | int width, |
228 | 0 | int height) { |
229 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
230 | 0 | src_stride_v, // Swap U and V |
231 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
232 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
233 | 0 | width, height); |
234 | 0 | } |
235 | | |
236 | | // Convert H420 to ARGB. |
237 | | LIBYUV_API |
238 | | int H420ToARGB(const uint8_t* src_y, |
239 | | int src_stride_y, |
240 | | const uint8_t* src_u, |
241 | | int src_stride_u, |
242 | | const uint8_t* src_v, |
243 | | int src_stride_v, |
244 | | uint8_t* dst_argb, |
245 | | int dst_stride_argb, |
246 | | int width, |
247 | 0 | int height) { |
248 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
249 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
250 | 0 | &kYuvH709Constants, width, height); |
251 | 0 | } |
252 | | |
253 | | // Convert H420 to ABGR. |
254 | | LIBYUV_API |
255 | | int H420ToABGR(const uint8_t* src_y, |
256 | | int src_stride_y, |
257 | | const uint8_t* src_u, |
258 | | int src_stride_u, |
259 | | const uint8_t* src_v, |
260 | | int src_stride_v, |
261 | | uint8_t* dst_abgr, |
262 | | int dst_stride_abgr, |
263 | | int width, |
264 | 0 | int height) { |
265 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
266 | 0 | src_stride_v, // Swap U and V |
267 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
268 | 0 | &kYvuH709Constants, // Use Yvu matrix |
269 | 0 | width, height); |
270 | 0 | } |
271 | | |
272 | | // Convert U420 to ARGB. |
273 | | LIBYUV_API |
274 | | int U420ToARGB(const uint8_t* src_y, |
275 | | int src_stride_y, |
276 | | const uint8_t* src_u, |
277 | | int src_stride_u, |
278 | | const uint8_t* src_v, |
279 | | int src_stride_v, |
280 | | uint8_t* dst_argb, |
281 | | int dst_stride_argb, |
282 | | int width, |
283 | 0 | int height) { |
284 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
285 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
286 | 0 | &kYuv2020Constants, width, height); |
287 | 0 | } |
288 | | |
289 | | // Convert U420 to ABGR. |
290 | | LIBYUV_API |
291 | | int U420ToABGR(const uint8_t* src_y, |
292 | | int src_stride_y, |
293 | | const uint8_t* src_u, |
294 | | int src_stride_u, |
295 | | const uint8_t* src_v, |
296 | | int src_stride_v, |
297 | | uint8_t* dst_abgr, |
298 | | int dst_stride_abgr, |
299 | | int width, |
300 | 0 | int height) { |
301 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
302 | 0 | src_stride_v, // Swap U and V |
303 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
304 | 0 | &kYvu2020Constants, // Use Yvu matrix |
305 | 0 | width, height); |
306 | 0 | } |
307 | | |
308 | | // Convert I422 to ARGB with matrix. |
309 | | LIBYUV_API |
310 | | int I422ToARGBMatrix(const uint8_t* src_y, |
311 | | int src_stride_y, |
312 | | const uint8_t* src_u, |
313 | | int src_stride_u, |
314 | | const uint8_t* src_v, |
315 | | int src_stride_v, |
316 | | uint8_t* dst_argb, |
317 | | int dst_stride_argb, |
318 | | const struct YuvConstants* yuvconstants, |
319 | | int width, |
320 | 108 | int height) { |
321 | 108 | int y; |
322 | 108 | void (*I422ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
323 | 108 | const uint8_t* v_buf, uint8_t* rgb_buf, |
324 | 108 | const struct YuvConstants* yuvconstants, int width) = |
325 | 108 | I422ToARGBRow_C; |
326 | 108 | assert(yuvconstants); |
327 | 108 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
328 | 0 | return -1; |
329 | 0 | } |
330 | | // Negative height means invert the image. |
331 | 108 | if (height < 0) { |
332 | 0 | height = -height; |
333 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
334 | 0 | dst_stride_argb = -dst_stride_argb; |
335 | 0 | } |
336 | | // Coalesce rows. |
337 | 108 | if (src_stride_y == width && src_stride_u * 2 == width && |
338 | 57 | src_stride_v * 2 == width && dst_stride_argb == width * 4) { |
339 | 57 | width *= height; |
340 | 57 | height = 1; |
341 | 57 | src_stride_y = src_stride_u = src_stride_v = dst_stride_argb = 0; |
342 | 57 | } |
343 | 108 | #if defined(HAS_I422TOARGBROW_SSSE3) |
344 | 108 | if (TestCpuFlag(kCpuHasSSSE3)) { |
345 | 108 | I422ToARGBRow = I422ToARGBRow_Any_SSSE3; |
346 | 108 | if (IS_ALIGNED(width, 8)) { |
347 | 37 | I422ToARGBRow = I422ToARGBRow_SSSE3; |
348 | 37 | } |
349 | 108 | } |
350 | 108 | #endif |
351 | 108 | #if defined(HAS_I422TOARGBROW_AVX2) |
352 | 108 | if (TestCpuFlag(kCpuHasAVX2)) { |
353 | 108 | I422ToARGBRow = I422ToARGBRow_Any_AVX2; |
354 | 108 | if (IS_ALIGNED(width, 16)) { |
355 | 17 | I422ToARGBRow = I422ToARGBRow_AVX2; |
356 | 17 | } |
357 | 108 | } |
358 | 108 | #endif |
359 | 108 | #if defined(HAS_I422TOARGBROW_AVX512BW) |
360 | 108 | if (TestCpuFlag(kCpuHasAVX512BW | kCpuHasAVX512VL) == |
361 | 108 | (kCpuHasAVX512BW | kCpuHasAVX512VL)) { |
362 | 0 | I422ToARGBRow = I422ToARGBRow_Any_AVX512BW; |
363 | 0 | if (IS_ALIGNED(width, 32)) { |
364 | 0 | I422ToARGBRow = I422ToARGBRow_AVX512BW; |
365 | 0 | } |
366 | 0 | } |
367 | 108 | #endif |
368 | | #if defined(HAS_I422TOARGBROW_NEON) |
369 | | if (TestCpuFlag(kCpuHasNEON)) { |
370 | | I422ToARGBRow = I422ToARGBRow_Any_NEON; |
371 | | if (IS_ALIGNED(width, 8)) { |
372 | | I422ToARGBRow = I422ToARGBRow_NEON; |
373 | | } |
374 | | } |
375 | | #endif |
376 | | #if defined(HAS_I422TOARGBROW_SVE2) |
377 | | if (TestCpuFlag(kCpuHasSVE2)) { |
378 | | I422ToARGBRow = I422ToARGBRow_SVE2; |
379 | | } |
380 | | #endif |
381 | | #if defined(HAS_I422TOARGBROW_SME) |
382 | | if (TestCpuFlag(kCpuHasSME)) { |
383 | | I422ToARGBRow = I422ToARGBRow_SME; |
384 | | } |
385 | | #endif |
386 | | #if defined(HAS_I422TOARGBROW_MSA) |
387 | | if (TestCpuFlag(kCpuHasMSA)) { |
388 | | I422ToARGBRow = I422ToARGBRow_Any_MSA; |
389 | | if (IS_ALIGNED(width, 8)) { |
390 | | I422ToARGBRow = I422ToARGBRow_MSA; |
391 | | } |
392 | | } |
393 | | #endif |
394 | | #if defined(HAS_I422TOARGBROW_LSX) |
395 | | if (TestCpuFlag(kCpuHasLSX)) { |
396 | | I422ToARGBRow = I422ToARGBRow_Any_LSX; |
397 | | if (IS_ALIGNED(width, 16)) { |
398 | | I422ToARGBRow = I422ToARGBRow_LSX; |
399 | | } |
400 | | } |
401 | | #endif |
402 | | #if defined(HAS_I422TOARGBROW_LASX) |
403 | | if (TestCpuFlag(kCpuHasLASX)) { |
404 | | I422ToARGBRow = I422ToARGBRow_Any_LASX; |
405 | | if (IS_ALIGNED(width, 32)) { |
406 | | I422ToARGBRow = I422ToARGBRow_LASX; |
407 | | } |
408 | | } |
409 | | #endif |
410 | | #if defined(HAS_I422TOARGBROW_RVV) |
411 | | if (TestCpuFlag(kCpuHasRVV)) { |
412 | | I422ToARGBRow = I422ToARGBRow_RVV; |
413 | | } |
414 | | #endif |
415 | | |
416 | 1.68k | for (y = 0; y < height; ++y) { |
417 | 1.58k | I422ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
418 | 1.58k | dst_argb += dst_stride_argb; |
419 | 1.58k | src_y += src_stride_y; |
420 | 1.58k | src_u += src_stride_u; |
421 | 1.58k | src_v += src_stride_v; |
422 | 1.58k | } |
423 | 108 | return 0; |
424 | 108 | } |
425 | | |
426 | | // Convert I422 to ARGB. |
427 | | LIBYUV_API |
428 | | int I422ToARGB(const uint8_t* src_y, |
429 | | int src_stride_y, |
430 | | const uint8_t* src_u, |
431 | | int src_stride_u, |
432 | | const uint8_t* src_v, |
433 | | int src_stride_v, |
434 | | uint8_t* dst_argb, |
435 | | int dst_stride_argb, |
436 | | int width, |
437 | 0 | int height) { |
438 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
439 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
440 | 0 | &kYuvI601Constants, width, height); |
441 | 0 | } |
442 | | |
443 | | // Convert I422 to ABGR. |
444 | | LIBYUV_API |
445 | | int I422ToABGR(const uint8_t* src_y, |
446 | | int src_stride_y, |
447 | | const uint8_t* src_u, |
448 | | int src_stride_u, |
449 | | const uint8_t* src_v, |
450 | | int src_stride_v, |
451 | | uint8_t* dst_abgr, |
452 | | int dst_stride_abgr, |
453 | | int width, |
454 | 0 | int height) { |
455 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
456 | 0 | src_stride_v, // Swap U and V |
457 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
458 | 0 | &kYvuI601Constants, // Use Yvu matrix |
459 | 0 | width, height); |
460 | 0 | } |
461 | | |
462 | | // Convert J422 to ARGB. |
463 | | LIBYUV_API |
464 | | int J422ToARGB(const uint8_t* src_y, |
465 | | int src_stride_y, |
466 | | const uint8_t* src_u, |
467 | | int src_stride_u, |
468 | | const uint8_t* src_v, |
469 | | int src_stride_v, |
470 | | uint8_t* dst_argb, |
471 | | int dst_stride_argb, |
472 | | int width, |
473 | 0 | int height) { |
474 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
475 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
476 | 0 | &kYuvJPEGConstants, width, height); |
477 | 0 | } |
478 | | |
479 | | // Convert J422 to ABGR. |
480 | | LIBYUV_API |
481 | | int J422ToABGR(const uint8_t* src_y, |
482 | | int src_stride_y, |
483 | | const uint8_t* src_u, |
484 | | int src_stride_u, |
485 | | const uint8_t* src_v, |
486 | | int src_stride_v, |
487 | | uint8_t* dst_abgr, |
488 | | int dst_stride_abgr, |
489 | | int width, |
490 | 0 | int height) { |
491 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
492 | 0 | src_stride_v, // Swap U and V |
493 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
494 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
495 | 0 | width, height); |
496 | 0 | } |
497 | | |
498 | | // Convert H422 to ARGB. |
499 | | LIBYUV_API |
500 | | int H422ToARGB(const uint8_t* src_y, |
501 | | int src_stride_y, |
502 | | const uint8_t* src_u, |
503 | | int src_stride_u, |
504 | | const uint8_t* src_v, |
505 | | int src_stride_v, |
506 | | uint8_t* dst_argb, |
507 | | int dst_stride_argb, |
508 | | int width, |
509 | 0 | int height) { |
510 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
511 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
512 | 0 | &kYuvH709Constants, width, height); |
513 | 0 | } |
514 | | |
515 | | // Convert H422 to ABGR. |
516 | | LIBYUV_API |
517 | | int H422ToABGR(const uint8_t* src_y, |
518 | | int src_stride_y, |
519 | | const uint8_t* src_u, |
520 | | int src_stride_u, |
521 | | const uint8_t* src_v, |
522 | | int src_stride_v, |
523 | | uint8_t* dst_abgr, |
524 | | int dst_stride_abgr, |
525 | | int width, |
526 | 0 | int height) { |
527 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
528 | 0 | src_stride_v, // Swap U and V |
529 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
530 | 0 | &kYvuH709Constants, // Use Yvu matrix |
531 | 0 | width, height); |
532 | 0 | } |
533 | | |
534 | | // Convert U422 to ARGB. |
535 | | LIBYUV_API |
536 | | int U422ToARGB(const uint8_t* src_y, |
537 | | int src_stride_y, |
538 | | const uint8_t* src_u, |
539 | | int src_stride_u, |
540 | | const uint8_t* src_v, |
541 | | int src_stride_v, |
542 | | uint8_t* dst_argb, |
543 | | int dst_stride_argb, |
544 | | int width, |
545 | 0 | int height) { |
546 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
547 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
548 | 0 | &kYuv2020Constants, width, height); |
549 | 0 | } |
550 | | |
551 | | // Convert U422 to ABGR. |
552 | | LIBYUV_API |
553 | | int U422ToABGR(const uint8_t* src_y, |
554 | | int src_stride_y, |
555 | | const uint8_t* src_u, |
556 | | int src_stride_u, |
557 | | const uint8_t* src_v, |
558 | | int src_stride_v, |
559 | | uint8_t* dst_abgr, |
560 | | int dst_stride_abgr, |
561 | | int width, |
562 | 0 | int height) { |
563 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
564 | 0 | src_stride_v, // Swap U and V |
565 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
566 | 0 | &kYvu2020Constants, // Use Yvu matrix |
567 | 0 | width, height); |
568 | 0 | } |
569 | | |
570 | | // Convert I444 to ARGB with matrix. |
571 | | LIBYUV_API |
572 | | int I444ToARGBMatrix(const uint8_t* src_y, |
573 | | int src_stride_y, |
574 | | const uint8_t* src_u, |
575 | | int src_stride_u, |
576 | | const uint8_t* src_v, |
577 | | int src_stride_v, |
578 | | uint8_t* dst_argb, |
579 | | int dst_stride_argb, |
580 | | const struct YuvConstants* yuvconstants, |
581 | | int width, |
582 | 76 | int height) { |
583 | 76 | int y; |
584 | 76 | void (*I444ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
585 | 76 | const uint8_t* v_buf, uint8_t* rgb_buf, |
586 | 76 | const struct YuvConstants* yuvconstants, int width) = |
587 | 76 | I444ToARGBRow_C; |
588 | 76 | assert(yuvconstants); |
589 | 76 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
590 | 0 | return -1; |
591 | 0 | } |
592 | | // Negative height means invert the image. |
593 | 76 | if (height < 0) { |
594 | 0 | height = -height; |
595 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
596 | 0 | dst_stride_argb = -dst_stride_argb; |
597 | 0 | } |
598 | | // Coalesce rows. |
599 | 76 | if (src_stride_y == width && src_stride_u == width && src_stride_v == width && |
600 | 76 | dst_stride_argb == width * 4) { |
601 | 76 | width *= height; |
602 | 76 | height = 1; |
603 | 76 | src_stride_y = src_stride_u = src_stride_v = dst_stride_argb = 0; |
604 | 76 | } |
605 | 76 | #if defined(HAS_I444TOARGBROW_SSSE3) |
606 | 76 | if (TestCpuFlag(kCpuHasSSSE3)) { |
607 | 76 | I444ToARGBRow = I444ToARGBRow_Any_SSSE3; |
608 | 76 | if (IS_ALIGNED(width, 8)) { |
609 | 37 | I444ToARGBRow = I444ToARGBRow_SSSE3; |
610 | 37 | } |
611 | 76 | } |
612 | 76 | #endif |
613 | 76 | #if defined(HAS_I444TOARGBROW_AVX2) |
614 | 76 | if (TestCpuFlag(kCpuHasAVX2)) { |
615 | 76 | I444ToARGBRow = I444ToARGBRow_Any_AVX2; |
616 | 76 | if (IS_ALIGNED(width, 16)) { |
617 | 17 | I444ToARGBRow = I444ToARGBRow_AVX2; |
618 | 17 | } |
619 | 76 | } |
620 | 76 | #endif |
621 | | #if defined(HAS_I444TOARGBROW_NEON) |
622 | | if (TestCpuFlag(kCpuHasNEON)) { |
623 | | I444ToARGBRow = I444ToARGBRow_Any_NEON; |
624 | | if (IS_ALIGNED(width, 8)) { |
625 | | I444ToARGBRow = I444ToARGBRow_NEON; |
626 | | } |
627 | | } |
628 | | #endif |
629 | | #if defined(HAS_I444TOARGBROW_SVE2) |
630 | | if (TestCpuFlag(kCpuHasSVE2)) { |
631 | | I444ToARGBRow = I444ToARGBRow_SVE2; |
632 | | } |
633 | | #endif |
634 | | #if defined(HAS_I444TOARGBROW_SME) |
635 | | if (TestCpuFlag(kCpuHasSME)) { |
636 | | I444ToARGBRow = I444ToARGBRow_SME; |
637 | | } |
638 | | #endif |
639 | | #if defined(HAS_I444TOARGBROW_MSA) |
640 | | if (TestCpuFlag(kCpuHasMSA)) { |
641 | | I444ToARGBRow = I444ToARGBRow_Any_MSA; |
642 | | if (IS_ALIGNED(width, 8)) { |
643 | | I444ToARGBRow = I444ToARGBRow_MSA; |
644 | | } |
645 | | } |
646 | | #endif |
647 | | #if defined(HAS_I444TOARGBROW_LSX) |
648 | | if (TestCpuFlag(kCpuHasLSX)) { |
649 | | I444ToARGBRow = I444ToARGBRow_Any_LSX; |
650 | | if (IS_ALIGNED(width, 16)) { |
651 | | I444ToARGBRow = I444ToARGBRow_LSX; |
652 | | } |
653 | | } |
654 | | #endif |
655 | | #if defined(HAS_I444TOARGBROW_RVV) |
656 | | if (TestCpuFlag(kCpuHasRVV)) { |
657 | | I444ToARGBRow = I444ToARGBRow_RVV; |
658 | | } |
659 | | #endif |
660 | | |
661 | 152 | for (y = 0; y < height; ++y) { |
662 | 76 | I444ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
663 | 76 | dst_argb += dst_stride_argb; |
664 | 76 | src_y += src_stride_y; |
665 | 76 | src_u += src_stride_u; |
666 | 76 | src_v += src_stride_v; |
667 | 76 | } |
668 | 76 | return 0; |
669 | 76 | } |
670 | | |
671 | | // Convert I444 to ARGB. |
672 | | LIBYUV_API |
673 | | int I444ToARGB(const uint8_t* src_y, |
674 | | int src_stride_y, |
675 | | const uint8_t* src_u, |
676 | | int src_stride_u, |
677 | | const uint8_t* src_v, |
678 | | int src_stride_v, |
679 | | uint8_t* dst_argb, |
680 | | int dst_stride_argb, |
681 | | int width, |
682 | 0 | int height) { |
683 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
684 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
685 | 0 | &kYuvI601Constants, width, height); |
686 | 0 | } |
687 | | |
688 | | // Convert I444 to ABGR. |
689 | | LIBYUV_API |
690 | | int I444ToABGR(const uint8_t* src_y, |
691 | | int src_stride_y, |
692 | | const uint8_t* src_u, |
693 | | int src_stride_u, |
694 | | const uint8_t* src_v, |
695 | | int src_stride_v, |
696 | | uint8_t* dst_abgr, |
697 | | int dst_stride_abgr, |
698 | | int width, |
699 | 0 | int height) { |
700 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
701 | 0 | src_stride_v, // Swap U and V |
702 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
703 | 0 | &kYvuI601Constants, // Use Yvu matrix |
704 | 0 | width, height); |
705 | 0 | } |
706 | | |
707 | | // Convert J444 to ARGB. |
708 | | LIBYUV_API |
709 | | int J444ToARGB(const uint8_t* src_y, |
710 | | int src_stride_y, |
711 | | const uint8_t* src_u, |
712 | | int src_stride_u, |
713 | | const uint8_t* src_v, |
714 | | int src_stride_v, |
715 | | uint8_t* dst_argb, |
716 | | int dst_stride_argb, |
717 | | int width, |
718 | 0 | int height) { |
719 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
720 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
721 | 0 | &kYuvJPEGConstants, width, height); |
722 | 0 | } |
723 | | |
724 | | // Convert J444 to ABGR. |
725 | | LIBYUV_API |
726 | | int J444ToABGR(const uint8_t* src_y, |
727 | | int src_stride_y, |
728 | | const uint8_t* src_u, |
729 | | int src_stride_u, |
730 | | const uint8_t* src_v, |
731 | | int src_stride_v, |
732 | | uint8_t* dst_abgr, |
733 | | int dst_stride_abgr, |
734 | | int width, |
735 | 0 | int height) { |
736 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
737 | 0 | src_stride_v, // Swap U and V |
738 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
739 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
740 | 0 | width, height); |
741 | 0 | } |
742 | | |
743 | | // Convert H444 to ARGB. |
744 | | LIBYUV_API |
745 | | int H444ToARGB(const uint8_t* src_y, |
746 | | int src_stride_y, |
747 | | const uint8_t* src_u, |
748 | | int src_stride_u, |
749 | | const uint8_t* src_v, |
750 | | int src_stride_v, |
751 | | uint8_t* dst_argb, |
752 | | int dst_stride_argb, |
753 | | int width, |
754 | 0 | int height) { |
755 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
756 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
757 | 0 | &kYuvH709Constants, width, height); |
758 | 0 | } |
759 | | |
760 | | // Convert H444 to ABGR. |
761 | | LIBYUV_API |
762 | | int H444ToABGR(const uint8_t* src_y, |
763 | | int src_stride_y, |
764 | | const uint8_t* src_u, |
765 | | int src_stride_u, |
766 | | const uint8_t* src_v, |
767 | | int src_stride_v, |
768 | | uint8_t* dst_abgr, |
769 | | int dst_stride_abgr, |
770 | | int width, |
771 | 0 | int height) { |
772 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
773 | 0 | src_stride_v, // Swap U and V |
774 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
775 | 0 | &kYvuH709Constants, // Use Yvu matrix |
776 | 0 | width, height); |
777 | 0 | } |
778 | | |
779 | | // Convert U444 to ARGB. |
780 | | LIBYUV_API |
781 | | int U444ToARGB(const uint8_t* src_y, |
782 | | int src_stride_y, |
783 | | const uint8_t* src_u, |
784 | | int src_stride_u, |
785 | | const uint8_t* src_v, |
786 | | int src_stride_v, |
787 | | uint8_t* dst_argb, |
788 | | int dst_stride_argb, |
789 | | int width, |
790 | 0 | int height) { |
791 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
792 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
793 | 0 | &kYuv2020Constants, width, height); |
794 | 0 | } |
795 | | |
796 | | // Convert U444 to ABGR. |
797 | | LIBYUV_API |
798 | | int U444ToABGR(const uint8_t* src_y, |
799 | | int src_stride_y, |
800 | | const uint8_t* src_u, |
801 | | int src_stride_u, |
802 | | const uint8_t* src_v, |
803 | | int src_stride_v, |
804 | | uint8_t* dst_abgr, |
805 | | int dst_stride_abgr, |
806 | | int width, |
807 | 0 | int height) { |
808 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
809 | 0 | src_stride_v, // Swap U and V |
810 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
811 | 0 | &kYvu2020Constants, // Use Yvu matrix |
812 | 0 | width, height); |
813 | 0 | } |
814 | | |
815 | | // Convert I444 to RGB24 with matrix. |
816 | | LIBYUV_API |
817 | | int I444ToRGB24Matrix(const uint8_t* src_y, |
818 | | int src_stride_y, |
819 | | const uint8_t* src_u, |
820 | | int src_stride_u, |
821 | | const uint8_t* src_v, |
822 | | int src_stride_v, |
823 | | uint8_t* dst_rgb24, |
824 | | int dst_stride_rgb24, |
825 | | const struct YuvConstants* yuvconstants, |
826 | | int width, |
827 | 123 | int height) { |
828 | 123 | int y; |
829 | 123 | void (*I444ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
830 | 123 | const uint8_t* v_buf, uint8_t* rgb_buf, |
831 | 123 | const struct YuvConstants* yuvconstants, int width) = |
832 | 123 | I444ToRGB24Row_C; |
833 | 123 | assert(yuvconstants); |
834 | 123 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
835 | 0 | return -1; |
836 | 0 | } |
837 | | // Negative height means invert the image. |
838 | 123 | if (height < 0) { |
839 | 0 | height = -height; |
840 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
841 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
842 | 0 | } |
843 | | // Coalesce rows. |
844 | 123 | if (src_stride_y == width && src_stride_u == width && src_stride_v == width && |
845 | 123 | dst_stride_rgb24 == width * 3) { |
846 | 123 | width *= height; |
847 | 123 | height = 1; |
848 | 123 | src_stride_y = src_stride_u = src_stride_v = dst_stride_rgb24 = 0; |
849 | 123 | } |
850 | 123 | #if defined(HAS_I444TORGB24ROW_SSSE3) |
851 | 123 | if (TestCpuFlag(kCpuHasSSSE3)) { |
852 | 123 | I444ToRGB24Row = I444ToRGB24Row_Any_SSSE3; |
853 | 123 | if (IS_ALIGNED(width, 16)) { |
854 | 60 | I444ToRGB24Row = I444ToRGB24Row_SSSE3; |
855 | 60 | } |
856 | 123 | } |
857 | 123 | #endif |
858 | 123 | #if defined(HAS_I444TORGB24ROW_AVX2) |
859 | 123 | if (TestCpuFlag(kCpuHasAVX2)) { |
860 | 123 | I444ToRGB24Row = I444ToRGB24Row_Any_AVX2; |
861 | 123 | if (IS_ALIGNED(width, 32)) { |
862 | 33 | I444ToRGB24Row = I444ToRGB24Row_AVX2; |
863 | 33 | } |
864 | 123 | } |
865 | 123 | #endif |
866 | | #if defined(HAS_I444TORGB24ROW_NEON) |
867 | | if (TestCpuFlag(kCpuHasNEON)) { |
868 | | I444ToRGB24Row = I444ToRGB24Row_Any_NEON; |
869 | | if (IS_ALIGNED(width, 8)) { |
870 | | I444ToRGB24Row = I444ToRGB24Row_NEON; |
871 | | } |
872 | | } |
873 | | #endif |
874 | | #if defined(HAS_I444TORGB24ROW_MSA) |
875 | | if (TestCpuFlag(kCpuHasMSA)) { |
876 | | I444ToRGB24Row = I444ToRGB24Row_Any_MSA; |
877 | | if (IS_ALIGNED(width, 8)) { |
878 | | I444ToRGB24Row = I444ToRGB24Row_MSA; |
879 | | } |
880 | | } |
881 | | #endif |
882 | | #if defined(HAS_I444TORGB24ROW_LSX) |
883 | | if (TestCpuFlag(kCpuHasLSX)) { |
884 | | I444ToRGB24Row = I444ToRGB24Row_Any_LSX; |
885 | | if (IS_ALIGNED(width, 16)) { |
886 | | I444ToRGB24Row = I444ToRGB24Row_LSX; |
887 | | } |
888 | | } |
889 | | #endif |
890 | | #if defined(HAS_I444TORGB24ROW_RVV) |
891 | | if (TestCpuFlag(kCpuHasRVV)) { |
892 | | I444ToRGB24Row = I444ToRGB24Row_RVV; |
893 | | } |
894 | | #endif |
895 | | |
896 | 246 | for (y = 0; y < height; ++y) { |
897 | 123 | I444ToRGB24Row(src_y, src_u, src_v, dst_rgb24, yuvconstants, width); |
898 | 123 | dst_rgb24 += dst_stride_rgb24; |
899 | 123 | src_y += src_stride_y; |
900 | 123 | src_u += src_stride_u; |
901 | 123 | src_v += src_stride_v; |
902 | 123 | } |
903 | 123 | return 0; |
904 | 123 | } |
905 | | |
906 | | // Convert I444 to RGB24. |
907 | | LIBYUV_API |
908 | | int I444ToRGB24(const uint8_t* src_y, |
909 | | int src_stride_y, |
910 | | const uint8_t* src_u, |
911 | | int src_stride_u, |
912 | | const uint8_t* src_v, |
913 | | int src_stride_v, |
914 | | uint8_t* dst_rgb24, |
915 | | int dst_stride_rgb24, |
916 | | int width, |
917 | 0 | int height) { |
918 | 0 | return I444ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
919 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
920 | 0 | &kYuvI601Constants, width, height); |
921 | 0 | } |
922 | | |
923 | | // Convert I444 to RAW. |
924 | | LIBYUV_API |
925 | | int I444ToRAW(const uint8_t* src_y, |
926 | | int src_stride_y, |
927 | | const uint8_t* src_u, |
928 | | int src_stride_u, |
929 | | const uint8_t* src_v, |
930 | | int src_stride_v, |
931 | | uint8_t* dst_raw, |
932 | | int dst_stride_raw, |
933 | | int width, |
934 | 0 | int height) { |
935 | 0 | return I444ToRGB24Matrix(src_y, src_stride_y, src_v, |
936 | 0 | src_stride_v, // Swap U and V |
937 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
938 | 0 | &kYvuI601Constants, // Use Yvu matrix |
939 | 0 | width, height); |
940 | 0 | } |
941 | | |
942 | | // Convert 10 bit YUV to ARGB with matrix. |
943 | | // TODO(fbarchard): Consider passing scale multiplier to I210ToARGB to |
944 | | // multiply 10 bit yuv into high bits to allow any number of bits. |
945 | | LIBYUV_API |
946 | | int I010ToAR30Matrix(const uint16_t* src_y, |
947 | | int src_stride_y, |
948 | | const uint16_t* src_u, |
949 | | int src_stride_u, |
950 | | const uint16_t* src_v, |
951 | | int src_stride_v, |
952 | | uint8_t* dst_ar30, |
953 | | int dst_stride_ar30, |
954 | | const struct YuvConstants* yuvconstants, |
955 | | int width, |
956 | 0 | int height) { |
957 | 0 | int y; |
958 | 0 | void (*I210ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
959 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
960 | 0 | const struct YuvConstants* yuvconstants, int width) = |
961 | 0 | I210ToAR30Row_C; |
962 | 0 | assert(yuvconstants); |
963 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
964 | 0 | return -1; |
965 | 0 | } |
966 | | // Negative height means invert the image. |
967 | 0 | if (height < 0) { |
968 | 0 | height = -height; |
969 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
970 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
971 | 0 | } |
972 | | #if defined(HAS_I210TOAR30ROW_NEON) |
973 | | if (TestCpuFlag(kCpuHasNEON)) { |
974 | | I210ToAR30Row = I210ToAR30Row_Any_NEON; |
975 | | if (IS_ALIGNED(width, 8)) { |
976 | | I210ToAR30Row = I210ToAR30Row_NEON; |
977 | | } |
978 | | } |
979 | | #endif |
980 | | #if defined(HAS_I210TOAR30ROW_SVE2) |
981 | | if (TestCpuFlag(kCpuHasSVE2)) { |
982 | | I210ToAR30Row = I210ToAR30Row_SVE2; |
983 | | } |
984 | | #endif |
985 | | #if defined(HAS_I210TOAR30ROW_SME) |
986 | | if (TestCpuFlag(kCpuHasSME)) { |
987 | | I210ToAR30Row = I210ToAR30Row_SME; |
988 | | } |
989 | | #endif |
990 | 0 | #if defined(HAS_I210TOAR30ROW_SSSE3) |
991 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
992 | 0 | I210ToAR30Row = I210ToAR30Row_Any_SSSE3; |
993 | 0 | if (IS_ALIGNED(width, 8)) { |
994 | 0 | I210ToAR30Row = I210ToAR30Row_SSSE3; |
995 | 0 | } |
996 | 0 | } |
997 | 0 | #endif |
998 | 0 | #if defined(HAS_I210TOAR30ROW_AVX2) |
999 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1000 | 0 | I210ToAR30Row = I210ToAR30Row_Any_AVX2; |
1001 | 0 | if (IS_ALIGNED(width, 16)) { |
1002 | 0 | I210ToAR30Row = I210ToAR30Row_AVX2; |
1003 | 0 | } |
1004 | 0 | } |
1005 | 0 | #endif |
1006 | 0 | for (y = 0; y < height; ++y) { |
1007 | 0 | I210ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
1008 | 0 | dst_ar30 += dst_stride_ar30; |
1009 | 0 | src_y += src_stride_y; |
1010 | 0 | if (y & 1) { |
1011 | 0 | src_u += src_stride_u; |
1012 | 0 | src_v += src_stride_v; |
1013 | 0 | } |
1014 | 0 | } |
1015 | 0 | return 0; |
1016 | 0 | } |
1017 | | |
1018 | | // Convert I010 to AR30. |
1019 | | LIBYUV_API |
1020 | | int I010ToAR30(const uint16_t* src_y, |
1021 | | int src_stride_y, |
1022 | | const uint16_t* src_u, |
1023 | | int src_stride_u, |
1024 | | const uint16_t* src_v, |
1025 | | int src_stride_v, |
1026 | | uint8_t* dst_ar30, |
1027 | | int dst_stride_ar30, |
1028 | | int width, |
1029 | 0 | int height) { |
1030 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1031 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1032 | 0 | &kYuvI601Constants, width, height); |
1033 | 0 | } |
1034 | | |
1035 | | // Convert H010 to AR30. |
1036 | | LIBYUV_API |
1037 | | int H010ToAR30(const uint16_t* src_y, |
1038 | | int src_stride_y, |
1039 | | const uint16_t* src_u, |
1040 | | int src_stride_u, |
1041 | | const uint16_t* src_v, |
1042 | | int src_stride_v, |
1043 | | uint8_t* dst_ar30, |
1044 | | int dst_stride_ar30, |
1045 | | int width, |
1046 | 0 | int height) { |
1047 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1048 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1049 | 0 | &kYuvH709Constants, width, height); |
1050 | 0 | } |
1051 | | |
1052 | | // Convert U010 to AR30. |
1053 | | LIBYUV_API |
1054 | | int U010ToAR30(const uint16_t* src_y, |
1055 | | int src_stride_y, |
1056 | | const uint16_t* src_u, |
1057 | | int src_stride_u, |
1058 | | const uint16_t* src_v, |
1059 | | int src_stride_v, |
1060 | | uint8_t* dst_ar30, |
1061 | | int dst_stride_ar30, |
1062 | | int width, |
1063 | 0 | int height) { |
1064 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1065 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1066 | 0 | &kYuv2020Constants, width, height); |
1067 | 0 | } |
1068 | | |
1069 | | // Convert I010 to AB30. |
1070 | | LIBYUV_API |
1071 | | int I010ToAB30(const uint16_t* src_y, |
1072 | | int src_stride_y, |
1073 | | const uint16_t* src_u, |
1074 | | int src_stride_u, |
1075 | | const uint16_t* src_v, |
1076 | | int src_stride_v, |
1077 | | uint8_t* dst_ab30, |
1078 | | int dst_stride_ab30, |
1079 | | int width, |
1080 | 0 | int height) { |
1081 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1082 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1083 | 0 | &kYvuI601Constants, width, height); |
1084 | 0 | } |
1085 | | |
1086 | | // Convert H010 to AB30. |
1087 | | LIBYUV_API |
1088 | | int H010ToAB30(const uint16_t* src_y, |
1089 | | int src_stride_y, |
1090 | | const uint16_t* src_u, |
1091 | | int src_stride_u, |
1092 | | const uint16_t* src_v, |
1093 | | int src_stride_v, |
1094 | | uint8_t* dst_ab30, |
1095 | | int dst_stride_ab30, |
1096 | | int width, |
1097 | 0 | int height) { |
1098 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1099 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1100 | 0 | &kYvuH709Constants, width, height); |
1101 | 0 | } |
1102 | | |
1103 | | // Convert U010 to AB30. |
1104 | | LIBYUV_API |
1105 | | int U010ToAB30(const uint16_t* src_y, |
1106 | | int src_stride_y, |
1107 | | const uint16_t* src_u, |
1108 | | int src_stride_u, |
1109 | | const uint16_t* src_v, |
1110 | | int src_stride_v, |
1111 | | uint8_t* dst_ab30, |
1112 | | int dst_stride_ab30, |
1113 | | int width, |
1114 | 0 | int height) { |
1115 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1116 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1117 | 0 | &kYuv2020Constants, width, height); |
1118 | 0 | } |
1119 | | |
1120 | | // Convert 12 bit YUV to ARGB with matrix. |
1121 | | // TODO(fbarchard): Consider passing scale multiplier to I212ToARGB to |
1122 | | // multiply 12 bit yuv into high bits to allow any number of bits. |
1123 | | LIBYUV_API |
1124 | | int I012ToAR30Matrix(const uint16_t* src_y, |
1125 | | int src_stride_y, |
1126 | | const uint16_t* src_u, |
1127 | | int src_stride_u, |
1128 | | const uint16_t* src_v, |
1129 | | int src_stride_v, |
1130 | | uint8_t* dst_ar30, |
1131 | | int dst_stride_ar30, |
1132 | | const struct YuvConstants* yuvconstants, |
1133 | | int width, |
1134 | 0 | int height) { |
1135 | 0 | int y; |
1136 | 0 | void (*I212ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
1137 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1138 | 0 | const struct YuvConstants* yuvconstants, int width) = |
1139 | 0 | I212ToAR30Row_C; |
1140 | 0 | assert(yuvconstants); |
1141 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
1142 | 0 | return -1; |
1143 | 0 | } |
1144 | | // Negative height means invert the image. |
1145 | 0 | if (height < 0) { |
1146 | 0 | height = -height; |
1147 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
1148 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
1149 | 0 | } |
1150 | 0 | #if defined(HAS_I212TOAR30ROW_SSSE3) |
1151 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1152 | 0 | I212ToAR30Row = I212ToAR30Row_Any_SSSE3; |
1153 | 0 | if (IS_ALIGNED(width, 8)) { |
1154 | 0 | I212ToAR30Row = I212ToAR30Row_SSSE3; |
1155 | 0 | } |
1156 | 0 | } |
1157 | 0 | #endif |
1158 | 0 | #if defined(HAS_I212TOAR30ROW_AVX2) |
1159 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1160 | 0 | I212ToAR30Row = I212ToAR30Row_Any_AVX2; |
1161 | 0 | if (IS_ALIGNED(width, 16)) { |
1162 | 0 | I212ToAR30Row = I212ToAR30Row_AVX2; |
1163 | 0 | } |
1164 | 0 | } |
1165 | 0 | #endif |
1166 | | #if defined(HAS_I212TOAR30ROW_NEON) |
1167 | | if (TestCpuFlag(kCpuHasNEON)) { |
1168 | | I212ToAR30Row = I212ToAR30Row_Any_NEON; |
1169 | | if (IS_ALIGNED(width, 8)) { |
1170 | | I212ToAR30Row = I212ToAR30Row_NEON; |
1171 | | } |
1172 | | } |
1173 | | #endif |
1174 | | #if defined(HAS_I212TOAR30ROW_SVE2) |
1175 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1176 | | I212ToAR30Row = I212ToAR30Row_SVE2; |
1177 | | } |
1178 | | #endif |
1179 | | #if defined(HAS_I212TOAR30ROW_SME) |
1180 | | if (TestCpuFlag(kCpuHasSME)) { |
1181 | | I212ToAR30Row = I212ToAR30Row_SME; |
1182 | | } |
1183 | | #endif |
1184 | 0 | for (y = 0; y < height; ++y) { |
1185 | 0 | I212ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
1186 | 0 | dst_ar30 += dst_stride_ar30; |
1187 | 0 | src_y += src_stride_y; |
1188 | 0 | if (y & 1) { |
1189 | 0 | src_u += src_stride_u; |
1190 | 0 | src_v += src_stride_v; |
1191 | 0 | } |
1192 | 0 | } |
1193 | 0 | return 0; |
1194 | 0 | } |
1195 | | |
1196 | | // Convert 10 bit YUV to ARGB with matrix. |
1197 | | // TODO(fbarchard): Consider passing scale multiplier to I210ToARGB to |
1198 | | // multiply 10 bit yuv into high bits to allow any number of bits. |
1199 | | LIBYUV_API |
1200 | | int I210ToAR30Matrix(const uint16_t* src_y, |
1201 | | int src_stride_y, |
1202 | | const uint16_t* src_u, |
1203 | | int src_stride_u, |
1204 | | const uint16_t* src_v, |
1205 | | int src_stride_v, |
1206 | | uint8_t* dst_ar30, |
1207 | | int dst_stride_ar30, |
1208 | | const struct YuvConstants* yuvconstants, |
1209 | | int width, |
1210 | 0 | int height) { |
1211 | 0 | int y; |
1212 | 0 | void (*I210ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
1213 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1214 | 0 | const struct YuvConstants* yuvconstants, int width) = |
1215 | 0 | I210ToAR30Row_C; |
1216 | 0 | assert(yuvconstants); |
1217 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
1218 | 0 | return -1; |
1219 | 0 | } |
1220 | | // Negative height means invert the image. |
1221 | 0 | if (height < 0) { |
1222 | 0 | height = -height; |
1223 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
1224 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
1225 | 0 | } |
1226 | | #if defined(HAS_I210TOAR30ROW_NEON) |
1227 | | if (TestCpuFlag(kCpuHasNEON)) { |
1228 | | I210ToAR30Row = I210ToAR30Row_Any_NEON; |
1229 | | if (IS_ALIGNED(width, 8)) { |
1230 | | I210ToAR30Row = I210ToAR30Row_NEON; |
1231 | | } |
1232 | | } |
1233 | | #endif |
1234 | | #if defined(HAS_I210TOAR30ROW_SVE2) |
1235 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1236 | | I210ToAR30Row = I210ToAR30Row_SVE2; |
1237 | | } |
1238 | | #endif |
1239 | | #if defined(HAS_I210TOAR30ROW_SME) |
1240 | | if (TestCpuFlag(kCpuHasSME)) { |
1241 | | I210ToAR30Row = I210ToAR30Row_SME; |
1242 | | } |
1243 | | #endif |
1244 | 0 | #if defined(HAS_I210TOAR30ROW_SSSE3) |
1245 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1246 | 0 | I210ToAR30Row = I210ToAR30Row_Any_SSSE3; |
1247 | 0 | if (IS_ALIGNED(width, 8)) { |
1248 | 0 | I210ToAR30Row = I210ToAR30Row_SSSE3; |
1249 | 0 | } |
1250 | 0 | } |
1251 | 0 | #endif |
1252 | 0 | #if defined(HAS_I210TOAR30ROW_AVX2) |
1253 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1254 | 0 | I210ToAR30Row = I210ToAR30Row_Any_AVX2; |
1255 | 0 | if (IS_ALIGNED(width, 16)) { |
1256 | 0 | I210ToAR30Row = I210ToAR30Row_AVX2; |
1257 | 0 | } |
1258 | 0 | } |
1259 | 0 | #endif |
1260 | 0 | for (y = 0; y < height; ++y) { |
1261 | 0 | I210ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
1262 | 0 | dst_ar30 += dst_stride_ar30; |
1263 | 0 | src_y += src_stride_y; |
1264 | 0 | src_u += src_stride_u; |
1265 | 0 | src_v += src_stride_v; |
1266 | 0 | } |
1267 | 0 | return 0; |
1268 | 0 | } |
1269 | | |
1270 | | // Convert I210 to AR30. |
1271 | | LIBYUV_API |
1272 | | int I210ToAR30(const uint16_t* src_y, |
1273 | | int src_stride_y, |
1274 | | const uint16_t* src_u, |
1275 | | int src_stride_u, |
1276 | | const uint16_t* src_v, |
1277 | | int src_stride_v, |
1278 | | uint8_t* dst_ar30, |
1279 | | int dst_stride_ar30, |
1280 | | int width, |
1281 | 0 | int height) { |
1282 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1283 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1284 | 0 | &kYuvI601Constants, width, height); |
1285 | 0 | } |
1286 | | |
1287 | | // Convert H210 to AR30. |
1288 | | LIBYUV_API |
1289 | | int H210ToAR30(const uint16_t* src_y, |
1290 | | int src_stride_y, |
1291 | | const uint16_t* src_u, |
1292 | | int src_stride_u, |
1293 | | const uint16_t* src_v, |
1294 | | int src_stride_v, |
1295 | | uint8_t* dst_ar30, |
1296 | | int dst_stride_ar30, |
1297 | | int width, |
1298 | 0 | int height) { |
1299 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1300 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1301 | 0 | &kYuvH709Constants, width, height); |
1302 | 0 | } |
1303 | | |
1304 | | // Convert U210 to AR30. |
1305 | | LIBYUV_API |
1306 | | int U210ToAR30(const uint16_t* src_y, |
1307 | | int src_stride_y, |
1308 | | const uint16_t* src_u, |
1309 | | int src_stride_u, |
1310 | | const uint16_t* src_v, |
1311 | | int src_stride_v, |
1312 | | uint8_t* dst_ar30, |
1313 | | int dst_stride_ar30, |
1314 | | int width, |
1315 | 0 | int height) { |
1316 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1317 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1318 | 0 | &kYuv2020Constants, width, height); |
1319 | 0 | } |
1320 | | |
1321 | | // Convert I210 to AB30. |
1322 | | LIBYUV_API |
1323 | | int I210ToAB30(const uint16_t* src_y, |
1324 | | int src_stride_y, |
1325 | | const uint16_t* src_u, |
1326 | | int src_stride_u, |
1327 | | const uint16_t* src_v, |
1328 | | int src_stride_v, |
1329 | | uint8_t* dst_ab30, |
1330 | | int dst_stride_ab30, |
1331 | | int width, |
1332 | 0 | int height) { |
1333 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1334 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1335 | 0 | &kYvuI601Constants, width, height); |
1336 | 0 | } |
1337 | | |
1338 | | // Convert H210 to AB30. |
1339 | | LIBYUV_API |
1340 | | int H210ToAB30(const uint16_t* src_y, |
1341 | | int src_stride_y, |
1342 | | const uint16_t* src_u, |
1343 | | int src_stride_u, |
1344 | | const uint16_t* src_v, |
1345 | | int src_stride_v, |
1346 | | uint8_t* dst_ab30, |
1347 | | int dst_stride_ab30, |
1348 | | int width, |
1349 | 0 | int height) { |
1350 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1351 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1352 | 0 | &kYvuH709Constants, width, height); |
1353 | 0 | } |
1354 | | |
1355 | | // Convert U210 to AB30. |
1356 | | LIBYUV_API |
1357 | | int U210ToAB30(const uint16_t* src_y, |
1358 | | int src_stride_y, |
1359 | | const uint16_t* src_u, |
1360 | | int src_stride_u, |
1361 | | const uint16_t* src_v, |
1362 | | int src_stride_v, |
1363 | | uint8_t* dst_ab30, |
1364 | | int dst_stride_ab30, |
1365 | | int width, |
1366 | 0 | int height) { |
1367 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1368 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1369 | 0 | &kYuv2020Constants, width, height); |
1370 | 0 | } |
1371 | | |
1372 | | LIBYUV_API |
1373 | | int I410ToAR30Matrix(const uint16_t* src_y, |
1374 | | int src_stride_y, |
1375 | | const uint16_t* src_u, |
1376 | | int src_stride_u, |
1377 | | const uint16_t* src_v, |
1378 | | int src_stride_v, |
1379 | | uint8_t* dst_ar30, |
1380 | | int dst_stride_ar30, |
1381 | | const struct YuvConstants* yuvconstants, |
1382 | | int width, |
1383 | 0 | int height) { |
1384 | 0 | int y; |
1385 | 0 | void (*I410ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
1386 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1387 | 0 | const struct YuvConstants* yuvconstants, int width) = |
1388 | 0 | I410ToAR30Row_C; |
1389 | 0 | assert(yuvconstants); |
1390 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
1391 | 0 | return -1; |
1392 | 0 | } |
1393 | | // Negative height means invert the image. |
1394 | 0 | if (height < 0) { |
1395 | 0 | height = -height; |
1396 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
1397 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
1398 | 0 | } |
1399 | | #if defined(HAS_I410TOAR30ROW_NEON) |
1400 | | if (TestCpuFlag(kCpuHasNEON)) { |
1401 | | I410ToAR30Row = I410ToAR30Row_Any_NEON; |
1402 | | if (IS_ALIGNED(width, 8)) { |
1403 | | I410ToAR30Row = I410ToAR30Row_NEON; |
1404 | | } |
1405 | | } |
1406 | | #endif |
1407 | | #if defined(HAS_I410TOAR30ROW_SVE2) |
1408 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1409 | | I410ToAR30Row = I410ToAR30Row_SVE2; |
1410 | | } |
1411 | | #endif |
1412 | | #if defined(HAS_I410TOAR30ROW_SME) |
1413 | | if (TestCpuFlag(kCpuHasSME)) { |
1414 | | I410ToAR30Row = I410ToAR30Row_SME; |
1415 | | } |
1416 | | #endif |
1417 | 0 | #if defined(HAS_I410TOAR30ROW_SSSE3) |
1418 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1419 | 0 | I410ToAR30Row = I410ToAR30Row_Any_SSSE3; |
1420 | 0 | if (IS_ALIGNED(width, 8)) { |
1421 | 0 | I410ToAR30Row = I410ToAR30Row_SSSE3; |
1422 | 0 | } |
1423 | 0 | } |
1424 | 0 | #endif |
1425 | 0 | #if defined(HAS_I410TOAR30ROW_AVX2) |
1426 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1427 | 0 | I410ToAR30Row = I410ToAR30Row_Any_AVX2; |
1428 | 0 | if (IS_ALIGNED(width, 16)) { |
1429 | 0 | I410ToAR30Row = I410ToAR30Row_AVX2; |
1430 | 0 | } |
1431 | 0 | } |
1432 | 0 | #endif |
1433 | 0 | for (y = 0; y < height; ++y) { |
1434 | 0 | I410ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
1435 | 0 | dst_ar30 += dst_stride_ar30; |
1436 | 0 | src_y += src_stride_y; |
1437 | 0 | src_u += src_stride_u; |
1438 | 0 | src_v += src_stride_v; |
1439 | 0 | } |
1440 | 0 | return 0; |
1441 | 0 | } |
1442 | | |
1443 | | // Convert 10 bit YUV to ARGB with matrix. |
1444 | | LIBYUV_API |
1445 | | int I010ToARGBMatrix(const uint16_t* src_y, |
1446 | | int src_stride_y, |
1447 | | const uint16_t* src_u, |
1448 | | int src_stride_u, |
1449 | | const uint16_t* src_v, |
1450 | | int src_stride_v, |
1451 | | uint8_t* dst_argb, |
1452 | | int dst_stride_argb, |
1453 | | const struct YuvConstants* yuvconstants, |
1454 | | int width, |
1455 | 128 | int height) { |
1456 | 128 | int y; |
1457 | 128 | void (*I210ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1458 | 128 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1459 | 128 | const struct YuvConstants* yuvconstants, int width) = |
1460 | 128 | I210ToARGBRow_C; |
1461 | 128 | assert(yuvconstants); |
1462 | 128 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1463 | 0 | return -1; |
1464 | 0 | } |
1465 | | // Negative height means invert the image. |
1466 | 128 | if (height < 0) { |
1467 | 0 | height = -height; |
1468 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1469 | 0 | dst_stride_argb = -dst_stride_argb; |
1470 | 0 | } |
1471 | 128 | #if defined(HAS_I210TOARGBROW_SSSE3) |
1472 | 128 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1473 | 128 | I210ToARGBRow = I210ToARGBRow_Any_SSSE3; |
1474 | 128 | if (IS_ALIGNED(width, 8)) { |
1475 | 34 | I210ToARGBRow = I210ToARGBRow_SSSE3; |
1476 | 34 | } |
1477 | 128 | } |
1478 | 128 | #endif |
1479 | | #if defined(HAS_I210TOARGBROW_NEON) |
1480 | | if (TestCpuFlag(kCpuHasNEON)) { |
1481 | | I210ToARGBRow = I210ToARGBRow_Any_NEON; |
1482 | | if (IS_ALIGNED(width, 8)) { |
1483 | | I210ToARGBRow = I210ToARGBRow_NEON; |
1484 | | } |
1485 | | } |
1486 | | #endif |
1487 | | #if defined(HAS_I210TOARGBROW_SVE2) |
1488 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1489 | | I210ToARGBRow = I210ToARGBRow_SVE2; |
1490 | | } |
1491 | | #endif |
1492 | | #if defined(HAS_I210TOARGBROW_SME) |
1493 | | if (TestCpuFlag(kCpuHasSME)) { |
1494 | | I210ToARGBRow = I210ToARGBRow_SME; |
1495 | | } |
1496 | | #endif |
1497 | 128 | #if defined(HAS_I210TOARGBROW_AVX2) |
1498 | 128 | if (TestCpuFlag(kCpuHasAVX2)) { |
1499 | 128 | I210ToARGBRow = I210ToARGBRow_Any_AVX2; |
1500 | 128 | if (IS_ALIGNED(width, 16)) { |
1501 | 17 | I210ToARGBRow = I210ToARGBRow_AVX2; |
1502 | 17 | } |
1503 | 128 | } |
1504 | 128 | #endif |
1505 | 2.69k | for (y = 0; y < height; ++y) { |
1506 | 2.57k | I210ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1507 | 2.57k | dst_argb += dst_stride_argb; |
1508 | 2.57k | src_y += src_stride_y; |
1509 | 2.57k | if (y & 1) { |
1510 | 1.27k | src_u += src_stride_u; |
1511 | 1.27k | src_v += src_stride_v; |
1512 | 1.27k | } |
1513 | 2.57k | } |
1514 | 128 | return 0; |
1515 | 128 | } |
1516 | | |
1517 | | // Convert I010 to ARGB. |
1518 | | LIBYUV_API |
1519 | | int I010ToARGB(const uint16_t* src_y, |
1520 | | int src_stride_y, |
1521 | | const uint16_t* src_u, |
1522 | | int src_stride_u, |
1523 | | const uint16_t* src_v, |
1524 | | int src_stride_v, |
1525 | | uint8_t* dst_argb, |
1526 | | int dst_stride_argb, |
1527 | | int width, |
1528 | 0 | int height) { |
1529 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1530 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1531 | 0 | &kYuvI601Constants, width, height); |
1532 | 0 | } |
1533 | | |
1534 | | // Convert I010 to ABGR. |
1535 | | LIBYUV_API |
1536 | | int I010ToABGR(const uint16_t* src_y, |
1537 | | int src_stride_y, |
1538 | | const uint16_t* src_u, |
1539 | | int src_stride_u, |
1540 | | const uint16_t* src_v, |
1541 | | int src_stride_v, |
1542 | | uint8_t* dst_abgr, |
1543 | | int dst_stride_abgr, |
1544 | | int width, |
1545 | 0 | int height) { |
1546 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_v, |
1547 | 0 | src_stride_v, // Swap U and V |
1548 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1549 | 0 | &kYvuI601Constants, // Use Yvu matrix |
1550 | 0 | width, height); |
1551 | 0 | } |
1552 | | |
1553 | | // Convert H010 to ARGB. |
1554 | | LIBYUV_API |
1555 | | int H010ToARGB(const uint16_t* src_y, |
1556 | | int src_stride_y, |
1557 | | const uint16_t* src_u, |
1558 | | int src_stride_u, |
1559 | | const uint16_t* src_v, |
1560 | | int src_stride_v, |
1561 | | uint8_t* dst_argb, |
1562 | | int dst_stride_argb, |
1563 | | int width, |
1564 | 0 | int height) { |
1565 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1566 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1567 | 0 | &kYuvH709Constants, width, height); |
1568 | 0 | } |
1569 | | |
1570 | | // Convert H010 to ABGR. |
1571 | | LIBYUV_API |
1572 | | int H010ToABGR(const uint16_t* src_y, |
1573 | | int src_stride_y, |
1574 | | const uint16_t* src_u, |
1575 | | int src_stride_u, |
1576 | | const uint16_t* src_v, |
1577 | | int src_stride_v, |
1578 | | uint8_t* dst_abgr, |
1579 | | int dst_stride_abgr, |
1580 | | int width, |
1581 | 0 | int height) { |
1582 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_v, |
1583 | 0 | src_stride_v, // Swap U and V |
1584 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1585 | 0 | &kYvuH709Constants, // Use Yvu matrix |
1586 | 0 | width, height); |
1587 | 0 | } |
1588 | | |
1589 | | // Convert U010 to ARGB. |
1590 | | LIBYUV_API |
1591 | | int U010ToARGB(const uint16_t* src_y, |
1592 | | int src_stride_y, |
1593 | | const uint16_t* src_u, |
1594 | | int src_stride_u, |
1595 | | const uint16_t* src_v, |
1596 | | int src_stride_v, |
1597 | | uint8_t* dst_argb, |
1598 | | int dst_stride_argb, |
1599 | | int width, |
1600 | 0 | int height) { |
1601 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1602 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1603 | 0 | &kYuv2020Constants, width, height); |
1604 | 0 | } |
1605 | | |
1606 | | // Convert U010 to ABGR. |
1607 | | LIBYUV_API |
1608 | | int U010ToABGR(const uint16_t* src_y, |
1609 | | int src_stride_y, |
1610 | | const uint16_t* src_u, |
1611 | | int src_stride_u, |
1612 | | const uint16_t* src_v, |
1613 | | int src_stride_v, |
1614 | | uint8_t* dst_abgr, |
1615 | | int dst_stride_abgr, |
1616 | | int width, |
1617 | 0 | int height) { |
1618 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_v, |
1619 | 0 | src_stride_v, // Swap U and V |
1620 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1621 | 0 | &kYvu2020Constants, // Use Yvu matrix |
1622 | 0 | width, height); |
1623 | 0 | } |
1624 | | |
1625 | | // Convert 12 bit YUV to ARGB with matrix. |
1626 | | LIBYUV_API |
1627 | | int I012ToARGBMatrix(const uint16_t* src_y, |
1628 | | int src_stride_y, |
1629 | | const uint16_t* src_u, |
1630 | | int src_stride_u, |
1631 | | const uint16_t* src_v, |
1632 | | int src_stride_v, |
1633 | | uint8_t* dst_argb, |
1634 | | int dst_stride_argb, |
1635 | | const struct YuvConstants* yuvconstants, |
1636 | | int width, |
1637 | 108 | int height) { |
1638 | 108 | int y; |
1639 | 108 | void (*I212ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1640 | 108 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1641 | 108 | const struct YuvConstants* yuvconstants, int width) = |
1642 | 108 | I212ToARGBRow_C; |
1643 | 108 | assert(yuvconstants); |
1644 | 108 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1645 | 0 | return -1; |
1646 | 0 | } |
1647 | | // Negative height means invert the image. |
1648 | 108 | if (height < 0) { |
1649 | 0 | height = -height; |
1650 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1651 | 0 | dst_stride_argb = -dst_stride_argb; |
1652 | 0 | } |
1653 | 108 | #if defined(HAS_I212TOARGBROW_SSSE3) |
1654 | 108 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1655 | 108 | I212ToARGBRow = I212ToARGBRow_Any_SSSE3; |
1656 | 108 | if (IS_ALIGNED(width, 8)) { |
1657 | 36 | I212ToARGBRow = I212ToARGBRow_SSSE3; |
1658 | 36 | } |
1659 | 108 | } |
1660 | 108 | #endif |
1661 | 108 | #if defined(HAS_I212TOARGBROW_AVX2) |
1662 | 108 | if (TestCpuFlag(kCpuHasAVX2)) { |
1663 | 108 | I212ToARGBRow = I212ToARGBRow_Any_AVX2; |
1664 | 108 | if (IS_ALIGNED(width, 16)) { |
1665 | 19 | I212ToARGBRow = I212ToARGBRow_AVX2; |
1666 | 19 | } |
1667 | 108 | } |
1668 | 108 | #endif |
1669 | | #if defined(HAS_I212TOARGBROW_NEON) |
1670 | | if (TestCpuFlag(kCpuHasNEON)) { |
1671 | | I212ToARGBRow = I212ToARGBRow_Any_NEON; |
1672 | | if (IS_ALIGNED(width, 8)) { |
1673 | | I212ToARGBRow = I212ToARGBRow_NEON; |
1674 | | } |
1675 | | } |
1676 | | #endif |
1677 | | #if defined(HAS_I212TOARGBROW_SVE2) |
1678 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1679 | | I212ToARGBRow = I212ToARGBRow_SVE2; |
1680 | | } |
1681 | | #endif |
1682 | | #if defined(HAS_I212TOARGBROW_SME) |
1683 | | if (TestCpuFlag(kCpuHasSME)) { |
1684 | | I212ToARGBRow = I212ToARGBRow_SME; |
1685 | | } |
1686 | | #endif |
1687 | 3.83k | for (y = 0; y < height; ++y) { |
1688 | 3.72k | I212ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1689 | 3.72k | dst_argb += dst_stride_argb; |
1690 | 3.72k | src_y += src_stride_y; |
1691 | 3.72k | if (y & 1) { |
1692 | 1.84k | src_u += src_stride_u; |
1693 | 1.84k | src_v += src_stride_v; |
1694 | 1.84k | } |
1695 | 3.72k | } |
1696 | 108 | return 0; |
1697 | 108 | } |
1698 | | |
1699 | | // Convert 10 bit 422 YUV to ARGB with matrix. |
1700 | | LIBYUV_API |
1701 | | int I210ToARGBMatrix(const uint16_t* src_y, |
1702 | | int src_stride_y, |
1703 | | const uint16_t* src_u, |
1704 | | int src_stride_u, |
1705 | | const uint16_t* src_v, |
1706 | | int src_stride_v, |
1707 | | uint8_t* dst_argb, |
1708 | | int dst_stride_argb, |
1709 | | const struct YuvConstants* yuvconstants, |
1710 | | int width, |
1711 | 65 | int height) { |
1712 | 65 | int y; |
1713 | 65 | void (*I210ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1714 | 65 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1715 | 65 | const struct YuvConstants* yuvconstants, int width) = |
1716 | 65 | I210ToARGBRow_C; |
1717 | 65 | assert(yuvconstants); |
1718 | 65 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1719 | 0 | return -1; |
1720 | 0 | } |
1721 | | // Negative height means invert the image. |
1722 | 65 | if (height < 0) { |
1723 | 0 | height = -height; |
1724 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1725 | 0 | dst_stride_argb = -dst_stride_argb; |
1726 | 0 | } |
1727 | 65 | #if defined(HAS_I210TOARGBROW_SSSE3) |
1728 | 65 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1729 | 65 | I210ToARGBRow = I210ToARGBRow_Any_SSSE3; |
1730 | 65 | if (IS_ALIGNED(width, 8)) { |
1731 | 34 | I210ToARGBRow = I210ToARGBRow_SSSE3; |
1732 | 34 | } |
1733 | 65 | } |
1734 | 65 | #endif |
1735 | | #if defined(HAS_I210TOARGBROW_NEON) |
1736 | | if (TestCpuFlag(kCpuHasNEON)) { |
1737 | | I210ToARGBRow = I210ToARGBRow_Any_NEON; |
1738 | | if (IS_ALIGNED(width, 8)) { |
1739 | | I210ToARGBRow = I210ToARGBRow_NEON; |
1740 | | } |
1741 | | } |
1742 | | #endif |
1743 | | #if defined(HAS_I210TOARGBROW_SVE2) |
1744 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1745 | | I210ToARGBRow = I210ToARGBRow_SVE2; |
1746 | | } |
1747 | | #endif |
1748 | | #if defined(HAS_I210TOARGBROW_SME) |
1749 | | if (TestCpuFlag(kCpuHasSME)) { |
1750 | | I210ToARGBRow = I210ToARGBRow_SME; |
1751 | | } |
1752 | | #endif |
1753 | 65 | #if defined(HAS_I210TOARGBROW_AVX2) |
1754 | 65 | if (TestCpuFlag(kCpuHasAVX2)) { |
1755 | 65 | I210ToARGBRow = I210ToARGBRow_Any_AVX2; |
1756 | 65 | if (IS_ALIGNED(width, 16)) { |
1757 | 17 | I210ToARGBRow = I210ToARGBRow_AVX2; |
1758 | 17 | } |
1759 | 65 | } |
1760 | 65 | #endif |
1761 | 540 | for (y = 0; y < height; ++y) { |
1762 | 475 | I210ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1763 | 475 | dst_argb += dst_stride_argb; |
1764 | 475 | src_y += src_stride_y; |
1765 | 475 | src_u += src_stride_u; |
1766 | 475 | src_v += src_stride_v; |
1767 | 475 | } |
1768 | 65 | return 0; |
1769 | 65 | } |
1770 | | |
1771 | | // Convert I210 to ARGB. |
1772 | | LIBYUV_API |
1773 | | int I210ToARGB(const uint16_t* src_y, |
1774 | | int src_stride_y, |
1775 | | const uint16_t* src_u, |
1776 | | int src_stride_u, |
1777 | | const uint16_t* src_v, |
1778 | | int src_stride_v, |
1779 | | uint8_t* dst_argb, |
1780 | | int dst_stride_argb, |
1781 | | int width, |
1782 | 0 | int height) { |
1783 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1784 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1785 | 0 | &kYuvI601Constants, width, height); |
1786 | 0 | } |
1787 | | |
1788 | | // Convert I210 to ABGR. |
1789 | | LIBYUV_API |
1790 | | int I210ToABGR(const uint16_t* src_y, |
1791 | | int src_stride_y, |
1792 | | const uint16_t* src_u, |
1793 | | int src_stride_u, |
1794 | | const uint16_t* src_v, |
1795 | | int src_stride_v, |
1796 | | uint8_t* dst_abgr, |
1797 | | int dst_stride_abgr, |
1798 | | int width, |
1799 | 0 | int height) { |
1800 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_v, |
1801 | 0 | src_stride_v, // Swap U and V |
1802 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1803 | 0 | &kYvuI601Constants, // Use Yvu matrix |
1804 | 0 | width, height); |
1805 | 0 | } |
1806 | | |
1807 | | // Convert H210 to ARGB. |
1808 | | LIBYUV_API |
1809 | | int H210ToARGB(const uint16_t* src_y, |
1810 | | int src_stride_y, |
1811 | | const uint16_t* src_u, |
1812 | | int src_stride_u, |
1813 | | const uint16_t* src_v, |
1814 | | int src_stride_v, |
1815 | | uint8_t* dst_argb, |
1816 | | int dst_stride_argb, |
1817 | | int width, |
1818 | 0 | int height) { |
1819 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1820 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1821 | 0 | &kYuvH709Constants, width, height); |
1822 | 0 | } |
1823 | | |
1824 | | // Convert H210 to ABGR. |
1825 | | LIBYUV_API |
1826 | | int H210ToABGR(const uint16_t* src_y, |
1827 | | int src_stride_y, |
1828 | | const uint16_t* src_u, |
1829 | | int src_stride_u, |
1830 | | const uint16_t* src_v, |
1831 | | int src_stride_v, |
1832 | | uint8_t* dst_abgr, |
1833 | | int dst_stride_abgr, |
1834 | | int width, |
1835 | 0 | int height) { |
1836 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_v, |
1837 | 0 | src_stride_v, // Swap U and V |
1838 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1839 | 0 | &kYvuH709Constants, // Use Yvu matrix |
1840 | 0 | width, height); |
1841 | 0 | } |
1842 | | |
1843 | | // Convert U210 to ARGB. |
1844 | | LIBYUV_API |
1845 | | int U210ToARGB(const uint16_t* src_y, |
1846 | | int src_stride_y, |
1847 | | const uint16_t* src_u, |
1848 | | int src_stride_u, |
1849 | | const uint16_t* src_v, |
1850 | | int src_stride_v, |
1851 | | uint8_t* dst_argb, |
1852 | | int dst_stride_argb, |
1853 | | int width, |
1854 | 0 | int height) { |
1855 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1856 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1857 | 0 | &kYuv2020Constants, width, height); |
1858 | 0 | } |
1859 | | |
1860 | | // Convert U210 to ABGR. |
1861 | | LIBYUV_API |
1862 | | int U210ToABGR(const uint16_t* src_y, |
1863 | | int src_stride_y, |
1864 | | const uint16_t* src_u, |
1865 | | int src_stride_u, |
1866 | | const uint16_t* src_v, |
1867 | | int src_stride_v, |
1868 | | uint8_t* dst_abgr, |
1869 | | int dst_stride_abgr, |
1870 | | int width, |
1871 | 0 | int height) { |
1872 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_v, |
1873 | 0 | src_stride_v, // Swap U and V |
1874 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1875 | 0 | &kYvu2020Constants, // Use Yvu matrix |
1876 | 0 | width, height); |
1877 | 0 | } |
1878 | | |
1879 | | LIBYUV_API |
1880 | | int I410ToARGBMatrix(const uint16_t* src_y, |
1881 | | int src_stride_y, |
1882 | | const uint16_t* src_u, |
1883 | | int src_stride_u, |
1884 | | const uint16_t* src_v, |
1885 | | int src_stride_v, |
1886 | | uint8_t* dst_argb, |
1887 | | int dst_stride_argb, |
1888 | | const struct YuvConstants* yuvconstants, |
1889 | | int width, |
1890 | 75 | int height) { |
1891 | 75 | int y; |
1892 | 75 | void (*I410ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1893 | 75 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1894 | 75 | const struct YuvConstants* yuvconstants, int width) = |
1895 | 75 | I410ToARGBRow_C; |
1896 | 75 | assert(yuvconstants); |
1897 | 75 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1898 | 0 | return -1; |
1899 | 0 | } |
1900 | | // Negative height means invert the image. |
1901 | 75 | if (height < 0) { |
1902 | 0 | height = -height; |
1903 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1904 | 0 | dst_stride_argb = -dst_stride_argb; |
1905 | 0 | } |
1906 | 75 | #if defined(HAS_I410TOARGBROW_SSSE3) |
1907 | 75 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1908 | 75 | I410ToARGBRow = I410ToARGBRow_Any_SSSE3; |
1909 | 75 | if (IS_ALIGNED(width, 8)) { |
1910 | 36 | I410ToARGBRow = I410ToARGBRow_SSSE3; |
1911 | 36 | } |
1912 | 75 | } |
1913 | 75 | #endif |
1914 | | #if defined(HAS_I410TOARGBROW_NEON) |
1915 | | if (TestCpuFlag(kCpuHasNEON)) { |
1916 | | I410ToARGBRow = I410ToARGBRow_Any_NEON; |
1917 | | if (IS_ALIGNED(width, 8)) { |
1918 | | I410ToARGBRow = I410ToARGBRow_NEON; |
1919 | | } |
1920 | | } |
1921 | | #endif |
1922 | | #if defined(HAS_I410TOARGBROW_SVE2) |
1923 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1924 | | I410ToARGBRow = I410ToARGBRow_SVE2; |
1925 | | } |
1926 | | #endif |
1927 | | #if defined(HAS_I410TOARGBROW_SME) |
1928 | | if (TestCpuFlag(kCpuHasSME)) { |
1929 | | I410ToARGBRow = I410ToARGBRow_SME; |
1930 | | } |
1931 | | #endif |
1932 | 75 | #if defined(HAS_I410TOARGBROW_AVX2) |
1933 | 75 | if (TestCpuFlag(kCpuHasAVX2)) { |
1934 | 75 | I410ToARGBRow = I410ToARGBRow_Any_AVX2; |
1935 | 75 | if (IS_ALIGNED(width, 16)) { |
1936 | 19 | I410ToARGBRow = I410ToARGBRow_AVX2; |
1937 | 19 | } |
1938 | 75 | } |
1939 | 75 | #endif |
1940 | 634 | for (y = 0; y < height; ++y) { |
1941 | 559 | I410ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1942 | 559 | dst_argb += dst_stride_argb; |
1943 | 559 | src_y += src_stride_y; |
1944 | 559 | src_u += src_stride_u; |
1945 | 559 | src_v += src_stride_v; |
1946 | 559 | } |
1947 | 75 | return 0; |
1948 | 75 | } |
1949 | | |
1950 | | LIBYUV_API |
1951 | | int P010ToARGBMatrix(const uint16_t* src_y, |
1952 | | int src_stride_y, |
1953 | | const uint16_t* src_uv, |
1954 | | int src_stride_uv, |
1955 | | uint8_t* dst_argb, |
1956 | | int dst_stride_argb, |
1957 | | const struct YuvConstants* yuvconstants, |
1958 | | int width, |
1959 | 0 | int height) { |
1960 | 0 | int y; |
1961 | 0 | void (*P210ToARGBRow)( |
1962 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
1963 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToARGBRow_C; |
1964 | 0 | assert(yuvconstants); |
1965 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
1966 | 0 | return -1; |
1967 | 0 | } |
1968 | | // Negative height means invert the image. |
1969 | 0 | if (height < 0) { |
1970 | 0 | height = -height; |
1971 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1972 | 0 | dst_stride_argb = -dst_stride_argb; |
1973 | 0 | } |
1974 | 0 | #if defined(HAS_P210TOARGBROW_SSSE3) |
1975 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1976 | 0 | P210ToARGBRow = P210ToARGBRow_Any_SSSE3; |
1977 | 0 | if (IS_ALIGNED(width, 8)) { |
1978 | 0 | P210ToARGBRow = P210ToARGBRow_SSSE3; |
1979 | 0 | } |
1980 | 0 | } |
1981 | 0 | #endif |
1982 | 0 | #if defined(HAS_P210TOARGBROW_AVX2) |
1983 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1984 | 0 | P210ToARGBRow = P210ToARGBRow_Any_AVX2; |
1985 | 0 | if (IS_ALIGNED(width, 16)) { |
1986 | 0 | P210ToARGBRow = P210ToARGBRow_AVX2; |
1987 | 0 | } |
1988 | 0 | } |
1989 | 0 | #endif |
1990 | | #if defined(HAS_P210TOARGBROW_NEON) |
1991 | | if (TestCpuFlag(kCpuHasNEON)) { |
1992 | | P210ToARGBRow = P210ToARGBRow_Any_NEON; |
1993 | | if (IS_ALIGNED(width, 8)) { |
1994 | | P210ToARGBRow = P210ToARGBRow_NEON; |
1995 | | } |
1996 | | } |
1997 | | #endif |
1998 | | #if defined(HAS_P210TOARGBROW_SVE2) |
1999 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2000 | | P210ToARGBRow = P210ToARGBRow_SVE2; |
2001 | | } |
2002 | | #endif |
2003 | | #if defined(HAS_P210TOARGBROW_SME) |
2004 | | if (TestCpuFlag(kCpuHasSME)) { |
2005 | | P210ToARGBRow = P210ToARGBRow_SME; |
2006 | | } |
2007 | | #endif |
2008 | 0 | for (y = 0; y < height; ++y) { |
2009 | 0 | P210ToARGBRow(src_y, src_uv, dst_argb, yuvconstants, width); |
2010 | 0 | dst_argb += dst_stride_argb; |
2011 | 0 | src_y += src_stride_y; |
2012 | 0 | if (y & 1) { |
2013 | 0 | src_uv += src_stride_uv; |
2014 | 0 | } |
2015 | 0 | } |
2016 | 0 | return 0; |
2017 | 0 | } |
2018 | | |
2019 | | LIBYUV_API |
2020 | | int P210ToARGBMatrix(const uint16_t* src_y, |
2021 | | int src_stride_y, |
2022 | | const uint16_t* src_uv, |
2023 | | int src_stride_uv, |
2024 | | uint8_t* dst_argb, |
2025 | | int dst_stride_argb, |
2026 | | const struct YuvConstants* yuvconstants, |
2027 | | int width, |
2028 | 0 | int height) { |
2029 | 0 | int y; |
2030 | 0 | void (*P210ToARGBRow)( |
2031 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
2032 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToARGBRow_C; |
2033 | 0 | assert(yuvconstants); |
2034 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
2035 | 0 | return -1; |
2036 | 0 | } |
2037 | | // Negative height means invert the image. |
2038 | 0 | if (height < 0) { |
2039 | 0 | height = -height; |
2040 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2041 | 0 | dst_stride_argb = -dst_stride_argb; |
2042 | 0 | } |
2043 | 0 | #if defined(HAS_P210TOARGBROW_SSSE3) |
2044 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2045 | 0 | P210ToARGBRow = P210ToARGBRow_Any_SSSE3; |
2046 | 0 | if (IS_ALIGNED(width, 8)) { |
2047 | 0 | P210ToARGBRow = P210ToARGBRow_SSSE3; |
2048 | 0 | } |
2049 | 0 | } |
2050 | 0 | #endif |
2051 | 0 | #if defined(HAS_P210TOARGBROW_AVX2) |
2052 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2053 | 0 | P210ToARGBRow = P210ToARGBRow_Any_AVX2; |
2054 | 0 | if (IS_ALIGNED(width, 16)) { |
2055 | 0 | P210ToARGBRow = P210ToARGBRow_AVX2; |
2056 | 0 | } |
2057 | 0 | } |
2058 | 0 | #endif |
2059 | | #if defined(HAS_P210TOARGBROW_NEON) |
2060 | | if (TestCpuFlag(kCpuHasNEON)) { |
2061 | | P210ToARGBRow = P210ToARGBRow_Any_NEON; |
2062 | | if (IS_ALIGNED(width, 8)) { |
2063 | | P210ToARGBRow = P210ToARGBRow_NEON; |
2064 | | } |
2065 | | } |
2066 | | #endif |
2067 | | #if defined(HAS_P210TOARGBROW_SVE2) |
2068 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2069 | | P210ToARGBRow = P210ToARGBRow_SVE2; |
2070 | | } |
2071 | | #endif |
2072 | | #if defined(HAS_P210TOARGBROW_SME) |
2073 | | if (TestCpuFlag(kCpuHasSME)) { |
2074 | | P210ToARGBRow = P210ToARGBRow_SME; |
2075 | | } |
2076 | | #endif |
2077 | 0 | for (y = 0; y < height; ++y) { |
2078 | 0 | P210ToARGBRow(src_y, src_uv, dst_argb, yuvconstants, width); |
2079 | 0 | dst_argb += dst_stride_argb; |
2080 | 0 | src_y += src_stride_y; |
2081 | 0 | src_uv += src_stride_uv; |
2082 | 0 | } |
2083 | 0 | return 0; |
2084 | 0 | } |
2085 | | |
2086 | | LIBYUV_API |
2087 | | int P010ToAR30Matrix(const uint16_t* src_y, |
2088 | | int src_stride_y, |
2089 | | const uint16_t* src_uv, |
2090 | | int src_stride_uv, |
2091 | | uint8_t* dst_ar30, |
2092 | | int dst_stride_ar30, |
2093 | | const struct YuvConstants* yuvconstants, |
2094 | | int width, |
2095 | 0 | int height) { |
2096 | 0 | int y; |
2097 | 0 | void (*P210ToAR30Row)( |
2098 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
2099 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToAR30Row_C; |
2100 | 0 | assert(yuvconstants); |
2101 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
2102 | 0 | return -1; |
2103 | 0 | } |
2104 | | // Negative height means invert the image. |
2105 | 0 | if (height < 0) { |
2106 | 0 | height = -height; |
2107 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
2108 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
2109 | 0 | } |
2110 | 0 | #if defined(HAS_P210TOAR30ROW_SSSE3) |
2111 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2112 | 0 | P210ToAR30Row = P210ToAR30Row_Any_SSSE3; |
2113 | 0 | if (IS_ALIGNED(width, 8)) { |
2114 | 0 | P210ToAR30Row = P210ToAR30Row_SSSE3; |
2115 | 0 | } |
2116 | 0 | } |
2117 | 0 | #endif |
2118 | 0 | #if defined(HAS_P210TOAR30ROW_AVX2) |
2119 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2120 | 0 | P210ToAR30Row = P210ToAR30Row_Any_AVX2; |
2121 | 0 | if (IS_ALIGNED(width, 16)) { |
2122 | 0 | P210ToAR30Row = P210ToAR30Row_AVX2; |
2123 | 0 | } |
2124 | 0 | } |
2125 | 0 | #endif |
2126 | | #if defined(HAS_P210TOAR30ROW_NEON) |
2127 | | if (TestCpuFlag(kCpuHasNEON)) { |
2128 | | P210ToAR30Row = P210ToAR30Row_Any_NEON; |
2129 | | if (IS_ALIGNED(width, 8)) { |
2130 | | P210ToAR30Row = P210ToAR30Row_NEON; |
2131 | | } |
2132 | | } |
2133 | | #endif |
2134 | | #if defined(HAS_P210TOAR30ROW_SVE2) |
2135 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2136 | | P210ToAR30Row = P210ToAR30Row_SVE2; |
2137 | | } |
2138 | | #endif |
2139 | | #if defined(HAS_P210TOAR30ROW_SME) |
2140 | | if (TestCpuFlag(kCpuHasSME)) { |
2141 | | P210ToAR30Row = P210ToAR30Row_SME; |
2142 | | } |
2143 | | #endif |
2144 | 0 | for (y = 0; y < height; ++y) { |
2145 | 0 | P210ToAR30Row(src_y, src_uv, dst_ar30, yuvconstants, width); |
2146 | 0 | dst_ar30 += dst_stride_ar30; |
2147 | 0 | src_y += src_stride_y; |
2148 | 0 | if (y & 1) { |
2149 | 0 | src_uv += src_stride_uv; |
2150 | 0 | } |
2151 | 0 | } |
2152 | 0 | return 0; |
2153 | 0 | } |
2154 | | |
2155 | | LIBYUV_API |
2156 | | int P210ToAR30Matrix(const uint16_t* src_y, |
2157 | | int src_stride_y, |
2158 | | const uint16_t* src_uv, |
2159 | | int src_stride_uv, |
2160 | | uint8_t* dst_ar30, |
2161 | | int dst_stride_ar30, |
2162 | | const struct YuvConstants* yuvconstants, |
2163 | | int width, |
2164 | 0 | int height) { |
2165 | 0 | int y; |
2166 | 0 | void (*P210ToAR30Row)( |
2167 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
2168 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToAR30Row_C; |
2169 | 0 | assert(yuvconstants); |
2170 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
2171 | 0 | return -1; |
2172 | 0 | } |
2173 | | // Negative height means invert the image. |
2174 | 0 | if (height < 0) { |
2175 | 0 | height = -height; |
2176 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
2177 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
2178 | 0 | } |
2179 | 0 | #if defined(HAS_P210TOAR30ROW_SSSE3) |
2180 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2181 | 0 | P210ToAR30Row = P210ToAR30Row_Any_SSSE3; |
2182 | 0 | if (IS_ALIGNED(width, 8)) { |
2183 | 0 | P210ToAR30Row = P210ToAR30Row_SSSE3; |
2184 | 0 | } |
2185 | 0 | } |
2186 | 0 | #endif |
2187 | 0 | #if defined(HAS_P210TOAR30ROW_AVX2) |
2188 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2189 | 0 | P210ToAR30Row = P210ToAR30Row_Any_AVX2; |
2190 | 0 | if (IS_ALIGNED(width, 16)) { |
2191 | 0 | P210ToAR30Row = P210ToAR30Row_AVX2; |
2192 | 0 | } |
2193 | 0 | } |
2194 | 0 | #endif |
2195 | | #if defined(HAS_P210TOAR30ROW_NEON) |
2196 | | if (TestCpuFlag(kCpuHasNEON)) { |
2197 | | P210ToAR30Row = P210ToAR30Row_Any_NEON; |
2198 | | if (IS_ALIGNED(width, 8)) { |
2199 | | P210ToAR30Row = P210ToAR30Row_NEON; |
2200 | | } |
2201 | | } |
2202 | | #endif |
2203 | | #if defined(HAS_P210TOAR30ROW_SVE2) |
2204 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2205 | | P210ToAR30Row = P210ToAR30Row_SVE2; |
2206 | | } |
2207 | | #endif |
2208 | | #if defined(HAS_P210TOAR30ROW_SME) |
2209 | | if (TestCpuFlag(kCpuHasSME)) { |
2210 | | P210ToAR30Row = P210ToAR30Row_SME; |
2211 | | } |
2212 | | #endif |
2213 | 0 | for (y = 0; y < height; ++y) { |
2214 | 0 | P210ToAR30Row(src_y, src_uv, dst_ar30, yuvconstants, width); |
2215 | 0 | dst_ar30 += dst_stride_ar30; |
2216 | 0 | src_y += src_stride_y; |
2217 | 0 | src_uv += src_stride_uv; |
2218 | 0 | } |
2219 | 0 | return 0; |
2220 | 0 | } |
2221 | | |
2222 | | // Convert I420 with Alpha to preattenuated ARGB with matrix. |
2223 | | LIBYUV_API |
2224 | | int I420AlphaToARGBMatrix(const uint8_t* src_y, |
2225 | | int src_stride_y, |
2226 | | const uint8_t* src_u, |
2227 | | int src_stride_u, |
2228 | | const uint8_t* src_v, |
2229 | | int src_stride_v, |
2230 | | const uint8_t* src_a, |
2231 | | int src_stride_a, |
2232 | | uint8_t* dst_argb, |
2233 | | int dst_stride_argb, |
2234 | | const struct YuvConstants* yuvconstants, |
2235 | | int width, |
2236 | | int height, |
2237 | 131 | int attenuate) { |
2238 | 131 | int y; |
2239 | 131 | void (*I422AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
2240 | 131 | const uint8_t* v_buf, const uint8_t* a_buf, |
2241 | 131 | uint8_t* dst_argb, |
2242 | 131 | const struct YuvConstants* yuvconstants, |
2243 | 131 | int width) = I422AlphaToARGBRow_C; |
2244 | 131 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2245 | 131 | int width) = ARGBAttenuateRow_C; |
2246 | 131 | assert(yuvconstants); |
2247 | 131 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2248 | 131 | height == 0) { |
2249 | 0 | return -1; |
2250 | 0 | } |
2251 | | // Negative height means invert the image. |
2252 | 131 | if (height < 0) { |
2253 | 0 | height = -height; |
2254 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2255 | 0 | dst_stride_argb = -dst_stride_argb; |
2256 | 0 | } |
2257 | 131 | #if defined(HAS_I422ALPHATOARGBROW_SSSE3) |
2258 | 131 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2259 | 131 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_SSSE3; |
2260 | 131 | if (IS_ALIGNED(width, 8)) { |
2261 | 34 | I422AlphaToARGBRow = I422AlphaToARGBRow_SSSE3; |
2262 | 34 | } |
2263 | 131 | } |
2264 | 131 | #endif |
2265 | 131 | #if defined(HAS_I422ALPHATOARGBROW_AVX2) |
2266 | 131 | if (TestCpuFlag(kCpuHasAVX2)) { |
2267 | 131 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_AVX2; |
2268 | 131 | if (IS_ALIGNED(width, 16)) { |
2269 | 17 | I422AlphaToARGBRow = I422AlphaToARGBRow_AVX2; |
2270 | 17 | } |
2271 | 131 | } |
2272 | 131 | #endif |
2273 | | #if defined(HAS_I422ALPHATOARGBROW_NEON) |
2274 | | if (TestCpuFlag(kCpuHasNEON)) { |
2275 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_NEON; |
2276 | | if (IS_ALIGNED(width, 8)) { |
2277 | | I422AlphaToARGBRow = I422AlphaToARGBRow_NEON; |
2278 | | } |
2279 | | } |
2280 | | #endif |
2281 | | #if defined(HAS_I422ALPHATOARGBROW_SVE2) |
2282 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2283 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SVE2; |
2284 | | } |
2285 | | #endif |
2286 | | #if defined(HAS_I422ALPHATOARGBROW_SME) |
2287 | | if (TestCpuFlag(kCpuHasSME)) { |
2288 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SME; |
2289 | | } |
2290 | | #endif |
2291 | | #if defined(HAS_I422ALPHATOARGBROW_MSA) |
2292 | | if (TestCpuFlag(kCpuHasMSA)) { |
2293 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_MSA; |
2294 | | if (IS_ALIGNED(width, 8)) { |
2295 | | I422AlphaToARGBRow = I422AlphaToARGBRow_MSA; |
2296 | | } |
2297 | | } |
2298 | | #endif |
2299 | | #if defined(HAS_I422ALPHATOARGBROW_LSX) |
2300 | | if (TestCpuFlag(kCpuHasLSX)) { |
2301 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LSX; |
2302 | | if (IS_ALIGNED(width, 16)) { |
2303 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LSX; |
2304 | | } |
2305 | | } |
2306 | | #endif |
2307 | | #if defined(HAS_I422ALPHATOARGBROW_LASX) |
2308 | | if (TestCpuFlag(kCpuHasLASX)) { |
2309 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LASX; |
2310 | | if (IS_ALIGNED(width, 16)) { |
2311 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LASX; |
2312 | | } |
2313 | | } |
2314 | | #endif |
2315 | | #if defined(HAS_I422ALPHATOARGBROW_RVV) |
2316 | | if (TestCpuFlag(kCpuHasRVV)) { |
2317 | | I422AlphaToARGBRow = I422AlphaToARGBRow_RVV; |
2318 | | } |
2319 | | #endif |
2320 | 131 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2321 | 131 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2322 | 131 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2323 | 131 | if (IS_ALIGNED(width, 4)) { |
2324 | 51 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2325 | 51 | } |
2326 | 131 | } |
2327 | 131 | #endif |
2328 | 131 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2329 | 131 | if (TestCpuFlag(kCpuHasAVX2)) { |
2330 | 131 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2331 | 131 | if (IS_ALIGNED(width, 8)) { |
2332 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2333 | 34 | } |
2334 | 131 | } |
2335 | 131 | #endif |
2336 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2337 | | if (TestCpuFlag(kCpuHasNEON)) { |
2338 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2339 | | if (IS_ALIGNED(width, 8)) { |
2340 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2341 | | } |
2342 | | } |
2343 | | #endif |
2344 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
2345 | | if (TestCpuFlag(kCpuHasMSA)) { |
2346 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
2347 | | if (IS_ALIGNED(width, 8)) { |
2348 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
2349 | | } |
2350 | | } |
2351 | | #endif |
2352 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2353 | | if (TestCpuFlag(kCpuHasRVV)) { |
2354 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2355 | | } |
2356 | | #endif |
2357 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2358 | | if (TestCpuFlag(kCpuHasLSX)) { |
2359 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2360 | | if (IS_ALIGNED(width, 8)) { |
2361 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2362 | | } |
2363 | | } |
2364 | | #endif |
2365 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2366 | | if (TestCpuFlag(kCpuHasLASX)) { |
2367 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2368 | | if (IS_ALIGNED(width, 16)) { |
2369 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2370 | | } |
2371 | | } |
2372 | | #endif |
2373 | | |
2374 | 2.80k | for (y = 0; y < height; ++y) { |
2375 | 2.67k | I422AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2376 | 2.67k | width); |
2377 | 2.67k | if (attenuate) { |
2378 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2379 | 0 | } |
2380 | 2.67k | dst_argb += dst_stride_argb; |
2381 | 2.67k | src_a += src_stride_a; |
2382 | 2.67k | src_y += src_stride_y; |
2383 | 2.67k | if (y & 1) { |
2384 | 1.32k | src_u += src_stride_u; |
2385 | 1.32k | src_v += src_stride_v; |
2386 | 1.32k | } |
2387 | 2.67k | } |
2388 | 131 | return 0; |
2389 | 131 | } |
2390 | | |
2391 | | // Convert I422 with Alpha to preattenuated ARGB with matrix. |
2392 | | LIBYUV_API |
2393 | | int I422AlphaToARGBMatrix(const uint8_t* src_y, |
2394 | | int src_stride_y, |
2395 | | const uint8_t* src_u, |
2396 | | int src_stride_u, |
2397 | | const uint8_t* src_v, |
2398 | | int src_stride_v, |
2399 | | const uint8_t* src_a, |
2400 | | int src_stride_a, |
2401 | | uint8_t* dst_argb, |
2402 | | int dst_stride_argb, |
2403 | | const struct YuvConstants* yuvconstants, |
2404 | | int width, |
2405 | | int height, |
2406 | 93 | int attenuate) { |
2407 | 93 | int y; |
2408 | 93 | void (*I422AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
2409 | 93 | const uint8_t* v_buf, const uint8_t* a_buf, |
2410 | 93 | uint8_t* dst_argb, |
2411 | 93 | const struct YuvConstants* yuvconstants, |
2412 | 93 | int width) = I422AlphaToARGBRow_C; |
2413 | 93 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2414 | 93 | int width) = ARGBAttenuateRow_C; |
2415 | 93 | assert(yuvconstants); |
2416 | 93 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2417 | 93 | height == 0) { |
2418 | 0 | return -1; |
2419 | 0 | } |
2420 | | // Negative height means invert the image. |
2421 | 93 | if (height < 0) { |
2422 | 0 | height = -height; |
2423 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2424 | 0 | dst_stride_argb = -dst_stride_argb; |
2425 | 0 | } |
2426 | 93 | #if defined(HAS_I422ALPHATOARGBROW_SSSE3) |
2427 | 93 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2428 | 93 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_SSSE3; |
2429 | 93 | if (IS_ALIGNED(width, 8)) { |
2430 | 34 | I422AlphaToARGBRow = I422AlphaToARGBRow_SSSE3; |
2431 | 34 | } |
2432 | 93 | } |
2433 | 93 | #endif |
2434 | 93 | #if defined(HAS_I422ALPHATOARGBROW_AVX2) |
2435 | 93 | if (TestCpuFlag(kCpuHasAVX2)) { |
2436 | 93 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_AVX2; |
2437 | 93 | if (IS_ALIGNED(width, 16)) { |
2438 | 17 | I422AlphaToARGBRow = I422AlphaToARGBRow_AVX2; |
2439 | 17 | } |
2440 | 93 | } |
2441 | 93 | #endif |
2442 | | #if defined(HAS_I422ALPHATOARGBROW_NEON) |
2443 | | if (TestCpuFlag(kCpuHasNEON)) { |
2444 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_NEON; |
2445 | | if (IS_ALIGNED(width, 8)) { |
2446 | | I422AlphaToARGBRow = I422AlphaToARGBRow_NEON; |
2447 | | } |
2448 | | } |
2449 | | #endif |
2450 | | #if defined(HAS_I422ALPHATOARGBROW_SVE2) |
2451 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2452 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SVE2; |
2453 | | } |
2454 | | #endif |
2455 | | #if defined(HAS_I422ALPHATOARGBROW_SME) |
2456 | | if (TestCpuFlag(kCpuHasSME)) { |
2457 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SME; |
2458 | | } |
2459 | | #endif |
2460 | | #if defined(HAS_I422ALPHATOARGBROW_MSA) |
2461 | | if (TestCpuFlag(kCpuHasMSA)) { |
2462 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_MSA; |
2463 | | if (IS_ALIGNED(width, 8)) { |
2464 | | I422AlphaToARGBRow = I422AlphaToARGBRow_MSA; |
2465 | | } |
2466 | | } |
2467 | | #endif |
2468 | | #if defined(HAS_I422ALPHATOARGBROW_LSX) |
2469 | | if (TestCpuFlag(kCpuHasLSX)) { |
2470 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LSX; |
2471 | | if (IS_ALIGNED(width, 16)) { |
2472 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LSX; |
2473 | | } |
2474 | | } |
2475 | | #endif |
2476 | | #if defined(HAS_I422ALPHATOARGBROW_LASX) |
2477 | | if (TestCpuFlag(kCpuHasLASX)) { |
2478 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LASX; |
2479 | | if (IS_ALIGNED(width, 16)) { |
2480 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LASX; |
2481 | | } |
2482 | | } |
2483 | | #endif |
2484 | | #if defined(HAS_I422ALPHATOARGBROW_RVV) |
2485 | | if (TestCpuFlag(kCpuHasRVV)) { |
2486 | | I422AlphaToARGBRow = I422AlphaToARGBRow_RVV; |
2487 | | } |
2488 | | #endif |
2489 | 93 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2490 | 93 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2491 | 93 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2492 | 93 | if (IS_ALIGNED(width, 4)) { |
2493 | 52 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2494 | 52 | } |
2495 | 93 | } |
2496 | 93 | #endif |
2497 | 93 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2498 | 93 | if (TestCpuFlag(kCpuHasAVX2)) { |
2499 | 93 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2500 | 93 | if (IS_ALIGNED(width, 8)) { |
2501 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2502 | 34 | } |
2503 | 93 | } |
2504 | 93 | #endif |
2505 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2506 | | if (TestCpuFlag(kCpuHasNEON)) { |
2507 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2508 | | if (IS_ALIGNED(width, 8)) { |
2509 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2510 | | } |
2511 | | } |
2512 | | #endif |
2513 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
2514 | | if (TestCpuFlag(kCpuHasMSA)) { |
2515 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
2516 | | if (IS_ALIGNED(width, 8)) { |
2517 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
2518 | | } |
2519 | | } |
2520 | | #endif |
2521 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2522 | | if (TestCpuFlag(kCpuHasRVV)) { |
2523 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2524 | | } |
2525 | | #endif |
2526 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2527 | | if (TestCpuFlag(kCpuHasLSX)) { |
2528 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2529 | | if (IS_ALIGNED(width, 8)) { |
2530 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2531 | | } |
2532 | | } |
2533 | | #endif |
2534 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2535 | | if (TestCpuFlag(kCpuHasLASX)) { |
2536 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2537 | | if (IS_ALIGNED(width, 16)) { |
2538 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2539 | | } |
2540 | | } |
2541 | | #endif |
2542 | | |
2543 | 1.51k | for (y = 0; y < height; ++y) { |
2544 | 1.42k | I422AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2545 | 1.42k | width); |
2546 | 1.42k | if (attenuate) { |
2547 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2548 | 0 | } |
2549 | 1.42k | dst_argb += dst_stride_argb; |
2550 | 1.42k | src_a += src_stride_a; |
2551 | 1.42k | src_y += src_stride_y; |
2552 | 1.42k | src_u += src_stride_u; |
2553 | 1.42k | src_v += src_stride_v; |
2554 | 1.42k | } |
2555 | 93 | return 0; |
2556 | 93 | } |
2557 | | |
2558 | | // Convert I444 with Alpha to preattenuated ARGB with matrix. |
2559 | | LIBYUV_API |
2560 | | int I444AlphaToARGBMatrix(const uint8_t* src_y, |
2561 | | int src_stride_y, |
2562 | | const uint8_t* src_u, |
2563 | | int src_stride_u, |
2564 | | const uint8_t* src_v, |
2565 | | int src_stride_v, |
2566 | | const uint8_t* src_a, |
2567 | | int src_stride_a, |
2568 | | uint8_t* dst_argb, |
2569 | | int dst_stride_argb, |
2570 | | const struct YuvConstants* yuvconstants, |
2571 | | int width, |
2572 | | int height, |
2573 | 123 | int attenuate) { |
2574 | 123 | int y; |
2575 | 123 | void (*I444AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
2576 | 123 | const uint8_t* v_buf, const uint8_t* a_buf, |
2577 | 123 | uint8_t* dst_argb, |
2578 | 123 | const struct YuvConstants* yuvconstants, |
2579 | 123 | int width) = I444AlphaToARGBRow_C; |
2580 | 123 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2581 | 123 | int width) = ARGBAttenuateRow_C; |
2582 | 123 | assert(yuvconstants); |
2583 | 123 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2584 | 123 | height == 0) { |
2585 | 0 | return -1; |
2586 | 0 | } |
2587 | | // Negative height means invert the image. |
2588 | 123 | if (height < 0) { |
2589 | 0 | height = -height; |
2590 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2591 | 0 | dst_stride_argb = -dst_stride_argb; |
2592 | 0 | } |
2593 | 123 | #if defined(HAS_I444ALPHATOARGBROW_SSSE3) |
2594 | 123 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2595 | 123 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_SSSE3; |
2596 | 123 | if (IS_ALIGNED(width, 8)) { |
2597 | 36 | I444AlphaToARGBRow = I444AlphaToARGBRow_SSSE3; |
2598 | 36 | } |
2599 | 123 | } |
2600 | 123 | #endif |
2601 | 123 | #if defined(HAS_I444ALPHATOARGBROW_AVX2) |
2602 | 123 | if (TestCpuFlag(kCpuHasAVX2)) { |
2603 | 123 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_AVX2; |
2604 | 123 | if (IS_ALIGNED(width, 16)) { |
2605 | 19 | I444AlphaToARGBRow = I444AlphaToARGBRow_AVX2; |
2606 | 19 | } |
2607 | 123 | } |
2608 | 123 | #endif |
2609 | | #if defined(HAS_I444ALPHATOARGBROW_NEON) |
2610 | | if (TestCpuFlag(kCpuHasNEON)) { |
2611 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_NEON; |
2612 | | if (IS_ALIGNED(width, 8)) { |
2613 | | I444AlphaToARGBRow = I444AlphaToARGBRow_NEON; |
2614 | | } |
2615 | | } |
2616 | | #endif |
2617 | | #if defined(HAS_I444ALPHATOARGBROW_SVE2) |
2618 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2619 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SVE2; |
2620 | | } |
2621 | | #endif |
2622 | | #if defined(HAS_I444ALPHATOARGBROW_SME) |
2623 | | if (TestCpuFlag(kCpuHasSME)) { |
2624 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SME; |
2625 | | } |
2626 | | #endif |
2627 | | #if defined(HAS_I444ALPHATOARGBROW_MSA) |
2628 | | if (TestCpuFlag(kCpuHasMSA)) { |
2629 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_MSA; |
2630 | | if (IS_ALIGNED(width, 8)) { |
2631 | | I444AlphaToARGBRow = I444AlphaToARGBRow_MSA; |
2632 | | } |
2633 | | } |
2634 | | #endif |
2635 | | #if defined(HAS_I444ALPHATOARGBROW_RVV) |
2636 | | if (TestCpuFlag(kCpuHasRVV)) { |
2637 | | I444AlphaToARGBRow = I444AlphaToARGBRow_RVV; |
2638 | | } |
2639 | | #endif |
2640 | 123 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2641 | 123 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2642 | 123 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2643 | 123 | if (IS_ALIGNED(width, 4)) { |
2644 | 61 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2645 | 61 | } |
2646 | 123 | } |
2647 | 123 | #endif |
2648 | 123 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2649 | 123 | if (TestCpuFlag(kCpuHasAVX2)) { |
2650 | 123 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2651 | 123 | if (IS_ALIGNED(width, 8)) { |
2652 | 36 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2653 | 36 | } |
2654 | 123 | } |
2655 | 123 | #endif |
2656 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2657 | | if (TestCpuFlag(kCpuHasNEON)) { |
2658 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2659 | | if (IS_ALIGNED(width, 8)) { |
2660 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2661 | | } |
2662 | | } |
2663 | | #endif |
2664 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
2665 | | if (TestCpuFlag(kCpuHasMSA)) { |
2666 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
2667 | | if (IS_ALIGNED(width, 8)) { |
2668 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
2669 | | } |
2670 | | } |
2671 | | #endif |
2672 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2673 | | if (TestCpuFlag(kCpuHasRVV)) { |
2674 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2675 | | } |
2676 | | #endif |
2677 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2678 | | if (TestCpuFlag(kCpuHasLSX)) { |
2679 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2680 | | if (IS_ALIGNED(width, 8)) { |
2681 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2682 | | } |
2683 | | } |
2684 | | #endif |
2685 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2686 | | if (TestCpuFlag(kCpuHasLASX)) { |
2687 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2688 | | if (IS_ALIGNED(width, 16)) { |
2689 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2690 | | } |
2691 | | } |
2692 | | #endif |
2693 | | |
2694 | 2.02k | for (y = 0; y < height; ++y) { |
2695 | 1.90k | I444AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2696 | 1.90k | width); |
2697 | 1.90k | if (attenuate) { |
2698 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2699 | 0 | } |
2700 | 1.90k | dst_argb += dst_stride_argb; |
2701 | 1.90k | src_a += src_stride_a; |
2702 | 1.90k | src_y += src_stride_y; |
2703 | 1.90k | src_u += src_stride_u; |
2704 | 1.90k | src_v += src_stride_v; |
2705 | 1.90k | } |
2706 | 123 | return 0; |
2707 | 123 | } |
2708 | | |
2709 | | // Convert I420 with Alpha to ARGB. |
2710 | | LIBYUV_API |
2711 | | int I420AlphaToARGB(const uint8_t* src_y, |
2712 | | int src_stride_y, |
2713 | | const uint8_t* src_u, |
2714 | | int src_stride_u, |
2715 | | const uint8_t* src_v, |
2716 | | int src_stride_v, |
2717 | | const uint8_t* src_a, |
2718 | | int src_stride_a, |
2719 | | uint8_t* dst_argb, |
2720 | | int dst_stride_argb, |
2721 | | int width, |
2722 | | int height, |
2723 | 0 | int attenuate) { |
2724 | 0 | return I420AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
2725 | 0 | src_stride_v, src_a, src_stride_a, dst_argb, |
2726 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
2727 | 0 | height, attenuate); |
2728 | 0 | } |
2729 | | |
2730 | | // Convert I420 with Alpha to ABGR. |
2731 | | LIBYUV_API |
2732 | | int I420AlphaToABGR(const uint8_t* src_y, |
2733 | | int src_stride_y, |
2734 | | const uint8_t* src_u, |
2735 | | int src_stride_u, |
2736 | | const uint8_t* src_v, |
2737 | | int src_stride_v, |
2738 | | const uint8_t* src_a, |
2739 | | int src_stride_a, |
2740 | | uint8_t* dst_abgr, |
2741 | | int dst_stride_abgr, |
2742 | | int width, |
2743 | | int height, |
2744 | 0 | int attenuate) { |
2745 | 0 | return I420AlphaToARGBMatrix( |
2746 | 0 | src_y, src_stride_y, src_v, src_stride_v, // Swap U and V |
2747 | 0 | src_u, src_stride_u, src_a, src_stride_a, dst_abgr, dst_stride_abgr, |
2748 | 0 | &kYvuI601Constants, // Use Yvu matrix |
2749 | 0 | width, height, attenuate); |
2750 | 0 | } |
2751 | | |
2752 | | // Convert I422 with Alpha to ARGB. |
2753 | | LIBYUV_API |
2754 | | int I422AlphaToARGB(const uint8_t* src_y, |
2755 | | int src_stride_y, |
2756 | | const uint8_t* src_u, |
2757 | | int src_stride_u, |
2758 | | const uint8_t* src_v, |
2759 | | int src_stride_v, |
2760 | | const uint8_t* src_a, |
2761 | | int src_stride_a, |
2762 | | uint8_t* dst_argb, |
2763 | | int dst_stride_argb, |
2764 | | int width, |
2765 | | int height, |
2766 | 0 | int attenuate) { |
2767 | 0 | return I422AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
2768 | 0 | src_stride_v, src_a, src_stride_a, dst_argb, |
2769 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
2770 | 0 | height, attenuate); |
2771 | 0 | } |
2772 | | |
2773 | | // Convert I422 with Alpha to ABGR. |
2774 | | LIBYUV_API |
2775 | | int I422AlphaToABGR(const uint8_t* src_y, |
2776 | | int src_stride_y, |
2777 | | const uint8_t* src_u, |
2778 | | int src_stride_u, |
2779 | | const uint8_t* src_v, |
2780 | | int src_stride_v, |
2781 | | const uint8_t* src_a, |
2782 | | int src_stride_a, |
2783 | | uint8_t* dst_abgr, |
2784 | | int dst_stride_abgr, |
2785 | | int width, |
2786 | | int height, |
2787 | 0 | int attenuate) { |
2788 | 0 | return I422AlphaToARGBMatrix( |
2789 | 0 | src_y, src_stride_y, src_v, src_stride_v, // Swap U and V |
2790 | 0 | src_u, src_stride_u, src_a, src_stride_a, dst_abgr, dst_stride_abgr, |
2791 | 0 | &kYvuI601Constants, // Use Yvu matrix |
2792 | 0 | width, height, attenuate); |
2793 | 0 | } |
2794 | | |
2795 | | // Convert I444 with Alpha to ARGB. |
2796 | | LIBYUV_API |
2797 | | int I444AlphaToARGB(const uint8_t* src_y, |
2798 | | int src_stride_y, |
2799 | | const uint8_t* src_u, |
2800 | | int src_stride_u, |
2801 | | const uint8_t* src_v, |
2802 | | int src_stride_v, |
2803 | | const uint8_t* src_a, |
2804 | | int src_stride_a, |
2805 | | uint8_t* dst_argb, |
2806 | | int dst_stride_argb, |
2807 | | int width, |
2808 | | int height, |
2809 | 0 | int attenuate) { |
2810 | 0 | return I444AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
2811 | 0 | src_stride_v, src_a, src_stride_a, dst_argb, |
2812 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
2813 | 0 | height, attenuate); |
2814 | 0 | } |
2815 | | |
2816 | | // Convert I444 with Alpha to ABGR. |
2817 | | LIBYUV_API |
2818 | | int I444AlphaToABGR(const uint8_t* src_y, |
2819 | | int src_stride_y, |
2820 | | const uint8_t* src_u, |
2821 | | int src_stride_u, |
2822 | | const uint8_t* src_v, |
2823 | | int src_stride_v, |
2824 | | const uint8_t* src_a, |
2825 | | int src_stride_a, |
2826 | | uint8_t* dst_abgr, |
2827 | | int dst_stride_abgr, |
2828 | | int width, |
2829 | | int height, |
2830 | 0 | int attenuate) { |
2831 | 0 | return I444AlphaToARGBMatrix( |
2832 | 0 | src_y, src_stride_y, src_v, src_stride_v, // Swap U and V |
2833 | 0 | src_u, src_stride_u, src_a, src_stride_a, dst_abgr, dst_stride_abgr, |
2834 | 0 | &kYvuI601Constants, // Use Yvu matrix |
2835 | 0 | width, height, attenuate); |
2836 | 0 | } |
2837 | | |
2838 | | // Convert I010 with Alpha to preattenuated ARGB with matrix. |
2839 | | LIBYUV_API |
2840 | | int I010AlphaToARGBMatrix(const uint16_t* src_y, |
2841 | | int src_stride_y, |
2842 | | const uint16_t* src_u, |
2843 | | int src_stride_u, |
2844 | | const uint16_t* src_v, |
2845 | | int src_stride_v, |
2846 | | const uint16_t* src_a, |
2847 | | int src_stride_a, |
2848 | | uint8_t* dst_argb, |
2849 | | int dst_stride_argb, |
2850 | | const struct YuvConstants* yuvconstants, |
2851 | | int width, |
2852 | | int height, |
2853 | 125 | int attenuate) { |
2854 | 125 | int y; |
2855 | 125 | void (*I210AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
2856 | 125 | const uint16_t* v_buf, const uint16_t* a_buf, |
2857 | 125 | uint8_t* dst_argb, |
2858 | 125 | const struct YuvConstants* yuvconstants, |
2859 | 125 | int width) = I210AlphaToARGBRow_C; |
2860 | 125 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2861 | 125 | int width) = ARGBAttenuateRow_C; |
2862 | 125 | assert(yuvconstants); |
2863 | 125 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2864 | 125 | height == 0) { |
2865 | 0 | return -1; |
2866 | 0 | } |
2867 | | // Negative height means invert the image. |
2868 | 125 | if (height < 0) { |
2869 | 0 | height = -height; |
2870 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2871 | 0 | dst_stride_argb = -dst_stride_argb; |
2872 | 0 | } |
2873 | | #if defined(HAS_I210ALPHATOARGBROW_NEON) |
2874 | | if (TestCpuFlag(kCpuHasNEON)) { |
2875 | | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_NEON; |
2876 | | if (IS_ALIGNED(width, 8)) { |
2877 | | I210AlphaToARGBRow = I210AlphaToARGBRow_NEON; |
2878 | | } |
2879 | | } |
2880 | | #endif |
2881 | | #if defined(HAS_I210ALPHATOARGBROW_SVE2) |
2882 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2883 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SVE2; |
2884 | | } |
2885 | | #endif |
2886 | | #if defined(HAS_I210ALPHATOARGBROW_SME) |
2887 | | if (TestCpuFlag(kCpuHasSME)) { |
2888 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SME; |
2889 | | } |
2890 | | #endif |
2891 | 125 | #if defined(HAS_I210ALPHATOARGBROW_SSSE3) |
2892 | 125 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2893 | 125 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_SSSE3; |
2894 | 125 | if (IS_ALIGNED(width, 8)) { |
2895 | 34 | I210AlphaToARGBRow = I210AlphaToARGBRow_SSSE3; |
2896 | 34 | } |
2897 | 125 | } |
2898 | 125 | #endif |
2899 | 125 | #if defined(HAS_I210ALPHATOARGBROW_AVX2) |
2900 | 125 | if (TestCpuFlag(kCpuHasAVX2)) { |
2901 | 125 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_AVX2; |
2902 | 125 | if (IS_ALIGNED(width, 16)) { |
2903 | 17 | I210AlphaToARGBRow = I210AlphaToARGBRow_AVX2; |
2904 | 17 | } |
2905 | 125 | } |
2906 | 125 | #endif |
2907 | 125 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2908 | 125 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2909 | 125 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2910 | 125 | if (IS_ALIGNED(width, 4)) { |
2911 | 52 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2912 | 52 | } |
2913 | 125 | } |
2914 | 125 | #endif |
2915 | 125 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2916 | 125 | if (TestCpuFlag(kCpuHasAVX2)) { |
2917 | 125 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2918 | 125 | if (IS_ALIGNED(width, 8)) { |
2919 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2920 | 34 | } |
2921 | 125 | } |
2922 | 125 | #endif |
2923 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2924 | | if (TestCpuFlag(kCpuHasNEON)) { |
2925 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2926 | | if (IS_ALIGNED(width, 8)) { |
2927 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2928 | | } |
2929 | | } |
2930 | | #endif |
2931 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
2932 | | if (TestCpuFlag(kCpuHasMSA)) { |
2933 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
2934 | | if (IS_ALIGNED(width, 8)) { |
2935 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
2936 | | } |
2937 | | } |
2938 | | #endif |
2939 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2940 | | if (TestCpuFlag(kCpuHasRVV)) { |
2941 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2942 | | } |
2943 | | #endif |
2944 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2945 | | if (TestCpuFlag(kCpuHasLSX)) { |
2946 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2947 | | if (IS_ALIGNED(width, 8)) { |
2948 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2949 | | } |
2950 | | } |
2951 | | #endif |
2952 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2953 | | if (TestCpuFlag(kCpuHasLASX)) { |
2954 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2955 | | if (IS_ALIGNED(width, 16)) { |
2956 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2957 | | } |
2958 | | } |
2959 | | #endif |
2960 | | |
2961 | 2.82k | for (y = 0; y < height; ++y) { |
2962 | 2.70k | I210AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2963 | 2.70k | width); |
2964 | 2.70k | if (attenuate) { |
2965 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2966 | 0 | } |
2967 | 2.70k | dst_argb += dst_stride_argb; |
2968 | 2.70k | src_a += src_stride_a; |
2969 | 2.70k | src_y += src_stride_y; |
2970 | 2.70k | if (y & 1) { |
2971 | 1.34k | src_u += src_stride_u; |
2972 | 1.34k | src_v += src_stride_v; |
2973 | 1.34k | } |
2974 | 2.70k | } |
2975 | 125 | return 0; |
2976 | 125 | } |
2977 | | |
2978 | | // Convert I210 with Alpha to preattenuated ARGB with matrix. |
2979 | | LIBYUV_API |
2980 | | int I210AlphaToARGBMatrix(const uint16_t* src_y, |
2981 | | int src_stride_y, |
2982 | | const uint16_t* src_u, |
2983 | | int src_stride_u, |
2984 | | const uint16_t* src_v, |
2985 | | int src_stride_v, |
2986 | | const uint16_t* src_a, |
2987 | | int src_stride_a, |
2988 | | uint8_t* dst_argb, |
2989 | | int dst_stride_argb, |
2990 | | const struct YuvConstants* yuvconstants, |
2991 | | int width, |
2992 | | int height, |
2993 | 84 | int attenuate) { |
2994 | 84 | int y; |
2995 | 84 | void (*I210AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
2996 | 84 | const uint16_t* v_buf, const uint16_t* a_buf, |
2997 | 84 | uint8_t* dst_argb, |
2998 | 84 | const struct YuvConstants* yuvconstants, |
2999 | 84 | int width) = I210AlphaToARGBRow_C; |
3000 | 84 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
3001 | 84 | int width) = ARGBAttenuateRow_C; |
3002 | 84 | assert(yuvconstants); |
3003 | 84 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
3004 | 84 | height == 0) { |
3005 | 0 | return -1; |
3006 | 0 | } |
3007 | | // Negative height means invert the image. |
3008 | 84 | if (height < 0) { |
3009 | 0 | height = -height; |
3010 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
3011 | 0 | dst_stride_argb = -dst_stride_argb; |
3012 | 0 | } |
3013 | | #if defined(HAS_I210ALPHATOARGBROW_NEON) |
3014 | | if (TestCpuFlag(kCpuHasNEON)) { |
3015 | | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_NEON; |
3016 | | if (IS_ALIGNED(width, 8)) { |
3017 | | I210AlphaToARGBRow = I210AlphaToARGBRow_NEON; |
3018 | | } |
3019 | | } |
3020 | | #endif |
3021 | | #if defined(HAS_I210ALPHATOARGBROW_SVE2) |
3022 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3023 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SVE2; |
3024 | | } |
3025 | | #endif |
3026 | | #if defined(HAS_I210ALPHATOARGBROW_SME) |
3027 | | if (TestCpuFlag(kCpuHasSME)) { |
3028 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SME; |
3029 | | } |
3030 | | #endif |
3031 | 84 | #if defined(HAS_I210ALPHATOARGBROW_SSSE3) |
3032 | 84 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3033 | 84 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_SSSE3; |
3034 | 84 | if (IS_ALIGNED(width, 8)) { |
3035 | 34 | I210AlphaToARGBRow = I210AlphaToARGBRow_SSSE3; |
3036 | 34 | } |
3037 | 84 | } |
3038 | 84 | #endif |
3039 | 84 | #if defined(HAS_I210ALPHATOARGBROW_AVX2) |
3040 | 84 | if (TestCpuFlag(kCpuHasAVX2)) { |
3041 | 84 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_AVX2; |
3042 | 84 | if (IS_ALIGNED(width, 16)) { |
3043 | 17 | I210AlphaToARGBRow = I210AlphaToARGBRow_AVX2; |
3044 | 17 | } |
3045 | 84 | } |
3046 | 84 | #endif |
3047 | 84 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
3048 | 84 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3049 | 84 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
3050 | 84 | if (IS_ALIGNED(width, 4)) { |
3051 | 51 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
3052 | 51 | } |
3053 | 84 | } |
3054 | 84 | #endif |
3055 | 84 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
3056 | 84 | if (TestCpuFlag(kCpuHasAVX2)) { |
3057 | 84 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
3058 | 84 | if (IS_ALIGNED(width, 8)) { |
3059 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
3060 | 34 | } |
3061 | 84 | } |
3062 | 84 | #endif |
3063 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
3064 | | if (TestCpuFlag(kCpuHasNEON)) { |
3065 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
3066 | | if (IS_ALIGNED(width, 8)) { |
3067 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
3068 | | } |
3069 | | } |
3070 | | #endif |
3071 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
3072 | | if (TestCpuFlag(kCpuHasMSA)) { |
3073 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
3074 | | if (IS_ALIGNED(width, 8)) { |
3075 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
3076 | | } |
3077 | | } |
3078 | | #endif |
3079 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
3080 | | if (TestCpuFlag(kCpuHasRVV)) { |
3081 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
3082 | | } |
3083 | | #endif |
3084 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
3085 | | if (TestCpuFlag(kCpuHasLSX)) { |
3086 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
3087 | | if (IS_ALIGNED(width, 8)) { |
3088 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
3089 | | } |
3090 | | } |
3091 | | #endif |
3092 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
3093 | | if (TestCpuFlag(kCpuHasLASX)) { |
3094 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
3095 | | if (IS_ALIGNED(width, 16)) { |
3096 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
3097 | | } |
3098 | | } |
3099 | | #endif |
3100 | | |
3101 | 1.49k | for (y = 0; y < height; ++y) { |
3102 | 1.41k | I210AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
3103 | 1.41k | width); |
3104 | 1.41k | if (attenuate) { |
3105 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
3106 | 0 | } |
3107 | 1.41k | dst_argb += dst_stride_argb; |
3108 | 1.41k | src_a += src_stride_a; |
3109 | 1.41k | src_y += src_stride_y; |
3110 | 1.41k | src_u += src_stride_u; |
3111 | 1.41k | src_v += src_stride_v; |
3112 | 1.41k | } |
3113 | 84 | return 0; |
3114 | 84 | } |
3115 | | |
3116 | | // Convert I410 with Alpha to preattenuated ARGB with matrix. |
3117 | | LIBYUV_API |
3118 | | int I410AlphaToARGBMatrix(const uint16_t* src_y, |
3119 | | int src_stride_y, |
3120 | | const uint16_t* src_u, |
3121 | | int src_stride_u, |
3122 | | const uint16_t* src_v, |
3123 | | int src_stride_v, |
3124 | | const uint16_t* src_a, |
3125 | | int src_stride_a, |
3126 | | uint8_t* dst_argb, |
3127 | | int dst_stride_argb, |
3128 | | const struct YuvConstants* yuvconstants, |
3129 | | int width, |
3130 | | int height, |
3131 | 87 | int attenuate) { |
3132 | 87 | int y; |
3133 | 87 | void (*I410AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
3134 | 87 | const uint16_t* v_buf, const uint16_t* a_buf, |
3135 | 87 | uint8_t* dst_argb, |
3136 | 87 | const struct YuvConstants* yuvconstants, |
3137 | 87 | int width) = I410AlphaToARGBRow_C; |
3138 | 87 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
3139 | 87 | int width) = ARGBAttenuateRow_C; |
3140 | 87 | assert(yuvconstants); |
3141 | 87 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
3142 | 87 | height == 0) { |
3143 | 0 | return -1; |
3144 | 0 | } |
3145 | | // Negative height means invert the image. |
3146 | 87 | if (height < 0) { |
3147 | 0 | height = -height; |
3148 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
3149 | 0 | dst_stride_argb = -dst_stride_argb; |
3150 | 0 | } |
3151 | | #if defined(HAS_I410ALPHATOARGBROW_NEON) |
3152 | | if (TestCpuFlag(kCpuHasNEON)) { |
3153 | | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_NEON; |
3154 | | if (IS_ALIGNED(width, 8)) { |
3155 | | I410AlphaToARGBRow = I410AlphaToARGBRow_NEON; |
3156 | | } |
3157 | | } |
3158 | | #endif |
3159 | | #if defined(HAS_I410ALPHATOARGBROW_SVE2) |
3160 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3161 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SVE2; |
3162 | | } |
3163 | | #endif |
3164 | | #if defined(HAS_I410ALPHATOARGBROW_SME) |
3165 | | if (TestCpuFlag(kCpuHasSME)) { |
3166 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SME; |
3167 | | } |
3168 | | #endif |
3169 | 87 | #if defined(HAS_I410ALPHATOARGBROW_SSSE3) |
3170 | 87 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3171 | 87 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_SSSE3; |
3172 | 87 | if (IS_ALIGNED(width, 8)) { |
3173 | 34 | I410AlphaToARGBRow = I410AlphaToARGBRow_SSSE3; |
3174 | 34 | } |
3175 | 87 | } |
3176 | 87 | #endif |
3177 | 87 | #if defined(HAS_I410ALPHATOARGBROW_AVX2) |
3178 | 87 | if (TestCpuFlag(kCpuHasAVX2)) { |
3179 | 87 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_AVX2; |
3180 | 87 | if (IS_ALIGNED(width, 16)) { |
3181 | 17 | I410AlphaToARGBRow = I410AlphaToARGBRow_AVX2; |
3182 | 17 | } |
3183 | 87 | } |
3184 | 87 | #endif |
3185 | 87 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
3186 | 87 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3187 | 87 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
3188 | 87 | if (IS_ALIGNED(width, 4)) { |
3189 | 51 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
3190 | 51 | } |
3191 | 87 | } |
3192 | 87 | #endif |
3193 | 87 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
3194 | 87 | if (TestCpuFlag(kCpuHasAVX2)) { |
3195 | 87 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
3196 | 87 | if (IS_ALIGNED(width, 8)) { |
3197 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
3198 | 34 | } |
3199 | 87 | } |
3200 | 87 | #endif |
3201 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
3202 | | if (TestCpuFlag(kCpuHasNEON)) { |
3203 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
3204 | | if (IS_ALIGNED(width, 8)) { |
3205 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
3206 | | } |
3207 | | } |
3208 | | #endif |
3209 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
3210 | | if (TestCpuFlag(kCpuHasMSA)) { |
3211 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
3212 | | if (IS_ALIGNED(width, 8)) { |
3213 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
3214 | | } |
3215 | | } |
3216 | | #endif |
3217 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
3218 | | if (TestCpuFlag(kCpuHasRVV)) { |
3219 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
3220 | | } |
3221 | | #endif |
3222 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
3223 | | if (TestCpuFlag(kCpuHasLSX)) { |
3224 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
3225 | | if (IS_ALIGNED(width, 8)) { |
3226 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
3227 | | } |
3228 | | } |
3229 | | #endif |
3230 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
3231 | | if (TestCpuFlag(kCpuHasLASX)) { |
3232 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
3233 | | if (IS_ALIGNED(width, 16)) { |
3234 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
3235 | | } |
3236 | | } |
3237 | | #endif |
3238 | | |
3239 | 1.74k | for (y = 0; y < height; ++y) { |
3240 | 1.66k | I410AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
3241 | 1.66k | width); |
3242 | 1.66k | if (attenuate) { |
3243 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
3244 | 0 | } |
3245 | 1.66k | dst_argb += dst_stride_argb; |
3246 | 1.66k | src_a += src_stride_a; |
3247 | 1.66k | src_y += src_stride_y; |
3248 | 1.66k | src_u += src_stride_u; |
3249 | 1.66k | src_v += src_stride_v; |
3250 | 1.66k | } |
3251 | 87 | return 0; |
3252 | 87 | } |
3253 | | |
3254 | | // Convert I400 to ARGB with matrix. |
3255 | | LIBYUV_API |
3256 | | int I400ToARGBMatrix(const uint8_t* src_y, |
3257 | | int src_stride_y, |
3258 | | uint8_t* dst_argb, |
3259 | | int dst_stride_argb, |
3260 | | const struct YuvConstants* yuvconstants, |
3261 | | int width, |
3262 | 347 | int height) { |
3263 | 347 | int y; |
3264 | 347 | void (*I400ToARGBRow)(const uint8_t* y_buf, uint8_t* rgb_buf, |
3265 | 347 | const struct YuvConstants* yuvconstants, int width) = |
3266 | 347 | I400ToARGBRow_C; |
3267 | 347 | assert(yuvconstants); |
3268 | 347 | if (!src_y || !dst_argb || width <= 0 || height == 0) { |
3269 | 0 | return -1; |
3270 | 0 | } |
3271 | | // Negative height means invert the image. |
3272 | 347 | if (height < 0) { |
3273 | 0 | height = -height; |
3274 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
3275 | 0 | dst_stride_argb = -dst_stride_argb; |
3276 | 0 | } |
3277 | | // Coalesce rows. |
3278 | 347 | if (src_stride_y == width && dst_stride_argb == width * 4) { |
3279 | 347 | width *= height; |
3280 | 347 | height = 1; |
3281 | 347 | src_stride_y = dst_stride_argb = 0; |
3282 | 347 | } |
3283 | 347 | #if defined(HAS_I400TOARGBROW_SSE2) |
3284 | 347 | if (TestCpuFlag(kCpuHasSSE2)) { |
3285 | 347 | I400ToARGBRow = I400ToARGBRow_Any_SSE2; |
3286 | 347 | if (IS_ALIGNED(width, 8)) { |
3287 | 76 | I400ToARGBRow = I400ToARGBRow_SSE2; |
3288 | 76 | } |
3289 | 347 | } |
3290 | 347 | #endif |
3291 | 347 | #if defined(HAS_I400TOARGBROW_AVX2) |
3292 | 347 | if (TestCpuFlag(kCpuHasAVX2)) { |
3293 | 347 | I400ToARGBRow = I400ToARGBRow_Any_AVX2; |
3294 | 347 | if (IS_ALIGNED(width, 16)) { |
3295 | 33 | I400ToARGBRow = I400ToARGBRow_AVX2; |
3296 | 33 | } |
3297 | 347 | } |
3298 | 347 | #endif |
3299 | | #if defined(HAS_I400TOARGBROW_NEON) |
3300 | | if (TestCpuFlag(kCpuHasNEON)) { |
3301 | | I400ToARGBRow = I400ToARGBRow_Any_NEON; |
3302 | | if (IS_ALIGNED(width, 8)) { |
3303 | | I400ToARGBRow = I400ToARGBRow_NEON; |
3304 | | } |
3305 | | } |
3306 | | #endif |
3307 | | #if defined(HAS_I400TOARGBROW_SVE2) |
3308 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3309 | | I400ToARGBRow = I400ToARGBRow_SVE2; |
3310 | | } |
3311 | | #endif |
3312 | | #if defined(HAS_I400TOARGBROW_SME) |
3313 | | if (TestCpuFlag(kCpuHasSME)) { |
3314 | | I400ToARGBRow = I400ToARGBRow_SME; |
3315 | | } |
3316 | | #endif |
3317 | | #if defined(HAS_I400TOARGBROW_MSA) |
3318 | | if (TestCpuFlag(kCpuHasMSA)) { |
3319 | | I400ToARGBRow = I400ToARGBRow_Any_MSA; |
3320 | | if (IS_ALIGNED(width, 16)) { |
3321 | | I400ToARGBRow = I400ToARGBRow_MSA; |
3322 | | } |
3323 | | } |
3324 | | #endif |
3325 | | #if defined(HAS_I400TOARGBROW_LSX) |
3326 | | if (TestCpuFlag(kCpuHasLSX)) { |
3327 | | I400ToARGBRow = I400ToARGBRow_Any_LSX; |
3328 | | if (IS_ALIGNED(width, 16)) { |
3329 | | I400ToARGBRow = I400ToARGBRow_LSX; |
3330 | | } |
3331 | | } |
3332 | | #endif |
3333 | | #if defined(HAS_I400TOARGBROW_RVV) |
3334 | | if (TestCpuFlag(kCpuHasRVV)) { |
3335 | | I400ToARGBRow = I400ToARGBRow_RVV; |
3336 | | } |
3337 | | #endif |
3338 | | |
3339 | 694 | for (y = 0; y < height; ++y) { |
3340 | 347 | I400ToARGBRow(src_y, dst_argb, yuvconstants, width); |
3341 | 347 | dst_argb += dst_stride_argb; |
3342 | 347 | src_y += src_stride_y; |
3343 | 347 | } |
3344 | 347 | return 0; |
3345 | 347 | } |
3346 | | |
3347 | | // Convert I400 to ARGB. |
3348 | | LIBYUV_API |
3349 | | int I400ToARGB(const uint8_t* src_y, |
3350 | | int src_stride_y, |
3351 | | uint8_t* dst_argb, |
3352 | | int dst_stride_argb, |
3353 | | int width, |
3354 | 0 | int height) { |
3355 | 0 | return I400ToARGBMatrix(src_y, src_stride_y, dst_argb, dst_stride_argb, |
3356 | 0 | &kYuvI601Constants, width, height); |
3357 | 0 | } |
3358 | | |
3359 | | // Convert J400 to ARGB. |
3360 | | LIBYUV_API |
3361 | | int J400ToARGB(const uint8_t* src_y, |
3362 | | int src_stride_y, |
3363 | | uint8_t* dst_argb, |
3364 | | int dst_stride_argb, |
3365 | | int width, |
3366 | 0 | int height) { |
3367 | 0 | int y; |
3368 | 0 | void (*J400ToARGBRow)(const uint8_t* src_y, uint8_t* dst_argb, int width) = |
3369 | 0 | J400ToARGBRow_C; |
3370 | 0 | if (!src_y || !dst_argb || width <= 0 || height == 0) { |
3371 | 0 | return -1; |
3372 | 0 | } |
3373 | | // Negative height means invert the image. |
3374 | 0 | if (height < 0) { |
3375 | 0 | height = -height; |
3376 | 0 | src_y = src_y + (height - 1) * src_stride_y; |
3377 | 0 | src_stride_y = -src_stride_y; |
3378 | 0 | } |
3379 | | // Coalesce rows. |
3380 | 0 | if (src_stride_y == width && dst_stride_argb == width * 4) { |
3381 | 0 | width *= height; |
3382 | 0 | height = 1; |
3383 | 0 | src_stride_y = dst_stride_argb = 0; |
3384 | 0 | } |
3385 | 0 | #if defined(HAS_J400TOARGBROW_SSE2) |
3386 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
3387 | 0 | J400ToARGBRow = J400ToARGBRow_Any_SSE2; |
3388 | 0 | if (IS_ALIGNED(width, 8)) { |
3389 | 0 | J400ToARGBRow = J400ToARGBRow_SSE2; |
3390 | 0 | } |
3391 | 0 | } |
3392 | 0 | #endif |
3393 | | #if defined(HAS_J400TOARGBROW_AVX2) |
3394 | | if (TestCpuFlag(kCpuHasAVX2)) { |
3395 | | J400ToARGBRow = J400ToARGBRow_Any_AVX2; |
3396 | | if (IS_ALIGNED(width, 16)) { |
3397 | | J400ToARGBRow = J400ToARGBRow_AVX2; |
3398 | | } |
3399 | | } |
3400 | | #endif |
3401 | | #if defined(HAS_J400TOARGBROW_NEON) |
3402 | | if (TestCpuFlag(kCpuHasNEON)) { |
3403 | | J400ToARGBRow = J400ToARGBRow_Any_NEON; |
3404 | | if (IS_ALIGNED(width, 8)) { |
3405 | | J400ToARGBRow = J400ToARGBRow_NEON; |
3406 | | } |
3407 | | } |
3408 | | #endif |
3409 | | #if defined(HAS_J400TOARGBROW_MSA) |
3410 | | if (TestCpuFlag(kCpuHasMSA)) { |
3411 | | J400ToARGBRow = J400ToARGBRow_Any_MSA; |
3412 | | if (IS_ALIGNED(width, 16)) { |
3413 | | J400ToARGBRow = J400ToARGBRow_MSA; |
3414 | | } |
3415 | | } |
3416 | | #endif |
3417 | | #if defined(HAS_J400TOARGBROW_LSX) |
3418 | | if (TestCpuFlag(kCpuHasLSX)) { |
3419 | | J400ToARGBRow = J400ToARGBRow_Any_LSX; |
3420 | | if (IS_ALIGNED(width, 16)) { |
3421 | | J400ToARGBRow = J400ToARGBRow_LSX; |
3422 | | } |
3423 | | } |
3424 | | #endif |
3425 | | #if defined(HAS_J400TOARGBROW_RVV) |
3426 | | if (TestCpuFlag(kCpuHasRVV)) { |
3427 | | J400ToARGBRow = J400ToARGBRow_RVV; |
3428 | | } |
3429 | | #endif |
3430 | |
|
3431 | 0 | for (y = 0; y < height; ++y) { |
3432 | 0 | J400ToARGBRow(src_y, dst_argb, width); |
3433 | 0 | src_y += src_stride_y; |
3434 | 0 | dst_argb += dst_stride_argb; |
3435 | 0 | } |
3436 | 0 | return 0; |
3437 | 0 | } |
3438 | | |
3439 | | #ifndef __riscv |
3440 | | // Shuffle table for converting BGRA to ARGB. |
3441 | | static const uvec8 kShuffleMaskBGRAToARGB = { |
3442 | | 3u, 2u, 1u, 0u, 7u, 6u, 5u, 4u, 11u, 10u, 9u, 8u, 15u, 14u, 13u, 12u}; |
3443 | | |
3444 | | // Shuffle table for converting ABGR to ARGB. |
3445 | | static const uvec8 kShuffleMaskABGRToARGB = { |
3446 | | 2u, 1u, 0u, 3u, 6u, 5u, 4u, 7u, 10u, 9u, 8u, 11u, 14u, 13u, 12u, 15u}; |
3447 | | |
3448 | | // Shuffle table for converting RGBA to ARGB. |
3449 | | static const uvec8 kShuffleMaskRGBAToARGB = { |
3450 | | 1u, 2u, 3u, 0u, 5u, 6u, 7u, 4u, 9u, 10u, 11u, 8u, 13u, 14u, 15u, 12u}; |
3451 | | |
3452 | | // Shuffle table for converting AR64 to AB64. |
3453 | | static const uvec8 kShuffleMaskAR64ToAB64 = { |
3454 | | 4u, 5u, 2u, 3u, 0u, 1u, 6u, 7u, 12u, 13u, 10u, 11u, 8u, 9u, 14u, 15u}; |
3455 | | |
3456 | | // Convert BGRA to ARGB. |
3457 | | LIBYUV_API |
3458 | | int BGRAToARGB(const uint8_t* src_bgra, |
3459 | | int src_stride_bgra, |
3460 | | uint8_t* dst_argb, |
3461 | | int dst_stride_argb, |
3462 | | int width, |
3463 | 0 | int height) { |
3464 | 0 | return ARGBShuffle(src_bgra, src_stride_bgra, dst_argb, dst_stride_argb, |
3465 | 0 | (const uint8_t*)&kShuffleMaskBGRAToARGB, width, height); |
3466 | 0 | } |
3467 | | |
3468 | | // Convert ARGB to BGRA (same as BGRAToARGB). |
3469 | | LIBYUV_API |
3470 | | int ARGBToBGRA(const uint8_t* src_bgra, |
3471 | | int src_stride_bgra, |
3472 | | uint8_t* dst_argb, |
3473 | | int dst_stride_argb, |
3474 | | int width, |
3475 | 0 | int height) { |
3476 | 0 | return ARGBShuffle(src_bgra, src_stride_bgra, dst_argb, dst_stride_argb, |
3477 | 0 | (const uint8_t*)&kShuffleMaskBGRAToARGB, width, height); |
3478 | 0 | } |
3479 | | |
3480 | | // Convert ABGR to ARGB. |
3481 | | LIBYUV_API |
3482 | | int ABGRToARGB(const uint8_t* src_abgr, |
3483 | | int src_stride_abgr, |
3484 | | uint8_t* dst_argb, |
3485 | | int dst_stride_argb, |
3486 | | int width, |
3487 | 0 | int height) { |
3488 | 0 | return ARGBShuffle(src_abgr, src_stride_abgr, dst_argb, dst_stride_argb, |
3489 | 0 | (const uint8_t*)&kShuffleMaskABGRToARGB, width, height); |
3490 | 0 | } |
3491 | | |
3492 | | // Convert ARGB to ABGR to (same as ABGRToARGB). |
3493 | | LIBYUV_API |
3494 | | int ARGBToABGR(const uint8_t* src_abgr, |
3495 | | int src_stride_abgr, |
3496 | | uint8_t* dst_argb, |
3497 | | int dst_stride_argb, |
3498 | | int width, |
3499 | 0 | int height) { |
3500 | 0 | return ARGBShuffle(src_abgr, src_stride_abgr, dst_argb, dst_stride_argb, |
3501 | 0 | (const uint8_t*)&kShuffleMaskABGRToARGB, width, height); |
3502 | 0 | } |
3503 | | |
3504 | | // Convert RGBA to ARGB. |
3505 | | LIBYUV_API |
3506 | | int RGBAToARGB(const uint8_t* src_rgba, |
3507 | | int src_stride_rgba, |
3508 | | uint8_t* dst_argb, |
3509 | | int dst_stride_argb, |
3510 | | int width, |
3511 | 0 | int height) { |
3512 | 0 | return ARGBShuffle(src_rgba, src_stride_rgba, dst_argb, dst_stride_argb, |
3513 | 0 | (const uint8_t*)&kShuffleMaskRGBAToARGB, width, height); |
3514 | 0 | } |
3515 | | |
3516 | | // Convert AR64 To AB64. |
3517 | | LIBYUV_API |
3518 | | int AR64ToAB64(const uint16_t* src_ar64, |
3519 | | int src_stride_ar64, |
3520 | | uint16_t* dst_ab64, |
3521 | | int dst_stride_ab64, |
3522 | | int width, |
3523 | 0 | int height) { |
3524 | 0 | return AR64Shuffle(src_ar64, src_stride_ar64, dst_ab64, dst_stride_ab64, |
3525 | 0 | (const uint8_t*)&kShuffleMaskAR64ToAB64, width, height); |
3526 | 0 | } |
3527 | | #else |
3528 | | // Convert BGRA to ARGB (same as ARGBToBGRA). |
3529 | | LIBYUV_API |
3530 | | int BGRAToARGB(const uint8_t* src_bgra, |
3531 | | int src_stride_bgra, |
3532 | | uint8_t* dst_argb, |
3533 | | int dst_stride_argb, |
3534 | | int width, |
3535 | | int height) { |
3536 | | return ARGBToBGRA(src_bgra, src_stride_bgra, dst_argb, dst_stride_argb, width, |
3537 | | height); |
3538 | | } |
3539 | | |
3540 | | // Convert ARGB to BGRA. |
3541 | | LIBYUV_API |
3542 | | int ARGBToBGRA(const uint8_t* src_argb, |
3543 | | int src_stride_argb, |
3544 | | uint8_t* dst_bgra, |
3545 | | int dst_stride_bgra, |
3546 | | int width, |
3547 | | int height) { |
3548 | | int y; |
3549 | | void (*ARGBToBGRARow)(const uint8_t* src_argb, uint8_t* dst_bgra, int width) = |
3550 | | ARGBToBGRARow_C; |
3551 | | if (!src_argb || !dst_bgra || width <= 0 || height == 0) { |
3552 | | return -1; |
3553 | | } |
3554 | | // Negative height means invert the image. |
3555 | | if (height < 0) { |
3556 | | height = -height; |
3557 | | src_argb = src_argb + (height - 1) * src_stride_argb; |
3558 | | src_stride_argb = -src_stride_argb; |
3559 | | } |
3560 | | // Coalesce rows. |
3561 | | if (src_stride_argb == width * 4 && dst_stride_bgra == width * 4) { |
3562 | | width *= height; |
3563 | | height = 1; |
3564 | | src_stride_argb = dst_stride_bgra = 0; |
3565 | | } |
3566 | | |
3567 | | #if defined(HAS_ARGBTOBGRAROW_RVV) |
3568 | | if (TestCpuFlag(kCpuHasRVV)) { |
3569 | | ARGBToBGRARow = ARGBToBGRARow_RVV; |
3570 | | } |
3571 | | #endif |
3572 | | |
3573 | | for (y = 0; y < height; ++y) { |
3574 | | ARGBToBGRARow(src_argb, dst_bgra, width); |
3575 | | src_argb += src_stride_argb; |
3576 | | dst_bgra += dst_stride_bgra; |
3577 | | } |
3578 | | return 0; |
3579 | | } |
3580 | | |
3581 | | // Convert ARGB to ABGR. |
3582 | | LIBYUV_API |
3583 | | int ARGBToABGR(const uint8_t* src_argb, |
3584 | | int src_stride_argb, |
3585 | | uint8_t* dst_abgr, |
3586 | | int dst_stride_abgr, |
3587 | | int width, |
3588 | | int height) { |
3589 | | int y; |
3590 | | void (*ARGBToABGRRow)(const uint8_t* src_argb, uint8_t* dst_abgr, int width) = |
3591 | | ARGBToABGRRow_C; |
3592 | | if (!src_argb || !dst_abgr || width <= 0 || height == 0) { |
3593 | | return -1; |
3594 | | } |
3595 | | // Negative height means invert the image. |
3596 | | if (height < 0) { |
3597 | | height = -height; |
3598 | | src_argb = src_argb + (height - 1) * src_stride_argb; |
3599 | | src_stride_argb = -src_stride_argb; |
3600 | | } |
3601 | | // Coalesce rows. |
3602 | | if (src_stride_argb == width * 4 && dst_stride_abgr == width * 4) { |
3603 | | width *= height; |
3604 | | height = 1; |
3605 | | src_stride_argb = dst_stride_abgr = 0; |
3606 | | } |
3607 | | |
3608 | | #if defined(HAS_ARGBTOABGRROW_RVV) |
3609 | | if (TestCpuFlag(kCpuHasRVV)) { |
3610 | | ARGBToABGRRow = ARGBToABGRRow_RVV; |
3611 | | } |
3612 | | #endif |
3613 | | |
3614 | | for (y = 0; y < height; ++y) { |
3615 | | ARGBToABGRRow(src_argb, dst_abgr, width); |
3616 | | src_argb += src_stride_argb; |
3617 | | dst_abgr += dst_stride_abgr; |
3618 | | } |
3619 | | return 0; |
3620 | | } |
3621 | | |
3622 | | // Convert ABGR to ARGB (same as ARGBToABGR). |
3623 | | LIBYUV_API |
3624 | | int ABGRToARGB(const uint8_t* src_abgr, |
3625 | | int src_stride_abgr, |
3626 | | uint8_t* dst_argb, |
3627 | | int dst_stride_argb, |
3628 | | int width, |
3629 | | int height) { |
3630 | | return ARGBToABGR(src_abgr, src_stride_abgr, dst_argb, dst_stride_argb, width, |
3631 | | height); |
3632 | | } |
3633 | | |
3634 | | // Convert RGBA to ARGB. |
3635 | | LIBYUV_API |
3636 | | int RGBAToARGB(const uint8_t* src_rgba, |
3637 | | int src_stride_rgba, |
3638 | | uint8_t* dst_argb, |
3639 | | int dst_stride_argb, |
3640 | | int width, |
3641 | | int height) { |
3642 | | int y; |
3643 | | void (*RGBAToARGBRow)(const uint8_t* src_rgba, uint8_t* dst_argb, int width) = |
3644 | | RGBAToARGBRow_C; |
3645 | | if (!src_rgba || !dst_argb || width <= 0 || height == 0) { |
3646 | | return -1; |
3647 | | } |
3648 | | // Negative height means invert the image. |
3649 | | if (height < 0) { |
3650 | | height = -height; |
3651 | | src_rgba = src_rgba + (height - 1) * src_stride_rgba; |
3652 | | src_stride_rgba = -src_stride_rgba; |
3653 | | } |
3654 | | // Coalesce rows. |
3655 | | if (src_stride_rgba == width * 4 && dst_stride_argb == width * 4) { |
3656 | | width *= height; |
3657 | | height = 1; |
3658 | | src_stride_rgba = dst_stride_argb = 0; |
3659 | | } |
3660 | | |
3661 | | #if defined(HAS_RGBATOARGBROW_RVV) |
3662 | | if (TestCpuFlag(kCpuHasRVV)) { |
3663 | | RGBAToARGBRow = RGBAToARGBRow_RVV; |
3664 | | } |
3665 | | #endif |
3666 | | |
3667 | | for (y = 0; y < height; ++y) { |
3668 | | RGBAToARGBRow(src_rgba, dst_argb, width); |
3669 | | src_rgba += src_stride_rgba; |
3670 | | dst_argb += dst_stride_argb; |
3671 | | } |
3672 | | return 0; |
3673 | | } |
3674 | | |
3675 | | // Convert AR64 To AB64. |
3676 | | LIBYUV_API |
3677 | | int AR64ToAB64(const uint16_t* src_ar64, |
3678 | | int src_stride_ar64, |
3679 | | uint16_t* dst_ab64, |
3680 | | int dst_stride_ab64, |
3681 | | int width, |
3682 | | int height) { |
3683 | | int y; |
3684 | | void (*AR64ToAB64Row)(const uint16_t* src_ar64, uint16_t* dst_ab64, |
3685 | | int width) = AR64ToAB64Row_C; |
3686 | | if (!src_ar64 || !dst_ab64 || width <= 0 || height == 0) { |
3687 | | return -1; |
3688 | | } |
3689 | | // Negative height means invert the image. |
3690 | | if (height < 0) { |
3691 | | height = -height; |
3692 | | src_ar64 = src_ar64 + (height - 1) * src_stride_ar64; |
3693 | | src_stride_ar64 = -src_stride_ar64; |
3694 | | } |
3695 | | // Coalesce rows. |
3696 | | if (src_stride_ar64 == width * 4 && dst_stride_ab64 == width * 4) { |
3697 | | width *= height; |
3698 | | height = 1; |
3699 | | src_stride_ar64 = dst_stride_ab64 = 0; |
3700 | | } |
3701 | | |
3702 | | #if defined(HAS_AR64TOAB64ROW_RVV) |
3703 | | if (TestCpuFlag(kCpuHasRVV)) { |
3704 | | AR64ToAB64Row = AR64ToAB64Row_RVV; |
3705 | | } |
3706 | | #endif |
3707 | | |
3708 | | for (y = 0; y < height; ++y) { |
3709 | | AR64ToAB64Row(src_ar64, dst_ab64, width); |
3710 | | src_ar64 += src_stride_ar64; |
3711 | | dst_ab64 += dst_stride_ab64; |
3712 | | } |
3713 | | return 0; |
3714 | | } |
3715 | | #endif |
3716 | | |
3717 | | // Convert RGB24 to ARGB. |
3718 | | LIBYUV_API |
3719 | | int RGB24ToARGB(const uint8_t* src_rgb24, |
3720 | | int src_stride_rgb24, |
3721 | | uint8_t* dst_argb, |
3722 | | int dst_stride_argb, |
3723 | | int width, |
3724 | 0 | int height) { |
3725 | 0 | int y; |
3726 | 0 | void (*RGB24ToARGBRow)(const uint8_t* src_rgb, uint8_t* dst_argb, int width) = |
3727 | 0 | RGB24ToARGBRow_C; |
3728 | 0 | if (!src_rgb24 || !dst_argb || width <= 0 || height == 0) { |
3729 | 0 | return -1; |
3730 | 0 | } |
3731 | | // Negative height means invert the image. |
3732 | 0 | if (height < 0) { |
3733 | 0 | height = -height; |
3734 | 0 | src_rgb24 = src_rgb24 + (height - 1) * src_stride_rgb24; |
3735 | 0 | src_stride_rgb24 = -src_stride_rgb24; |
3736 | 0 | } |
3737 | | // Coalesce rows. |
3738 | 0 | if (src_stride_rgb24 == width * 3 && dst_stride_argb == width * 4) { |
3739 | 0 | width *= height; |
3740 | 0 | height = 1; |
3741 | 0 | src_stride_rgb24 = dst_stride_argb = 0; |
3742 | 0 | } |
3743 | 0 | #if defined(HAS_RGB24TOARGBROW_SSSE3) |
3744 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3745 | 0 | RGB24ToARGBRow = RGB24ToARGBRow_Any_SSSE3; |
3746 | 0 | if (IS_ALIGNED(width, 16)) { |
3747 | 0 | RGB24ToARGBRow = RGB24ToARGBRow_SSSE3; |
3748 | 0 | } |
3749 | 0 | } |
3750 | 0 | #endif |
3751 | | #if defined(HAS_RGB24TOARGBROW_NEON) |
3752 | | if (TestCpuFlag(kCpuHasNEON)) { |
3753 | | RGB24ToARGBRow = RGB24ToARGBRow_Any_NEON; |
3754 | | if (IS_ALIGNED(width, 8)) { |
3755 | | RGB24ToARGBRow = RGB24ToARGBRow_NEON; |
3756 | | } |
3757 | | } |
3758 | | #endif |
3759 | | #if defined(HAS_RGB24TOARGBROW_SVE2) |
3760 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3761 | | RGB24ToARGBRow = RGB24ToARGBRow_SVE2; |
3762 | | } |
3763 | | #endif |
3764 | | #if defined(HAS_RGB24TOARGBROW_MSA) |
3765 | | if (TestCpuFlag(kCpuHasMSA)) { |
3766 | | RGB24ToARGBRow = RGB24ToARGBRow_Any_MSA; |
3767 | | if (IS_ALIGNED(width, 16)) { |
3768 | | RGB24ToARGBRow = RGB24ToARGBRow_MSA; |
3769 | | } |
3770 | | } |
3771 | | #endif |
3772 | | #if defined(HAS_RGB24TOARGBROW_LSX) |
3773 | | if (TestCpuFlag(kCpuHasLSX)) { |
3774 | | RGB24ToARGBRow = RGB24ToARGBRow_Any_LSX; |
3775 | | if (IS_ALIGNED(width, 16)) { |
3776 | | RGB24ToARGBRow = RGB24ToARGBRow_LSX; |
3777 | | } |
3778 | | } |
3779 | | #endif |
3780 | | #if defined(HAS_RGB24TOARGBROW_LASX) |
3781 | | if (TestCpuFlag(kCpuHasLASX)) { |
3782 | | RGB24ToARGBRow = RGB24ToARGBRow_Any_LASX; |
3783 | | if (IS_ALIGNED(width, 32)) { |
3784 | | RGB24ToARGBRow = RGB24ToARGBRow_LASX; |
3785 | | } |
3786 | | } |
3787 | | #endif |
3788 | | #if defined(HAS_RGB24TOARGBROW_RVV) |
3789 | | if (TestCpuFlag(kCpuHasRVV)) { |
3790 | | RGB24ToARGBRow = RGB24ToARGBRow_RVV; |
3791 | | } |
3792 | | #endif |
3793 | |
|
3794 | 0 | for (y = 0; y < height; ++y) { |
3795 | 0 | RGB24ToARGBRow(src_rgb24, dst_argb, width); |
3796 | 0 | src_rgb24 += src_stride_rgb24; |
3797 | 0 | dst_argb += dst_stride_argb; |
3798 | 0 | } |
3799 | 0 | return 0; |
3800 | 0 | } |
3801 | | |
3802 | | // Convert RAW to ARGB. |
3803 | | LIBYUV_API |
3804 | | int RAWToARGB(const uint8_t* src_raw, |
3805 | | int src_stride_raw, |
3806 | | uint8_t* dst_argb, |
3807 | | int dst_stride_argb, |
3808 | | int width, |
3809 | 216k | int height) { |
3810 | 216k | int y; |
3811 | 216k | void (*RAWToARGBRow)(const uint8_t* src_rgb, uint8_t* dst_argb, int width) = |
3812 | 216k | RAWToARGBRow_C; |
3813 | 216k | if (!src_raw || !dst_argb || width <= 0 || height == 0) { |
3814 | 0 | return -1; |
3815 | 0 | } |
3816 | | // Negative height means invert the image. |
3817 | 216k | if (height < 0) { |
3818 | 0 | height = -height; |
3819 | 0 | src_raw = src_raw + (height - 1) * src_stride_raw; |
3820 | 0 | src_stride_raw = -src_stride_raw; |
3821 | 0 | } |
3822 | | // Coalesce rows. |
3823 | 216k | if (src_stride_raw == width * 3 && dst_stride_argb == width * 4) { |
3824 | 216k | width *= height; |
3825 | 216k | height = 1; |
3826 | 216k | src_stride_raw = dst_stride_argb = 0; |
3827 | 216k | } |
3828 | 216k | #if defined(HAS_RAWTOARGBROW_SSSE3) |
3829 | 216k | if (TestCpuFlag(kCpuHasSSSE3)) { |
3830 | 216k | RAWToARGBRow = RAWToARGBRow_Any_SSSE3; |
3831 | 216k | if (IS_ALIGNED(width, 16)) { |
3832 | 29.7k | RAWToARGBRow = RAWToARGBRow_SSSE3; |
3833 | 29.7k | } |
3834 | 216k | } |
3835 | 216k | #endif |
3836 | 216k | #if defined(HAS_RAWTOARGBROW_AVX2) |
3837 | 216k | if (TestCpuFlag(kCpuHasAVX2)) { |
3838 | 216k | RAWToARGBRow = RAWToARGBRow_Any_AVX2; |
3839 | 216k | if (IS_ALIGNED(width, 32)) { |
3840 | 7.60k | RAWToARGBRow = RAWToARGBRow_AVX2; |
3841 | 7.60k | } |
3842 | 216k | } |
3843 | 216k | #endif |
3844 | | #if defined(HAS_RAWTOARGBROW_NEON) |
3845 | | if (TestCpuFlag(kCpuHasNEON)) { |
3846 | | RAWToARGBRow = RAWToARGBRow_Any_NEON; |
3847 | | if (IS_ALIGNED(width, 8)) { |
3848 | | RAWToARGBRow = RAWToARGBRow_NEON; |
3849 | | } |
3850 | | } |
3851 | | #endif |
3852 | | #if defined(HAS_RAWTOARGBROW_SVE2) |
3853 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3854 | | RAWToARGBRow = RAWToARGBRow_SVE2; |
3855 | | } |
3856 | | #endif |
3857 | | #if defined(HAS_RAWTOARGBROW_MSA) |
3858 | | if (TestCpuFlag(kCpuHasMSA)) { |
3859 | | RAWToARGBRow = RAWToARGBRow_Any_MSA; |
3860 | | if (IS_ALIGNED(width, 16)) { |
3861 | | RAWToARGBRow = RAWToARGBRow_MSA; |
3862 | | } |
3863 | | } |
3864 | | #endif |
3865 | | #if defined(HAS_RAWTOARGBROW_LSX) |
3866 | | if (TestCpuFlag(kCpuHasLSX)) { |
3867 | | RAWToARGBRow = RAWToARGBRow_Any_LSX; |
3868 | | if (IS_ALIGNED(width, 16)) { |
3869 | | RAWToARGBRow = RAWToARGBRow_LSX; |
3870 | | } |
3871 | | } |
3872 | | #endif |
3873 | | #if defined(HAS_RAWTOARGBROW_LASX) |
3874 | | if (TestCpuFlag(kCpuHasLASX)) { |
3875 | | RAWToARGBRow = RAWToARGBRow_Any_LASX; |
3876 | | if (IS_ALIGNED(width, 32)) { |
3877 | | RAWToARGBRow = RAWToARGBRow_LASX; |
3878 | | } |
3879 | | } |
3880 | | #endif |
3881 | | #if defined(HAS_RAWTOARGBROW_RVV) |
3882 | | if (TestCpuFlag(kCpuHasRVV)) { |
3883 | | RAWToARGBRow = RAWToARGBRow_RVV; |
3884 | | } |
3885 | | #endif |
3886 | | |
3887 | 432k | for (y = 0; y < height; ++y) { |
3888 | 216k | RAWToARGBRow(src_raw, dst_argb, width); |
3889 | 216k | src_raw += src_stride_raw; |
3890 | 216k | dst_argb += dst_stride_argb; |
3891 | 216k | } |
3892 | 216k | return 0; |
3893 | 216k | } |
3894 | | |
3895 | | // Convert RAW to RGBA. |
3896 | | LIBYUV_API |
3897 | | int RAWToRGBA(const uint8_t* src_raw, |
3898 | | int src_stride_raw, |
3899 | | uint8_t* dst_rgba, |
3900 | | int dst_stride_rgba, |
3901 | | int width, |
3902 | 0 | int height) { |
3903 | 0 | int y; |
3904 | 0 | void (*RAWToRGBARow)(const uint8_t* src_rgb, uint8_t* dst_rgba, int width) = |
3905 | 0 | RAWToRGBARow_C; |
3906 | 0 | if (!src_raw || !dst_rgba || width <= 0 || height == 0) { |
3907 | 0 | return -1; |
3908 | 0 | } |
3909 | | // Negative height means invert the image. |
3910 | 0 | if (height < 0) { |
3911 | 0 | height = -height; |
3912 | 0 | src_raw = src_raw + (height - 1) * src_stride_raw; |
3913 | 0 | src_stride_raw = -src_stride_raw; |
3914 | 0 | } |
3915 | | // Coalesce rows. |
3916 | 0 | if (src_stride_raw == width * 3 && dst_stride_rgba == width * 4) { |
3917 | 0 | width *= height; |
3918 | 0 | height = 1; |
3919 | 0 | src_stride_raw = dst_stride_rgba = 0; |
3920 | 0 | } |
3921 | 0 | #if defined(HAS_RAWTORGBAROW_SSSE3) |
3922 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3923 | 0 | RAWToRGBARow = RAWToRGBARow_Any_SSSE3; |
3924 | 0 | if (IS_ALIGNED(width, 16)) { |
3925 | 0 | RAWToRGBARow = RAWToRGBARow_SSSE3; |
3926 | 0 | } |
3927 | 0 | } |
3928 | 0 | #endif |
3929 | | #if defined(HAS_RAWTORGBAROW_NEON) |
3930 | | if (TestCpuFlag(kCpuHasNEON)) { |
3931 | | RAWToRGBARow = RAWToRGBARow_Any_NEON; |
3932 | | if (IS_ALIGNED(width, 8)) { |
3933 | | RAWToRGBARow = RAWToRGBARow_NEON; |
3934 | | } |
3935 | | } |
3936 | | #endif |
3937 | | #if defined(HAS_RAWTORGBAROW_SVE2) |
3938 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3939 | | RAWToRGBARow = RAWToRGBARow_SVE2; |
3940 | | } |
3941 | | #endif |
3942 | | #if defined(HAS_RAWTORGBAROW_RVV) |
3943 | | if (TestCpuFlag(kCpuHasRVV)) { |
3944 | | RAWToRGBARow = RAWToRGBARow_RVV; |
3945 | | } |
3946 | | #endif |
3947 | |
|
3948 | 0 | for (y = 0; y < height; ++y) { |
3949 | 0 | RAWToRGBARow(src_raw, dst_rgba, width); |
3950 | 0 | src_raw += src_stride_raw; |
3951 | 0 | dst_rgba += dst_stride_rgba; |
3952 | 0 | } |
3953 | 0 | return 0; |
3954 | 0 | } |
3955 | | |
3956 | | // Convert RGB565 to ARGB. |
3957 | | LIBYUV_API |
3958 | | int RGB565ToARGB(const uint8_t* src_rgb565, |
3959 | | int src_stride_rgb565, |
3960 | | uint8_t* dst_argb, |
3961 | | int dst_stride_argb, |
3962 | | int width, |
3963 | 0 | int height) { |
3964 | 0 | int y; |
3965 | 0 | void (*RGB565ToARGBRow)(const uint8_t* src_rgb565, uint8_t* dst_argb, |
3966 | 0 | int width) = RGB565ToARGBRow_C; |
3967 | 0 | if (!src_rgb565 || !dst_argb || width <= 0 || height == 0) { |
3968 | 0 | return -1; |
3969 | 0 | } |
3970 | | // Negative height means invert the image. |
3971 | 0 | if (height < 0) { |
3972 | 0 | height = -height; |
3973 | 0 | src_rgb565 = src_rgb565 + (height - 1) * src_stride_rgb565; |
3974 | 0 | src_stride_rgb565 = -src_stride_rgb565; |
3975 | 0 | } |
3976 | | // Coalesce rows. |
3977 | 0 | if (src_stride_rgb565 == width * 2 && dst_stride_argb == width * 4) { |
3978 | 0 | width *= height; |
3979 | 0 | height = 1; |
3980 | 0 | src_stride_rgb565 = dst_stride_argb = 0; |
3981 | 0 | } |
3982 | 0 | #if defined(HAS_RGB565TOARGBROW_SSE2) |
3983 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
3984 | 0 | RGB565ToARGBRow = RGB565ToARGBRow_Any_SSE2; |
3985 | 0 | if (IS_ALIGNED(width, 8)) { |
3986 | 0 | RGB565ToARGBRow = RGB565ToARGBRow_SSE2; |
3987 | 0 | } |
3988 | 0 | } |
3989 | 0 | #endif |
3990 | | #if defined(HAS_RGB565TOARGBROW_AVX2) |
3991 | | if (TestCpuFlag(kCpuHasAVX2)) { |
3992 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_AVX2; |
3993 | | if (IS_ALIGNED(width, 16)) { |
3994 | | RGB565ToARGBRow = RGB565ToARGBRow_AVX2; |
3995 | | } |
3996 | | } |
3997 | | #endif |
3998 | | #if defined(HAS_RGB565TOARGBROW_NEON) |
3999 | | if (TestCpuFlag(kCpuHasNEON)) { |
4000 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_NEON; |
4001 | | if (IS_ALIGNED(width, 16)) { |
4002 | | RGB565ToARGBRow = RGB565ToARGBRow_NEON; |
4003 | | } |
4004 | | } |
4005 | | #endif |
4006 | | #if defined(HAS_RGB565TOARGBROW_MSA) |
4007 | | if (TestCpuFlag(kCpuHasMSA)) { |
4008 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_MSA; |
4009 | | if (IS_ALIGNED(width, 16)) { |
4010 | | RGB565ToARGBRow = RGB565ToARGBRow_MSA; |
4011 | | } |
4012 | | } |
4013 | | #endif |
4014 | | #if defined(HAS_RGB565TOARGBROW_LSX) |
4015 | | if (TestCpuFlag(kCpuHasLSX)) { |
4016 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_LSX; |
4017 | | if (IS_ALIGNED(width, 16)) { |
4018 | | RGB565ToARGBRow = RGB565ToARGBRow_LSX; |
4019 | | } |
4020 | | } |
4021 | | #endif |
4022 | | #if defined(HAS_RGB565TOARGBROW_LASX) |
4023 | | if (TestCpuFlag(kCpuHasLASX)) { |
4024 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_LASX; |
4025 | | if (IS_ALIGNED(width, 32)) { |
4026 | | RGB565ToARGBRow = RGB565ToARGBRow_LASX; |
4027 | | } |
4028 | | } |
4029 | | #endif |
4030 | |
|
4031 | 0 | for (y = 0; y < height; ++y) { |
4032 | 0 | RGB565ToARGBRow(src_rgb565, dst_argb, width); |
4033 | 0 | src_rgb565 += src_stride_rgb565; |
4034 | 0 | dst_argb += dst_stride_argb; |
4035 | 0 | } |
4036 | 0 | return 0; |
4037 | 0 | } |
4038 | | |
4039 | | // Convert ARGB1555 to ARGB. |
4040 | | LIBYUV_API |
4041 | | int ARGB1555ToARGB(const uint8_t* src_argb1555, |
4042 | | int src_stride_argb1555, |
4043 | | uint8_t* dst_argb, |
4044 | | int dst_stride_argb, |
4045 | | int width, |
4046 | 0 | int height) { |
4047 | 0 | int y; |
4048 | 0 | void (*ARGB1555ToARGBRow)(const uint8_t* src_argb1555, uint8_t* dst_argb, |
4049 | 0 | int width) = ARGB1555ToARGBRow_C; |
4050 | 0 | if (!src_argb1555 || !dst_argb || width <= 0 || height == 0) { |
4051 | 0 | return -1; |
4052 | 0 | } |
4053 | | // Negative height means invert the image. |
4054 | 0 | if (height < 0) { |
4055 | 0 | height = -height; |
4056 | 0 | src_argb1555 = src_argb1555 + (height - 1) * src_stride_argb1555; |
4057 | 0 | src_stride_argb1555 = -src_stride_argb1555; |
4058 | 0 | } |
4059 | | // Coalesce rows. |
4060 | 0 | if (src_stride_argb1555 == width * 2 && dst_stride_argb == width * 4) { |
4061 | 0 | width *= height; |
4062 | 0 | height = 1; |
4063 | 0 | src_stride_argb1555 = dst_stride_argb = 0; |
4064 | 0 | } |
4065 | 0 | #if defined(HAS_ARGB1555TOARGBROW_SSE2) |
4066 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
4067 | 0 | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_SSE2; |
4068 | 0 | if (IS_ALIGNED(width, 8)) { |
4069 | 0 | ARGB1555ToARGBRow = ARGB1555ToARGBRow_SSE2; |
4070 | 0 | } |
4071 | 0 | } |
4072 | 0 | #endif |
4073 | | #if defined(HAS_ARGB1555TOARGBROW_AVX2) |
4074 | | if (TestCpuFlag(kCpuHasAVX2)) { |
4075 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_AVX2; |
4076 | | if (IS_ALIGNED(width, 16)) { |
4077 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_AVX2; |
4078 | | } |
4079 | | } |
4080 | | #endif |
4081 | | #if defined(HAS_ARGB1555TOARGBROW_NEON) |
4082 | | if (TestCpuFlag(kCpuHasNEON)) { |
4083 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_NEON; |
4084 | | if (IS_ALIGNED(width, 16)) { |
4085 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_NEON; |
4086 | | } |
4087 | | } |
4088 | | #endif |
4089 | | #if defined(HAS_ARGB1555TOARGBROW_SVE2) |
4090 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4091 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_SVE2; |
4092 | | } |
4093 | | #endif |
4094 | | #if defined(HAS_ARGB1555TOARGBROW_MSA) |
4095 | | if (TestCpuFlag(kCpuHasMSA)) { |
4096 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_MSA; |
4097 | | if (IS_ALIGNED(width, 16)) { |
4098 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_MSA; |
4099 | | } |
4100 | | } |
4101 | | #endif |
4102 | | #if defined(HAS_ARGB1555TOARGBROW_LSX) |
4103 | | if (TestCpuFlag(kCpuHasLSX)) { |
4104 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_LSX; |
4105 | | if (IS_ALIGNED(width, 16)) { |
4106 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_LSX; |
4107 | | } |
4108 | | } |
4109 | | #endif |
4110 | | #if defined(HAS_ARGB1555TOARGBROW_LASX) |
4111 | | if (TestCpuFlag(kCpuHasLASX)) { |
4112 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_LASX; |
4113 | | if (IS_ALIGNED(width, 32)) { |
4114 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_LASX; |
4115 | | } |
4116 | | } |
4117 | | #endif |
4118 | |
|
4119 | 0 | for (y = 0; y < height; ++y) { |
4120 | 0 | ARGB1555ToARGBRow(src_argb1555, dst_argb, width); |
4121 | 0 | src_argb1555 += src_stride_argb1555; |
4122 | 0 | dst_argb += dst_stride_argb; |
4123 | 0 | } |
4124 | 0 | return 0; |
4125 | 0 | } |
4126 | | |
4127 | | // Convert ARGB4444 to ARGB. |
4128 | | LIBYUV_API |
4129 | | int ARGB4444ToARGB(const uint8_t* src_argb4444, |
4130 | | int src_stride_argb4444, |
4131 | | uint8_t* dst_argb, |
4132 | | int dst_stride_argb, |
4133 | | int width, |
4134 | 0 | int height) { |
4135 | 0 | int y; |
4136 | 0 | void (*ARGB4444ToARGBRow)(const uint8_t* src_argb4444, uint8_t* dst_argb, |
4137 | 0 | int width) = ARGB4444ToARGBRow_C; |
4138 | 0 | if (!src_argb4444 || !dst_argb || width <= 0 || height == 0) { |
4139 | 0 | return -1; |
4140 | 0 | } |
4141 | | // Negative height means invert the image. |
4142 | 0 | if (height < 0) { |
4143 | 0 | height = -height; |
4144 | 0 | src_argb4444 = src_argb4444 + (height - 1) * src_stride_argb4444; |
4145 | 0 | src_stride_argb4444 = -src_stride_argb4444; |
4146 | 0 | } |
4147 | | // Coalesce rows. |
4148 | 0 | if (src_stride_argb4444 == width * 2 && dst_stride_argb == width * 4) { |
4149 | 0 | width *= height; |
4150 | 0 | height = 1; |
4151 | 0 | src_stride_argb4444 = dst_stride_argb = 0; |
4152 | 0 | } |
4153 | 0 | #if defined(HAS_ARGB4444TOARGBROW_SSE2) |
4154 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
4155 | 0 | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_SSE2; |
4156 | 0 | if (IS_ALIGNED(width, 8)) { |
4157 | 0 | ARGB4444ToARGBRow = ARGB4444ToARGBRow_SSE2; |
4158 | 0 | } |
4159 | 0 | } |
4160 | 0 | #endif |
4161 | | #if defined(HAS_ARGB4444TOARGBROW_AVX2) |
4162 | | if (TestCpuFlag(kCpuHasAVX2)) { |
4163 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_AVX2; |
4164 | | if (IS_ALIGNED(width, 16)) { |
4165 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_AVX2; |
4166 | | } |
4167 | | } |
4168 | | #endif |
4169 | | #if defined(HAS_ARGB4444TOARGBROW_NEON) |
4170 | | if (TestCpuFlag(kCpuHasNEON)) { |
4171 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_NEON; |
4172 | | if (IS_ALIGNED(width, 8)) { |
4173 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_NEON; |
4174 | | } |
4175 | | } |
4176 | | #endif |
4177 | | #if defined(HAS_ARGB4444TOARGBROW_MSA) |
4178 | | if (TestCpuFlag(kCpuHasMSA)) { |
4179 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_MSA; |
4180 | | if (IS_ALIGNED(width, 16)) { |
4181 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_MSA; |
4182 | | } |
4183 | | } |
4184 | | #endif |
4185 | | #if defined(HAS_ARGB4444TOARGBROW_LSX) |
4186 | | if (TestCpuFlag(kCpuHasLSX)) { |
4187 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_LSX; |
4188 | | if (IS_ALIGNED(width, 16)) { |
4189 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_LSX; |
4190 | | } |
4191 | | } |
4192 | | #endif |
4193 | | #if defined(HAS_ARGB4444TOARGBROW_LASX) |
4194 | | if (TestCpuFlag(kCpuHasLASX)) { |
4195 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_LASX; |
4196 | | if (IS_ALIGNED(width, 32)) { |
4197 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_LASX; |
4198 | | } |
4199 | | } |
4200 | | #endif |
4201 | |
|
4202 | 0 | for (y = 0; y < height; ++y) { |
4203 | 0 | ARGB4444ToARGBRow(src_argb4444, dst_argb, width); |
4204 | 0 | src_argb4444 += src_stride_argb4444; |
4205 | 0 | dst_argb += dst_stride_argb; |
4206 | 0 | } |
4207 | 0 | return 0; |
4208 | 0 | } |
4209 | | |
4210 | | // Convert AR30 to ARGB. |
4211 | | LIBYUV_API |
4212 | | int AR30ToARGB(const uint8_t* src_ar30, |
4213 | | int src_stride_ar30, |
4214 | | uint8_t* dst_argb, |
4215 | | int dst_stride_argb, |
4216 | | int width, |
4217 | 0 | int height) { |
4218 | 0 | int y; |
4219 | 0 | if (!src_ar30 || !dst_argb || width <= 0 || height == 0) { |
4220 | 0 | return -1; |
4221 | 0 | } |
4222 | | // Negative height means invert the image. |
4223 | 0 | if (height < 0) { |
4224 | 0 | height = -height; |
4225 | 0 | src_ar30 = src_ar30 + (height - 1) * src_stride_ar30; |
4226 | 0 | src_stride_ar30 = -src_stride_ar30; |
4227 | 0 | } |
4228 | | // Coalesce rows. |
4229 | 0 | if (src_stride_ar30 == width * 4 && dst_stride_argb == width * 4) { |
4230 | 0 | width *= height; |
4231 | 0 | height = 1; |
4232 | 0 | src_stride_ar30 = dst_stride_argb = 0; |
4233 | 0 | } |
4234 | 0 | for (y = 0; y < height; ++y) { |
4235 | 0 | AR30ToARGBRow_C(src_ar30, dst_argb, width); |
4236 | 0 | src_ar30 += src_stride_ar30; |
4237 | 0 | dst_argb += dst_stride_argb; |
4238 | 0 | } |
4239 | 0 | return 0; |
4240 | 0 | } |
4241 | | |
4242 | | // Convert AR30 to ABGR. |
4243 | | LIBYUV_API |
4244 | | int AR30ToABGR(const uint8_t* src_ar30, |
4245 | | int src_stride_ar30, |
4246 | | uint8_t* dst_abgr, |
4247 | | int dst_stride_abgr, |
4248 | | int width, |
4249 | 0 | int height) { |
4250 | 0 | int y; |
4251 | 0 | if (!src_ar30 || !dst_abgr || width <= 0 || height == 0) { |
4252 | 0 | return -1; |
4253 | 0 | } |
4254 | | // Negative height means invert the image. |
4255 | 0 | if (height < 0) { |
4256 | 0 | height = -height; |
4257 | 0 | src_ar30 = src_ar30 + (height - 1) * src_stride_ar30; |
4258 | 0 | src_stride_ar30 = -src_stride_ar30; |
4259 | 0 | } |
4260 | | // Coalesce rows. |
4261 | 0 | if (src_stride_ar30 == width * 4 && dst_stride_abgr == width * 4) { |
4262 | 0 | width *= height; |
4263 | 0 | height = 1; |
4264 | 0 | src_stride_ar30 = dst_stride_abgr = 0; |
4265 | 0 | } |
4266 | 0 | for (y = 0; y < height; ++y) { |
4267 | 0 | AR30ToABGRRow_C(src_ar30, dst_abgr, width); |
4268 | 0 | src_ar30 += src_stride_ar30; |
4269 | 0 | dst_abgr += dst_stride_abgr; |
4270 | 0 | } |
4271 | 0 | return 0; |
4272 | 0 | } |
4273 | | |
4274 | | // Convert AR30 to AB30. |
4275 | | LIBYUV_API |
4276 | | int AR30ToAB30(const uint8_t* src_ar30, |
4277 | | int src_stride_ar30, |
4278 | | uint8_t* dst_ab30, |
4279 | | int dst_stride_ab30, |
4280 | | int width, |
4281 | 0 | int height) { |
4282 | 0 | int y; |
4283 | 0 | if (!src_ar30 || !dst_ab30 || width <= 0 || height == 0) { |
4284 | 0 | return -1; |
4285 | 0 | } |
4286 | | // Negative height means invert the image. |
4287 | 0 | if (height < 0) { |
4288 | 0 | height = -height; |
4289 | 0 | src_ar30 = src_ar30 + (height - 1) * src_stride_ar30; |
4290 | 0 | src_stride_ar30 = -src_stride_ar30; |
4291 | 0 | } |
4292 | | // Coalesce rows. |
4293 | 0 | if (src_stride_ar30 == width * 4 && dst_stride_ab30 == width * 4) { |
4294 | 0 | width *= height; |
4295 | 0 | height = 1; |
4296 | 0 | src_stride_ar30 = dst_stride_ab30 = 0; |
4297 | 0 | } |
4298 | 0 | for (y = 0; y < height; ++y) { |
4299 | 0 | AR30ToAB30Row_C(src_ar30, dst_ab30, width); |
4300 | 0 | src_ar30 += src_stride_ar30; |
4301 | 0 | dst_ab30 += dst_stride_ab30; |
4302 | 0 | } |
4303 | 0 | return 0; |
4304 | 0 | } |
4305 | | |
4306 | | // Convert AR64 to ARGB. |
4307 | | LIBYUV_API |
4308 | | int AR64ToARGB(const uint16_t* src_ar64, |
4309 | | int src_stride_ar64, |
4310 | | uint8_t* dst_argb, |
4311 | | int dst_stride_argb, |
4312 | | int width, |
4313 | 0 | int height) { |
4314 | 0 | int y; |
4315 | 0 | void (*AR64ToARGBRow)(const uint16_t* src_ar64, uint8_t* dst_argb, |
4316 | 0 | int width) = AR64ToARGBRow_C; |
4317 | 0 | if (!src_ar64 || !dst_argb || width <= 0 || height == 0) { |
4318 | 0 | return -1; |
4319 | 0 | } |
4320 | | // Negative height means invert the image. |
4321 | 0 | if (height < 0) { |
4322 | 0 | height = -height; |
4323 | 0 | src_ar64 = src_ar64 + (height - 1) * src_stride_ar64; |
4324 | 0 | src_stride_ar64 = -src_stride_ar64; |
4325 | 0 | } |
4326 | | // Coalesce rows. |
4327 | 0 | if (src_stride_ar64 == width * 4 && dst_stride_argb == width * 4) { |
4328 | 0 | width *= height; |
4329 | 0 | height = 1; |
4330 | 0 | src_stride_ar64 = dst_stride_argb = 0; |
4331 | 0 | } |
4332 | 0 | #if defined(HAS_AR64TOARGBROW_SSSE3) |
4333 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4334 | 0 | AR64ToARGBRow = AR64ToARGBRow_Any_SSSE3; |
4335 | 0 | if (IS_ALIGNED(width, 4)) { |
4336 | 0 | AR64ToARGBRow = AR64ToARGBRow_SSSE3; |
4337 | 0 | } |
4338 | 0 | } |
4339 | 0 | #endif |
4340 | 0 | #if defined(HAS_AR64TOARGBROW_AVX2) |
4341 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4342 | 0 | AR64ToARGBRow = AR64ToARGBRow_Any_AVX2; |
4343 | 0 | if (IS_ALIGNED(width, 8)) { |
4344 | 0 | AR64ToARGBRow = AR64ToARGBRow_AVX2; |
4345 | 0 | } |
4346 | 0 | } |
4347 | 0 | #endif |
4348 | | #if defined(HAS_AR64TOARGBROW_NEON) |
4349 | | if (TestCpuFlag(kCpuHasNEON)) { |
4350 | | AR64ToARGBRow = AR64ToARGBRow_Any_NEON; |
4351 | | if (IS_ALIGNED(width, 8)) { |
4352 | | AR64ToARGBRow = AR64ToARGBRow_NEON; |
4353 | | } |
4354 | | } |
4355 | | #endif |
4356 | | #if defined(HAS_AR64TOARGBROW_RVV) |
4357 | | if (TestCpuFlag(kCpuHasRVV)) { |
4358 | | AR64ToARGBRow = AR64ToARGBRow_RVV; |
4359 | | } |
4360 | | #endif |
4361 | |
|
4362 | 0 | for (y = 0; y < height; ++y) { |
4363 | 0 | AR64ToARGBRow(src_ar64, dst_argb, width); |
4364 | 0 | src_ar64 += src_stride_ar64; |
4365 | 0 | dst_argb += dst_stride_argb; |
4366 | 0 | } |
4367 | 0 | return 0; |
4368 | 0 | } |
4369 | | |
4370 | | // Convert AB64 to ARGB. |
4371 | | LIBYUV_API |
4372 | | int AB64ToARGB(const uint16_t* src_ab64, |
4373 | | int src_stride_ab64, |
4374 | | uint8_t* dst_argb, |
4375 | | int dst_stride_argb, |
4376 | | int width, |
4377 | 0 | int height) { |
4378 | 0 | int y; |
4379 | 0 | void (*AB64ToARGBRow)(const uint16_t* src_ar64, uint8_t* dst_argb, |
4380 | 0 | int width) = AB64ToARGBRow_C; |
4381 | 0 | if (!src_ab64 || !dst_argb || width <= 0 || height == 0) { |
4382 | 0 | return -1; |
4383 | 0 | } |
4384 | | // Negative height means invert the image. |
4385 | 0 | if (height < 0) { |
4386 | 0 | height = -height; |
4387 | 0 | src_ab64 = src_ab64 + (height - 1) * src_stride_ab64; |
4388 | 0 | src_stride_ab64 = -src_stride_ab64; |
4389 | 0 | } |
4390 | | // Coalesce rows. |
4391 | 0 | if (src_stride_ab64 == width * 4 && dst_stride_argb == width * 4) { |
4392 | 0 | width *= height; |
4393 | 0 | height = 1; |
4394 | 0 | src_stride_ab64 = dst_stride_argb = 0; |
4395 | 0 | } |
4396 | 0 | #if defined(HAS_AB64TOARGBROW_SSSE3) |
4397 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4398 | 0 | AB64ToARGBRow = AB64ToARGBRow_Any_SSSE3; |
4399 | 0 | if (IS_ALIGNED(width, 4)) { |
4400 | 0 | AB64ToARGBRow = AB64ToARGBRow_SSSE3; |
4401 | 0 | } |
4402 | 0 | } |
4403 | 0 | #endif |
4404 | 0 | #if defined(HAS_AB64TOARGBROW_AVX2) |
4405 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4406 | 0 | AB64ToARGBRow = AB64ToARGBRow_Any_AVX2; |
4407 | 0 | if (IS_ALIGNED(width, 8)) { |
4408 | 0 | AB64ToARGBRow = AB64ToARGBRow_AVX2; |
4409 | 0 | } |
4410 | 0 | } |
4411 | 0 | #endif |
4412 | | #if defined(HAS_AB64TOARGBROW_NEON) |
4413 | | if (TestCpuFlag(kCpuHasNEON)) { |
4414 | | AB64ToARGBRow = AB64ToARGBRow_Any_NEON; |
4415 | | if (IS_ALIGNED(width, 8)) { |
4416 | | AB64ToARGBRow = AB64ToARGBRow_NEON; |
4417 | | } |
4418 | | } |
4419 | | #endif |
4420 | | #if defined(HAS_AB64TOARGBROW_RVV) |
4421 | | if (TestCpuFlag(kCpuHasRVV)) { |
4422 | | AB64ToARGBRow = AB64ToARGBRow_RVV; |
4423 | | } |
4424 | | #endif |
4425 | |
|
4426 | 0 | for (y = 0; y < height; ++y) { |
4427 | 0 | AB64ToARGBRow(src_ab64, dst_argb, width); |
4428 | 0 | src_ab64 += src_stride_ab64; |
4429 | 0 | dst_argb += dst_stride_argb; |
4430 | 0 | } |
4431 | 0 | return 0; |
4432 | 0 | } |
4433 | | |
4434 | | // Convert NV12 to ARGB with matrix. |
4435 | | LIBYUV_API |
4436 | | int NV12ToARGBMatrix(const uint8_t* src_y, |
4437 | | int src_stride_y, |
4438 | | const uint8_t* src_uv, |
4439 | | int src_stride_uv, |
4440 | | uint8_t* dst_argb, |
4441 | | int dst_stride_argb, |
4442 | | const struct YuvConstants* yuvconstants, |
4443 | | int width, |
4444 | 0 | int height) { |
4445 | 0 | int y; |
4446 | 0 | void (*NV12ToARGBRow)( |
4447 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4448 | 0 | const struct YuvConstants* yuvconstants, int width) = NV12ToARGBRow_C; |
4449 | 0 | assert(yuvconstants); |
4450 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
4451 | 0 | return -1; |
4452 | 0 | } |
4453 | | // Negative height means invert the image. |
4454 | 0 | if (height < 0) { |
4455 | 0 | height = -height; |
4456 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
4457 | 0 | dst_stride_argb = -dst_stride_argb; |
4458 | 0 | } |
4459 | 0 | #if defined(HAS_NV12TOARGBROW_SSSE3) |
4460 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4461 | 0 | NV12ToARGBRow = NV12ToARGBRow_Any_SSSE3; |
4462 | 0 | if (IS_ALIGNED(width, 8)) { |
4463 | 0 | NV12ToARGBRow = NV12ToARGBRow_SSSE3; |
4464 | 0 | } |
4465 | 0 | } |
4466 | 0 | #endif |
4467 | 0 | #if defined(HAS_NV12TOARGBROW_AVX2) |
4468 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4469 | 0 | NV12ToARGBRow = NV12ToARGBRow_Any_AVX2; |
4470 | 0 | if (IS_ALIGNED(width, 16)) { |
4471 | 0 | NV12ToARGBRow = NV12ToARGBRow_AVX2; |
4472 | 0 | } |
4473 | 0 | } |
4474 | 0 | #endif |
4475 | | #if defined(HAS_NV12TOARGBROW_NEON) |
4476 | | if (TestCpuFlag(kCpuHasNEON)) { |
4477 | | NV12ToARGBRow = NV12ToARGBRow_Any_NEON; |
4478 | | if (IS_ALIGNED(width, 8)) { |
4479 | | NV12ToARGBRow = NV12ToARGBRow_NEON; |
4480 | | } |
4481 | | } |
4482 | | #endif |
4483 | | #if defined(HAS_NV12TOARGBROW_SVE2) |
4484 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4485 | | NV12ToARGBRow = NV12ToARGBRow_SVE2; |
4486 | | } |
4487 | | #endif |
4488 | | #if defined(HAS_NV12TOARGBROW_SME) |
4489 | | if (TestCpuFlag(kCpuHasSME)) { |
4490 | | NV12ToARGBRow = NV12ToARGBRow_SME; |
4491 | | } |
4492 | | #endif |
4493 | | #if defined(HAS_NV12TOARGBROW_MSA) |
4494 | | if (TestCpuFlag(kCpuHasMSA)) { |
4495 | | NV12ToARGBRow = NV12ToARGBRow_Any_MSA; |
4496 | | if (IS_ALIGNED(width, 8)) { |
4497 | | NV12ToARGBRow = NV12ToARGBRow_MSA; |
4498 | | } |
4499 | | } |
4500 | | #endif |
4501 | | #if defined(HAS_NV12TOARGBROW_LSX) |
4502 | | if (TestCpuFlag(kCpuHasLSX)) { |
4503 | | NV12ToARGBRow = NV12ToARGBRow_Any_LSX; |
4504 | | if (IS_ALIGNED(width, 8)) { |
4505 | | NV12ToARGBRow = NV12ToARGBRow_LSX; |
4506 | | } |
4507 | | } |
4508 | | #endif |
4509 | | #if defined(HAS_NV12TOARGBROW_LASX) |
4510 | | if (TestCpuFlag(kCpuHasLASX)) { |
4511 | | NV12ToARGBRow = NV12ToARGBRow_Any_LASX; |
4512 | | if (IS_ALIGNED(width, 16)) { |
4513 | | NV12ToARGBRow = NV12ToARGBRow_LASX; |
4514 | | } |
4515 | | } |
4516 | | #endif |
4517 | | #if defined(HAS_NV12TOARGBROW_RVV) |
4518 | | if (TestCpuFlag(kCpuHasRVV)) { |
4519 | | NV12ToARGBRow = NV12ToARGBRow_RVV; |
4520 | | } |
4521 | | #endif |
4522 | |
|
4523 | 0 | for (y = 0; y < height; ++y) { |
4524 | 0 | NV12ToARGBRow(src_y, src_uv, dst_argb, yuvconstants, width); |
4525 | 0 | dst_argb += dst_stride_argb; |
4526 | 0 | src_y += src_stride_y; |
4527 | 0 | if (y & 1) { |
4528 | 0 | src_uv += src_stride_uv; |
4529 | 0 | } |
4530 | 0 | } |
4531 | 0 | return 0; |
4532 | 0 | } |
4533 | | |
4534 | | // Convert NV21 to ARGB with matrix. |
4535 | | LIBYUV_API |
4536 | | int NV21ToARGBMatrix(const uint8_t* src_y, |
4537 | | int src_stride_y, |
4538 | | const uint8_t* src_vu, |
4539 | | int src_stride_vu, |
4540 | | uint8_t* dst_argb, |
4541 | | int dst_stride_argb, |
4542 | | const struct YuvConstants* yuvconstants, |
4543 | | int width, |
4544 | 0 | int height) { |
4545 | 0 | int y; |
4546 | 0 | void (*NV21ToARGBRow)( |
4547 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4548 | 0 | const struct YuvConstants* yuvconstants, int width) = NV21ToARGBRow_C; |
4549 | 0 | assert(yuvconstants); |
4550 | 0 | if (!src_y || !src_vu || !dst_argb || width <= 0 || height == 0) { |
4551 | 0 | return -1; |
4552 | 0 | } |
4553 | | // Negative height means invert the image. |
4554 | 0 | if (height < 0) { |
4555 | 0 | height = -height; |
4556 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
4557 | 0 | dst_stride_argb = -dst_stride_argb; |
4558 | 0 | } |
4559 | 0 | #if defined(HAS_NV21TOARGBROW_SSSE3) |
4560 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4561 | 0 | NV21ToARGBRow = NV21ToARGBRow_Any_SSSE3; |
4562 | 0 | if (IS_ALIGNED(width, 8)) { |
4563 | 0 | NV21ToARGBRow = NV21ToARGBRow_SSSE3; |
4564 | 0 | } |
4565 | 0 | } |
4566 | 0 | #endif |
4567 | 0 | #if defined(HAS_NV21TOARGBROW_AVX2) |
4568 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4569 | 0 | NV21ToARGBRow = NV21ToARGBRow_Any_AVX2; |
4570 | 0 | if (IS_ALIGNED(width, 16)) { |
4571 | 0 | NV21ToARGBRow = NV21ToARGBRow_AVX2; |
4572 | 0 | } |
4573 | 0 | } |
4574 | 0 | #endif |
4575 | | #if defined(HAS_NV21TOARGBROW_NEON) |
4576 | | if (TestCpuFlag(kCpuHasNEON)) { |
4577 | | NV21ToARGBRow = NV21ToARGBRow_Any_NEON; |
4578 | | if (IS_ALIGNED(width, 8)) { |
4579 | | NV21ToARGBRow = NV21ToARGBRow_NEON; |
4580 | | } |
4581 | | } |
4582 | | #endif |
4583 | | #if defined(HAS_NV21TOARGBROW_SVE2) |
4584 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4585 | | NV21ToARGBRow = NV21ToARGBRow_SVE2; |
4586 | | } |
4587 | | #endif |
4588 | | #if defined(HAS_NV21TOARGBROW_SME) |
4589 | | if (TestCpuFlag(kCpuHasSME)) { |
4590 | | NV21ToARGBRow = NV21ToARGBRow_SME; |
4591 | | } |
4592 | | #endif |
4593 | | #if defined(HAS_NV21TOARGBROW_MSA) |
4594 | | if (TestCpuFlag(kCpuHasMSA)) { |
4595 | | NV21ToARGBRow = NV21ToARGBRow_Any_MSA; |
4596 | | if (IS_ALIGNED(width, 8)) { |
4597 | | NV21ToARGBRow = NV21ToARGBRow_MSA; |
4598 | | } |
4599 | | } |
4600 | | #endif |
4601 | | #if defined(HAS_NV21TOARGBROW_LSX) |
4602 | | if (TestCpuFlag(kCpuHasLSX)) { |
4603 | | NV21ToARGBRow = NV21ToARGBRow_Any_LSX; |
4604 | | if (IS_ALIGNED(width, 8)) { |
4605 | | NV21ToARGBRow = NV21ToARGBRow_LSX; |
4606 | | } |
4607 | | } |
4608 | | #endif |
4609 | | #if defined(HAS_NV21TOARGBROW_LASX) |
4610 | | if (TestCpuFlag(kCpuHasLASX)) { |
4611 | | NV21ToARGBRow = NV21ToARGBRow_Any_LASX; |
4612 | | if (IS_ALIGNED(width, 16)) { |
4613 | | NV21ToARGBRow = NV21ToARGBRow_LASX; |
4614 | | } |
4615 | | } |
4616 | | #endif |
4617 | | #if defined(HAS_NV21TOARGBROW_RVV) |
4618 | | if (TestCpuFlag(kCpuHasRVV)) { |
4619 | | NV21ToARGBRow = NV21ToARGBRow_RVV; |
4620 | | } |
4621 | | #endif |
4622 | |
|
4623 | 0 | for (y = 0; y < height; ++y) { |
4624 | 0 | NV21ToARGBRow(src_y, src_vu, dst_argb, yuvconstants, width); |
4625 | 0 | dst_argb += dst_stride_argb; |
4626 | 0 | src_y += src_stride_y; |
4627 | 0 | if (y & 1) { |
4628 | 0 | src_vu += src_stride_vu; |
4629 | 0 | } |
4630 | 0 | } |
4631 | 0 | return 0; |
4632 | 0 | } |
4633 | | |
4634 | | // Convert NV12 to ARGB. |
4635 | | LIBYUV_API |
4636 | | int NV12ToARGB(const uint8_t* src_y, |
4637 | | int src_stride_y, |
4638 | | const uint8_t* src_uv, |
4639 | | int src_stride_uv, |
4640 | | uint8_t* dst_argb, |
4641 | | int dst_stride_argb, |
4642 | | int width, |
4643 | 0 | int height) { |
4644 | 0 | return NV12ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, dst_argb, |
4645 | 0 | dst_stride_argb, &kYuvI601Constants, width, height); |
4646 | 0 | } |
4647 | | |
4648 | | // Convert NV21 to ARGB. |
4649 | | LIBYUV_API |
4650 | | int NV21ToARGB(const uint8_t* src_y, |
4651 | | int src_stride_y, |
4652 | | const uint8_t* src_vu, |
4653 | | int src_stride_vu, |
4654 | | uint8_t* dst_argb, |
4655 | | int dst_stride_argb, |
4656 | | int width, |
4657 | 0 | int height) { |
4658 | 0 | return NV21ToARGBMatrix(src_y, src_stride_y, src_vu, src_stride_vu, dst_argb, |
4659 | 0 | dst_stride_argb, &kYuvI601Constants, width, height); |
4660 | 0 | } |
4661 | | |
4662 | | // Convert NV12 to ABGR. |
4663 | | // To output ABGR instead of ARGB swap the UV and use a mirrored yuv matrix. |
4664 | | // To swap the UV use NV12 instead of NV21.LIBYUV_API |
4665 | | LIBYUV_API |
4666 | | int NV12ToABGR(const uint8_t* src_y, |
4667 | | int src_stride_y, |
4668 | | const uint8_t* src_uv, |
4669 | | int src_stride_uv, |
4670 | | uint8_t* dst_abgr, |
4671 | | int dst_stride_abgr, |
4672 | | int width, |
4673 | 0 | int height) { |
4674 | 0 | return NV21ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, dst_abgr, |
4675 | 0 | dst_stride_abgr, &kYvuI601Constants, width, height); |
4676 | 0 | } |
4677 | | |
4678 | | // Convert NV21 to ABGR. |
4679 | | LIBYUV_API |
4680 | | int NV21ToABGR(const uint8_t* src_y, |
4681 | | int src_stride_y, |
4682 | | const uint8_t* src_vu, |
4683 | | int src_stride_vu, |
4684 | | uint8_t* dst_abgr, |
4685 | | int dst_stride_abgr, |
4686 | | int width, |
4687 | 0 | int height) { |
4688 | 0 | return NV12ToARGBMatrix(src_y, src_stride_y, src_vu, src_stride_vu, dst_abgr, |
4689 | 0 | dst_stride_abgr, &kYvuI601Constants, width, height); |
4690 | 0 | } |
4691 | | |
4692 | | // TODO(fbarchard): Consider SSSE3 2 step conversion. |
4693 | | // Convert NV12 to RGB24 with matrix. |
4694 | | LIBYUV_API |
4695 | | int NV12ToRGB24Matrix(const uint8_t* src_y, |
4696 | | int src_stride_y, |
4697 | | const uint8_t* src_uv, |
4698 | | int src_stride_uv, |
4699 | | uint8_t* dst_rgb24, |
4700 | | int dst_stride_rgb24, |
4701 | | const struct YuvConstants* yuvconstants, |
4702 | | int width, |
4703 | 0 | int height) { |
4704 | 0 | int y; |
4705 | 0 | void (*NV12ToRGB24Row)( |
4706 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4707 | 0 | const struct YuvConstants* yuvconstants, int width) = NV12ToRGB24Row_C; |
4708 | 0 | assert(yuvconstants); |
4709 | 0 | if (!src_y || !src_uv || !dst_rgb24 || width <= 0 || height == 0) { |
4710 | 0 | return -1; |
4711 | 0 | } |
4712 | | // Negative height means invert the image. |
4713 | 0 | if (height < 0) { |
4714 | 0 | height = -height; |
4715 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
4716 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
4717 | 0 | } |
4718 | | #if defined(HAS_NV12TORGB24ROW_NEON) |
4719 | | if (TestCpuFlag(kCpuHasNEON)) { |
4720 | | NV12ToRGB24Row = NV12ToRGB24Row_Any_NEON; |
4721 | | if (IS_ALIGNED(width, 8)) { |
4722 | | NV12ToRGB24Row = NV12ToRGB24Row_NEON; |
4723 | | } |
4724 | | } |
4725 | | #endif |
4726 | | #if defined(HAS_NV12TORGB24ROW_SVE2) |
4727 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4728 | | NV12ToRGB24Row = NV12ToRGB24Row_SVE2; |
4729 | | } |
4730 | | #endif |
4731 | | #if defined(HAS_NV12TORGB24ROW_SME) |
4732 | | if (TestCpuFlag(kCpuHasSME)) { |
4733 | | NV12ToRGB24Row = NV12ToRGB24Row_SME; |
4734 | | } |
4735 | | #endif |
4736 | 0 | #if defined(HAS_NV12TORGB24ROW_SSSE3) |
4737 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4738 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_Any_SSSE3; |
4739 | 0 | if (IS_ALIGNED(width, 16)) { |
4740 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_SSSE3; |
4741 | 0 | } |
4742 | 0 | } |
4743 | 0 | #endif |
4744 | 0 | #if defined(HAS_NV12TORGB24ROW_AVX2) |
4745 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4746 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_Any_AVX2; |
4747 | 0 | if (IS_ALIGNED(width, 32)) { |
4748 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_AVX2; |
4749 | 0 | } |
4750 | 0 | } |
4751 | 0 | #endif |
4752 | | #if defined(HAS_NV12TORGB24ROW_RVV) |
4753 | | if (TestCpuFlag(kCpuHasRVV)) { |
4754 | | NV12ToRGB24Row = NV12ToRGB24Row_RVV; |
4755 | | } |
4756 | | #endif |
4757 | |
|
4758 | 0 | for (y = 0; y < height; ++y) { |
4759 | 0 | NV12ToRGB24Row(src_y, src_uv, dst_rgb24, yuvconstants, width); |
4760 | 0 | dst_rgb24 += dst_stride_rgb24; |
4761 | 0 | src_y += src_stride_y; |
4762 | 0 | if (y & 1) { |
4763 | 0 | src_uv += src_stride_uv; |
4764 | 0 | } |
4765 | 0 | } |
4766 | 0 | return 0; |
4767 | 0 | } |
4768 | | |
4769 | | // Convert NV21 to RGB24 with matrix. |
4770 | | LIBYUV_API |
4771 | | int NV21ToRGB24Matrix(const uint8_t* src_y, |
4772 | | int src_stride_y, |
4773 | | const uint8_t* src_vu, |
4774 | | int src_stride_vu, |
4775 | | uint8_t* dst_rgb24, |
4776 | | int dst_stride_rgb24, |
4777 | | const struct YuvConstants* yuvconstants, |
4778 | | int width, |
4779 | 0 | int height) { |
4780 | 0 | int y; |
4781 | 0 | void (*NV21ToRGB24Row)( |
4782 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4783 | 0 | const struct YuvConstants* yuvconstants, int width) = NV21ToRGB24Row_C; |
4784 | 0 | assert(yuvconstants); |
4785 | 0 | if (!src_y || !src_vu || !dst_rgb24 || width <= 0 || height == 0) { |
4786 | 0 | return -1; |
4787 | 0 | } |
4788 | | // Negative height means invert the image. |
4789 | 0 | if (height < 0) { |
4790 | 0 | height = -height; |
4791 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
4792 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
4793 | 0 | } |
4794 | | #if defined(HAS_NV21TORGB24ROW_NEON) |
4795 | | if (TestCpuFlag(kCpuHasNEON)) { |
4796 | | NV21ToRGB24Row = NV21ToRGB24Row_Any_NEON; |
4797 | | if (IS_ALIGNED(width, 8)) { |
4798 | | NV21ToRGB24Row = NV21ToRGB24Row_NEON; |
4799 | | } |
4800 | | } |
4801 | | #endif |
4802 | | #if defined(HAS_NV21TORGB24ROW_SVE2) |
4803 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4804 | | NV21ToRGB24Row = NV21ToRGB24Row_SVE2; |
4805 | | } |
4806 | | #endif |
4807 | | #if defined(HAS_NV21TORGB24ROW_SME) |
4808 | | if (TestCpuFlag(kCpuHasSME)) { |
4809 | | NV21ToRGB24Row = NV21ToRGB24Row_SME; |
4810 | | } |
4811 | | #endif |
4812 | 0 | #if defined(HAS_NV21TORGB24ROW_SSSE3) |
4813 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4814 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_Any_SSSE3; |
4815 | 0 | if (IS_ALIGNED(width, 16)) { |
4816 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_SSSE3; |
4817 | 0 | } |
4818 | 0 | } |
4819 | 0 | #endif |
4820 | 0 | #if defined(HAS_NV21TORGB24ROW_AVX2) |
4821 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4822 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_Any_AVX2; |
4823 | 0 | if (IS_ALIGNED(width, 32)) { |
4824 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_AVX2; |
4825 | 0 | } |
4826 | 0 | } |
4827 | 0 | #endif |
4828 | | #if defined(HAS_NV21TORGB24ROW_RVV) |
4829 | | if (TestCpuFlag(kCpuHasRVV)) { |
4830 | | NV21ToRGB24Row = NV21ToRGB24Row_RVV; |
4831 | | } |
4832 | | #endif |
4833 | |
|
4834 | 0 | for (y = 0; y < height; ++y) { |
4835 | 0 | NV21ToRGB24Row(src_y, src_vu, dst_rgb24, yuvconstants, width); |
4836 | 0 | dst_rgb24 += dst_stride_rgb24; |
4837 | 0 | src_y += src_stride_y; |
4838 | 0 | if (y & 1) { |
4839 | 0 | src_vu += src_stride_vu; |
4840 | 0 | } |
4841 | 0 | } |
4842 | 0 | return 0; |
4843 | 0 | } |
4844 | | |
4845 | | // Convert NV12 to RGB24. |
4846 | | LIBYUV_API |
4847 | | int NV12ToRGB24(const uint8_t* src_y, |
4848 | | int src_stride_y, |
4849 | | const uint8_t* src_uv, |
4850 | | int src_stride_uv, |
4851 | | uint8_t* dst_rgb24, |
4852 | | int dst_stride_rgb24, |
4853 | | int width, |
4854 | 0 | int height) { |
4855 | 0 | return NV12ToRGB24Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
4856 | 0 | dst_rgb24, dst_stride_rgb24, &kYuvI601Constants, |
4857 | 0 | width, height); |
4858 | 0 | } |
4859 | | |
4860 | | // Convert NV21 to RGB24. |
4861 | | LIBYUV_API |
4862 | | int NV21ToRGB24(const uint8_t* src_y, |
4863 | | int src_stride_y, |
4864 | | const uint8_t* src_vu, |
4865 | | int src_stride_vu, |
4866 | | uint8_t* dst_rgb24, |
4867 | | int dst_stride_rgb24, |
4868 | | int width, |
4869 | 0 | int height) { |
4870 | 0 | return NV21ToRGB24Matrix(src_y, src_stride_y, src_vu, src_stride_vu, |
4871 | 0 | dst_rgb24, dst_stride_rgb24, &kYuvI601Constants, |
4872 | 0 | width, height); |
4873 | 0 | } |
4874 | | |
4875 | | // Convert NV12 to RAW. |
4876 | | LIBYUV_API |
4877 | | int NV12ToRAW(const uint8_t* src_y, |
4878 | | int src_stride_y, |
4879 | | const uint8_t* src_uv, |
4880 | | int src_stride_uv, |
4881 | | uint8_t* dst_raw, |
4882 | | int dst_stride_raw, |
4883 | | int width, |
4884 | 0 | int height) { |
4885 | 0 | return NV21ToRGB24Matrix(src_y, src_stride_y, src_uv, src_stride_uv, dst_raw, |
4886 | 0 | dst_stride_raw, &kYvuI601Constants, width, height); |
4887 | 0 | } |
4888 | | |
4889 | | // Convert NV21 to RAW. |
4890 | | LIBYUV_API |
4891 | | int NV21ToRAW(const uint8_t* src_y, |
4892 | | int src_stride_y, |
4893 | | const uint8_t* src_vu, |
4894 | | int src_stride_vu, |
4895 | | uint8_t* dst_raw, |
4896 | | int dst_stride_raw, |
4897 | | int width, |
4898 | 0 | int height) { |
4899 | 0 | return NV12ToRGB24Matrix(src_y, src_stride_y, src_vu, src_stride_vu, dst_raw, |
4900 | 0 | dst_stride_raw, &kYvuI601Constants, width, height); |
4901 | 0 | } |
4902 | | |
4903 | | // Convert NV21 to YUV24 |
4904 | | int NV21ToYUV24(const uint8_t* src_y, |
4905 | | int src_stride_y, |
4906 | | const uint8_t* src_vu, |
4907 | | int src_stride_vu, |
4908 | | uint8_t* dst_yuv24, |
4909 | | int dst_stride_yuv24, |
4910 | | int width, |
4911 | 0 | int height) { |
4912 | 0 | int y; |
4913 | 0 | void (*NV21ToYUV24Row)(const uint8_t* src_y, const uint8_t* src_vu, |
4914 | 0 | uint8_t* dst_yuv24, int width) = NV21ToYUV24Row_C; |
4915 | 0 | if (!src_y || !src_vu || !dst_yuv24 || width <= 0 || height == 0) { |
4916 | 0 | return -1; |
4917 | 0 | } |
4918 | | // Negative height means invert the image. |
4919 | 0 | if (height < 0) { |
4920 | 0 | height = -height; |
4921 | 0 | dst_yuv24 = dst_yuv24 + (height - 1) * dst_stride_yuv24; |
4922 | 0 | dst_stride_yuv24 = -dst_stride_yuv24; |
4923 | 0 | } |
4924 | | #if defined(HAS_NV21TOYUV24ROW_NEON) |
4925 | | if (TestCpuFlag(kCpuHasNEON)) { |
4926 | | NV21ToYUV24Row = NV21ToYUV24Row_Any_NEON; |
4927 | | if (IS_ALIGNED(width, 16)) { |
4928 | | NV21ToYUV24Row = NV21ToYUV24Row_NEON; |
4929 | | } |
4930 | | } |
4931 | | #endif |
4932 | 0 | #if defined(HAS_NV21TOYUV24ROW_SSSE3) |
4933 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4934 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_Any_SSSE3; |
4935 | 0 | if (IS_ALIGNED(width, 16)) { |
4936 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_SSSE3; |
4937 | 0 | } |
4938 | 0 | } |
4939 | 0 | #endif |
4940 | 0 | #if defined(HAS_NV21TOYUV24ROW_AVX2) |
4941 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4942 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_Any_AVX2; |
4943 | 0 | if (IS_ALIGNED(width, 32)) { |
4944 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_AVX2; |
4945 | 0 | } |
4946 | 0 | } |
4947 | 0 | #endif |
4948 | 0 | for (y = 0; y < height; ++y) { |
4949 | 0 | NV21ToYUV24Row(src_y, src_vu, dst_yuv24, width); |
4950 | 0 | dst_yuv24 += dst_stride_yuv24; |
4951 | 0 | src_y += src_stride_y; |
4952 | 0 | if (y & 1) { |
4953 | 0 | src_vu += src_stride_vu; |
4954 | 0 | } |
4955 | 0 | } |
4956 | 0 | return 0; |
4957 | 0 | } |
4958 | | |
4959 | | // Convert YUY2 to ARGB with matrix. |
4960 | | LIBYUV_API |
4961 | | int YUY2ToARGBMatrix(const uint8_t* src_yuy2, |
4962 | | int src_stride_yuy2, |
4963 | | uint8_t* dst_argb, |
4964 | | int dst_stride_argb, |
4965 | | const struct YuvConstants* yuvconstants, |
4966 | | int width, |
4967 | 0 | int height) { |
4968 | 0 | int y; |
4969 | 0 | void (*YUY2ToARGBRow)(const uint8_t* src_yuy2, uint8_t* dst_argb, |
4970 | 0 | const struct YuvConstants* yuvconstants, int width) = |
4971 | 0 | YUY2ToARGBRow_C; |
4972 | 0 | if (!src_yuy2 || !dst_argb || width <= 0 || height == 0) { |
4973 | 0 | return -1; |
4974 | 0 | } |
4975 | | // Negative height means invert the image. |
4976 | 0 | if (height < 0) { |
4977 | 0 | height = -height; |
4978 | 0 | src_yuy2 = src_yuy2 + (height - 1) * src_stride_yuy2; |
4979 | 0 | src_stride_yuy2 = -src_stride_yuy2; |
4980 | 0 | } |
4981 | | // Coalesce rows. |
4982 | 0 | if (src_stride_yuy2 == width * 2 && dst_stride_argb == width * 4) { |
4983 | 0 | width *= height; |
4984 | 0 | height = 1; |
4985 | 0 | src_stride_yuy2 = dst_stride_argb = 0; |
4986 | 0 | } |
4987 | 0 | #if defined(HAS_YUY2TOARGBROW_SSSE3) |
4988 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4989 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_Any_SSSE3; |
4990 | 0 | if (IS_ALIGNED(width, 16)) { |
4991 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_SSSE3; |
4992 | 0 | } |
4993 | 0 | } |
4994 | 0 | #endif |
4995 | 0 | #if defined(HAS_YUY2TOARGBROW_AVX2) |
4996 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4997 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_Any_AVX2; |
4998 | 0 | if (IS_ALIGNED(width, 32)) { |
4999 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_AVX2; |
5000 | 0 | } |
5001 | 0 | } |
5002 | 0 | #endif |
5003 | | #if defined(HAS_YUY2TOARGBROW_NEON) |
5004 | | if (TestCpuFlag(kCpuHasNEON)) { |
5005 | | YUY2ToARGBRow = YUY2ToARGBRow_Any_NEON; |
5006 | | if (IS_ALIGNED(width, 8)) { |
5007 | | YUY2ToARGBRow = YUY2ToARGBRow_NEON; |
5008 | | } |
5009 | | } |
5010 | | #endif |
5011 | | #if defined(HAS_YUY2TOARGBROW_SVE2) |
5012 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5013 | | YUY2ToARGBRow = YUY2ToARGBRow_SVE2; |
5014 | | } |
5015 | | #endif |
5016 | | #if defined(HAS_YUY2TOARGBROW_SME) |
5017 | | if (TestCpuFlag(kCpuHasSME)) { |
5018 | | YUY2ToARGBRow = YUY2ToARGBRow_SME; |
5019 | | } |
5020 | | #endif |
5021 | | #if defined(HAS_YUY2TOARGBROW_MSA) |
5022 | | if (TestCpuFlag(kCpuHasMSA)) { |
5023 | | YUY2ToARGBRow = YUY2ToARGBRow_Any_MSA; |
5024 | | if (IS_ALIGNED(width, 8)) { |
5025 | | YUY2ToARGBRow = YUY2ToARGBRow_MSA; |
5026 | | } |
5027 | | } |
5028 | | #endif |
5029 | | #if defined(HAS_YUY2TOARGBROW_LSX) |
5030 | | if (TestCpuFlag(kCpuHasLSX)) { |
5031 | | YUY2ToARGBRow = YUY2ToARGBRow_Any_LSX; |
5032 | | if (IS_ALIGNED(width, 8)) { |
5033 | | YUY2ToARGBRow = YUY2ToARGBRow_LSX; |
5034 | | } |
5035 | | } |
5036 | | #endif |
5037 | 0 | for (y = 0; y < height; ++y) { |
5038 | 0 | YUY2ToARGBRow(src_yuy2, dst_argb, yuvconstants, width); |
5039 | 0 | src_yuy2 += src_stride_yuy2; |
5040 | 0 | dst_argb += dst_stride_argb; |
5041 | 0 | } |
5042 | 0 | return 0; |
5043 | 0 | } |
5044 | | |
5045 | | // Convert YUY2 to ARGB. |
5046 | | LIBYUV_API |
5047 | | int YUY2ToARGB(const uint8_t* src_yuy2, |
5048 | | int src_stride_yuy2, |
5049 | | uint8_t* dst_argb, |
5050 | | int dst_stride_argb, |
5051 | | int width, |
5052 | 0 | int height) { |
5053 | 0 | return YUY2ToARGBMatrix(src_yuy2, src_stride_yuy2, dst_argb, dst_stride_argb, |
5054 | 0 | &kYuvI601Constants, width, height); |
5055 | 0 | } |
5056 | | |
5057 | | // Convert UYVY to ARGB with matrix. |
5058 | | LIBYUV_API |
5059 | | int UYVYToARGBMatrix(const uint8_t* src_uyvy, |
5060 | | int src_stride_uyvy, |
5061 | | uint8_t* dst_argb, |
5062 | | int dst_stride_argb, |
5063 | | const struct YuvConstants* yuvconstants, |
5064 | | int width, |
5065 | 0 | int height) { |
5066 | 0 | int y; |
5067 | 0 | void (*UYVYToARGBRow)(const uint8_t* src_uyvy, uint8_t* dst_argb, |
5068 | 0 | const struct YuvConstants* yuvconstants, int width) = |
5069 | 0 | UYVYToARGBRow_C; |
5070 | 0 | if (!src_uyvy || !dst_argb || width <= 0 || height == 0) { |
5071 | 0 | return -1; |
5072 | 0 | } |
5073 | | // Negative height means invert the image. |
5074 | 0 | if (height < 0) { |
5075 | 0 | height = -height; |
5076 | 0 | src_uyvy = src_uyvy + (height - 1) * src_stride_uyvy; |
5077 | 0 | src_stride_uyvy = -src_stride_uyvy; |
5078 | 0 | } |
5079 | | // Coalesce rows. |
5080 | 0 | if (src_stride_uyvy == width * 2 && dst_stride_argb == width * 4) { |
5081 | 0 | width *= height; |
5082 | 0 | height = 1; |
5083 | 0 | src_stride_uyvy = dst_stride_argb = 0; |
5084 | 0 | } |
5085 | 0 | #if defined(HAS_UYVYTOARGBROW_SSSE3) |
5086 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5087 | 0 | UYVYToARGBRow = UYVYToARGBRow_Any_SSSE3; |
5088 | 0 | if (IS_ALIGNED(width, 16)) { |
5089 | 0 | UYVYToARGBRow = UYVYToARGBRow_SSSE3; |
5090 | 0 | } |
5091 | 0 | } |
5092 | 0 | #endif |
5093 | 0 | #if defined(HAS_UYVYTOARGBROW_AVX2) |
5094 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
5095 | 0 | UYVYToARGBRow = UYVYToARGBRow_Any_AVX2; |
5096 | 0 | if (IS_ALIGNED(width, 32)) { |
5097 | 0 | UYVYToARGBRow = UYVYToARGBRow_AVX2; |
5098 | 0 | } |
5099 | 0 | } |
5100 | 0 | #endif |
5101 | | #if defined(HAS_UYVYTOARGBROW_NEON) |
5102 | | if (TestCpuFlag(kCpuHasNEON)) { |
5103 | | UYVYToARGBRow = UYVYToARGBRow_Any_NEON; |
5104 | | if (IS_ALIGNED(width, 8)) { |
5105 | | UYVYToARGBRow = UYVYToARGBRow_NEON; |
5106 | | } |
5107 | | } |
5108 | | #endif |
5109 | | #if defined(HAS_UYVYTOARGBROW_SVE2) |
5110 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5111 | | UYVYToARGBRow = UYVYToARGBRow_SVE2; |
5112 | | } |
5113 | | #endif |
5114 | | #if defined(HAS_UYVYTOARGBROW_SME) |
5115 | | if (TestCpuFlag(kCpuHasSME)) { |
5116 | | UYVYToARGBRow = UYVYToARGBRow_SME; |
5117 | | } |
5118 | | #endif |
5119 | | #if defined(HAS_UYVYTOARGBROW_MSA) |
5120 | | if (TestCpuFlag(kCpuHasMSA)) { |
5121 | | UYVYToARGBRow = UYVYToARGBRow_Any_MSA; |
5122 | | if (IS_ALIGNED(width, 8)) { |
5123 | | UYVYToARGBRow = UYVYToARGBRow_MSA; |
5124 | | } |
5125 | | } |
5126 | | #endif |
5127 | | #if defined(HAS_UYVYTOARGBROW_LSX) |
5128 | | if (TestCpuFlag(kCpuHasLSX)) { |
5129 | | UYVYToARGBRow = UYVYToARGBRow_Any_LSX; |
5130 | | if (IS_ALIGNED(width, 8)) { |
5131 | | UYVYToARGBRow = UYVYToARGBRow_LSX; |
5132 | | } |
5133 | | } |
5134 | | #endif |
5135 | 0 | for (y = 0; y < height; ++y) { |
5136 | 0 | UYVYToARGBRow(src_uyvy, dst_argb, yuvconstants, width); |
5137 | 0 | src_uyvy += src_stride_uyvy; |
5138 | 0 | dst_argb += dst_stride_argb; |
5139 | 0 | } |
5140 | 0 | return 0; |
5141 | 0 | } |
5142 | | |
5143 | | // Convert UYVY to ARGB. |
5144 | | LIBYUV_API |
5145 | | int UYVYToARGB(const uint8_t* src_uyvy, |
5146 | | int src_stride_uyvy, |
5147 | | uint8_t* dst_argb, |
5148 | | int dst_stride_argb, |
5149 | | int width, |
5150 | 0 | int height) { |
5151 | 0 | return UYVYToARGBMatrix(src_uyvy, src_stride_uyvy, dst_argb, dst_stride_argb, |
5152 | 0 | &kYuvI601Constants, width, height); |
5153 | 0 | } |
5154 | | |
5155 | | static void WeavePixels(const uint8_t* src_u, |
5156 | | const uint8_t* src_v, |
5157 | | int src_pixel_stride_uv, |
5158 | | uint8_t* dst_uv, |
5159 | 0 | int width) { |
5160 | 0 | int i; |
5161 | 0 | for (i = 0; i < width; ++i) { |
5162 | 0 | dst_uv[0] = *src_u; |
5163 | 0 | dst_uv[1] = *src_v; |
5164 | 0 | dst_uv += 2; |
5165 | 0 | src_u += src_pixel_stride_uv; |
5166 | 0 | src_v += src_pixel_stride_uv; |
5167 | 0 | } |
5168 | 0 | } |
5169 | | |
5170 | | // Convert Android420 to ARGB with matrix. |
5171 | | LIBYUV_API |
5172 | | int Android420ToARGBMatrix(const uint8_t* src_y, |
5173 | | int src_stride_y, |
5174 | | const uint8_t* src_u, |
5175 | | int src_stride_u, |
5176 | | const uint8_t* src_v, |
5177 | | int src_stride_v, |
5178 | | int src_pixel_stride_uv, |
5179 | | uint8_t* dst_argb, |
5180 | | int dst_stride_argb, |
5181 | | const struct YuvConstants* yuvconstants, |
5182 | | int width, |
5183 | 0 | int height) { |
5184 | 0 | int y; |
5185 | 0 | uint8_t* dst_uv; |
5186 | 0 | const ptrdiff_t vu_off = src_v - src_u; |
5187 | 0 | int halfwidth = (width + 1) >> 1; |
5188 | 0 | int halfheight = (height + 1) >> 1; |
5189 | 0 | assert(yuvconstants); |
5190 | 0 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
5191 | 0 | return -1; |
5192 | 0 | } |
5193 | | // Negative height means invert the image. |
5194 | 0 | if (height < 0) { |
5195 | 0 | height = -height; |
5196 | 0 | halfheight = (height + 1) >> 1; |
5197 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
5198 | 0 | dst_stride_argb = -dst_stride_argb; |
5199 | 0 | } |
5200 | | |
5201 | | // I420 |
5202 | 0 | if (src_pixel_stride_uv == 1) { |
5203 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5204 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
5205 | 0 | yuvconstants, width, height); |
5206 | | // NV21 |
5207 | 0 | } |
5208 | 0 | if (src_pixel_stride_uv == 2 && vu_off == -1 && |
5209 | 0 | src_stride_u == src_stride_v) { |
5210 | 0 | return NV21ToARGBMatrix(src_y, src_stride_y, src_v, src_stride_v, dst_argb, |
5211 | 0 | dst_stride_argb, yuvconstants, width, height); |
5212 | | // NV12 |
5213 | 0 | } |
5214 | 0 | if (src_pixel_stride_uv == 2 && vu_off == 1 && src_stride_u == src_stride_v) { |
5215 | 0 | return NV12ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, dst_argb, |
5216 | 0 | dst_stride_argb, yuvconstants, width, height); |
5217 | 0 | } |
5218 | | |
5219 | | // General case fallback creates NV12 |
5220 | 0 | align_buffer_64(plane_uv, halfwidth * 2 * halfheight); |
5221 | 0 | if (!plane_uv) |
5222 | 0 | return 1; |
5223 | 0 | dst_uv = plane_uv; |
5224 | 0 | for (y = 0; y < halfheight; ++y) { |
5225 | 0 | WeavePixels(src_u, src_v, src_pixel_stride_uv, dst_uv, halfwidth); |
5226 | 0 | src_u += src_stride_u; |
5227 | 0 | src_v += src_stride_v; |
5228 | 0 | dst_uv += halfwidth * 2; |
5229 | 0 | } |
5230 | 0 | NV12ToARGBMatrix(src_y, src_stride_y, plane_uv, halfwidth * 2, dst_argb, |
5231 | 0 | dst_stride_argb, yuvconstants, width, height); |
5232 | 0 | free_aligned_buffer_64(plane_uv); |
5233 | 0 | return 0; |
5234 | 0 | } |
5235 | | |
5236 | | // Convert Android420 to ARGB. |
5237 | | LIBYUV_API |
5238 | | int Android420ToARGB(const uint8_t* src_y, |
5239 | | int src_stride_y, |
5240 | | const uint8_t* src_u, |
5241 | | int src_stride_u, |
5242 | | const uint8_t* src_v, |
5243 | | int src_stride_v, |
5244 | | int src_pixel_stride_uv, |
5245 | | uint8_t* dst_argb, |
5246 | | int dst_stride_argb, |
5247 | | int width, |
5248 | 0 | int height) { |
5249 | 0 | return Android420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5250 | 0 | src_stride_v, src_pixel_stride_uv, dst_argb, |
5251 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
5252 | 0 | height); |
5253 | 0 | } |
5254 | | |
5255 | | // Convert Android420 to ABGR. |
5256 | | LIBYUV_API |
5257 | | int Android420ToABGR(const uint8_t* src_y, |
5258 | | int src_stride_y, |
5259 | | const uint8_t* src_u, |
5260 | | int src_stride_u, |
5261 | | const uint8_t* src_v, |
5262 | | int src_stride_v, |
5263 | | int src_pixel_stride_uv, |
5264 | | uint8_t* dst_abgr, |
5265 | | int dst_stride_abgr, |
5266 | | int width, |
5267 | 0 | int height) { |
5268 | 0 | return Android420ToARGBMatrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
5269 | 0 | src_stride_u, src_pixel_stride_uv, dst_abgr, |
5270 | 0 | dst_stride_abgr, &kYvuI601Constants, width, |
5271 | 0 | height); |
5272 | 0 | } |
5273 | | |
5274 | | // Convert I422 to RGBA with matrix. |
5275 | | LIBYUV_API |
5276 | | int I422ToRGBAMatrix(const uint8_t* src_y, |
5277 | | int src_stride_y, |
5278 | | const uint8_t* src_u, |
5279 | | int src_stride_u, |
5280 | | const uint8_t* src_v, |
5281 | | int src_stride_v, |
5282 | | uint8_t* dst_rgba, |
5283 | | int dst_stride_rgba, |
5284 | | const struct YuvConstants* yuvconstants, |
5285 | | int width, |
5286 | 71 | int height) { |
5287 | 71 | int y; |
5288 | 71 | void (*I422ToRGBARow)(const uint8_t* y_buf, const uint8_t* u_buf, |
5289 | 71 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5290 | 71 | const struct YuvConstants* yuvconstants, int width) = |
5291 | 71 | I422ToRGBARow_C; |
5292 | 71 | assert(yuvconstants); |
5293 | 71 | if (!src_y || !src_u || !src_v || !dst_rgba || width <= 0 || height == 0) { |
5294 | 0 | return -1; |
5295 | 0 | } |
5296 | | // Negative height means invert the image. |
5297 | 71 | if (height < 0) { |
5298 | 0 | height = -height; |
5299 | 0 | dst_rgba = dst_rgba + (height - 1) * dst_stride_rgba; |
5300 | 0 | dst_stride_rgba = -dst_stride_rgba; |
5301 | 0 | } |
5302 | 71 | #if defined(HAS_I422TORGBAROW_SSSE3) |
5303 | 71 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5304 | 71 | I422ToRGBARow = I422ToRGBARow_Any_SSSE3; |
5305 | 71 | if (IS_ALIGNED(width, 8)) { |
5306 | 34 | I422ToRGBARow = I422ToRGBARow_SSSE3; |
5307 | 34 | } |
5308 | 71 | } |
5309 | 71 | #endif |
5310 | 71 | #if defined(HAS_I422TORGBAROW_AVX2) |
5311 | 71 | if (TestCpuFlag(kCpuHasAVX2)) { |
5312 | 71 | I422ToRGBARow = I422ToRGBARow_Any_AVX2; |
5313 | 71 | if (IS_ALIGNED(width, 16)) { |
5314 | 17 | I422ToRGBARow = I422ToRGBARow_AVX2; |
5315 | 17 | } |
5316 | 71 | } |
5317 | 71 | #endif |
5318 | | #if defined(HAS_I422TORGBAROW_NEON) |
5319 | | if (TestCpuFlag(kCpuHasNEON)) { |
5320 | | I422ToRGBARow = I422ToRGBARow_Any_NEON; |
5321 | | if (IS_ALIGNED(width, 8)) { |
5322 | | I422ToRGBARow = I422ToRGBARow_NEON; |
5323 | | } |
5324 | | } |
5325 | | #endif |
5326 | | #if defined(HAS_I422TORGBAROW_SVE2) |
5327 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5328 | | I422ToRGBARow = I422ToRGBARow_SVE2; |
5329 | | } |
5330 | | #endif |
5331 | | #if defined(HAS_I422TORGBAROW_SME) |
5332 | | if (TestCpuFlag(kCpuHasSME)) { |
5333 | | I422ToRGBARow = I422ToRGBARow_SME; |
5334 | | } |
5335 | | #endif |
5336 | | #if defined(HAS_I422TORGBAROW_MSA) |
5337 | | if (TestCpuFlag(kCpuHasMSA)) { |
5338 | | I422ToRGBARow = I422ToRGBARow_Any_MSA; |
5339 | | if (IS_ALIGNED(width, 8)) { |
5340 | | I422ToRGBARow = I422ToRGBARow_MSA; |
5341 | | } |
5342 | | } |
5343 | | #endif |
5344 | | #if defined(HAS_I422TORGBAROW_LSX) |
5345 | | if (TestCpuFlag(kCpuHasLSX)) { |
5346 | | I422ToRGBARow = I422ToRGBARow_Any_LSX; |
5347 | | if (IS_ALIGNED(width, 16)) { |
5348 | | I422ToRGBARow = I422ToRGBARow_LSX; |
5349 | | } |
5350 | | } |
5351 | | #endif |
5352 | | #if defined(HAS_I422TORGBAROW_LASX) |
5353 | | if (TestCpuFlag(kCpuHasLASX)) { |
5354 | | I422ToRGBARow = I422ToRGBARow_Any_LASX; |
5355 | | if (IS_ALIGNED(width, 32)) { |
5356 | | I422ToRGBARow = I422ToRGBARow_LASX; |
5357 | | } |
5358 | | } |
5359 | | #endif |
5360 | | #if defined(HAS_I422TORGBAROW_RVV) |
5361 | | if (TestCpuFlag(kCpuHasRVV)) { |
5362 | | I422ToRGBARow = I422ToRGBARow_RVV; |
5363 | | } |
5364 | | #endif |
5365 | | |
5366 | 1.92k | for (y = 0; y < height; ++y) { |
5367 | 1.85k | I422ToRGBARow(src_y, src_u, src_v, dst_rgba, yuvconstants, width); |
5368 | 1.85k | dst_rgba += dst_stride_rgba; |
5369 | 1.85k | src_y += src_stride_y; |
5370 | 1.85k | src_u += src_stride_u; |
5371 | 1.85k | src_v += src_stride_v; |
5372 | 1.85k | } |
5373 | 71 | return 0; |
5374 | 71 | } |
5375 | | |
5376 | | // Convert I422 to RGBA. |
5377 | | LIBYUV_API |
5378 | | int I422ToRGBA(const uint8_t* src_y, |
5379 | | int src_stride_y, |
5380 | | const uint8_t* src_u, |
5381 | | int src_stride_u, |
5382 | | const uint8_t* src_v, |
5383 | | int src_stride_v, |
5384 | | uint8_t* dst_rgba, |
5385 | | int dst_stride_rgba, |
5386 | | int width, |
5387 | 0 | int height) { |
5388 | 0 | return I422ToRGBAMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5389 | 0 | src_stride_v, dst_rgba, dst_stride_rgba, |
5390 | 0 | &kYuvI601Constants, width, height); |
5391 | 0 | } |
5392 | | |
5393 | | // Convert I422 to BGRA. |
5394 | | LIBYUV_API |
5395 | | int I422ToBGRA(const uint8_t* src_y, |
5396 | | int src_stride_y, |
5397 | | const uint8_t* src_u, |
5398 | | int src_stride_u, |
5399 | | const uint8_t* src_v, |
5400 | | int src_stride_v, |
5401 | | uint8_t* dst_bgra, |
5402 | | int dst_stride_bgra, |
5403 | | int width, |
5404 | 0 | int height) { |
5405 | 0 | return I422ToRGBAMatrix(src_y, src_stride_y, src_v, |
5406 | 0 | src_stride_v, // Swap U and V |
5407 | 0 | src_u, src_stride_u, dst_bgra, dst_stride_bgra, |
5408 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5409 | 0 | width, height); |
5410 | 0 | } |
5411 | | |
5412 | | // Convert NV12 to RGB565 with matrix. |
5413 | | LIBYUV_API |
5414 | | int NV12ToRGB565Matrix(const uint8_t* src_y, |
5415 | | int src_stride_y, |
5416 | | const uint8_t* src_uv, |
5417 | | int src_stride_uv, |
5418 | | uint8_t* dst_rgb565, |
5419 | | int dst_stride_rgb565, |
5420 | | const struct YuvConstants* yuvconstants, |
5421 | | int width, |
5422 | 0 | int height) { |
5423 | 0 | int y; |
5424 | 0 | void (*NV12ToRGB565Row)( |
5425 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
5426 | 0 | const struct YuvConstants* yuvconstants, int width) = NV12ToRGB565Row_C; |
5427 | 0 | assert(yuvconstants); |
5428 | 0 | if (!src_y || !src_uv || !dst_rgb565 || width <= 0 || height == 0) { |
5429 | 0 | return -1; |
5430 | 0 | } |
5431 | | // Negative height means invert the image. |
5432 | 0 | if (height < 0) { |
5433 | 0 | height = -height; |
5434 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
5435 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
5436 | 0 | } |
5437 | 0 | #if defined(HAS_NV12TORGB565ROW_SSSE3) |
5438 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5439 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_Any_SSSE3; |
5440 | 0 | if (IS_ALIGNED(width, 8)) { |
5441 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_SSSE3; |
5442 | 0 | } |
5443 | 0 | } |
5444 | 0 | #endif |
5445 | 0 | #if defined(HAS_NV12TORGB565ROW_AVX2) |
5446 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
5447 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_Any_AVX2; |
5448 | 0 | if (IS_ALIGNED(width, 16)) { |
5449 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_AVX2; |
5450 | 0 | } |
5451 | 0 | } |
5452 | 0 | #endif |
5453 | | #if defined(HAS_NV12TORGB565ROW_NEON) |
5454 | | if (TestCpuFlag(kCpuHasNEON)) { |
5455 | | NV12ToRGB565Row = NV12ToRGB565Row_Any_NEON; |
5456 | | if (IS_ALIGNED(width, 8)) { |
5457 | | NV12ToRGB565Row = NV12ToRGB565Row_NEON; |
5458 | | } |
5459 | | } |
5460 | | #endif |
5461 | | #if defined(HAS_NV12TORGB565ROW_MSA) |
5462 | | if (TestCpuFlag(kCpuHasMSA)) { |
5463 | | NV12ToRGB565Row = NV12ToRGB565Row_Any_MSA; |
5464 | | if (IS_ALIGNED(width, 8)) { |
5465 | | NV12ToRGB565Row = NV12ToRGB565Row_MSA; |
5466 | | } |
5467 | | } |
5468 | | #endif |
5469 | | #if defined(HAS_NV12TORGB565ROW_LSX) |
5470 | | if (TestCpuFlag(kCpuHasLSX)) { |
5471 | | NV12ToRGB565Row = NV12ToRGB565Row_Any_LSX; |
5472 | | if (IS_ALIGNED(width, 8)) { |
5473 | | NV12ToRGB565Row = NV12ToRGB565Row_LSX; |
5474 | | } |
5475 | | } |
5476 | | #endif |
5477 | | #if defined(HAS_NV12TORGB565ROW_LASX) |
5478 | | if (TestCpuFlag(kCpuHasLASX)) { |
5479 | | NV12ToRGB565Row = NV12ToRGB565Row_Any_LASX; |
5480 | | if (IS_ALIGNED(width, 16)) { |
5481 | | NV12ToRGB565Row = NV12ToRGB565Row_LASX; |
5482 | | } |
5483 | | } |
5484 | | #endif |
5485 | |
|
5486 | 0 | for (y = 0; y < height; ++y) { |
5487 | 0 | NV12ToRGB565Row(src_y, src_uv, dst_rgb565, yuvconstants, width); |
5488 | 0 | dst_rgb565 += dst_stride_rgb565; |
5489 | 0 | src_y += src_stride_y; |
5490 | 0 | if (y & 1) { |
5491 | 0 | src_uv += src_stride_uv; |
5492 | 0 | } |
5493 | 0 | } |
5494 | 0 | return 0; |
5495 | 0 | } |
5496 | | |
5497 | | // Convert NV12 to RGB565. |
5498 | | LIBYUV_API |
5499 | | int NV12ToRGB565(const uint8_t* src_y, |
5500 | | int src_stride_y, |
5501 | | const uint8_t* src_uv, |
5502 | | int src_stride_uv, |
5503 | | uint8_t* dst_rgb565, |
5504 | | int dst_stride_rgb565, |
5505 | | int width, |
5506 | 0 | int height) { |
5507 | 0 | return NV12ToRGB565Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
5508 | 0 | dst_rgb565, dst_stride_rgb565, &kYuvI601Constants, |
5509 | 0 | width, height); |
5510 | 0 | } |
5511 | | |
5512 | | // Convert I422 to RGBA with matrix. |
5513 | | LIBYUV_API |
5514 | | int I420ToRGBAMatrix(const uint8_t* src_y, |
5515 | | int src_stride_y, |
5516 | | const uint8_t* src_u, |
5517 | | int src_stride_u, |
5518 | | const uint8_t* src_v, |
5519 | | int src_stride_v, |
5520 | | uint8_t* dst_rgba, |
5521 | | int dst_stride_rgba, |
5522 | | const struct YuvConstants* yuvconstants, |
5523 | | int width, |
5524 | 143 | int height) { |
5525 | 143 | int y; |
5526 | 143 | void (*I422ToRGBARow)(const uint8_t* y_buf, const uint8_t* u_buf, |
5527 | 143 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5528 | 143 | const struct YuvConstants* yuvconstants, int width) = |
5529 | 143 | I422ToRGBARow_C; |
5530 | 143 | assert(yuvconstants); |
5531 | 143 | if (!src_y || !src_u || !src_v || !dst_rgba || width <= 0 || height == 0) { |
5532 | 0 | return -1; |
5533 | 0 | } |
5534 | | // Negative height means invert the image. |
5535 | 143 | if (height < 0) { |
5536 | 0 | height = -height; |
5537 | 0 | dst_rgba = dst_rgba + (height - 1) * dst_stride_rgba; |
5538 | 0 | dst_stride_rgba = -dst_stride_rgba; |
5539 | 0 | } |
5540 | 143 | #if defined(HAS_I422TORGBAROW_SSSE3) |
5541 | 143 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5542 | 143 | I422ToRGBARow = I422ToRGBARow_Any_SSSE3; |
5543 | 143 | if (IS_ALIGNED(width, 8)) { |
5544 | 35 | I422ToRGBARow = I422ToRGBARow_SSSE3; |
5545 | 35 | } |
5546 | 143 | } |
5547 | 143 | #endif |
5548 | 143 | #if defined(HAS_I422TORGBAROW_AVX2) |
5549 | 143 | if (TestCpuFlag(kCpuHasAVX2)) { |
5550 | 143 | I422ToRGBARow = I422ToRGBARow_Any_AVX2; |
5551 | 143 | if (IS_ALIGNED(width, 16)) { |
5552 | 18 | I422ToRGBARow = I422ToRGBARow_AVX2; |
5553 | 18 | } |
5554 | 143 | } |
5555 | 143 | #endif |
5556 | | #if defined(HAS_I422TORGBAROW_NEON) |
5557 | | if (TestCpuFlag(kCpuHasNEON)) { |
5558 | | I422ToRGBARow = I422ToRGBARow_Any_NEON; |
5559 | | if (IS_ALIGNED(width, 8)) { |
5560 | | I422ToRGBARow = I422ToRGBARow_NEON; |
5561 | | } |
5562 | | } |
5563 | | #endif |
5564 | | #if defined(HAS_I422TORGBAROW_SVE2) |
5565 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5566 | | I422ToRGBARow = I422ToRGBARow_SVE2; |
5567 | | } |
5568 | | #endif |
5569 | | #if defined(HAS_I422TORGBAROW_SME) |
5570 | | if (TestCpuFlag(kCpuHasSME)) { |
5571 | | I422ToRGBARow = I422ToRGBARow_SME; |
5572 | | } |
5573 | | #endif |
5574 | | #if defined(HAS_I422TORGBAROW_MSA) |
5575 | | if (TestCpuFlag(kCpuHasMSA)) { |
5576 | | I422ToRGBARow = I422ToRGBARow_Any_MSA; |
5577 | | if (IS_ALIGNED(width, 8)) { |
5578 | | I422ToRGBARow = I422ToRGBARow_MSA; |
5579 | | } |
5580 | | } |
5581 | | #endif |
5582 | | #if defined(HAS_I422TORGBAROW_LSX) |
5583 | | if (TestCpuFlag(kCpuHasLSX)) { |
5584 | | I422ToRGBARow = I422ToRGBARow_Any_LSX; |
5585 | | if (IS_ALIGNED(width, 16)) { |
5586 | | I422ToRGBARow = I422ToRGBARow_LSX; |
5587 | | } |
5588 | | } |
5589 | | #endif |
5590 | | #if defined(HAS_I422TORGBAROW_LASX) |
5591 | | if (TestCpuFlag(kCpuHasLASX)) { |
5592 | | I422ToRGBARow = I422ToRGBARow_Any_LASX; |
5593 | | if (IS_ALIGNED(width, 32)) { |
5594 | | I422ToRGBARow = I422ToRGBARow_LASX; |
5595 | | } |
5596 | | } |
5597 | | #endif |
5598 | | #if defined(HAS_I422TORGBAROW_RVV) |
5599 | | if (TestCpuFlag(kCpuHasRVV)) { |
5600 | | I422ToRGBARow = I422ToRGBARow_RVV; |
5601 | | } |
5602 | | #endif |
5603 | | |
5604 | 2.64k | for (y = 0; y < height; ++y) { |
5605 | 2.50k | I422ToRGBARow(src_y, src_u, src_v, dst_rgba, yuvconstants, width); |
5606 | 2.50k | dst_rgba += dst_stride_rgba; |
5607 | 2.50k | src_y += src_stride_y; |
5608 | 2.50k | if (y & 1) { |
5609 | 1.23k | src_u += src_stride_u; |
5610 | 1.23k | src_v += src_stride_v; |
5611 | 1.23k | } |
5612 | 2.50k | } |
5613 | 143 | return 0; |
5614 | 143 | } |
5615 | | |
5616 | | // Convert I420 to RGBA. |
5617 | | LIBYUV_API |
5618 | | int I420ToRGBA(const uint8_t* src_y, |
5619 | | int src_stride_y, |
5620 | | const uint8_t* src_u, |
5621 | | int src_stride_u, |
5622 | | const uint8_t* src_v, |
5623 | | int src_stride_v, |
5624 | | uint8_t* dst_rgba, |
5625 | | int dst_stride_rgba, |
5626 | | int width, |
5627 | 0 | int height) { |
5628 | 0 | return I420ToRGBAMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5629 | 0 | src_stride_v, dst_rgba, dst_stride_rgba, |
5630 | 0 | &kYuvI601Constants, width, height); |
5631 | 0 | } |
5632 | | |
5633 | | // Convert I420 to BGRA. |
5634 | | LIBYUV_API |
5635 | | int I420ToBGRA(const uint8_t* src_y, |
5636 | | int src_stride_y, |
5637 | | const uint8_t* src_u, |
5638 | | int src_stride_u, |
5639 | | const uint8_t* src_v, |
5640 | | int src_stride_v, |
5641 | | uint8_t* dst_bgra, |
5642 | | int dst_stride_bgra, |
5643 | | int width, |
5644 | 0 | int height) { |
5645 | 0 | return I420ToRGBAMatrix(src_y, src_stride_y, src_v, |
5646 | 0 | src_stride_v, // Swap U and V |
5647 | 0 | src_u, src_stride_u, dst_bgra, dst_stride_bgra, |
5648 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5649 | 0 | width, height); |
5650 | 0 | } |
5651 | | |
5652 | | // Convert I420 to RGB24 with matrix. |
5653 | | LIBYUV_API |
5654 | | int I420ToRGB24Matrix(const uint8_t* src_y, |
5655 | | int src_stride_y, |
5656 | | const uint8_t* src_u, |
5657 | | int src_stride_u, |
5658 | | const uint8_t* src_v, |
5659 | | int src_stride_v, |
5660 | | uint8_t* dst_rgb24, |
5661 | | int dst_stride_rgb24, |
5662 | | const struct YuvConstants* yuvconstants, |
5663 | | int width, |
5664 | 184 | int height) { |
5665 | 184 | int y; |
5666 | 184 | void (*I422ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
5667 | 184 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5668 | 184 | const struct YuvConstants* yuvconstants, int width) = |
5669 | 184 | I422ToRGB24Row_C; |
5670 | 184 | assert(yuvconstants); |
5671 | 184 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
5672 | 0 | return -1; |
5673 | 0 | } |
5674 | | // Negative height means invert the image. |
5675 | 184 | if (height < 0) { |
5676 | 0 | height = -height; |
5677 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
5678 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
5679 | 0 | } |
5680 | 184 | #if defined(HAS_I422TORGB24ROW_SSSE3) |
5681 | 184 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5682 | 184 | I422ToRGB24Row = I422ToRGB24Row_Any_SSSE3; |
5683 | 184 | if (IS_ALIGNED(width, 16)) { |
5684 | 37 | I422ToRGB24Row = I422ToRGB24Row_SSSE3; |
5685 | 37 | } |
5686 | 184 | } |
5687 | 184 | #endif |
5688 | 184 | #if defined(HAS_I422TORGB24ROW_AVX2) |
5689 | 184 | if (TestCpuFlag(kCpuHasAVX2)) { |
5690 | 184 | I422ToRGB24Row = I422ToRGB24Row_Any_AVX2; |
5691 | 184 | if (IS_ALIGNED(width, 32)) { |
5692 | 19 | I422ToRGB24Row = I422ToRGB24Row_AVX2; |
5693 | 19 | } |
5694 | 184 | } |
5695 | 184 | #endif |
5696 | | #if defined(HAS_I422TORGB24ROW_NEON) |
5697 | | if (TestCpuFlag(kCpuHasNEON)) { |
5698 | | I422ToRGB24Row = I422ToRGB24Row_Any_NEON; |
5699 | | if (IS_ALIGNED(width, 8)) { |
5700 | | I422ToRGB24Row = I422ToRGB24Row_NEON; |
5701 | | } |
5702 | | } |
5703 | | #endif |
5704 | | #if defined(HAS_I422TORGB24ROW_SVE2) |
5705 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5706 | | I422ToRGB24Row = I422ToRGB24Row_SVE2; |
5707 | | } |
5708 | | #endif |
5709 | | #if defined(HAS_I422TORGB24ROW_SME) |
5710 | | if (TestCpuFlag(kCpuHasSME)) { |
5711 | | I422ToRGB24Row = I422ToRGB24Row_SME; |
5712 | | } |
5713 | | #endif |
5714 | | #if defined(HAS_I422TORGB24ROW_MSA) |
5715 | | if (TestCpuFlag(kCpuHasMSA)) { |
5716 | | I422ToRGB24Row = I422ToRGB24Row_Any_MSA; |
5717 | | if (IS_ALIGNED(width, 16)) { |
5718 | | I422ToRGB24Row = I422ToRGB24Row_MSA; |
5719 | | } |
5720 | | } |
5721 | | #endif |
5722 | | #if defined(HAS_I422TORGB24ROW_LSX) |
5723 | | if (TestCpuFlag(kCpuHasLSX)) { |
5724 | | I422ToRGB24Row = I422ToRGB24Row_Any_LSX; |
5725 | | if (IS_ALIGNED(width, 16)) { |
5726 | | I422ToRGB24Row = I422ToRGB24Row_LSX; |
5727 | | } |
5728 | | } |
5729 | | #endif |
5730 | | #if defined(HAS_I422TORGB24ROW_LASX) |
5731 | | if (TestCpuFlag(kCpuHasLASX)) { |
5732 | | I422ToRGB24Row = I422ToRGB24Row_Any_LASX; |
5733 | | if (IS_ALIGNED(width, 32)) { |
5734 | | I422ToRGB24Row = I422ToRGB24Row_LASX; |
5735 | | } |
5736 | | } |
5737 | | #endif |
5738 | | #if defined(HAS_I422TORGB24ROW_RVV) |
5739 | | if (TestCpuFlag(kCpuHasRVV)) { |
5740 | | I422ToRGB24Row = I422ToRGB24Row_RVV; |
5741 | | } |
5742 | | #endif |
5743 | | |
5744 | 3.67k | for (y = 0; y < height; ++y) { |
5745 | 3.49k | I422ToRGB24Row(src_y, src_u, src_v, dst_rgb24, yuvconstants, width); |
5746 | 3.49k | dst_rgb24 += dst_stride_rgb24; |
5747 | 3.49k | src_y += src_stride_y; |
5748 | 3.49k | if (y & 1) { |
5749 | 1.71k | src_u += src_stride_u; |
5750 | 1.71k | src_v += src_stride_v; |
5751 | 1.71k | } |
5752 | 3.49k | } |
5753 | 184 | return 0; |
5754 | 184 | } |
5755 | | |
5756 | | // Convert I420 to RGB24. |
5757 | | LIBYUV_API |
5758 | | int I420ToRGB24(const uint8_t* src_y, |
5759 | | int src_stride_y, |
5760 | | const uint8_t* src_u, |
5761 | | int src_stride_u, |
5762 | | const uint8_t* src_v, |
5763 | | int src_stride_v, |
5764 | | uint8_t* dst_rgb24, |
5765 | | int dst_stride_rgb24, |
5766 | | int width, |
5767 | 0 | int height) { |
5768 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5769 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5770 | 0 | &kYuvI601Constants, width, height); |
5771 | 0 | } |
5772 | | |
5773 | | // Convert I420 to RAW. |
5774 | | LIBYUV_API |
5775 | | int I420ToRAW(const uint8_t* src_y, |
5776 | | int src_stride_y, |
5777 | | const uint8_t* src_u, |
5778 | | int src_stride_u, |
5779 | | const uint8_t* src_v, |
5780 | | int src_stride_v, |
5781 | | uint8_t* dst_raw, |
5782 | | int dst_stride_raw, |
5783 | | int width, |
5784 | 0 | int height) { |
5785 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_v, |
5786 | 0 | src_stride_v, // Swap U and V |
5787 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5788 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5789 | 0 | width, height); |
5790 | 0 | } |
5791 | | |
5792 | | // Convert J420 to RGB24. |
5793 | | LIBYUV_API |
5794 | | int J420ToRGB24(const uint8_t* src_y, |
5795 | | int src_stride_y, |
5796 | | const uint8_t* src_u, |
5797 | | int src_stride_u, |
5798 | | const uint8_t* src_v, |
5799 | | int src_stride_v, |
5800 | | uint8_t* dst_rgb24, |
5801 | | int dst_stride_rgb24, |
5802 | | int width, |
5803 | 0 | int height) { |
5804 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5805 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5806 | 0 | &kYuvJPEGConstants, width, height); |
5807 | 0 | } |
5808 | | |
5809 | | // Convert J420 to RAW. |
5810 | | LIBYUV_API |
5811 | | int J420ToRAW(const uint8_t* src_y, |
5812 | | int src_stride_y, |
5813 | | const uint8_t* src_u, |
5814 | | int src_stride_u, |
5815 | | const uint8_t* src_v, |
5816 | | int src_stride_v, |
5817 | | uint8_t* dst_raw, |
5818 | | int dst_stride_raw, |
5819 | | int width, |
5820 | 0 | int height) { |
5821 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_v, |
5822 | 0 | src_stride_v, // Swap U and V |
5823 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5824 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
5825 | 0 | width, height); |
5826 | 0 | } |
5827 | | |
5828 | | // Convert H420 to RGB24. |
5829 | | LIBYUV_API |
5830 | | int H420ToRGB24(const uint8_t* src_y, |
5831 | | int src_stride_y, |
5832 | | const uint8_t* src_u, |
5833 | | int src_stride_u, |
5834 | | const uint8_t* src_v, |
5835 | | int src_stride_v, |
5836 | | uint8_t* dst_rgb24, |
5837 | | int dst_stride_rgb24, |
5838 | | int width, |
5839 | 0 | int height) { |
5840 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5841 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5842 | 0 | &kYuvH709Constants, width, height); |
5843 | 0 | } |
5844 | | |
5845 | | // Convert H420 to RAW. |
5846 | | LIBYUV_API |
5847 | | int H420ToRAW(const uint8_t* src_y, |
5848 | | int src_stride_y, |
5849 | | const uint8_t* src_u, |
5850 | | int src_stride_u, |
5851 | | const uint8_t* src_v, |
5852 | | int src_stride_v, |
5853 | | uint8_t* dst_raw, |
5854 | | int dst_stride_raw, |
5855 | | int width, |
5856 | 0 | int height) { |
5857 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_v, |
5858 | 0 | src_stride_v, // Swap U and V |
5859 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5860 | 0 | &kYvuH709Constants, // Use Yvu matrix |
5861 | 0 | width, height); |
5862 | 0 | } |
5863 | | |
5864 | | // Convert I422 to RGB24 with matrix. |
5865 | | LIBYUV_API |
5866 | | int I422ToRGB24Matrix(const uint8_t* src_y, |
5867 | | int src_stride_y, |
5868 | | const uint8_t* src_u, |
5869 | | int src_stride_u, |
5870 | | const uint8_t* src_v, |
5871 | | int src_stride_v, |
5872 | | uint8_t* dst_rgb24, |
5873 | | int dst_stride_rgb24, |
5874 | | const struct YuvConstants* yuvconstants, |
5875 | | int width, |
5876 | 80 | int height) { |
5877 | 80 | int y; |
5878 | 80 | void (*I422ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
5879 | 80 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5880 | 80 | const struct YuvConstants* yuvconstants, int width) = |
5881 | 80 | I422ToRGB24Row_C; |
5882 | 80 | assert(yuvconstants); |
5883 | 80 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
5884 | 0 | return -1; |
5885 | 0 | } |
5886 | | // Negative height means invert the image. |
5887 | 80 | if (height < 0) { |
5888 | 0 | height = -height; |
5889 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
5890 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
5891 | 0 | } |
5892 | 80 | #if defined(HAS_I422TORGB24ROW_SSSE3) |
5893 | 80 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5894 | 80 | I422ToRGB24Row = I422ToRGB24Row_Any_SSSE3; |
5895 | 80 | if (IS_ALIGNED(width, 16)) { |
5896 | 36 | I422ToRGB24Row = I422ToRGB24Row_SSSE3; |
5897 | 36 | } |
5898 | 80 | } |
5899 | 80 | #endif |
5900 | 80 | #if defined(HAS_I422TORGB24ROW_AVX2) |
5901 | 80 | if (TestCpuFlag(kCpuHasAVX2)) { |
5902 | 80 | I422ToRGB24Row = I422ToRGB24Row_Any_AVX2; |
5903 | 80 | if (IS_ALIGNED(width, 32)) { |
5904 | 19 | I422ToRGB24Row = I422ToRGB24Row_AVX2; |
5905 | 19 | } |
5906 | 80 | } |
5907 | 80 | #endif |
5908 | | #if defined(HAS_I422TORGB24ROW_NEON) |
5909 | | if (TestCpuFlag(kCpuHasNEON)) { |
5910 | | I422ToRGB24Row = I422ToRGB24Row_Any_NEON; |
5911 | | if (IS_ALIGNED(width, 8)) { |
5912 | | I422ToRGB24Row = I422ToRGB24Row_NEON; |
5913 | | } |
5914 | | } |
5915 | | #endif |
5916 | | #if defined(HAS_I422TORGB24ROW_SVE2) |
5917 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5918 | | I422ToRGB24Row = I422ToRGB24Row_SVE2; |
5919 | | } |
5920 | | #endif |
5921 | | #if defined(HAS_I422TORGB24ROW_SME) |
5922 | | if (TestCpuFlag(kCpuHasSME)) { |
5923 | | I422ToRGB24Row = I422ToRGB24Row_SME; |
5924 | | } |
5925 | | #endif |
5926 | | #if defined(HAS_I422TORGB24ROW_MSA) |
5927 | | if (TestCpuFlag(kCpuHasMSA)) { |
5928 | | I422ToRGB24Row = I422ToRGB24Row_Any_MSA; |
5929 | | if (IS_ALIGNED(width, 16)) { |
5930 | | I422ToRGB24Row = I422ToRGB24Row_MSA; |
5931 | | } |
5932 | | } |
5933 | | #endif |
5934 | | #if defined(HAS_I422TORGB24ROW_LSX) |
5935 | | if (TestCpuFlag(kCpuHasLSX)) { |
5936 | | I422ToRGB24Row = I422ToRGB24Row_Any_LSX; |
5937 | | if (IS_ALIGNED(width, 16)) { |
5938 | | I422ToRGB24Row = I422ToRGB24Row_LSX; |
5939 | | } |
5940 | | } |
5941 | | #endif |
5942 | | #if defined(HAS_I422TORGB24ROW_LASX) |
5943 | | if (TestCpuFlag(kCpuHasLASX)) { |
5944 | | I422ToRGB24Row = I422ToRGB24Row_Any_LASX; |
5945 | | if (IS_ALIGNED(width, 32)) { |
5946 | | I422ToRGB24Row = I422ToRGB24Row_LASX; |
5947 | | } |
5948 | | } |
5949 | | #endif |
5950 | | #if defined(HAS_I422TORGB24ROW_RVV) |
5951 | | if (TestCpuFlag(kCpuHasRVV)) { |
5952 | | I422ToRGB24Row = I422ToRGB24Row_RVV; |
5953 | | } |
5954 | | #endif |
5955 | | |
5956 | 1.57k | for (y = 0; y < height; ++y) { |
5957 | 1.49k | I422ToRGB24Row(src_y, src_u, src_v, dst_rgb24, yuvconstants, width); |
5958 | 1.49k | dst_rgb24 += dst_stride_rgb24; |
5959 | 1.49k | src_y += src_stride_y; |
5960 | 1.49k | src_u += src_stride_u; |
5961 | 1.49k | src_v += src_stride_v; |
5962 | 1.49k | } |
5963 | 80 | return 0; |
5964 | 80 | } |
5965 | | |
5966 | | // Convert I422 to RGB24. |
5967 | | LIBYUV_API |
5968 | | int I422ToRGB24(const uint8_t* src_y, |
5969 | | int src_stride_y, |
5970 | | const uint8_t* src_u, |
5971 | | int src_stride_u, |
5972 | | const uint8_t* src_v, |
5973 | | int src_stride_v, |
5974 | | uint8_t* dst_rgb24, |
5975 | | int dst_stride_rgb24, |
5976 | | int width, |
5977 | 0 | int height) { |
5978 | 0 | return I422ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5979 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5980 | 0 | &kYuvI601Constants, width, height); |
5981 | 0 | } |
5982 | | |
5983 | | // Convert I422 to RAW. |
5984 | | LIBYUV_API |
5985 | | int I422ToRAW(const uint8_t* src_y, |
5986 | | int src_stride_y, |
5987 | | const uint8_t* src_u, |
5988 | | int src_stride_u, |
5989 | | const uint8_t* src_v, |
5990 | | int src_stride_v, |
5991 | | uint8_t* dst_raw, |
5992 | | int dst_stride_raw, |
5993 | | int width, |
5994 | 0 | int height) { |
5995 | 0 | return I422ToRGB24Matrix(src_y, src_stride_y, src_v, |
5996 | 0 | src_stride_v, // Swap U and V |
5997 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5998 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5999 | 0 | width, height); |
6000 | 0 | } |
6001 | | |
6002 | | // Convert I420 to ARGB1555. |
6003 | | LIBYUV_API |
6004 | | int I420ToARGB1555(const uint8_t* src_y, |
6005 | | int src_stride_y, |
6006 | | const uint8_t* src_u, |
6007 | | int src_stride_u, |
6008 | | const uint8_t* src_v, |
6009 | | int src_stride_v, |
6010 | | uint8_t* dst_argb1555, |
6011 | | int dst_stride_argb1555, |
6012 | | int width, |
6013 | 0 | int height) { |
6014 | 0 | int y; |
6015 | 0 | void (*I422ToARGB1555Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6016 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6017 | 0 | const struct YuvConstants* yuvconstants, |
6018 | 0 | int width) = I422ToARGB1555Row_C; |
6019 | 0 | if (!src_y || !src_u || !src_v || !dst_argb1555 || width <= 0 || |
6020 | 0 | height == 0) { |
6021 | 0 | return -1; |
6022 | 0 | } |
6023 | | // Negative height means invert the image. |
6024 | 0 | if (height < 0) { |
6025 | 0 | height = -height; |
6026 | 0 | dst_argb1555 = dst_argb1555 + (height - 1) * dst_stride_argb1555; |
6027 | 0 | dst_stride_argb1555 = -dst_stride_argb1555; |
6028 | 0 | } |
6029 | 0 | #if defined(HAS_I422TOARGB1555ROW_SSSE3) |
6030 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6031 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_Any_SSSE3; |
6032 | 0 | if (IS_ALIGNED(width, 8)) { |
6033 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_SSSE3; |
6034 | 0 | } |
6035 | 0 | } |
6036 | 0 | #endif |
6037 | 0 | #if defined(HAS_I422TOARGB1555ROW_AVX2) |
6038 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6039 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_Any_AVX2; |
6040 | 0 | if (IS_ALIGNED(width, 16)) { |
6041 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_AVX2; |
6042 | 0 | } |
6043 | 0 | } |
6044 | 0 | #endif |
6045 | | #if defined(HAS_I422TOARGB1555ROW_NEON) |
6046 | | if (TestCpuFlag(kCpuHasNEON)) { |
6047 | | I422ToARGB1555Row = I422ToARGB1555Row_Any_NEON; |
6048 | | if (IS_ALIGNED(width, 8)) { |
6049 | | I422ToARGB1555Row = I422ToARGB1555Row_NEON; |
6050 | | } |
6051 | | } |
6052 | | #endif |
6053 | | #if defined(HAS_I422TOARGB1555ROW_SVE2) |
6054 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6055 | | I422ToARGB1555Row = I422ToARGB1555Row_SVE2; |
6056 | | } |
6057 | | #endif |
6058 | | #if defined(HAS_I422TOARGB1555ROW_SME) |
6059 | | if (TestCpuFlag(kCpuHasSME)) { |
6060 | | I422ToARGB1555Row = I422ToARGB1555Row_SME; |
6061 | | } |
6062 | | #endif |
6063 | | #if defined(HAS_I422TOARGB1555ROW_MSA) |
6064 | | if (TestCpuFlag(kCpuHasMSA)) { |
6065 | | I422ToARGB1555Row = I422ToARGB1555Row_Any_MSA; |
6066 | | if (IS_ALIGNED(width, 8)) { |
6067 | | I422ToARGB1555Row = I422ToARGB1555Row_MSA; |
6068 | | } |
6069 | | } |
6070 | | #endif |
6071 | | #if defined(HAS_I422TOARGB1555ROW_LSX) |
6072 | | if (TestCpuFlag(kCpuHasLSX)) { |
6073 | | I422ToARGB1555Row = I422ToARGB1555Row_Any_LSX; |
6074 | | if (IS_ALIGNED(width, 16)) { |
6075 | | I422ToARGB1555Row = I422ToARGB1555Row_LSX; |
6076 | | } |
6077 | | } |
6078 | | #endif |
6079 | | #if defined(HAS_I422TOARGB1555ROW_LASX) |
6080 | | if (TestCpuFlag(kCpuHasLASX)) { |
6081 | | I422ToARGB1555Row = I422ToARGB1555Row_Any_LASX; |
6082 | | if (IS_ALIGNED(width, 8)) { |
6083 | | I422ToARGB1555Row = I422ToARGB1555Row_LASX; |
6084 | | } |
6085 | | } |
6086 | | #endif |
6087 | |
|
6088 | 0 | for (y = 0; y < height; ++y) { |
6089 | 0 | I422ToARGB1555Row(src_y, src_u, src_v, dst_argb1555, &kYuvI601Constants, |
6090 | 0 | width); |
6091 | 0 | dst_argb1555 += dst_stride_argb1555; |
6092 | 0 | src_y += src_stride_y; |
6093 | 0 | if (y & 1) { |
6094 | 0 | src_u += src_stride_u; |
6095 | 0 | src_v += src_stride_v; |
6096 | 0 | } |
6097 | 0 | } |
6098 | 0 | return 0; |
6099 | 0 | } |
6100 | | |
6101 | | // Convert I420 to ARGB4444. |
6102 | | LIBYUV_API |
6103 | | int I420ToARGB4444(const uint8_t* src_y, |
6104 | | int src_stride_y, |
6105 | | const uint8_t* src_u, |
6106 | | int src_stride_u, |
6107 | | const uint8_t* src_v, |
6108 | | int src_stride_v, |
6109 | | uint8_t* dst_argb4444, |
6110 | | int dst_stride_argb4444, |
6111 | | int width, |
6112 | 0 | int height) { |
6113 | 0 | int y; |
6114 | 0 | void (*I422ToARGB4444Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6115 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6116 | 0 | const struct YuvConstants* yuvconstants, |
6117 | 0 | int width) = I422ToARGB4444Row_C; |
6118 | 0 | if (!src_y || !src_u || !src_v || !dst_argb4444 || width <= 0 || |
6119 | 0 | height == 0) { |
6120 | 0 | return -1; |
6121 | 0 | } |
6122 | | // Negative height means invert the image. |
6123 | 0 | if (height < 0) { |
6124 | 0 | height = -height; |
6125 | 0 | dst_argb4444 = dst_argb4444 + (height - 1) * dst_stride_argb4444; |
6126 | 0 | dst_stride_argb4444 = -dst_stride_argb4444; |
6127 | 0 | } |
6128 | 0 | #if defined(HAS_I422TOARGB4444ROW_SSSE3) |
6129 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6130 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_Any_SSSE3; |
6131 | 0 | if (IS_ALIGNED(width, 8)) { |
6132 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_SSSE3; |
6133 | 0 | } |
6134 | 0 | } |
6135 | 0 | #endif |
6136 | 0 | #if defined(HAS_I422TOARGB4444ROW_AVX2) |
6137 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6138 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_Any_AVX2; |
6139 | 0 | if (IS_ALIGNED(width, 16)) { |
6140 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_AVX2; |
6141 | 0 | } |
6142 | 0 | } |
6143 | 0 | #endif |
6144 | | #if defined(HAS_I422TOARGB4444ROW_NEON) |
6145 | | if (TestCpuFlag(kCpuHasNEON)) { |
6146 | | I422ToARGB4444Row = I422ToARGB4444Row_Any_NEON; |
6147 | | if (IS_ALIGNED(width, 8)) { |
6148 | | I422ToARGB4444Row = I422ToARGB4444Row_NEON; |
6149 | | } |
6150 | | } |
6151 | | #endif |
6152 | | #if defined(HAS_I422TOARGB4444ROW_SVE2) |
6153 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6154 | | I422ToARGB4444Row = I422ToARGB4444Row_SVE2; |
6155 | | } |
6156 | | #endif |
6157 | | #if defined(HAS_I422TOARGB4444ROW_SME) |
6158 | | if (TestCpuFlag(kCpuHasSME)) { |
6159 | | I422ToARGB4444Row = I422ToARGB4444Row_SME; |
6160 | | } |
6161 | | #endif |
6162 | | #if defined(HAS_I422TOARGB4444ROW_MSA) |
6163 | | if (TestCpuFlag(kCpuHasMSA)) { |
6164 | | I422ToARGB4444Row = I422ToARGB4444Row_Any_MSA; |
6165 | | if (IS_ALIGNED(width, 8)) { |
6166 | | I422ToARGB4444Row = I422ToARGB4444Row_MSA; |
6167 | | } |
6168 | | } |
6169 | | #endif |
6170 | | #if defined(HAS_I422TOARGB4444ROW_LSX) |
6171 | | if (TestCpuFlag(kCpuHasLSX)) { |
6172 | | I422ToARGB4444Row = I422ToARGB4444Row_Any_LSX; |
6173 | | if (IS_ALIGNED(width, 16)) { |
6174 | | I422ToARGB4444Row = I422ToARGB4444Row_LSX; |
6175 | | } |
6176 | | } |
6177 | | #endif |
6178 | | #if defined(HAS_I422TOARGB4444ROW_LASX) |
6179 | | if (TestCpuFlag(kCpuHasLASX)) { |
6180 | | I422ToARGB4444Row = I422ToARGB4444Row_Any_LASX; |
6181 | | if (IS_ALIGNED(width, 8)) { |
6182 | | I422ToARGB4444Row = I422ToARGB4444Row_LASX; |
6183 | | } |
6184 | | } |
6185 | | #endif |
6186 | |
|
6187 | 0 | for (y = 0; y < height; ++y) { |
6188 | 0 | I422ToARGB4444Row(src_y, src_u, src_v, dst_argb4444, &kYuvI601Constants, |
6189 | 0 | width); |
6190 | 0 | dst_argb4444 += dst_stride_argb4444; |
6191 | 0 | src_y += src_stride_y; |
6192 | 0 | if (y & 1) { |
6193 | 0 | src_u += src_stride_u; |
6194 | 0 | src_v += src_stride_v; |
6195 | 0 | } |
6196 | 0 | } |
6197 | 0 | return 0; |
6198 | 0 | } |
6199 | | |
6200 | | // Convert I420 to RGB565 with specified color matrix. |
6201 | | LIBYUV_API |
6202 | | int I420ToRGB565Matrix(const uint8_t* src_y, |
6203 | | int src_stride_y, |
6204 | | const uint8_t* src_u, |
6205 | | int src_stride_u, |
6206 | | const uint8_t* src_v, |
6207 | | int src_stride_v, |
6208 | | uint8_t* dst_rgb565, |
6209 | | int dst_stride_rgb565, |
6210 | | const struct YuvConstants* yuvconstants, |
6211 | | int width, |
6212 | 140 | int height) { |
6213 | 140 | int y; |
6214 | 140 | void (*I422ToRGB565Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6215 | 140 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6216 | 140 | const struct YuvConstants* yuvconstants, int width) = |
6217 | 140 | I422ToRGB565Row_C; |
6218 | 140 | assert(yuvconstants); |
6219 | 140 | if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) { |
6220 | 0 | return -1; |
6221 | 0 | } |
6222 | | // Negative height means invert the image. |
6223 | 140 | if (height < 0) { |
6224 | 0 | height = -height; |
6225 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
6226 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
6227 | 0 | } |
6228 | 140 | #if defined(HAS_I422TORGB565ROW_SSSE3) |
6229 | 140 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6230 | 140 | I422ToRGB565Row = I422ToRGB565Row_Any_SSSE3; |
6231 | 140 | if (IS_ALIGNED(width, 8)) { |
6232 | 34 | I422ToRGB565Row = I422ToRGB565Row_SSSE3; |
6233 | 34 | } |
6234 | 140 | } |
6235 | 140 | #endif |
6236 | 140 | #if defined(HAS_I422TORGB565ROW_AVX2) |
6237 | 140 | if (TestCpuFlag(kCpuHasAVX2)) { |
6238 | 140 | I422ToRGB565Row = I422ToRGB565Row_Any_AVX2; |
6239 | 140 | if (IS_ALIGNED(width, 16)) { |
6240 | 17 | I422ToRGB565Row = I422ToRGB565Row_AVX2; |
6241 | 17 | } |
6242 | 140 | } |
6243 | 140 | #endif |
6244 | | #if defined(HAS_I422TORGB565ROW_NEON) |
6245 | | if (TestCpuFlag(kCpuHasNEON)) { |
6246 | | I422ToRGB565Row = I422ToRGB565Row_Any_NEON; |
6247 | | if (IS_ALIGNED(width, 8)) { |
6248 | | I422ToRGB565Row = I422ToRGB565Row_NEON; |
6249 | | } |
6250 | | } |
6251 | | #endif |
6252 | | #if defined(HAS_I422TORGB565ROW_SVE2) |
6253 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6254 | | I422ToRGB565Row = I422ToRGB565Row_SVE2; |
6255 | | } |
6256 | | #endif |
6257 | | #if defined(HAS_I422TORGB565ROW_SME) |
6258 | | if (TestCpuFlag(kCpuHasSME)) { |
6259 | | I422ToRGB565Row = I422ToRGB565Row_SME; |
6260 | | } |
6261 | | #endif |
6262 | | #if defined(HAS_I422TORGB565ROW_MSA) |
6263 | | if (TestCpuFlag(kCpuHasMSA)) { |
6264 | | I422ToRGB565Row = I422ToRGB565Row_Any_MSA; |
6265 | | if (IS_ALIGNED(width, 8)) { |
6266 | | I422ToRGB565Row = I422ToRGB565Row_MSA; |
6267 | | } |
6268 | | } |
6269 | | #endif |
6270 | | #if defined(HAS_I422TORGB565ROW_LSX) |
6271 | | if (TestCpuFlag(kCpuHasLSX)) { |
6272 | | I422ToRGB565Row = I422ToRGB565Row_Any_LSX; |
6273 | | if (IS_ALIGNED(width, 16)) { |
6274 | | I422ToRGB565Row = I422ToRGB565Row_LSX; |
6275 | | } |
6276 | | } |
6277 | | #endif |
6278 | | #if defined(HAS_I422TORGB565ROW_LASX) |
6279 | | if (TestCpuFlag(kCpuHasLASX)) { |
6280 | | I422ToRGB565Row = I422ToRGB565Row_Any_LASX; |
6281 | | if (IS_ALIGNED(width, 32)) { |
6282 | | I422ToRGB565Row = I422ToRGB565Row_LASX; |
6283 | | } |
6284 | | } |
6285 | | #endif |
6286 | | |
6287 | 2.86k | for (y = 0; y < height; ++y) { |
6288 | 2.72k | I422ToRGB565Row(src_y, src_u, src_v, dst_rgb565, yuvconstants, width); |
6289 | 2.72k | dst_rgb565 += dst_stride_rgb565; |
6290 | 2.72k | src_y += src_stride_y; |
6291 | 2.72k | if (y & 1) { |
6292 | 1.34k | src_u += src_stride_u; |
6293 | 1.34k | src_v += src_stride_v; |
6294 | 1.34k | } |
6295 | 2.72k | } |
6296 | 140 | return 0; |
6297 | 140 | } |
6298 | | |
6299 | | // Convert I420 to RGB565. |
6300 | | LIBYUV_API |
6301 | | int I420ToRGB565(const uint8_t* src_y, |
6302 | | int src_stride_y, |
6303 | | const uint8_t* src_u, |
6304 | | int src_stride_u, |
6305 | | const uint8_t* src_v, |
6306 | | int src_stride_v, |
6307 | | uint8_t* dst_rgb565, |
6308 | | int dst_stride_rgb565, |
6309 | | int width, |
6310 | 0 | int height) { |
6311 | 0 | return I420ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6312 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6313 | 0 | &kYuvI601Constants, width, height); |
6314 | 0 | } |
6315 | | |
6316 | | // Convert J420 to RGB565. |
6317 | | LIBYUV_API |
6318 | | int J420ToRGB565(const uint8_t* src_y, |
6319 | | int src_stride_y, |
6320 | | const uint8_t* src_u, |
6321 | | int src_stride_u, |
6322 | | const uint8_t* src_v, |
6323 | | int src_stride_v, |
6324 | | uint8_t* dst_rgb565, |
6325 | | int dst_stride_rgb565, |
6326 | | int width, |
6327 | 0 | int height) { |
6328 | 0 | return I420ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6329 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6330 | 0 | &kYuvJPEGConstants, width, height); |
6331 | 0 | } |
6332 | | |
6333 | | // Convert H420 to RGB565. |
6334 | | LIBYUV_API |
6335 | | int H420ToRGB565(const uint8_t* src_y, |
6336 | | int src_stride_y, |
6337 | | const uint8_t* src_u, |
6338 | | int src_stride_u, |
6339 | | const uint8_t* src_v, |
6340 | | int src_stride_v, |
6341 | | uint8_t* dst_rgb565, |
6342 | | int dst_stride_rgb565, |
6343 | | int width, |
6344 | 0 | int height) { |
6345 | 0 | return I420ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6346 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6347 | 0 | &kYuvH709Constants, width, height); |
6348 | 0 | } |
6349 | | |
6350 | | // Convert I422 to RGB565 with specified color matrix. |
6351 | | LIBYUV_API |
6352 | | int I422ToRGB565Matrix(const uint8_t* src_y, |
6353 | | int src_stride_y, |
6354 | | const uint8_t* src_u, |
6355 | | int src_stride_u, |
6356 | | const uint8_t* src_v, |
6357 | | int src_stride_v, |
6358 | | uint8_t* dst_rgb565, |
6359 | | int dst_stride_rgb565, |
6360 | | const struct YuvConstants* yuvconstants, |
6361 | | int width, |
6362 | 88 | int height) { |
6363 | 88 | int y; |
6364 | 88 | void (*I422ToRGB565Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6365 | 88 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6366 | 88 | const struct YuvConstants* yuvconstants, int width) = |
6367 | 88 | I422ToRGB565Row_C; |
6368 | 88 | assert(yuvconstants); |
6369 | 88 | if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) { |
6370 | 0 | return -1; |
6371 | 0 | } |
6372 | | // Negative height means invert the image. |
6373 | 88 | if (height < 0) { |
6374 | 0 | height = -height; |
6375 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
6376 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
6377 | 0 | } |
6378 | 88 | #if defined(HAS_I422TORGB565ROW_SSSE3) |
6379 | 88 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6380 | 88 | I422ToRGB565Row = I422ToRGB565Row_Any_SSSE3; |
6381 | 88 | if (IS_ALIGNED(width, 8)) { |
6382 | 34 | I422ToRGB565Row = I422ToRGB565Row_SSSE3; |
6383 | 34 | } |
6384 | 88 | } |
6385 | 88 | #endif |
6386 | 88 | #if defined(HAS_I422TORGB565ROW_AVX2) |
6387 | 88 | if (TestCpuFlag(kCpuHasAVX2)) { |
6388 | 88 | I422ToRGB565Row = I422ToRGB565Row_Any_AVX2; |
6389 | 88 | if (IS_ALIGNED(width, 16)) { |
6390 | 17 | I422ToRGB565Row = I422ToRGB565Row_AVX2; |
6391 | 17 | } |
6392 | 88 | } |
6393 | 88 | #endif |
6394 | | #if defined(HAS_I422TORGB565ROW_NEON) |
6395 | | if (TestCpuFlag(kCpuHasNEON)) { |
6396 | | I422ToRGB565Row = I422ToRGB565Row_Any_NEON; |
6397 | | if (IS_ALIGNED(width, 8)) { |
6398 | | I422ToRGB565Row = I422ToRGB565Row_NEON; |
6399 | | } |
6400 | | } |
6401 | | #endif |
6402 | | #if defined(HAS_I422TORGB565ROW_SVE2) |
6403 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6404 | | I422ToRGB565Row = I422ToRGB565Row_SVE2; |
6405 | | } |
6406 | | #endif |
6407 | | #if defined(HAS_I422TORGB565ROW_SME) |
6408 | | if (TestCpuFlag(kCpuHasSME)) { |
6409 | | I422ToRGB565Row = I422ToRGB565Row_SME; |
6410 | | } |
6411 | | #endif |
6412 | | #if defined(HAS_I422TORGB565ROW_MSA) |
6413 | | if (TestCpuFlag(kCpuHasMSA)) { |
6414 | | I422ToRGB565Row = I422ToRGB565Row_Any_MSA; |
6415 | | if (IS_ALIGNED(width, 8)) { |
6416 | | I422ToRGB565Row = I422ToRGB565Row_MSA; |
6417 | | } |
6418 | | } |
6419 | | #endif |
6420 | | #if defined(HAS_I422TORGB565ROW_LSX) |
6421 | | if (TestCpuFlag(kCpuHasLSX)) { |
6422 | | I422ToRGB565Row = I422ToRGB565Row_Any_LSX; |
6423 | | if (IS_ALIGNED(width, 16)) { |
6424 | | I422ToRGB565Row = I422ToRGB565Row_LSX; |
6425 | | } |
6426 | | } |
6427 | | #endif |
6428 | | #if defined(HAS_I422TORGB565ROW_LASX) |
6429 | | if (TestCpuFlag(kCpuHasLASX)) { |
6430 | | I422ToRGB565Row = I422ToRGB565Row_Any_LASX; |
6431 | | if (IS_ALIGNED(width, 32)) { |
6432 | | I422ToRGB565Row = I422ToRGB565Row_LASX; |
6433 | | } |
6434 | | } |
6435 | | #endif |
6436 | | |
6437 | 1.61k | for (y = 0; y < height; ++y) { |
6438 | 1.52k | I422ToRGB565Row(src_y, src_u, src_v, dst_rgb565, yuvconstants, width); |
6439 | 1.52k | dst_rgb565 += dst_stride_rgb565; |
6440 | 1.52k | src_y += src_stride_y; |
6441 | 1.52k | src_u += src_stride_u; |
6442 | 1.52k | src_v += src_stride_v; |
6443 | 1.52k | } |
6444 | 88 | return 0; |
6445 | 88 | } |
6446 | | |
6447 | | // Convert I422 to RGB565. |
6448 | | LIBYUV_API |
6449 | | int I422ToRGB565(const uint8_t* src_y, |
6450 | | int src_stride_y, |
6451 | | const uint8_t* src_u, |
6452 | | int src_stride_u, |
6453 | | const uint8_t* src_v, |
6454 | | int src_stride_v, |
6455 | | uint8_t* dst_rgb565, |
6456 | | int dst_stride_rgb565, |
6457 | | int width, |
6458 | 0 | int height) { |
6459 | 0 | return I422ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6460 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6461 | 0 | &kYuvI601Constants, width, height); |
6462 | 0 | } |
6463 | | |
6464 | | // Ordered 8x8 dither for 888 to 565. Values from 0 to 7. |
6465 | | static const uint8_t kDither565_4x4[16] = { |
6466 | | 0, 4, 1, 5, 6, 2, 7, 3, 1, 5, 0, 4, 7, 3, 6, 2, |
6467 | | }; |
6468 | | |
6469 | | // Convert I420 to RGB565 with dithering. |
6470 | | LIBYUV_API |
6471 | | int I420ToRGB565Dither(const uint8_t* src_y, |
6472 | | int src_stride_y, |
6473 | | const uint8_t* src_u, |
6474 | | int src_stride_u, |
6475 | | const uint8_t* src_v, |
6476 | | int src_stride_v, |
6477 | | uint8_t* dst_rgb565, |
6478 | | int dst_stride_rgb565, |
6479 | | const uint8_t* dither4x4, |
6480 | | int width, |
6481 | 0 | int height) { |
6482 | 0 | int y; |
6483 | 0 | void (*I422ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
6484 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6485 | 0 | const struct YuvConstants* yuvconstants, int width) = |
6486 | 0 | I422ToARGBRow_C; |
6487 | 0 | void (*ARGBToRGB565DitherRow)(const uint8_t* src_argb, uint8_t* dst_rgb, |
6488 | 0 | uint32_t dither4, int width) = |
6489 | 0 | ARGBToRGB565DitherRow_C; |
6490 | 0 | if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) { |
6491 | 0 | return -1; |
6492 | 0 | } |
6493 | | // Negative height means invert the image. |
6494 | 0 | if (height < 0) { |
6495 | 0 | height = -height; |
6496 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
6497 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
6498 | 0 | } |
6499 | 0 | if (!dither4x4) { |
6500 | 0 | dither4x4 = kDither565_4x4; |
6501 | 0 | } |
6502 | 0 | #if defined(HAS_I422TOARGBROW_SSSE3) |
6503 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6504 | 0 | I422ToARGBRow = I422ToARGBRow_Any_SSSE3; |
6505 | 0 | if (IS_ALIGNED(width, 8)) { |
6506 | 0 | I422ToARGBRow = I422ToARGBRow_SSSE3; |
6507 | 0 | } |
6508 | 0 | } |
6509 | 0 | #endif |
6510 | 0 | #if defined(HAS_I422TOARGBROW_AVX2) |
6511 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6512 | 0 | I422ToARGBRow = I422ToARGBRow_Any_AVX2; |
6513 | 0 | if (IS_ALIGNED(width, 16)) { |
6514 | 0 | I422ToARGBRow = I422ToARGBRow_AVX2; |
6515 | 0 | } |
6516 | 0 | } |
6517 | 0 | #endif |
6518 | 0 | #if defined(HAS_I422TOARGBROW_AVX512BW) |
6519 | 0 | if (TestCpuFlag(kCpuHasAVX512BW | kCpuHasAVX512VL) == |
6520 | 0 | (kCpuHasAVX512BW | kCpuHasAVX512VL)) { |
6521 | 0 | I422ToARGBRow = I422ToARGBRow_Any_AVX512BW; |
6522 | 0 | if (IS_ALIGNED(width, 32)) { |
6523 | 0 | I422ToARGBRow = I422ToARGBRow_AVX512BW; |
6524 | 0 | } |
6525 | 0 | } |
6526 | 0 | #endif |
6527 | | #if defined(HAS_I422TOARGBROW_NEON) |
6528 | | if (TestCpuFlag(kCpuHasNEON)) { |
6529 | | I422ToARGBRow = I422ToARGBRow_Any_NEON; |
6530 | | if (IS_ALIGNED(width, 8)) { |
6531 | | I422ToARGBRow = I422ToARGBRow_NEON; |
6532 | | } |
6533 | | } |
6534 | | #endif |
6535 | | #if defined(HAS_I422TOARGBROW_SVE2) |
6536 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6537 | | I422ToARGBRow = I422ToARGBRow_SVE2; |
6538 | | } |
6539 | | #endif |
6540 | | #if defined(HAS_I422TOARGBROW_SME) |
6541 | | if (TestCpuFlag(kCpuHasSME)) { |
6542 | | I422ToARGBRow = I422ToARGBRow_SME; |
6543 | | } |
6544 | | #endif |
6545 | | #if defined(HAS_I422TOARGBROW_MSA) |
6546 | | if (TestCpuFlag(kCpuHasMSA)) { |
6547 | | I422ToARGBRow = I422ToARGBRow_Any_MSA; |
6548 | | if (IS_ALIGNED(width, 8)) { |
6549 | | I422ToARGBRow = I422ToARGBRow_MSA; |
6550 | | } |
6551 | | } |
6552 | | #endif |
6553 | | #if defined(HAS_I422TOARGBROW_LSX) |
6554 | | if (TestCpuFlag(kCpuHasLSX)) { |
6555 | | I422ToARGBRow = I422ToARGBRow_Any_LSX; |
6556 | | if (IS_ALIGNED(width, 16)) { |
6557 | | I422ToARGBRow = I422ToARGBRow_LSX; |
6558 | | } |
6559 | | } |
6560 | | #endif |
6561 | | #if defined(HAS_I422TOARGBROW_LASX) |
6562 | | if (TestCpuFlag(kCpuHasLASX)) { |
6563 | | I422ToARGBRow = I422ToARGBRow_Any_LASX; |
6564 | | if (IS_ALIGNED(width, 32)) { |
6565 | | I422ToARGBRow = I422ToARGBRow_LASX; |
6566 | | } |
6567 | | } |
6568 | | #endif |
6569 | | #if defined(HAS_I422TOARGBROW_RVV) |
6570 | | if (TestCpuFlag(kCpuHasRVV)) { |
6571 | | I422ToARGBRow = I422ToARGBRow_RVV; |
6572 | | } |
6573 | | #endif |
6574 | 0 | #if defined(HAS_ARGBTORGB565DITHERROW_SSE2) |
6575 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
6576 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_SSE2; |
6577 | 0 | if (IS_ALIGNED(width, 4)) { |
6578 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_SSE2; |
6579 | 0 | } |
6580 | 0 | } |
6581 | 0 | #endif |
6582 | 0 | #if defined(HAS_ARGBTORGB565DITHERROW_AVX2) |
6583 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6584 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_AVX2; |
6585 | 0 | if (IS_ALIGNED(width, 8)) { |
6586 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_AVX2; |
6587 | 0 | } |
6588 | 0 | } |
6589 | 0 | #endif |
6590 | | #if defined(HAS_ARGBTORGB565DITHERROW_NEON) |
6591 | | if (TestCpuFlag(kCpuHasNEON)) { |
6592 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_NEON; |
6593 | | if (IS_ALIGNED(width, 8)) { |
6594 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_NEON; |
6595 | | } |
6596 | | } |
6597 | | #endif |
6598 | | #if defined(HAS_ARGBTORGB565DITHERROW_SVE2) |
6599 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6600 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_SVE2; |
6601 | | } |
6602 | | #endif |
6603 | | #if defined(HAS_ARGBTORGB565DITHERROW_MSA) |
6604 | | if (TestCpuFlag(kCpuHasMSA)) { |
6605 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_MSA; |
6606 | | if (IS_ALIGNED(width, 8)) { |
6607 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_MSA; |
6608 | | } |
6609 | | } |
6610 | | #endif |
6611 | | #if defined(HAS_ARGBTORGB565DITHERROW_LSX) |
6612 | | if (TestCpuFlag(kCpuHasLSX)) { |
6613 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LSX; |
6614 | | if (IS_ALIGNED(width, 8)) { |
6615 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LSX; |
6616 | | } |
6617 | | } |
6618 | | #endif |
6619 | | #if defined(HAS_ARGBTORGB565DITHERROW_LASX) |
6620 | | if (TestCpuFlag(kCpuHasLASX)) { |
6621 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LASX; |
6622 | | if (IS_ALIGNED(width, 16)) { |
6623 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LASX; |
6624 | | } |
6625 | | } |
6626 | | #endif |
6627 | 0 | { |
6628 | | // Allocate a row of argb. |
6629 | 0 | align_buffer_64(row_argb, width * 4); |
6630 | 0 | if (!row_argb) |
6631 | 0 | return 1; |
6632 | 0 | for (y = 0; y < height; ++y) { |
6633 | 0 | I422ToARGBRow(src_y, src_u, src_v, row_argb, &kYuvI601Constants, width); |
6634 | 0 | ARGBToRGB565DitherRow(row_argb, dst_rgb565, |
6635 | 0 | *(const uint32_t*)(dither4x4 + ((y & 3) << 2)), |
6636 | 0 | width); |
6637 | 0 | dst_rgb565 += dst_stride_rgb565; |
6638 | 0 | src_y += src_stride_y; |
6639 | 0 | if (y & 1) { |
6640 | 0 | src_u += src_stride_u; |
6641 | 0 | src_v += src_stride_v; |
6642 | 0 | } |
6643 | 0 | } |
6644 | 0 | free_aligned_buffer_64(row_argb); |
6645 | 0 | } |
6646 | 0 | return 0; |
6647 | 0 | } |
6648 | | |
6649 | | // Convert I420 to AR30 with matrix. |
6650 | | LIBYUV_API |
6651 | | int I420ToAR30Matrix(const uint8_t* src_y, |
6652 | | int src_stride_y, |
6653 | | const uint8_t* src_u, |
6654 | | int src_stride_u, |
6655 | | const uint8_t* src_v, |
6656 | | int src_stride_v, |
6657 | | uint8_t* dst_ar30, |
6658 | | int dst_stride_ar30, |
6659 | | const struct YuvConstants* yuvconstants, |
6660 | | int width, |
6661 | 0 | int height) { |
6662 | 0 | int y; |
6663 | 0 | void (*I422ToAR30Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6664 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6665 | 0 | const struct YuvConstants* yuvconstants, int width) = |
6666 | 0 | I422ToAR30Row_C; |
6667 | |
|
6668 | 0 | assert(yuvconstants); |
6669 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
6670 | 0 | return -1; |
6671 | 0 | } |
6672 | | // Negative height means invert the image. |
6673 | 0 | if (height < 0) { |
6674 | 0 | height = -height; |
6675 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
6676 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
6677 | 0 | } |
6678 | |
|
6679 | 0 | #if defined(HAS_I422TOAR30ROW_SSSE3) |
6680 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6681 | 0 | I422ToAR30Row = I422ToAR30Row_Any_SSSE3; |
6682 | 0 | if (IS_ALIGNED(width, 8)) { |
6683 | 0 | I422ToAR30Row = I422ToAR30Row_SSSE3; |
6684 | 0 | } |
6685 | 0 | } |
6686 | 0 | #endif |
6687 | 0 | #if defined(HAS_I422TOAR30ROW_AVX2) |
6688 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6689 | 0 | I422ToAR30Row = I422ToAR30Row_Any_AVX2; |
6690 | 0 | if (IS_ALIGNED(width, 16)) { |
6691 | 0 | I422ToAR30Row = I422ToAR30Row_AVX2; |
6692 | 0 | } |
6693 | 0 | } |
6694 | 0 | #endif |
6695 | | #if defined(HAS_I422TOAR30ROW_NEON) |
6696 | | if (TestCpuFlag(kCpuHasNEON)) { |
6697 | | I422ToAR30Row = I422ToAR30Row_Any_NEON; |
6698 | | if (IS_ALIGNED(width, 8)) { |
6699 | | I422ToAR30Row = I422ToAR30Row_NEON; |
6700 | | } |
6701 | | } |
6702 | | #endif |
6703 | |
|
6704 | 0 | for (y = 0; y < height; ++y) { |
6705 | 0 | I422ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
6706 | 0 | dst_ar30 += dst_stride_ar30; |
6707 | 0 | src_y += src_stride_y; |
6708 | 0 | if (y & 1) { |
6709 | 0 | src_u += src_stride_u; |
6710 | 0 | src_v += src_stride_v; |
6711 | 0 | } |
6712 | 0 | } |
6713 | 0 | return 0; |
6714 | 0 | } |
6715 | | |
6716 | | // Convert I420 to AR30. |
6717 | | LIBYUV_API |
6718 | | int I420ToAR30(const uint8_t* src_y, |
6719 | | int src_stride_y, |
6720 | | const uint8_t* src_u, |
6721 | | int src_stride_u, |
6722 | | const uint8_t* src_v, |
6723 | | int src_stride_v, |
6724 | | uint8_t* dst_ar30, |
6725 | | int dst_stride_ar30, |
6726 | | int width, |
6727 | 0 | int height) { |
6728 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6729 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
6730 | 0 | &kYuvI601Constants, width, height); |
6731 | 0 | } |
6732 | | |
6733 | | // Convert H420 to AR30. |
6734 | | LIBYUV_API |
6735 | | int H420ToAR30(const uint8_t* src_y, |
6736 | | int src_stride_y, |
6737 | | const uint8_t* src_u, |
6738 | | int src_stride_u, |
6739 | | const uint8_t* src_v, |
6740 | | int src_stride_v, |
6741 | | uint8_t* dst_ar30, |
6742 | | int dst_stride_ar30, |
6743 | | int width, |
6744 | 0 | int height) { |
6745 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6746 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
6747 | 0 | &kYvuH709Constants, width, height); |
6748 | 0 | } |
6749 | | |
6750 | | // Convert I420 to AB30. |
6751 | | LIBYUV_API |
6752 | | int I420ToAB30(const uint8_t* src_y, |
6753 | | int src_stride_y, |
6754 | | const uint8_t* src_u, |
6755 | | int src_stride_u, |
6756 | | const uint8_t* src_v, |
6757 | | int src_stride_v, |
6758 | | uint8_t* dst_ab30, |
6759 | | int dst_stride_ab30, |
6760 | | int width, |
6761 | 0 | int height) { |
6762 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
6763 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
6764 | 0 | &kYvuI601Constants, width, height); |
6765 | 0 | } |
6766 | | |
6767 | | // Convert H420 to AB30. |
6768 | | LIBYUV_API |
6769 | | int H420ToAB30(const uint8_t* src_y, |
6770 | | int src_stride_y, |
6771 | | const uint8_t* src_u, |
6772 | | int src_stride_u, |
6773 | | const uint8_t* src_v, |
6774 | | int src_stride_v, |
6775 | | uint8_t* dst_ab30, |
6776 | | int dst_stride_ab30, |
6777 | | int width, |
6778 | 0 | int height) { |
6779 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
6780 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
6781 | 0 | &kYvuH709Constants, width, height); |
6782 | 0 | } |
6783 | | |
6784 | | static int I420ToARGBMatrixBilinear(const uint8_t* src_y, |
6785 | | int src_stride_y, |
6786 | | const uint8_t* src_u, |
6787 | | int src_stride_u, |
6788 | | const uint8_t* src_v, |
6789 | | int src_stride_v, |
6790 | | uint8_t* dst_argb, |
6791 | | int dst_stride_argb, |
6792 | | const struct YuvConstants* yuvconstants, |
6793 | | int width, |
6794 | 155 | int height) { |
6795 | 155 | int y; |
6796 | 155 | void (*I444ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
6797 | 155 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6798 | 155 | const struct YuvConstants* yuvconstants, int width) = |
6799 | 155 | I444ToARGBRow_C; |
6800 | 155 | void (*Scale2RowUp_Bilinear)(const uint8_t* src_ptr, ptrdiff_t src_stride, |
6801 | 155 | uint8_t* dst_ptr, ptrdiff_t dst_stride, |
6802 | 155 | int dst_width) = ScaleRowUp2_Bilinear_Any_C; |
6803 | 155 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
6804 | 155 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
6805 | 155 | assert(yuvconstants); |
6806 | 155 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
6807 | 0 | return -1; |
6808 | 0 | } |
6809 | | // Negative height means invert the image. |
6810 | 155 | if (height < 0) { |
6811 | 0 | height = -height; |
6812 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
6813 | 0 | dst_stride_argb = -dst_stride_argb; |
6814 | 0 | } |
6815 | 155 | #if defined(HAS_I444TOARGBROW_SSSE3) |
6816 | 155 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6817 | 155 | I444ToARGBRow = I444ToARGBRow_Any_SSSE3; |
6818 | 155 | if (IS_ALIGNED(width, 8)) { |
6819 | 5 | I444ToARGBRow = I444ToARGBRow_SSSE3; |
6820 | 5 | } |
6821 | 155 | } |
6822 | 155 | #endif |
6823 | 155 | #if defined(HAS_I444TOARGBROW_AVX2) |
6824 | 155 | if (TestCpuFlag(kCpuHasAVX2)) { |
6825 | 155 | I444ToARGBRow = I444ToARGBRow_Any_AVX2; |
6826 | 155 | if (IS_ALIGNED(width, 16)) { |
6827 | 3 | I444ToARGBRow = I444ToARGBRow_AVX2; |
6828 | 3 | } |
6829 | 155 | } |
6830 | 155 | #endif |
6831 | | #if defined(HAS_I444TOARGBROW_NEON) |
6832 | | if (TestCpuFlag(kCpuHasNEON)) { |
6833 | | I444ToARGBRow = I444ToARGBRow_Any_NEON; |
6834 | | if (IS_ALIGNED(width, 8)) { |
6835 | | I444ToARGBRow = I444ToARGBRow_NEON; |
6836 | | } |
6837 | | } |
6838 | | #endif |
6839 | | #if defined(HAS_I444TOARGBROW_SVE2) |
6840 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6841 | | I444ToARGBRow = I444ToARGBRow_SVE2; |
6842 | | } |
6843 | | #endif |
6844 | | #if defined(HAS_I444TOARGBROW_SME) |
6845 | | if (TestCpuFlag(kCpuHasSME)) { |
6846 | | I444ToARGBRow = I444ToARGBRow_SME; |
6847 | | } |
6848 | | #endif |
6849 | | #if defined(HAS_I444TOARGBROW_MSA) |
6850 | | if (TestCpuFlag(kCpuHasMSA)) { |
6851 | | I444ToARGBRow = I444ToARGBRow_Any_MSA; |
6852 | | if (IS_ALIGNED(width, 8)) { |
6853 | | I444ToARGBRow = I444ToARGBRow_MSA; |
6854 | | } |
6855 | | } |
6856 | | #endif |
6857 | | #if defined(HAS_I444TOARGBROW_LASX) |
6858 | | if (TestCpuFlag(kCpuHasLASX)) { |
6859 | | I444ToARGBRow = I444ToARGBRow_Any_LASX; |
6860 | | if (IS_ALIGNED(width, 32)) { |
6861 | | I444ToARGBRow = I444ToARGBRow_LASX; |
6862 | | } |
6863 | | } |
6864 | | #endif |
6865 | | #if defined(HAS_I444TOARGBROW_RVV) |
6866 | | if (TestCpuFlag(kCpuHasRVV)) { |
6867 | | I444ToARGBRow = I444ToARGBRow_RVV; |
6868 | | } |
6869 | | #endif |
6870 | | |
6871 | 155 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSE2) |
6872 | 155 | if (TestCpuFlag(kCpuHasSSE2)) { |
6873 | 155 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSE2; |
6874 | 155 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
6875 | 155 | } |
6876 | 155 | #endif |
6877 | | |
6878 | 155 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSSE3) |
6879 | 155 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6880 | 155 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSSE3; |
6881 | 155 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
6882 | 155 | } |
6883 | 155 | #endif |
6884 | | |
6885 | 155 | #if defined(HAS_SCALEROWUP2_BILINEAR_AVX2) |
6886 | 155 | if (TestCpuFlag(kCpuHasAVX2)) { |
6887 | 155 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_AVX2; |
6888 | 155 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
6889 | 155 | } |
6890 | 155 | #endif |
6891 | | |
6892 | | #if defined(HAS_SCALEROWUP2_BILINEAR_NEON) |
6893 | | if (TestCpuFlag(kCpuHasNEON)) { |
6894 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_NEON; |
6895 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
6896 | | } |
6897 | | #endif |
6898 | | #if defined(HAS_SCALEROWUP2_BILINEAR_RVV) |
6899 | | if (TestCpuFlag(kCpuHasRVV)) { |
6900 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_RVV; |
6901 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
6902 | | } |
6903 | | #endif |
6904 | | |
6905 | | // alloc 4 lines temp |
6906 | 155 | const int row_size = (width + 31) & ~31; |
6907 | 155 | align_buffer_64(row, row_size * 4); |
6908 | 155 | uint8_t* temp_u_1 = row; |
6909 | 155 | uint8_t* temp_u_2 = row + row_size; |
6910 | 155 | uint8_t* temp_v_1 = row + row_size * 2; |
6911 | 155 | uint8_t* temp_v_2 = row + row_size * 3; |
6912 | 155 | if (!row) |
6913 | 0 | return 1; |
6914 | | |
6915 | 155 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
6916 | 155 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
6917 | 155 | I444ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
6918 | 155 | dst_argb += dst_stride_argb; |
6919 | 155 | src_y += src_stride_y; |
6920 | | |
6921 | 2.91k | for (y = 0; y < height - 2; y += 2) { |
6922 | 2.76k | Scale2RowUp_Bilinear(src_u, src_stride_u, temp_u_1, row_size, width); |
6923 | 2.76k | Scale2RowUp_Bilinear(src_v, src_stride_v, temp_v_1, row_size, width); |
6924 | 2.76k | I444ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
6925 | 2.76k | dst_argb += dst_stride_argb; |
6926 | 2.76k | src_y += src_stride_y; |
6927 | 2.76k | I444ToARGBRow(src_y, temp_u_2, temp_v_2, dst_argb, yuvconstants, width); |
6928 | 2.76k | dst_argb += dst_stride_argb; |
6929 | 2.76k | src_y += src_stride_y; |
6930 | 2.76k | src_u += src_stride_u; |
6931 | 2.76k | src_v += src_stride_v; |
6932 | 2.76k | } |
6933 | | |
6934 | 155 | if (!(height & 1)) { |
6935 | 38 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
6936 | 38 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
6937 | 38 | I444ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
6938 | 38 | } |
6939 | | |
6940 | 155 | free_aligned_buffer_64(row); |
6941 | 155 | return 0; |
6942 | 155 | } |
6943 | | |
6944 | | static int I422ToARGBMatrixLinear(const uint8_t* src_y, |
6945 | | int src_stride_y, |
6946 | | const uint8_t* src_u, |
6947 | | int src_stride_u, |
6948 | | const uint8_t* src_v, |
6949 | | int src_stride_v, |
6950 | | uint8_t* dst_argb, |
6951 | | int dst_stride_argb, |
6952 | | const struct YuvConstants* yuvconstants, |
6953 | | int width, |
6954 | 156 | int height) { |
6955 | 156 | int y; |
6956 | 156 | void (*I444ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
6957 | 156 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6958 | 156 | const struct YuvConstants* yuvconstants, int width) = |
6959 | 156 | I444ToARGBRow_C; |
6960 | 156 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
6961 | 156 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
6962 | 156 | assert(yuvconstants); |
6963 | 156 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
6964 | 0 | return -1; |
6965 | 0 | } |
6966 | | // Negative height means invert the image. |
6967 | 156 | if (height < 0) { |
6968 | 0 | height = -height; |
6969 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
6970 | 0 | dst_stride_argb = -dst_stride_argb; |
6971 | 0 | } |
6972 | 156 | #if defined(HAS_I444TOARGBROW_SSSE3) |
6973 | 156 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6974 | 156 | I444ToARGBRow = I444ToARGBRow_Any_SSSE3; |
6975 | 156 | if (IS_ALIGNED(width, 8)) { |
6976 | 34 | I444ToARGBRow = I444ToARGBRow_SSSE3; |
6977 | 34 | } |
6978 | 156 | } |
6979 | 156 | #endif |
6980 | 156 | #if defined(HAS_I444TOARGBROW_AVX2) |
6981 | 156 | if (TestCpuFlag(kCpuHasAVX2)) { |
6982 | 156 | I444ToARGBRow = I444ToARGBRow_Any_AVX2; |
6983 | 156 | if (IS_ALIGNED(width, 16)) { |
6984 | 17 | I444ToARGBRow = I444ToARGBRow_AVX2; |
6985 | 17 | } |
6986 | 156 | } |
6987 | 156 | #endif |
6988 | | #if defined(HAS_I444TOARGBROW_NEON) |
6989 | | if (TestCpuFlag(kCpuHasNEON)) { |
6990 | | I444ToARGBRow = I444ToARGBRow_Any_NEON; |
6991 | | if (IS_ALIGNED(width, 8)) { |
6992 | | I444ToARGBRow = I444ToARGBRow_NEON; |
6993 | | } |
6994 | | } |
6995 | | #endif |
6996 | | #if defined(HAS_I444TOARGBROW_SVE2) |
6997 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6998 | | I444ToARGBRow = I444ToARGBRow_SVE2; |
6999 | | } |
7000 | | #endif |
7001 | | #if defined(HAS_I444TOARGBROW_SME) |
7002 | | if (TestCpuFlag(kCpuHasSME)) { |
7003 | | I444ToARGBRow = I444ToARGBRow_SME; |
7004 | | } |
7005 | | #endif |
7006 | | #if defined(HAS_I444TOARGBROW_MSA) |
7007 | | if (TestCpuFlag(kCpuHasMSA)) { |
7008 | | I444ToARGBRow = I444ToARGBRow_Any_MSA; |
7009 | | if (IS_ALIGNED(width, 8)) { |
7010 | | I444ToARGBRow = I444ToARGBRow_MSA; |
7011 | | } |
7012 | | } |
7013 | | #endif |
7014 | | #if defined(HAS_I444TOARGBROW_LASX) |
7015 | | if (TestCpuFlag(kCpuHasLASX)) { |
7016 | | I444ToARGBRow = I444ToARGBRow_Any_LASX; |
7017 | | if (IS_ALIGNED(width, 32)) { |
7018 | | I444ToARGBRow = I444ToARGBRow_LASX; |
7019 | | } |
7020 | | } |
7021 | | #endif |
7022 | | #if defined(HAS_I444TOARGBROW_RVV) |
7023 | | if (TestCpuFlag(kCpuHasRVV)) { |
7024 | | I444ToARGBRow = I444ToARGBRow_RVV; |
7025 | | } |
7026 | | #endif |
7027 | 156 | #if defined(HAS_SCALEROWUP2_LINEAR_SSE2) |
7028 | 156 | if (TestCpuFlag(kCpuHasSSE2)) { |
7029 | 156 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
7030 | 156 | } |
7031 | 156 | #endif |
7032 | 156 | #if defined(HAS_SCALEROWUP2_LINEAR_SSSE3) |
7033 | 156 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7034 | 156 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
7035 | 156 | } |
7036 | 156 | #endif |
7037 | 156 | #if defined(HAS_SCALEROWUP2_LINEAR_AVX2) |
7038 | 156 | if (TestCpuFlag(kCpuHasAVX2)) { |
7039 | 155 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
7040 | 155 | } |
7041 | 156 | #endif |
7042 | | #if defined(HAS_SCALEROWUP2_LINEAR_NEON) |
7043 | | if (TestCpuFlag(kCpuHasNEON)) { |
7044 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
7045 | | } |
7046 | | #endif |
7047 | | #if defined(HAS_SCALEROWUP2_LINEAR_RVV) |
7048 | | if (TestCpuFlag(kCpuHasRVV)) { |
7049 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
7050 | | } |
7051 | | #endif |
7052 | | |
7053 | | // alloc 2 lines temp |
7054 | 156 | const int row_size = (width + 31) & ~31; |
7055 | 156 | align_buffer_64(row, row_size * 2); |
7056 | 156 | uint8_t* temp_u = row; |
7057 | 156 | uint8_t* temp_v = row + row_size; |
7058 | 156 | if (!row) |
7059 | 0 | return 1; |
7060 | | |
7061 | 7.46k | for (y = 0; y < height; ++y) { |
7062 | 7.30k | ScaleRowUp2_Linear(src_u, temp_u, width); |
7063 | 7.30k | ScaleRowUp2_Linear(src_v, temp_v, width); |
7064 | 7.30k | I444ToARGBRow(src_y, temp_u, temp_v, dst_argb, yuvconstants, width); |
7065 | 7.30k | dst_argb += dst_stride_argb; |
7066 | 7.30k | src_y += src_stride_y; |
7067 | 7.30k | src_u += src_stride_u; |
7068 | 7.30k | src_v += src_stride_v; |
7069 | 7.30k | } |
7070 | | |
7071 | 156 | free_aligned_buffer_64(row); |
7072 | 156 | return 0; |
7073 | 156 | } |
7074 | | |
7075 | | static int I420ToRGB24MatrixBilinear(const uint8_t* src_y, |
7076 | | int src_stride_y, |
7077 | | const uint8_t* src_u, |
7078 | | int src_stride_u, |
7079 | | const uint8_t* src_v, |
7080 | | int src_stride_v, |
7081 | | uint8_t* dst_rgb24, |
7082 | | int dst_stride_rgb24, |
7083 | | const struct YuvConstants* yuvconstants, |
7084 | | int width, |
7085 | 88 | int height) { |
7086 | 88 | int y; |
7087 | 88 | void (*I444ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
7088 | 88 | const uint8_t* v_buf, uint8_t* rgb_buf, |
7089 | 88 | const struct YuvConstants* yuvconstants, int width) = |
7090 | 88 | I444ToRGB24Row_C; |
7091 | 88 | void (*Scale2RowUp_Bilinear)(const uint8_t* src_ptr, ptrdiff_t src_stride, |
7092 | 88 | uint8_t* dst_ptr, ptrdiff_t dst_stride, |
7093 | 88 | int dst_width) = ScaleRowUp2_Bilinear_Any_C; |
7094 | 88 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
7095 | 88 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
7096 | 88 | assert(yuvconstants); |
7097 | 88 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
7098 | 0 | return -1; |
7099 | 0 | } |
7100 | | // Negative height means invert the image. |
7101 | 88 | if (height < 0) { |
7102 | 0 | height = -height; |
7103 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
7104 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
7105 | 0 | } |
7106 | 88 | #if defined(HAS_I444TORGB24ROW_SSSE3) |
7107 | 88 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7108 | 88 | I444ToRGB24Row = I444ToRGB24Row_Any_SSSE3; |
7109 | 88 | if (IS_ALIGNED(width, 16)) { |
7110 | 7 | I444ToRGB24Row = I444ToRGB24Row_SSSE3; |
7111 | 7 | } |
7112 | 88 | } |
7113 | 88 | #endif |
7114 | 88 | #if defined(HAS_I444TORGB24ROW_AVX2) |
7115 | 88 | if (TestCpuFlag(kCpuHasAVX2)) { |
7116 | 88 | I444ToRGB24Row = I444ToRGB24Row_Any_AVX2; |
7117 | 88 | if (IS_ALIGNED(width, 32)) { |
7118 | 5 | I444ToRGB24Row = I444ToRGB24Row_AVX2; |
7119 | 5 | } |
7120 | 88 | } |
7121 | 88 | #endif |
7122 | | #if defined(HAS_I444TORGB24ROW_NEON) |
7123 | | if (TestCpuFlag(kCpuHasNEON)) { |
7124 | | I444ToRGB24Row = I444ToRGB24Row_Any_NEON; |
7125 | | if (IS_ALIGNED(width, 8)) { |
7126 | | I444ToRGB24Row = I444ToRGB24Row_NEON; |
7127 | | } |
7128 | | } |
7129 | | #endif |
7130 | | #if defined(HAS_I444TORGB24ROW_MSA) |
7131 | | if (TestCpuFlag(kCpuHasMSA)) { |
7132 | | I444ToRGB24Row = I444ToRGB24Row_Any_MSA; |
7133 | | if (IS_ALIGNED(width, 8)) { |
7134 | | I444ToRGB24Row = I444ToRGB24Row_MSA; |
7135 | | } |
7136 | | } |
7137 | | #endif |
7138 | | #if defined(HAS_I444TORGB24ROW_LASX) |
7139 | | if (TestCpuFlag(kCpuHasLASX)) { |
7140 | | I444ToRGB24Row = I444ToRGB24Row_Any_LASX; |
7141 | | if (IS_ALIGNED(width, 32)) { |
7142 | | I444ToRGB24Row = I444ToRGB24Row_LASX; |
7143 | | } |
7144 | | } |
7145 | | #endif |
7146 | | #if defined(HAS_I444TORGB24ROW_RVV) |
7147 | | if (TestCpuFlag(kCpuHasRVV)) { |
7148 | | I444ToRGB24Row = I444ToRGB24Row_RVV; |
7149 | | } |
7150 | | #endif |
7151 | | |
7152 | 88 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSE2) |
7153 | 88 | if (TestCpuFlag(kCpuHasSSE2)) { |
7154 | 88 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSE2; |
7155 | 88 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
7156 | 88 | } |
7157 | 88 | #endif |
7158 | | |
7159 | 88 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSSE3) |
7160 | 88 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7161 | 88 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSSE3; |
7162 | 88 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
7163 | 88 | } |
7164 | 88 | #endif |
7165 | | |
7166 | 88 | #if defined(HAS_SCALEROWUP2_BILINEAR_AVX2) |
7167 | 88 | if (TestCpuFlag(kCpuHasAVX2)) { |
7168 | 88 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_AVX2; |
7169 | 88 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
7170 | 88 | } |
7171 | 88 | #endif |
7172 | | |
7173 | | #if defined(HAS_SCALEROWUP2_BILINEAR_NEON) |
7174 | | if (TestCpuFlag(kCpuHasNEON)) { |
7175 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_NEON; |
7176 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
7177 | | } |
7178 | | #endif |
7179 | | #if defined(HAS_SCALEROWUP2_BILINEAR_RVV) |
7180 | | if (TestCpuFlag(kCpuHasRVV)) { |
7181 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_RVV; |
7182 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
7183 | | } |
7184 | | #endif |
7185 | | |
7186 | | // alloc 4 lines temp |
7187 | 88 | const int row_size = (width + 31) & ~31; |
7188 | 88 | align_buffer_64(row, row_size * 4); |
7189 | 88 | uint8_t* temp_u_1 = row; |
7190 | 88 | uint8_t* temp_u_2 = row + row_size; |
7191 | 88 | uint8_t* temp_v_1 = row + row_size * 2; |
7192 | 88 | uint8_t* temp_v_2 = row + row_size * 3; |
7193 | 88 | if (!row) |
7194 | 0 | return 1; |
7195 | | |
7196 | 88 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
7197 | 88 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
7198 | 88 | I444ToRGB24Row(src_y, temp_u_1, temp_v_1, dst_rgb24, yuvconstants, width); |
7199 | 88 | dst_rgb24 += dst_stride_rgb24; |
7200 | 88 | src_y += src_stride_y; |
7201 | | |
7202 | 1.81k | for (y = 0; y < height - 2; y += 2) { |
7203 | 1.72k | Scale2RowUp_Bilinear(src_u, src_stride_u, temp_u_1, row_size, width); |
7204 | 1.72k | Scale2RowUp_Bilinear(src_v, src_stride_v, temp_v_1, row_size, width); |
7205 | 1.72k | I444ToRGB24Row(src_y, temp_u_1, temp_v_1, dst_rgb24, yuvconstants, width); |
7206 | 1.72k | dst_rgb24 += dst_stride_rgb24; |
7207 | 1.72k | src_y += src_stride_y; |
7208 | 1.72k | I444ToRGB24Row(src_y, temp_u_2, temp_v_2, dst_rgb24, yuvconstants, width); |
7209 | 1.72k | dst_rgb24 += dst_stride_rgb24; |
7210 | 1.72k | src_y += src_stride_y; |
7211 | 1.72k | src_u += src_stride_u; |
7212 | 1.72k | src_v += src_stride_v; |
7213 | 1.72k | } |
7214 | | |
7215 | 88 | if (!(height & 1)) { |
7216 | 24 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
7217 | 24 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
7218 | 24 | I444ToRGB24Row(src_y, temp_u_1, temp_v_1, dst_rgb24, yuvconstants, width); |
7219 | 24 | } |
7220 | | |
7221 | 88 | free_aligned_buffer_64(row); |
7222 | 88 | return 0; |
7223 | 88 | } |
7224 | | |
7225 | | static int I010ToAR30MatrixBilinear(const uint16_t* src_y, |
7226 | | int src_stride_y, |
7227 | | const uint16_t* src_u, |
7228 | | int src_stride_u, |
7229 | | const uint16_t* src_v, |
7230 | | int src_stride_v, |
7231 | | uint8_t* dst_ar30, |
7232 | | int dst_stride_ar30, |
7233 | | const struct YuvConstants* yuvconstants, |
7234 | | int width, |
7235 | 0 | int height) { |
7236 | 0 | int y; |
7237 | 0 | void (*I410ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
7238 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
7239 | 0 | const struct YuvConstants* yuvconstants, int width) = |
7240 | 0 | I410ToAR30Row_C; |
7241 | 0 | void (*Scale2RowUp_Bilinear_12)( |
7242 | 0 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
7243 | 0 | ptrdiff_t dst_stride, int dst_width) = ScaleRowUp2_Bilinear_16_Any_C; |
7244 | 0 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7245 | 0 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7246 | 0 | assert(yuvconstants); |
7247 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
7248 | 0 | return -1; |
7249 | 0 | } |
7250 | | // Negative height means invert the image. |
7251 | 0 | if (height < 0) { |
7252 | 0 | height = -height; |
7253 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
7254 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
7255 | 0 | } |
7256 | | #if defined(HAS_I410TOAR30ROW_NEON) |
7257 | | if (TestCpuFlag(kCpuHasNEON)) { |
7258 | | I410ToAR30Row = I410ToAR30Row_Any_NEON; |
7259 | | if (IS_ALIGNED(width, 8)) { |
7260 | | I410ToAR30Row = I410ToAR30Row_NEON; |
7261 | | } |
7262 | | } |
7263 | | #endif |
7264 | | #if defined(HAS_I410TOAR30ROW_SVE2) |
7265 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7266 | | I410ToAR30Row = I410ToAR30Row_SVE2; |
7267 | | } |
7268 | | #endif |
7269 | | #if defined(HAS_I410TOAR30ROW_SME) |
7270 | | if (TestCpuFlag(kCpuHasSME)) { |
7271 | | I410ToAR30Row = I410ToAR30Row_SME; |
7272 | | } |
7273 | | #endif |
7274 | 0 | #if defined(HAS_I410TOAR30ROW_SSSE3) |
7275 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7276 | 0 | I410ToAR30Row = I410ToAR30Row_Any_SSSE3; |
7277 | 0 | if (IS_ALIGNED(width, 8)) { |
7278 | 0 | I410ToAR30Row = I410ToAR30Row_SSSE3; |
7279 | 0 | } |
7280 | 0 | } |
7281 | 0 | #endif |
7282 | 0 | #if defined(HAS_I410TOAR30ROW_AVX2) |
7283 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7284 | 0 | I410ToAR30Row = I410ToAR30Row_Any_AVX2; |
7285 | 0 | if (IS_ALIGNED(width, 16)) { |
7286 | 0 | I410ToAR30Row = I410ToAR30Row_AVX2; |
7287 | 0 | } |
7288 | 0 | } |
7289 | 0 | #endif |
7290 | |
|
7291 | 0 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_SSSE3) |
7292 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7293 | 0 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_SSSE3; |
7294 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7295 | 0 | } |
7296 | 0 | #endif |
7297 | |
|
7298 | 0 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_AVX2) |
7299 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7300 | 0 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_AVX2; |
7301 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7302 | 0 | } |
7303 | 0 | #endif |
7304 | |
|
7305 | | #if defined(HAS_SCALEROWUP2_BILINEAR_12_NEON) |
7306 | | if (TestCpuFlag(kCpuHasNEON)) { |
7307 | | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_NEON; |
7308 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7309 | | } |
7310 | | #endif |
7311 | | |
7312 | | // alloc 4 lines temp |
7313 | 0 | const int row_size = (width + 31) & ~31; |
7314 | 0 | align_buffer_64(row, row_size * 4 * sizeof(uint16_t)); |
7315 | 0 | uint16_t* temp_u_1 = (uint16_t*)(row); |
7316 | 0 | uint16_t* temp_u_2 = (uint16_t*)(row) + row_size; |
7317 | 0 | uint16_t* temp_v_1 = (uint16_t*)(row) + row_size * 2; |
7318 | 0 | uint16_t* temp_v_2 = (uint16_t*)(row) + row_size * 3; |
7319 | 0 | if (!row) |
7320 | 0 | return 1; |
7321 | | |
7322 | 0 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7323 | 0 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7324 | 0 | I410ToAR30Row(src_y, temp_u_1, temp_v_1, dst_ar30, yuvconstants, width); |
7325 | 0 | dst_ar30 += dst_stride_ar30; |
7326 | 0 | src_y += src_stride_y; |
7327 | |
|
7328 | 0 | for (y = 0; y < height - 2; y += 2) { |
7329 | 0 | Scale2RowUp_Bilinear_12(src_u, src_stride_u, temp_u_1, row_size, width); |
7330 | 0 | Scale2RowUp_Bilinear_12(src_v, src_stride_v, temp_v_1, row_size, width); |
7331 | 0 | I410ToAR30Row(src_y, temp_u_1, temp_v_1, dst_ar30, yuvconstants, width); |
7332 | 0 | dst_ar30 += dst_stride_ar30; |
7333 | 0 | src_y += src_stride_y; |
7334 | 0 | I410ToAR30Row(src_y, temp_u_2, temp_v_2, dst_ar30, yuvconstants, width); |
7335 | 0 | dst_ar30 += dst_stride_ar30; |
7336 | 0 | src_y += src_stride_y; |
7337 | 0 | src_u += src_stride_u; |
7338 | 0 | src_v += src_stride_v; |
7339 | 0 | } |
7340 | |
|
7341 | 0 | if (!(height & 1)) { |
7342 | 0 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7343 | 0 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7344 | 0 | I410ToAR30Row(src_y, temp_u_1, temp_v_1, dst_ar30, yuvconstants, width); |
7345 | 0 | } |
7346 | |
|
7347 | 0 | free_aligned_buffer_64(row); |
7348 | |
|
7349 | 0 | return 0; |
7350 | 0 | } |
7351 | | |
7352 | | static int I210ToAR30MatrixLinear(const uint16_t* src_y, |
7353 | | int src_stride_y, |
7354 | | const uint16_t* src_u, |
7355 | | int src_stride_u, |
7356 | | const uint16_t* src_v, |
7357 | | int src_stride_v, |
7358 | | uint8_t* dst_ar30, |
7359 | | int dst_stride_ar30, |
7360 | | const struct YuvConstants* yuvconstants, |
7361 | | int width, |
7362 | 0 | int height) { |
7363 | 0 | int y; |
7364 | 0 | void (*I410ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
7365 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
7366 | 0 | const struct YuvConstants* yuvconstants, int width) = |
7367 | 0 | I410ToAR30Row_C; |
7368 | 0 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7369 | 0 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7370 | 0 | assert(yuvconstants); |
7371 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
7372 | 0 | return -1; |
7373 | 0 | } |
7374 | | // Negative height means invert the image. |
7375 | 0 | if (height < 0) { |
7376 | 0 | height = -height; |
7377 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
7378 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
7379 | 0 | } |
7380 | | #if defined(HAS_I410TOAR30ROW_NEON) |
7381 | | if (TestCpuFlag(kCpuHasNEON)) { |
7382 | | I410ToAR30Row = I410ToAR30Row_Any_NEON; |
7383 | | if (IS_ALIGNED(width, 8)) { |
7384 | | I410ToAR30Row = I410ToAR30Row_NEON; |
7385 | | } |
7386 | | } |
7387 | | #endif |
7388 | | #if defined(HAS_I410TOAR30ROW_SVE2) |
7389 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7390 | | I410ToAR30Row = I410ToAR30Row_SVE2; |
7391 | | } |
7392 | | #endif |
7393 | | #if defined(HAS_I410TOAR30ROW_SME) |
7394 | | if (TestCpuFlag(kCpuHasSME)) { |
7395 | | I410ToAR30Row = I410ToAR30Row_SME; |
7396 | | } |
7397 | | #endif |
7398 | 0 | #if defined(HAS_I410TOAR30ROW_SSSE3) |
7399 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7400 | 0 | I410ToAR30Row = I410ToAR30Row_Any_SSSE3; |
7401 | 0 | if (IS_ALIGNED(width, 8)) { |
7402 | 0 | I410ToAR30Row = I410ToAR30Row_SSSE3; |
7403 | 0 | } |
7404 | 0 | } |
7405 | 0 | #endif |
7406 | 0 | #if defined(HAS_I410TOAR30ROW_AVX2) |
7407 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7408 | 0 | I410ToAR30Row = I410ToAR30Row_Any_AVX2; |
7409 | 0 | if (IS_ALIGNED(width, 16)) { |
7410 | 0 | I410ToAR30Row = I410ToAR30Row_AVX2; |
7411 | 0 | } |
7412 | 0 | } |
7413 | 0 | #endif |
7414 | |
|
7415 | 0 | #if defined(HAS_SCALEROWUP2_LINEAR_12_SSSE3) |
7416 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7417 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7418 | 0 | } |
7419 | 0 | #endif |
7420 | 0 | #if defined(HAS_SCALEROWUP2_LINEAR_12_AVX2) |
7421 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7422 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7423 | 0 | } |
7424 | 0 | #endif |
7425 | | #if defined(HAS_SCALEROWUP2_LINEAR_12_NEON) |
7426 | | if (TestCpuFlag(kCpuHasNEON)) { |
7427 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7428 | | } |
7429 | | #endif |
7430 | | |
7431 | | // alloc 2 lines temp |
7432 | 0 | const int row_size = (width + 31) & ~31; |
7433 | 0 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
7434 | 0 | uint16_t* temp_u = (uint16_t*)(row); |
7435 | 0 | uint16_t* temp_v = (uint16_t*)(row) + row_size; |
7436 | 0 | if (!row) |
7437 | 0 | return 1; |
7438 | | |
7439 | 0 | for (y = 0; y < height; ++y) { |
7440 | 0 | ScaleRowUp2_Linear_12(src_u, temp_u, width); |
7441 | 0 | ScaleRowUp2_Linear_12(src_v, temp_v, width); |
7442 | 0 | I410ToAR30Row(src_y, temp_u, temp_v, dst_ar30, yuvconstants, width); |
7443 | 0 | dst_ar30 += dst_stride_ar30; |
7444 | 0 | src_y += src_stride_y; |
7445 | 0 | src_u += src_stride_u; |
7446 | 0 | src_v += src_stride_v; |
7447 | 0 | } |
7448 | 0 | free_aligned_buffer_64(row); |
7449 | 0 | return 0; |
7450 | 0 | } |
7451 | | |
7452 | | static int I010ToARGBMatrixBilinear(const uint16_t* src_y, |
7453 | | int src_stride_y, |
7454 | | const uint16_t* src_u, |
7455 | | int src_stride_u, |
7456 | | const uint16_t* src_v, |
7457 | | int src_stride_v, |
7458 | | uint8_t* dst_argb, |
7459 | | int dst_stride_argb, |
7460 | | const struct YuvConstants* yuvconstants, |
7461 | | int width, |
7462 | 89 | int height) { |
7463 | 89 | int y; |
7464 | 89 | void (*I410ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
7465 | 89 | const uint16_t* v_buf, uint8_t* rgb_buf, |
7466 | 89 | const struct YuvConstants* yuvconstants, int width) = |
7467 | 89 | I410ToARGBRow_C; |
7468 | 89 | void (*Scale2RowUp_Bilinear_12)( |
7469 | 89 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
7470 | 89 | ptrdiff_t dst_stride, int dst_width) = ScaleRowUp2_Bilinear_16_Any_C; |
7471 | 89 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7472 | 89 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7473 | 89 | assert(yuvconstants); |
7474 | 89 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
7475 | 0 | return -1; |
7476 | 0 | } |
7477 | | // Negative height means invert the image. |
7478 | 89 | if (height < 0) { |
7479 | 0 | height = -height; |
7480 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7481 | 0 | dst_stride_argb = -dst_stride_argb; |
7482 | 0 | } |
7483 | 89 | #if defined(HAS_I410TOARGBROW_SSSE3) |
7484 | 89 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7485 | 89 | I410ToARGBRow = I410ToARGBRow_Any_SSSE3; |
7486 | 89 | if (IS_ALIGNED(width, 8)) { |
7487 | 5 | I410ToARGBRow = I410ToARGBRow_SSSE3; |
7488 | 5 | } |
7489 | 89 | } |
7490 | 89 | #endif |
7491 | | #if defined(HAS_I410TOARGBROW_NEON) |
7492 | | if (TestCpuFlag(kCpuHasNEON)) { |
7493 | | I410ToARGBRow = I410ToARGBRow_Any_NEON; |
7494 | | if (IS_ALIGNED(width, 8)) { |
7495 | | I410ToARGBRow = I410ToARGBRow_NEON; |
7496 | | } |
7497 | | } |
7498 | | #endif |
7499 | | #if defined(HAS_I410TOARGBROW_SVE2) |
7500 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7501 | | I410ToARGBRow = I410ToARGBRow_SVE2; |
7502 | | } |
7503 | | #endif |
7504 | | #if defined(HAS_I410TOARGBROW_SME) |
7505 | | if (TestCpuFlag(kCpuHasSME)) { |
7506 | | I410ToARGBRow = I410ToARGBRow_SME; |
7507 | | } |
7508 | | #endif |
7509 | 89 | #if defined(HAS_I410TOARGBROW_AVX2) |
7510 | 89 | if (TestCpuFlag(kCpuHasAVX2)) { |
7511 | 89 | I410ToARGBRow = I410ToARGBRow_Any_AVX2; |
7512 | 89 | if (IS_ALIGNED(width, 16)) { |
7513 | 3 | I410ToARGBRow = I410ToARGBRow_AVX2; |
7514 | 3 | } |
7515 | 89 | } |
7516 | 89 | #endif |
7517 | | |
7518 | 89 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_SSSE3) |
7519 | 89 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7520 | 89 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_SSSE3; |
7521 | 89 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7522 | 89 | } |
7523 | 89 | #endif |
7524 | | |
7525 | 89 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_AVX2) |
7526 | 89 | if (TestCpuFlag(kCpuHasAVX2)) { |
7527 | 89 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_AVX2; |
7528 | 89 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7529 | 89 | } |
7530 | 89 | #endif |
7531 | | |
7532 | | #if defined(HAS_SCALEROWUP2_BILINEAR_12_NEON) |
7533 | | if (TestCpuFlag(kCpuHasNEON)) { |
7534 | | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_NEON; |
7535 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7536 | | } |
7537 | | #endif |
7538 | | |
7539 | | // alloc 4 lines temp |
7540 | 89 | const int row_size = (width + 31) & ~31; |
7541 | 89 | align_buffer_64(row, row_size * 4 * sizeof(uint16_t)); |
7542 | 89 | uint16_t* temp_u_1 = (uint16_t*)(row); |
7543 | 89 | uint16_t* temp_u_2 = (uint16_t*)(row) + row_size; |
7544 | 89 | uint16_t* temp_v_1 = (uint16_t*)(row) + row_size * 2; |
7545 | 89 | uint16_t* temp_v_2 = (uint16_t*)(row) + row_size * 3; |
7546 | 89 | if (!row) |
7547 | 0 | return 1; |
7548 | | |
7549 | 89 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7550 | 89 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7551 | 89 | I410ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
7552 | 89 | dst_argb += dst_stride_argb; |
7553 | 89 | src_y += src_stride_y; |
7554 | | |
7555 | 1.53k | for (y = 0; y < height - 2; y += 2) { |
7556 | 1.44k | Scale2RowUp_Bilinear_12(src_u, src_stride_u, temp_u_1, row_size, width); |
7557 | 1.44k | Scale2RowUp_Bilinear_12(src_v, src_stride_v, temp_v_1, row_size, width); |
7558 | 1.44k | I410ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
7559 | 1.44k | dst_argb += dst_stride_argb; |
7560 | 1.44k | src_y += src_stride_y; |
7561 | 1.44k | I410ToARGBRow(src_y, temp_u_2, temp_v_2, dst_argb, yuvconstants, width); |
7562 | 1.44k | dst_argb += dst_stride_argb; |
7563 | 1.44k | src_y += src_stride_y; |
7564 | 1.44k | src_u += src_stride_u; |
7565 | 1.44k | src_v += src_stride_v; |
7566 | 1.44k | } |
7567 | | |
7568 | 89 | if (!(height & 1)) { |
7569 | 22 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7570 | 22 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7571 | 22 | I410ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
7572 | 22 | } |
7573 | | |
7574 | 89 | free_aligned_buffer_64(row); |
7575 | 89 | return 0; |
7576 | 89 | } |
7577 | | |
7578 | | static int I210ToARGBMatrixLinear(const uint16_t* src_y, |
7579 | | int src_stride_y, |
7580 | | const uint16_t* src_u, |
7581 | | int src_stride_u, |
7582 | | const uint16_t* src_v, |
7583 | | int src_stride_v, |
7584 | | uint8_t* dst_argb, |
7585 | | int dst_stride_argb, |
7586 | | const struct YuvConstants* yuvconstants, |
7587 | | int width, |
7588 | 175 | int height) { |
7589 | 175 | int y; |
7590 | 175 | void (*I410ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
7591 | 175 | const uint16_t* v_buf, uint8_t* rgb_buf, |
7592 | 175 | const struct YuvConstants* yuvconstants, int width) = |
7593 | 175 | I410ToARGBRow_C; |
7594 | 175 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7595 | 175 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7596 | 175 | assert(yuvconstants); |
7597 | 175 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
7598 | 0 | return -1; |
7599 | 0 | } |
7600 | | // Negative height means invert the image. |
7601 | 175 | if (height < 0) { |
7602 | 0 | height = -height; |
7603 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7604 | 0 | dst_stride_argb = -dst_stride_argb; |
7605 | 0 | } |
7606 | 175 | #if defined(HAS_I410TOARGBROW_SSSE3) |
7607 | 175 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7608 | 175 | I410ToARGBRow = I410ToARGBRow_Any_SSSE3; |
7609 | 175 | if (IS_ALIGNED(width, 8)) { |
7610 | 34 | I410ToARGBRow = I410ToARGBRow_SSSE3; |
7611 | 34 | } |
7612 | 175 | } |
7613 | 175 | #endif |
7614 | | #if defined(HAS_I410TOARGBROW_NEON) |
7615 | | if (TestCpuFlag(kCpuHasNEON)) { |
7616 | | I410ToARGBRow = I410ToARGBRow_Any_NEON; |
7617 | | if (IS_ALIGNED(width, 8)) { |
7618 | | I410ToARGBRow = I410ToARGBRow_NEON; |
7619 | | } |
7620 | | } |
7621 | | #endif |
7622 | | #if defined(HAS_I410TOARGBROW_SVE2) |
7623 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7624 | | I410ToARGBRow = I410ToARGBRow_SVE2; |
7625 | | } |
7626 | | #endif |
7627 | | #if defined(HAS_I410TOARGBROW_SME) |
7628 | | if (TestCpuFlag(kCpuHasSME)) { |
7629 | | I410ToARGBRow = I410ToARGBRow_SME; |
7630 | | } |
7631 | | #endif |
7632 | 175 | #if defined(HAS_I410TOARGBROW_AVX2) |
7633 | 175 | if (TestCpuFlag(kCpuHasAVX2)) { |
7634 | 175 | I410ToARGBRow = I410ToARGBRow_Any_AVX2; |
7635 | 175 | if (IS_ALIGNED(width, 16)) { |
7636 | 17 | I410ToARGBRow = I410ToARGBRow_AVX2; |
7637 | 17 | } |
7638 | 175 | } |
7639 | 175 | #endif |
7640 | | |
7641 | 175 | #if defined(HAS_SCALEROWUP2_LINEAR_12_SSSE3) |
7642 | 175 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7643 | 175 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7644 | 175 | } |
7645 | 175 | #endif |
7646 | 175 | #if defined(HAS_SCALEROWUP2_LINEAR_12_AVX2) |
7647 | 175 | if (TestCpuFlag(kCpuHasAVX2)) { |
7648 | 175 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7649 | 175 | } |
7650 | 175 | #endif |
7651 | | #if defined(HAS_SCALEROWUP2_LINEAR_12_NEON) |
7652 | | if (TestCpuFlag(kCpuHasNEON)) { |
7653 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7654 | | } |
7655 | | #endif |
7656 | | |
7657 | | // alloc 2 lines temp |
7658 | 175 | const int row_size = (width + 31) & ~31; |
7659 | 175 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
7660 | 175 | uint16_t* temp_u = (uint16_t*)(row); |
7661 | 175 | uint16_t* temp_v = (uint16_t*)(row) + row_size; |
7662 | 175 | if (!row) |
7663 | 0 | return 1; |
7664 | | |
7665 | 3.57k | for (y = 0; y < height; ++y) { |
7666 | 3.39k | ScaleRowUp2_Linear_12(src_u, temp_u, width); |
7667 | 3.39k | ScaleRowUp2_Linear_12(src_v, temp_v, width); |
7668 | 3.39k | I410ToARGBRow(src_y, temp_u, temp_v, dst_argb, yuvconstants, width); |
7669 | 3.39k | dst_argb += dst_stride_argb; |
7670 | 3.39k | src_y += src_stride_y; |
7671 | 3.39k | src_u += src_stride_u; |
7672 | 3.39k | src_v += src_stride_v; |
7673 | 3.39k | } |
7674 | | |
7675 | 175 | free_aligned_buffer_64(row); |
7676 | 175 | return 0; |
7677 | 175 | } |
7678 | | |
7679 | | static int I420AlphaToARGBMatrixBilinear( |
7680 | | const uint8_t* src_y, |
7681 | | int src_stride_y, |
7682 | | const uint8_t* src_u, |
7683 | | int src_stride_u, |
7684 | | const uint8_t* src_v, |
7685 | | int src_stride_v, |
7686 | | const uint8_t* src_a, |
7687 | | int src_stride_a, |
7688 | | uint8_t* dst_argb, |
7689 | | int dst_stride_argb, |
7690 | | const struct YuvConstants* yuvconstants, |
7691 | | int width, |
7692 | | int height, |
7693 | 48 | int attenuate) { |
7694 | 48 | int y; |
7695 | 48 | void (*I444AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
7696 | 48 | const uint8_t* v_buf, const uint8_t* a_buf, |
7697 | 48 | uint8_t* dst_argb, |
7698 | 48 | const struct YuvConstants* yuvconstants, |
7699 | 48 | int width) = I444AlphaToARGBRow_C; |
7700 | 48 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
7701 | 48 | int width) = ARGBAttenuateRow_C; |
7702 | 48 | void (*Scale2RowUp_Bilinear)(const uint8_t* src_ptr, ptrdiff_t src_stride, |
7703 | 48 | uint8_t* dst_ptr, ptrdiff_t dst_stride, |
7704 | 48 | int dst_width) = ScaleRowUp2_Bilinear_Any_C; |
7705 | 48 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
7706 | 48 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
7707 | 48 | assert(yuvconstants); |
7708 | 48 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
7709 | 48 | height == 0) { |
7710 | 0 | return -1; |
7711 | 0 | } |
7712 | | // Negative height means invert the image. |
7713 | 48 | if (height < 0) { |
7714 | 0 | height = -height; |
7715 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7716 | 0 | dst_stride_argb = -dst_stride_argb; |
7717 | 0 | } |
7718 | 48 | #if defined(HAS_I444ALPHATOARGBROW_SSSE3) |
7719 | 48 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7720 | 48 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_SSSE3; |
7721 | 48 | if (IS_ALIGNED(width, 8)) { |
7722 | 3 | I444AlphaToARGBRow = I444AlphaToARGBRow_SSSE3; |
7723 | 3 | } |
7724 | 48 | } |
7725 | 48 | #endif |
7726 | 48 | #if defined(HAS_I444ALPHATOARGBROW_AVX2) |
7727 | 48 | if (TestCpuFlag(kCpuHasAVX2)) { |
7728 | 48 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_AVX2; |
7729 | 48 | if (IS_ALIGNED(width, 16)) { |
7730 | 2 | I444AlphaToARGBRow = I444AlphaToARGBRow_AVX2; |
7731 | 2 | } |
7732 | 48 | } |
7733 | 48 | #endif |
7734 | | #if defined(HAS_I444ALPHATOARGBROW_NEON) |
7735 | | if (TestCpuFlag(kCpuHasNEON)) { |
7736 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_NEON; |
7737 | | if (IS_ALIGNED(width, 8)) { |
7738 | | I444AlphaToARGBRow = I444AlphaToARGBRow_NEON; |
7739 | | } |
7740 | | } |
7741 | | #endif |
7742 | | #if defined(HAS_I444ALPHATOARGBROW_SVE2) |
7743 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7744 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SVE2; |
7745 | | } |
7746 | | #endif |
7747 | | #if defined(HAS_I444ALPHATOARGBROW_SME) |
7748 | | if (TestCpuFlag(kCpuHasSME)) { |
7749 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SME; |
7750 | | } |
7751 | | #endif |
7752 | | #if defined(HAS_I444ALPHATOARGBROW_MSA) |
7753 | | if (TestCpuFlag(kCpuHasMSA)) { |
7754 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_MSA; |
7755 | | if (IS_ALIGNED(width, 8)) { |
7756 | | I444AlphaToARGBRow = I444AlphaToARGBRow_MSA; |
7757 | | } |
7758 | | } |
7759 | | #endif |
7760 | | #if defined(HAS_I444ALPHATOARGBROW_LASX) |
7761 | | if (TestCpuFlag(kCpuHasLASX)) { |
7762 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_LASX; |
7763 | | if (IS_ALIGNED(width, 16)) { |
7764 | | I444AlphaToARGBRow = I444AlphaToARGBRow_LASX; |
7765 | | } |
7766 | | } |
7767 | | #endif |
7768 | | #if defined(HAS_I444ALPHATOARGBROW_RVV) |
7769 | | if (TestCpuFlag(kCpuHasRVV)) { |
7770 | | I444AlphaToARGBRow = I444AlphaToARGBRow_RVV; |
7771 | | } |
7772 | | #endif |
7773 | 48 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
7774 | 48 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7775 | 48 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
7776 | 48 | if (IS_ALIGNED(width, 4)) { |
7777 | 4 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
7778 | 4 | } |
7779 | 48 | } |
7780 | 48 | #endif |
7781 | 48 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
7782 | 48 | if (TestCpuFlag(kCpuHasAVX2)) { |
7783 | 48 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
7784 | 48 | if (IS_ALIGNED(width, 8)) { |
7785 | 3 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
7786 | 3 | } |
7787 | 48 | } |
7788 | 48 | #endif |
7789 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
7790 | | if (TestCpuFlag(kCpuHasNEON)) { |
7791 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
7792 | | if (IS_ALIGNED(width, 8)) { |
7793 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
7794 | | } |
7795 | | } |
7796 | | #endif |
7797 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
7798 | | if (TestCpuFlag(kCpuHasMSA)) { |
7799 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
7800 | | if (IS_ALIGNED(width, 8)) { |
7801 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
7802 | | } |
7803 | | } |
7804 | | #endif |
7805 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
7806 | | if (TestCpuFlag(kCpuHasRVV)) { |
7807 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
7808 | | } |
7809 | | #endif |
7810 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
7811 | | if (TestCpuFlag(kCpuHasLSX)) { |
7812 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
7813 | | if (IS_ALIGNED(width, 8)) { |
7814 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
7815 | | } |
7816 | | } |
7817 | | #endif |
7818 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
7819 | | if (TestCpuFlag(kCpuHasLASX)) { |
7820 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
7821 | | if (IS_ALIGNED(width, 16)) { |
7822 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
7823 | | } |
7824 | | } |
7825 | | #endif |
7826 | | |
7827 | 48 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSE2) |
7828 | 48 | if (TestCpuFlag(kCpuHasSSE2)) { |
7829 | 48 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSE2; |
7830 | 48 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
7831 | 48 | } |
7832 | 48 | #endif |
7833 | | |
7834 | 48 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSSE3) |
7835 | 48 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7836 | 48 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSSE3; |
7837 | 48 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
7838 | 48 | } |
7839 | 48 | #endif |
7840 | | |
7841 | 48 | #if defined(HAS_SCALEROWUP2_BILINEAR_AVX2) |
7842 | 48 | if (TestCpuFlag(kCpuHasAVX2)) { |
7843 | 48 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_AVX2; |
7844 | 48 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
7845 | 48 | } |
7846 | 48 | #endif |
7847 | | |
7848 | | #if defined(HAS_SCALEROWUP2_BILINEAR_NEON) |
7849 | | if (TestCpuFlag(kCpuHasNEON)) { |
7850 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_NEON; |
7851 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
7852 | | } |
7853 | | #endif |
7854 | | #if defined(HAS_SCALEROWUP2_BILINEAR_RVV) |
7855 | | if (TestCpuFlag(kCpuHasRVV)) { |
7856 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_RVV; |
7857 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
7858 | | } |
7859 | | #endif |
7860 | | |
7861 | | // alloc 4 lines temp |
7862 | 48 | const int row_size = (width + 31) & ~31; |
7863 | 48 | align_buffer_64(row, row_size * 4); |
7864 | 48 | uint8_t* temp_u_1 = row; |
7865 | 48 | uint8_t* temp_u_2 = row + row_size; |
7866 | 48 | uint8_t* temp_v_1 = row + row_size * 2; |
7867 | 48 | uint8_t* temp_v_2 = row + row_size * 3; |
7868 | 48 | if (!row) |
7869 | 0 | return 1; |
7870 | | |
7871 | 48 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
7872 | 48 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
7873 | 48 | I444AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7874 | 48 | width); |
7875 | 48 | if (attenuate) { |
7876 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7877 | 0 | } |
7878 | 48 | dst_argb += dst_stride_argb; |
7879 | 48 | src_y += src_stride_y; |
7880 | 48 | src_a += src_stride_a; |
7881 | | |
7882 | 1.46k | for (y = 0; y < height - 2; y += 2) { |
7883 | 1.41k | Scale2RowUp_Bilinear(src_u, src_stride_u, temp_u_1, row_size, width); |
7884 | 1.41k | Scale2RowUp_Bilinear(src_v, src_stride_v, temp_v_1, row_size, width); |
7885 | 1.41k | I444AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7886 | 1.41k | width); |
7887 | 1.41k | if (attenuate) { |
7888 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7889 | 0 | } |
7890 | 1.41k | dst_argb += dst_stride_argb; |
7891 | 1.41k | src_y += src_stride_y; |
7892 | 1.41k | src_a += src_stride_a; |
7893 | 1.41k | I444AlphaToARGBRow(src_y, temp_u_2, temp_v_2, src_a, dst_argb, yuvconstants, |
7894 | 1.41k | width); |
7895 | 1.41k | if (attenuate) { |
7896 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7897 | 0 | } |
7898 | 1.41k | dst_argb += dst_stride_argb; |
7899 | 1.41k | src_y += src_stride_y; |
7900 | 1.41k | src_u += src_stride_u; |
7901 | 1.41k | src_v += src_stride_v; |
7902 | 1.41k | src_a += src_stride_a; |
7903 | 1.41k | } |
7904 | | |
7905 | 48 | if (!(height & 1)) { |
7906 | 18 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
7907 | 18 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
7908 | 18 | I444AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7909 | 18 | width); |
7910 | 18 | if (attenuate) { |
7911 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7912 | 0 | } |
7913 | 18 | } |
7914 | | |
7915 | 48 | free_aligned_buffer_64(row); |
7916 | 48 | return 0; |
7917 | 48 | } |
7918 | | |
7919 | | static int I422AlphaToARGBMatrixLinear(const uint8_t* src_y, |
7920 | | int src_stride_y, |
7921 | | const uint8_t* src_u, |
7922 | | int src_stride_u, |
7923 | | const uint8_t* src_v, |
7924 | | int src_stride_v, |
7925 | | const uint8_t* src_a, |
7926 | | int src_stride_a, |
7927 | | uint8_t* dst_argb, |
7928 | | int dst_stride_argb, |
7929 | | const struct YuvConstants* yuvconstants, |
7930 | | int width, |
7931 | | int height, |
7932 | 123 | int attenuate) { |
7933 | 123 | int y; |
7934 | 123 | void (*I444AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
7935 | 123 | const uint8_t* v_buf, const uint8_t* a_buf, |
7936 | 123 | uint8_t* dst_argb, |
7937 | 123 | const struct YuvConstants* yuvconstants, |
7938 | 123 | int width) = I444AlphaToARGBRow_C; |
7939 | 123 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
7940 | 123 | int width) = ARGBAttenuateRow_C; |
7941 | 123 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
7942 | 123 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
7943 | 123 | assert(yuvconstants); |
7944 | 123 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
7945 | 123 | height == 0) { |
7946 | 0 | return -1; |
7947 | 0 | } |
7948 | | // Negative height means invert the image. |
7949 | 123 | if (height < 0) { |
7950 | 0 | height = -height; |
7951 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7952 | 0 | dst_stride_argb = -dst_stride_argb; |
7953 | 0 | } |
7954 | 123 | #if defined(HAS_I444ALPHATOARGBROW_SSSE3) |
7955 | 123 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7956 | 123 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_SSSE3; |
7957 | 123 | if (IS_ALIGNED(width, 8)) { |
7958 | 34 | I444AlphaToARGBRow = I444AlphaToARGBRow_SSSE3; |
7959 | 34 | } |
7960 | 123 | } |
7961 | 123 | #endif |
7962 | 123 | #if defined(HAS_I444ALPHATOARGBROW_AVX2) |
7963 | 123 | if (TestCpuFlag(kCpuHasAVX2)) { |
7964 | 123 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_AVX2; |
7965 | 123 | if (IS_ALIGNED(width, 16)) { |
7966 | 17 | I444AlphaToARGBRow = I444AlphaToARGBRow_AVX2; |
7967 | 17 | } |
7968 | 123 | } |
7969 | 123 | #endif |
7970 | | #if defined(HAS_I444ALPHATOARGBROW_NEON) |
7971 | | if (TestCpuFlag(kCpuHasNEON)) { |
7972 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_NEON; |
7973 | | if (IS_ALIGNED(width, 8)) { |
7974 | | I444AlphaToARGBRow = I444AlphaToARGBRow_NEON; |
7975 | | } |
7976 | | } |
7977 | | #endif |
7978 | | #if defined(HAS_I444ALPHATOARGBROW_SVE2) |
7979 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7980 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SVE2; |
7981 | | } |
7982 | | #endif |
7983 | | #if defined(HAS_I444ALPHATOARGBROW_SME) |
7984 | | if (TestCpuFlag(kCpuHasSME)) { |
7985 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SME; |
7986 | | } |
7987 | | #endif |
7988 | | #if defined(HAS_I444ALPHATOARGBROW_MSA) |
7989 | | if (TestCpuFlag(kCpuHasMSA)) { |
7990 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_MSA; |
7991 | | if (IS_ALIGNED(width, 8)) { |
7992 | | I444AlphaToARGBRow = I444AlphaToARGBRow_MSA; |
7993 | | } |
7994 | | } |
7995 | | #endif |
7996 | | #if defined(HAS_I444ALPHATOARGBROW_LASX) |
7997 | | if (TestCpuFlag(kCpuHasLASX)) { |
7998 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_LASX; |
7999 | | if (IS_ALIGNED(width, 16)) { |
8000 | | I444AlphaToARGBRow = I444AlphaToARGBRow_LASX; |
8001 | | } |
8002 | | } |
8003 | | #endif |
8004 | | #if defined(HAS_I444ALPHATOARGBROW_RVV) |
8005 | | if (TestCpuFlag(kCpuHasRVV)) { |
8006 | | I444AlphaToARGBRow = I444AlphaToARGBRow_RVV; |
8007 | | } |
8008 | | #endif |
8009 | 123 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
8010 | 123 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8011 | 123 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
8012 | 123 | if (IS_ALIGNED(width, 4)) { |
8013 | 52 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
8014 | 52 | } |
8015 | 123 | } |
8016 | 123 | #endif |
8017 | 123 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
8018 | 123 | if (TestCpuFlag(kCpuHasAVX2)) { |
8019 | 123 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
8020 | 123 | if (IS_ALIGNED(width, 8)) { |
8021 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
8022 | 34 | } |
8023 | 123 | } |
8024 | 123 | #endif |
8025 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
8026 | | if (TestCpuFlag(kCpuHasNEON)) { |
8027 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
8028 | | if (IS_ALIGNED(width, 8)) { |
8029 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
8030 | | } |
8031 | | } |
8032 | | #endif |
8033 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
8034 | | if (TestCpuFlag(kCpuHasMSA)) { |
8035 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
8036 | | if (IS_ALIGNED(width, 8)) { |
8037 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
8038 | | } |
8039 | | } |
8040 | | #endif |
8041 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
8042 | | if (TestCpuFlag(kCpuHasRVV)) { |
8043 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
8044 | | } |
8045 | | #endif |
8046 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
8047 | | if (TestCpuFlag(kCpuHasLSX)) { |
8048 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
8049 | | if (IS_ALIGNED(width, 8)) { |
8050 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
8051 | | } |
8052 | | } |
8053 | | #endif |
8054 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
8055 | | if (TestCpuFlag(kCpuHasLASX)) { |
8056 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
8057 | | if (IS_ALIGNED(width, 16)) { |
8058 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
8059 | | } |
8060 | | } |
8061 | | #endif |
8062 | | |
8063 | 123 | #if defined(HAS_SCALEROWUP2_LINEAR_SSE2) |
8064 | 123 | if (TestCpuFlag(kCpuHasSSE2)) { |
8065 | 123 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
8066 | 123 | } |
8067 | 123 | #endif |
8068 | 123 | #if defined(HAS_SCALEROWUP2_LINEAR_SSSE3) |
8069 | 123 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8070 | 123 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
8071 | 123 | } |
8072 | 123 | #endif |
8073 | 123 | #if defined(HAS_SCALEROWUP2_LINEAR_AVX2) |
8074 | 123 | if (TestCpuFlag(kCpuHasAVX2)) { |
8075 | 123 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
8076 | 123 | } |
8077 | 123 | #endif |
8078 | | #if defined(HAS_SCALEROWUP2_LINEAR_NEON) |
8079 | | if (TestCpuFlag(kCpuHasNEON)) { |
8080 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
8081 | | } |
8082 | | #endif |
8083 | | #if defined(HAS_SCALEROWUP2_LINEAR_RVV) |
8084 | | if (TestCpuFlag(kCpuHasRVV)) { |
8085 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
8086 | | } |
8087 | | #endif |
8088 | | |
8089 | | // alloc 2 lines temp |
8090 | 123 | const int row_size = (width + 31) & ~31; |
8091 | 123 | align_buffer_64(row, row_size * 2); |
8092 | 123 | uint8_t* temp_u = row; |
8093 | 123 | uint8_t* temp_v = row + row_size; |
8094 | 123 | if (!row) |
8095 | 0 | return 1; |
8096 | | |
8097 | 1.35k | for (y = 0; y < height; ++y) { |
8098 | 1.23k | ScaleRowUp2_Linear(src_u, temp_u, width); |
8099 | 1.23k | ScaleRowUp2_Linear(src_v, temp_v, width); |
8100 | 1.23k | I444AlphaToARGBRow(src_y, temp_u, temp_v, src_a, dst_argb, yuvconstants, |
8101 | 1.23k | width); |
8102 | 1.23k | if (attenuate) { |
8103 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8104 | 0 | } |
8105 | 1.23k | dst_argb += dst_stride_argb; |
8106 | 1.23k | src_a += src_stride_a; |
8107 | 1.23k | src_y += src_stride_y; |
8108 | 1.23k | src_u += src_stride_u; |
8109 | 1.23k | src_v += src_stride_v; |
8110 | 1.23k | } |
8111 | | |
8112 | 123 | free_aligned_buffer_64(row); |
8113 | 123 | return 0; |
8114 | 123 | } |
8115 | | |
8116 | | static int I010AlphaToARGBMatrixBilinear( |
8117 | | const uint16_t* src_y, |
8118 | | int src_stride_y, |
8119 | | const uint16_t* src_u, |
8120 | | int src_stride_u, |
8121 | | const uint16_t* src_v, |
8122 | | int src_stride_v, |
8123 | | const uint16_t* src_a, |
8124 | | int src_stride_a, |
8125 | | uint8_t* dst_argb, |
8126 | | int dst_stride_argb, |
8127 | | const struct YuvConstants* yuvconstants, |
8128 | | int width, |
8129 | | int height, |
8130 | 37 | int attenuate) { |
8131 | 37 | int y; |
8132 | 37 | void (*I410AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
8133 | 37 | const uint16_t* v_buf, const uint16_t* a_buf, |
8134 | 37 | uint8_t* dst_argb, |
8135 | 37 | const struct YuvConstants* yuvconstants, |
8136 | 37 | int width) = I410AlphaToARGBRow_C; |
8137 | 37 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
8138 | 37 | int width) = ARGBAttenuateRow_C; |
8139 | 37 | void (*Scale2RowUp_Bilinear_12)( |
8140 | 37 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
8141 | 37 | ptrdiff_t dst_stride, int dst_width) = ScaleRowUp2_Bilinear_16_Any_C; |
8142 | 37 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
8143 | 37 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
8144 | 37 | assert(yuvconstants); |
8145 | 37 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
8146 | 37 | height == 0) { |
8147 | 0 | return -1; |
8148 | 0 | } |
8149 | | // Negative height means invert the image. |
8150 | 37 | if (height < 0) { |
8151 | 0 | height = -height; |
8152 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
8153 | 0 | dst_stride_argb = -dst_stride_argb; |
8154 | 0 | } |
8155 | | #if defined(HAS_I410ALPHATOARGBROW_NEON) |
8156 | | if (TestCpuFlag(kCpuHasNEON)) { |
8157 | | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_NEON; |
8158 | | if (IS_ALIGNED(width, 8)) { |
8159 | | I410AlphaToARGBRow = I410AlphaToARGBRow_NEON; |
8160 | | } |
8161 | | } |
8162 | | #endif |
8163 | | #if defined(HAS_I410ALPHATOARGBROW_SVE2) |
8164 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8165 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SVE2; |
8166 | | } |
8167 | | #endif |
8168 | | #if defined(HAS_I410ALPHATOARGBROW_SME) |
8169 | | if (TestCpuFlag(kCpuHasSME)) { |
8170 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SME; |
8171 | | } |
8172 | | #endif |
8173 | 37 | #if defined(HAS_I410ALPHATOARGBROW_SSSE3) |
8174 | 37 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8175 | 37 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_SSSE3; |
8176 | 37 | if (IS_ALIGNED(width, 8)) { |
8177 | 6 | I410AlphaToARGBRow = I410AlphaToARGBRow_SSSE3; |
8178 | 6 | } |
8179 | 37 | } |
8180 | 37 | #endif |
8181 | 37 | #if defined(HAS_I410ALPHATOARGBROW_AVX2) |
8182 | 37 | if (TestCpuFlag(kCpuHasAVX2)) { |
8183 | 37 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_AVX2; |
8184 | 37 | if (IS_ALIGNED(width, 16)) { |
8185 | 5 | I410AlphaToARGBRow = I410AlphaToARGBRow_AVX2; |
8186 | 5 | } |
8187 | 37 | } |
8188 | 37 | #endif |
8189 | 37 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
8190 | 37 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8191 | 37 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
8192 | 37 | if (IS_ALIGNED(width, 4)) { |
8193 | 8 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
8194 | 8 | } |
8195 | 37 | } |
8196 | 37 | #endif |
8197 | 37 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
8198 | 37 | if (TestCpuFlag(kCpuHasAVX2)) { |
8199 | 37 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
8200 | 37 | if (IS_ALIGNED(width, 8)) { |
8201 | 6 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
8202 | 6 | } |
8203 | 37 | } |
8204 | 37 | #endif |
8205 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
8206 | | if (TestCpuFlag(kCpuHasNEON)) { |
8207 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
8208 | | if (IS_ALIGNED(width, 8)) { |
8209 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
8210 | | } |
8211 | | } |
8212 | | #endif |
8213 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
8214 | | if (TestCpuFlag(kCpuHasMSA)) { |
8215 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
8216 | | if (IS_ALIGNED(width, 8)) { |
8217 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
8218 | | } |
8219 | | } |
8220 | | #endif |
8221 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
8222 | | if (TestCpuFlag(kCpuHasRVV)) { |
8223 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
8224 | | } |
8225 | | #endif |
8226 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
8227 | | if (TestCpuFlag(kCpuHasLSX)) { |
8228 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
8229 | | if (IS_ALIGNED(width, 8)) { |
8230 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
8231 | | } |
8232 | | } |
8233 | | #endif |
8234 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
8235 | | if (TestCpuFlag(kCpuHasLASX)) { |
8236 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
8237 | | if (IS_ALIGNED(width, 16)) { |
8238 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
8239 | | } |
8240 | | } |
8241 | | #endif |
8242 | | |
8243 | 37 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_SSSE3) |
8244 | 37 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8245 | 37 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_SSSE3; |
8246 | 37 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
8247 | 37 | } |
8248 | 37 | #endif |
8249 | | |
8250 | 37 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_AVX2) |
8251 | 37 | if (TestCpuFlag(kCpuHasAVX2)) { |
8252 | 37 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_AVX2; |
8253 | 37 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
8254 | 37 | } |
8255 | 37 | #endif |
8256 | | |
8257 | | #if defined(HAS_SCALEROWUP2_BILINEAR_12_NEON) |
8258 | | if (TestCpuFlag(kCpuHasNEON)) { |
8259 | | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_NEON; |
8260 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
8261 | | } |
8262 | | #endif |
8263 | | |
8264 | | // alloc 4 lines temp |
8265 | 37 | const int row_size = (width + 31) & ~31; |
8266 | 37 | align_buffer_64(row, row_size * 4 * sizeof(uint16_t)); |
8267 | 37 | uint16_t* temp_u_1 = (uint16_t*)(row); |
8268 | 37 | uint16_t* temp_u_2 = (uint16_t*)(row) + row_size; |
8269 | 37 | uint16_t* temp_v_1 = (uint16_t*)(row) + row_size * 2; |
8270 | 37 | uint16_t* temp_v_2 = (uint16_t*)(row) + row_size * 3; |
8271 | 37 | if (!row) |
8272 | 0 | return 1; |
8273 | | |
8274 | 37 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
8275 | 37 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
8276 | 37 | I410AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
8277 | 37 | width); |
8278 | 37 | if (attenuate) { |
8279 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8280 | 0 | } |
8281 | 37 | dst_argb += dst_stride_argb; |
8282 | 37 | src_y += src_stride_y; |
8283 | 37 | src_a += src_stride_a; |
8284 | | |
8285 | 579 | for (y = 0; y < height - 2; y += 2) { |
8286 | 542 | Scale2RowUp_Bilinear_12(src_u, src_stride_u, temp_u_1, row_size, width); |
8287 | 542 | Scale2RowUp_Bilinear_12(src_v, src_stride_v, temp_v_1, row_size, width); |
8288 | 542 | I410AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
8289 | 542 | width); |
8290 | 542 | if (attenuate) { |
8291 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8292 | 0 | } |
8293 | 542 | dst_argb += dst_stride_argb; |
8294 | 542 | src_y += src_stride_y; |
8295 | 542 | src_a += src_stride_a; |
8296 | 542 | I410AlphaToARGBRow(src_y, temp_u_2, temp_v_2, src_a, dst_argb, yuvconstants, |
8297 | 542 | width); |
8298 | 542 | if (attenuate) { |
8299 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8300 | 0 | } |
8301 | 542 | dst_argb += dst_stride_argb; |
8302 | 542 | src_y += src_stride_y; |
8303 | 542 | src_a += src_stride_a; |
8304 | 542 | src_u += src_stride_u; |
8305 | 542 | src_v += src_stride_v; |
8306 | 542 | } |
8307 | | |
8308 | 37 | if (!(height & 1)) { |
8309 | 14 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
8310 | 14 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
8311 | 14 | I410AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
8312 | 14 | width); |
8313 | 14 | if (attenuate) { |
8314 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8315 | 0 | } |
8316 | 14 | } |
8317 | | |
8318 | 37 | free_aligned_buffer_64(row); |
8319 | 37 | return 0; |
8320 | 37 | } |
8321 | | |
8322 | | static int I210AlphaToARGBMatrixLinear(const uint16_t* src_y, |
8323 | | int src_stride_y, |
8324 | | const uint16_t* src_u, |
8325 | | int src_stride_u, |
8326 | | const uint16_t* src_v, |
8327 | | int src_stride_v, |
8328 | | const uint16_t* src_a, |
8329 | | int src_stride_a, |
8330 | | uint8_t* dst_argb, |
8331 | | int dst_stride_argb, |
8332 | | const struct YuvConstants* yuvconstants, |
8333 | | int width, |
8334 | | int height, |
8335 | 96 | int attenuate) { |
8336 | 96 | int y; |
8337 | 96 | void (*I410AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
8338 | 96 | const uint16_t* v_buf, const uint16_t* a_buf, |
8339 | 96 | uint8_t* dst_argb, |
8340 | 96 | const struct YuvConstants* yuvconstants, |
8341 | 96 | int width) = I410AlphaToARGBRow_C; |
8342 | 96 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
8343 | 96 | int width) = ARGBAttenuateRow_C; |
8344 | 96 | void (*ScaleRowUp2_Linear)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
8345 | 96 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
8346 | 96 | assert(yuvconstants); |
8347 | 96 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
8348 | 96 | height == 0) { |
8349 | 0 | return -1; |
8350 | 0 | } |
8351 | | // Negative height means invert the image. |
8352 | 96 | if (height < 0) { |
8353 | 0 | height = -height; |
8354 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
8355 | 0 | dst_stride_argb = -dst_stride_argb; |
8356 | 0 | } |
8357 | | #if defined(HAS_I410ALPHATOARGBROW_NEON) |
8358 | | if (TestCpuFlag(kCpuHasNEON)) { |
8359 | | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_NEON; |
8360 | | if (IS_ALIGNED(width, 8)) { |
8361 | | I410AlphaToARGBRow = I410AlphaToARGBRow_NEON; |
8362 | | } |
8363 | | } |
8364 | | #endif |
8365 | | #if defined(HAS_I410ALPHATOARGBROW_SVE2) |
8366 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8367 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SVE2; |
8368 | | } |
8369 | | #endif |
8370 | | #if defined(HAS_I410ALPHATOARGBROW_SME) |
8371 | | if (TestCpuFlag(kCpuHasSME)) { |
8372 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SME; |
8373 | | } |
8374 | | #endif |
8375 | 96 | #if defined(HAS_I410ALPHATOARGBROW_SSSE3) |
8376 | 96 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8377 | 96 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_SSSE3; |
8378 | 96 | if (IS_ALIGNED(width, 8)) { |
8379 | 34 | I410AlphaToARGBRow = I410AlphaToARGBRow_SSSE3; |
8380 | 34 | } |
8381 | 96 | } |
8382 | 96 | #endif |
8383 | 96 | #if defined(HAS_I410ALPHATOARGBROW_AVX2) |
8384 | 96 | if (TestCpuFlag(kCpuHasAVX2)) { |
8385 | 96 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_AVX2; |
8386 | 96 | if (IS_ALIGNED(width, 16)) { |
8387 | 17 | I410AlphaToARGBRow = I410AlphaToARGBRow_AVX2; |
8388 | 17 | } |
8389 | 96 | } |
8390 | 96 | #endif |
8391 | 96 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
8392 | 96 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8393 | 96 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
8394 | 96 | if (IS_ALIGNED(width, 4)) { |
8395 | 59 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
8396 | 59 | } |
8397 | 96 | } |
8398 | 96 | #endif |
8399 | 96 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
8400 | 96 | if (TestCpuFlag(kCpuHasAVX2)) { |
8401 | 96 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
8402 | 96 | if (IS_ALIGNED(width, 8)) { |
8403 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
8404 | 34 | } |
8405 | 96 | } |
8406 | 96 | #endif |
8407 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
8408 | | if (TestCpuFlag(kCpuHasNEON)) { |
8409 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
8410 | | if (IS_ALIGNED(width, 8)) { |
8411 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
8412 | | } |
8413 | | } |
8414 | | #endif |
8415 | | #if defined(HAS_ARGBATTENUATEROW_MSA) |
8416 | | if (TestCpuFlag(kCpuHasMSA)) { |
8417 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_MSA; |
8418 | | if (IS_ALIGNED(width, 8)) { |
8419 | | ARGBAttenuateRow = ARGBAttenuateRow_MSA; |
8420 | | } |
8421 | | } |
8422 | | #endif |
8423 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
8424 | | if (TestCpuFlag(kCpuHasRVV)) { |
8425 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
8426 | | } |
8427 | | #endif |
8428 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
8429 | | if (TestCpuFlag(kCpuHasLSX)) { |
8430 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
8431 | | if (IS_ALIGNED(width, 8)) { |
8432 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
8433 | | } |
8434 | | } |
8435 | | #endif |
8436 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
8437 | | if (TestCpuFlag(kCpuHasLASX)) { |
8438 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
8439 | | if (IS_ALIGNED(width, 16)) { |
8440 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
8441 | | } |
8442 | | } |
8443 | | #endif |
8444 | | |
8445 | 96 | #if defined(HAS_SCALEROWUP2_LINEAR_12_SSSE3) |
8446 | 96 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8447 | 96 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_12_Any_SSSE3; |
8448 | 96 | } |
8449 | 96 | #endif |
8450 | 96 | #if defined(HAS_SCALEROWUP2_LINEAR_12_AVX2) |
8451 | 96 | if (TestCpuFlag(kCpuHasAVX2)) { |
8452 | 96 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_12_Any_AVX2; |
8453 | 96 | } |
8454 | 96 | #endif |
8455 | | #if defined(HAS_SCALEROWUP2_LINEAR_12_NEON) |
8456 | | if (TestCpuFlag(kCpuHasNEON)) { |
8457 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_12_Any_NEON; |
8458 | | } |
8459 | | #endif |
8460 | | |
8461 | | // alloc 2 lines temp |
8462 | 96 | const int row_size = (width + 31) & ~31; |
8463 | 96 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
8464 | 96 | uint16_t* temp_u = (uint16_t*)(row); |
8465 | 96 | uint16_t* temp_v = (uint16_t*)(row) + row_size; |
8466 | 96 | if (!row) |
8467 | 0 | return 1; |
8468 | | |
8469 | 1.87k | for (y = 0; y < height; ++y) { |
8470 | 1.78k | ScaleRowUp2_Linear(src_u, temp_u, width); |
8471 | 1.78k | ScaleRowUp2_Linear(src_v, temp_v, width); |
8472 | 1.78k | I410AlphaToARGBRow(src_y, temp_u, temp_v, src_a, dst_argb, yuvconstants, |
8473 | 1.78k | width); |
8474 | 1.78k | if (attenuate) { |
8475 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8476 | 0 | } |
8477 | 1.78k | dst_argb += dst_stride_argb; |
8478 | 1.78k | src_a += src_stride_a; |
8479 | 1.78k | src_y += src_stride_y; |
8480 | 1.78k | src_u += src_stride_u; |
8481 | 1.78k | src_v += src_stride_v; |
8482 | 1.78k | } |
8483 | 96 | free_aligned_buffer_64(row); |
8484 | 96 | return 0; |
8485 | 96 | } |
8486 | | |
8487 | | static int P010ToARGBMatrixBilinear(const uint16_t* src_y, |
8488 | | int src_stride_y, |
8489 | | const uint16_t* src_uv, |
8490 | | int src_stride_uv, |
8491 | | uint8_t* dst_argb, |
8492 | | int dst_stride_argb, |
8493 | | const struct YuvConstants* yuvconstants, |
8494 | | int width, |
8495 | 0 | int height) { |
8496 | 0 | int y; |
8497 | 0 | void (*P410ToARGBRow)( |
8498 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8499 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToARGBRow_C; |
8500 | 0 | void (*Scale2RowUp_Bilinear_16)( |
8501 | 0 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
8502 | 0 | ptrdiff_t dst_stride, int dst_width) = ScaleUVRowUp2_Bilinear_16_Any_C; |
8503 | 0 | assert(yuvconstants); |
8504 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
8505 | 0 | return -1; |
8506 | 0 | } |
8507 | | // Negative height means invert the image. |
8508 | 0 | if (height < 0) { |
8509 | 0 | height = -height; |
8510 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
8511 | 0 | dst_stride_argb = -dst_stride_argb; |
8512 | 0 | } |
8513 | 0 | #if defined(HAS_P410TOARGBROW_SSSE3) |
8514 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8515 | 0 | P410ToARGBRow = P410ToARGBRow_Any_SSSE3; |
8516 | 0 | if (IS_ALIGNED(width, 8)) { |
8517 | 0 | P410ToARGBRow = P410ToARGBRow_SSSE3; |
8518 | 0 | } |
8519 | 0 | } |
8520 | 0 | #endif |
8521 | 0 | #if defined(HAS_P410TOARGBROW_AVX2) |
8522 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8523 | 0 | P410ToARGBRow = P410ToARGBRow_Any_AVX2; |
8524 | 0 | if (IS_ALIGNED(width, 16)) { |
8525 | 0 | P410ToARGBRow = P410ToARGBRow_AVX2; |
8526 | 0 | } |
8527 | 0 | } |
8528 | 0 | #endif |
8529 | | #if defined(HAS_P410TOARGBROW_NEON) |
8530 | | if (TestCpuFlag(kCpuHasNEON)) { |
8531 | | P410ToARGBRow = P410ToARGBRow_Any_NEON; |
8532 | | if (IS_ALIGNED(width, 8)) { |
8533 | | P410ToARGBRow = P410ToARGBRow_NEON; |
8534 | | } |
8535 | | } |
8536 | | #endif |
8537 | | #if defined(HAS_P410TOARGBROW_SVE2) |
8538 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8539 | | P410ToARGBRow = P410ToARGBRow_SVE2; |
8540 | | } |
8541 | | #endif |
8542 | | #if defined(HAS_P410TOARGBROW_SME) |
8543 | | if (TestCpuFlag(kCpuHasSME)) { |
8544 | | P410ToARGBRow = P410ToARGBRow_SME; |
8545 | | } |
8546 | | #endif |
8547 | |
|
8548 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_SSE41 |
8549 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8550 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_SSE41; |
8551 | 0 | } |
8552 | 0 | #endif |
8553 | |
|
8554 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_AVX2 |
8555 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8556 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_AVX2; |
8557 | 0 | } |
8558 | 0 | #endif |
8559 | |
|
8560 | | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_NEON |
8561 | | if (TestCpuFlag(kCpuHasNEON)) { |
8562 | | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_NEON; |
8563 | | } |
8564 | | #endif |
8565 | | |
8566 | | // alloc 2 lines temp |
8567 | 0 | const int row_size = (2 * width + 31) & ~31; |
8568 | 0 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
8569 | 0 | uint16_t* temp_uv_1 = (uint16_t*)(row); |
8570 | 0 | uint16_t* temp_uv_2 = (uint16_t*)(row) + row_size; |
8571 | 0 | if (!row) |
8572 | 0 | return 1; |
8573 | | |
8574 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8575 | 0 | P410ToARGBRow(src_y, temp_uv_1, dst_argb, yuvconstants, width); |
8576 | 0 | dst_argb += dst_stride_argb; |
8577 | 0 | src_y += src_stride_y; |
8578 | |
|
8579 | 0 | for (y = 0; y < height - 2; y += 2) { |
8580 | 0 | Scale2RowUp_Bilinear_16(src_uv, src_stride_uv, temp_uv_1, row_size, width); |
8581 | 0 | P410ToARGBRow(src_y, temp_uv_1, dst_argb, yuvconstants, width); |
8582 | 0 | dst_argb += dst_stride_argb; |
8583 | 0 | src_y += src_stride_y; |
8584 | 0 | P410ToARGBRow(src_y, temp_uv_2, dst_argb, yuvconstants, width); |
8585 | 0 | dst_argb += dst_stride_argb; |
8586 | 0 | src_y += src_stride_y; |
8587 | 0 | src_uv += src_stride_uv; |
8588 | 0 | } |
8589 | |
|
8590 | 0 | if (!(height & 1)) { |
8591 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8592 | 0 | P410ToARGBRow(src_y, temp_uv_1, dst_argb, yuvconstants, width); |
8593 | 0 | } |
8594 | |
|
8595 | 0 | free_aligned_buffer_64(row); |
8596 | 0 | return 0; |
8597 | 0 | } |
8598 | | |
8599 | | static int P210ToARGBMatrixLinear(const uint16_t* src_y, |
8600 | | int src_stride_y, |
8601 | | const uint16_t* src_uv, |
8602 | | int src_stride_uv, |
8603 | | uint8_t* dst_argb, |
8604 | | int dst_stride_argb, |
8605 | | const struct YuvConstants* yuvconstants, |
8606 | | int width, |
8607 | 0 | int height) { |
8608 | 0 | int y; |
8609 | 0 | void (*P410ToARGBRow)( |
8610 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8611 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToARGBRow_C; |
8612 | 0 | void (*ScaleRowUp2_Linear)(const uint16_t* src_uv, uint16_t* dst_uv, |
8613 | 0 | int dst_width) = ScaleUVRowUp2_Linear_16_Any_C; |
8614 | 0 | assert(yuvconstants); |
8615 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
8616 | 0 | return -1; |
8617 | 0 | } |
8618 | | // Negative height means invert the image. |
8619 | 0 | if (height < 0) { |
8620 | 0 | height = -height; |
8621 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
8622 | 0 | dst_stride_argb = -dst_stride_argb; |
8623 | 0 | } |
8624 | 0 | #if defined(HAS_P410TOARGBROW_SSSE3) |
8625 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8626 | 0 | P410ToARGBRow = P410ToARGBRow_Any_SSSE3; |
8627 | 0 | if (IS_ALIGNED(width, 8)) { |
8628 | 0 | P410ToARGBRow = P410ToARGBRow_SSSE3; |
8629 | 0 | } |
8630 | 0 | } |
8631 | 0 | #endif |
8632 | 0 | #if defined(HAS_P410TOARGBROW_AVX2) |
8633 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8634 | 0 | P410ToARGBRow = P410ToARGBRow_Any_AVX2; |
8635 | 0 | if (IS_ALIGNED(width, 16)) { |
8636 | 0 | P410ToARGBRow = P410ToARGBRow_AVX2; |
8637 | 0 | } |
8638 | 0 | } |
8639 | 0 | #endif |
8640 | | #if defined(HAS_P410TOARGBROW_NEON) |
8641 | | if (TestCpuFlag(kCpuHasNEON)) { |
8642 | | P410ToARGBRow = P410ToARGBRow_Any_NEON; |
8643 | | if (IS_ALIGNED(width, 8)) { |
8644 | | P410ToARGBRow = P410ToARGBRow_NEON; |
8645 | | } |
8646 | | } |
8647 | | #endif |
8648 | | #if defined(HAS_P410TOARGBROW_SVE2) |
8649 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8650 | | P410ToARGBRow = P410ToARGBRow_SVE2; |
8651 | | } |
8652 | | #endif |
8653 | | #if defined(HAS_P410TOARGBROW_SME) |
8654 | | if (TestCpuFlag(kCpuHasSME)) { |
8655 | | P410ToARGBRow = P410ToARGBRow_SME; |
8656 | | } |
8657 | | #endif |
8658 | |
|
8659 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_SSE41 |
8660 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8661 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_SSE41; |
8662 | 0 | } |
8663 | 0 | #endif |
8664 | |
|
8665 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_AVX2 |
8666 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8667 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_AVX2; |
8668 | 0 | } |
8669 | 0 | #endif |
8670 | |
|
8671 | | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_NEON |
8672 | | if (TestCpuFlag(kCpuHasNEON)) { |
8673 | | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_NEON; |
8674 | | } |
8675 | | #endif |
8676 | |
|
8677 | 0 | const int row_size = (2 * width + 31) & ~31; |
8678 | 0 | align_buffer_64(row, row_size * sizeof(uint16_t)); |
8679 | 0 | uint16_t* temp_uv = (uint16_t*)(row); |
8680 | 0 | if (!row) |
8681 | 0 | return 1; |
8682 | | |
8683 | 0 | for (y = 0; y < height; ++y) { |
8684 | 0 | ScaleRowUp2_Linear(src_uv, temp_uv, width); |
8685 | 0 | P410ToARGBRow(src_y, temp_uv, dst_argb, yuvconstants, width); |
8686 | 0 | dst_argb += dst_stride_argb; |
8687 | 0 | src_y += src_stride_y; |
8688 | 0 | src_uv += src_stride_uv; |
8689 | 0 | } |
8690 | |
|
8691 | 0 | free_aligned_buffer_64(row); |
8692 | 0 | return 0; |
8693 | 0 | } |
8694 | | |
8695 | | static int P010ToAR30MatrixBilinear(const uint16_t* src_y, |
8696 | | int src_stride_y, |
8697 | | const uint16_t* src_uv, |
8698 | | int src_stride_uv, |
8699 | | uint8_t* dst_ar30, |
8700 | | int dst_stride_ar30, |
8701 | | const struct YuvConstants* yuvconstants, |
8702 | | int width, |
8703 | 0 | int height) { |
8704 | 0 | int y; |
8705 | 0 | void (*P410ToAR30Row)( |
8706 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8707 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToAR30Row_C; |
8708 | 0 | void (*Scale2RowUp_Bilinear_16)( |
8709 | 0 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
8710 | 0 | ptrdiff_t dst_stride, int dst_width) = ScaleUVRowUp2_Bilinear_16_Any_C; |
8711 | 0 | assert(yuvconstants); |
8712 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
8713 | 0 | return -1; |
8714 | 0 | } |
8715 | | // Negative height means invert the image. |
8716 | 0 | if (height < 0) { |
8717 | 0 | height = -height; |
8718 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
8719 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
8720 | 0 | } |
8721 | 0 | #if defined(HAS_P410TOAR30ROW_SSSE3) |
8722 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8723 | 0 | P410ToAR30Row = P410ToAR30Row_Any_SSSE3; |
8724 | 0 | if (IS_ALIGNED(width, 8)) { |
8725 | 0 | P410ToAR30Row = P410ToAR30Row_SSSE3; |
8726 | 0 | } |
8727 | 0 | } |
8728 | 0 | #endif |
8729 | 0 | #if defined(HAS_P410TOAR30ROW_AVX2) |
8730 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8731 | 0 | P410ToAR30Row = P410ToAR30Row_Any_AVX2; |
8732 | 0 | if (IS_ALIGNED(width, 16)) { |
8733 | 0 | P410ToAR30Row = P410ToAR30Row_AVX2; |
8734 | 0 | } |
8735 | 0 | } |
8736 | 0 | #endif |
8737 | | #if defined(HAS_P410TOAR30ROW_NEON) |
8738 | | if (TestCpuFlag(kCpuHasNEON)) { |
8739 | | P410ToAR30Row = P410ToAR30Row_Any_NEON; |
8740 | | if (IS_ALIGNED(width, 8)) { |
8741 | | P410ToAR30Row = P410ToAR30Row_NEON; |
8742 | | } |
8743 | | } |
8744 | | #endif |
8745 | | #if defined(HAS_P410TOAR30ROW_SVE2) |
8746 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8747 | | P410ToAR30Row = P410ToAR30Row_SVE2; |
8748 | | } |
8749 | | #endif |
8750 | | #if defined(HAS_P410TOAR30ROW_SME) |
8751 | | if (TestCpuFlag(kCpuHasSME)) { |
8752 | | P410ToAR30Row = P410ToAR30Row_SME; |
8753 | | } |
8754 | | #endif |
8755 | |
|
8756 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_SSE41 |
8757 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8758 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_SSE41; |
8759 | 0 | } |
8760 | 0 | #endif |
8761 | |
|
8762 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_AVX2 |
8763 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8764 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_AVX2; |
8765 | 0 | } |
8766 | 0 | #endif |
8767 | |
|
8768 | | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_NEON |
8769 | | if (TestCpuFlag(kCpuHasNEON)) { |
8770 | | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_NEON; |
8771 | | } |
8772 | | #endif |
8773 | | |
8774 | | // alloc 2 lines temp |
8775 | 0 | const int row_size = (2 * width + 31) & ~31; |
8776 | 0 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
8777 | 0 | uint16_t* temp_uv_1 = (uint16_t*)(row); |
8778 | 0 | uint16_t* temp_uv_2 = (uint16_t*)(row) + row_size; |
8779 | 0 | if (!row) |
8780 | 0 | return 1; |
8781 | | |
8782 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8783 | 0 | P410ToAR30Row(src_y, temp_uv_1, dst_ar30, yuvconstants, width); |
8784 | 0 | dst_ar30 += dst_stride_ar30; |
8785 | 0 | src_y += src_stride_y; |
8786 | |
|
8787 | 0 | for (y = 0; y < height - 2; y += 2) { |
8788 | 0 | Scale2RowUp_Bilinear_16(src_uv, src_stride_uv, temp_uv_1, row_size, width); |
8789 | 0 | P410ToAR30Row(src_y, temp_uv_1, dst_ar30, yuvconstants, width); |
8790 | 0 | dst_ar30 += dst_stride_ar30; |
8791 | 0 | src_y += src_stride_y; |
8792 | 0 | P410ToAR30Row(src_y, temp_uv_2, dst_ar30, yuvconstants, width); |
8793 | 0 | dst_ar30 += dst_stride_ar30; |
8794 | 0 | src_y += src_stride_y; |
8795 | 0 | src_uv += src_stride_uv; |
8796 | 0 | } |
8797 | |
|
8798 | 0 | if (!(height & 1)) { |
8799 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8800 | 0 | P410ToAR30Row(src_y, temp_uv_1, dst_ar30, yuvconstants, width); |
8801 | 0 | } |
8802 | |
|
8803 | 0 | free_aligned_buffer_64(row); |
8804 | 0 | return 0; |
8805 | 0 | } |
8806 | | |
8807 | | static int P210ToAR30MatrixLinear(const uint16_t* src_y, |
8808 | | int src_stride_y, |
8809 | | const uint16_t* src_uv, |
8810 | | int src_stride_uv, |
8811 | | uint8_t* dst_ar30, |
8812 | | int dst_stride_ar30, |
8813 | | const struct YuvConstants* yuvconstants, |
8814 | | int width, |
8815 | 0 | int height) { |
8816 | 0 | int y; |
8817 | 0 | void (*P410ToAR30Row)( |
8818 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8819 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToAR30Row_C; |
8820 | 0 | void (*ScaleRowUp2_Linear)(const uint16_t* src_uv, uint16_t* dst_uv, |
8821 | 0 | int dst_width) = ScaleUVRowUp2_Linear_16_Any_C; |
8822 | 0 | assert(yuvconstants); |
8823 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
8824 | 0 | return -1; |
8825 | 0 | } |
8826 | | // Negative height means invert the image. |
8827 | 0 | if (height < 0) { |
8828 | 0 | height = -height; |
8829 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
8830 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
8831 | 0 | } |
8832 | 0 | #if defined(HAS_P410TOAR30ROW_SSSE3) |
8833 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8834 | 0 | P410ToAR30Row = P410ToAR30Row_Any_SSSE3; |
8835 | 0 | if (IS_ALIGNED(width, 8)) { |
8836 | 0 | P410ToAR30Row = P410ToAR30Row_SSSE3; |
8837 | 0 | } |
8838 | 0 | } |
8839 | 0 | #endif |
8840 | 0 | #if defined(HAS_P410TOAR30ROW_AVX2) |
8841 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8842 | 0 | P410ToAR30Row = P410ToAR30Row_Any_AVX2; |
8843 | 0 | if (IS_ALIGNED(width, 16)) { |
8844 | 0 | P410ToAR30Row = P410ToAR30Row_AVX2; |
8845 | 0 | } |
8846 | 0 | } |
8847 | 0 | #endif |
8848 | | #if defined(HAS_P410TOAR30ROW_NEON) |
8849 | | if (TestCpuFlag(kCpuHasNEON)) { |
8850 | | P410ToAR30Row = P410ToAR30Row_Any_NEON; |
8851 | | if (IS_ALIGNED(width, 8)) { |
8852 | | P410ToAR30Row = P410ToAR30Row_NEON; |
8853 | | } |
8854 | | } |
8855 | | #endif |
8856 | | #if defined(HAS_P410TOAR30ROW_SVE2) |
8857 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8858 | | P410ToAR30Row = P410ToAR30Row_SVE2; |
8859 | | } |
8860 | | #endif |
8861 | | #if defined(HAS_P410TOAR30ROW_SME) |
8862 | | if (TestCpuFlag(kCpuHasSME)) { |
8863 | | P410ToAR30Row = P410ToAR30Row_SME; |
8864 | | } |
8865 | | #endif |
8866 | |
|
8867 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_SSE41 |
8868 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8869 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_SSE41; |
8870 | 0 | } |
8871 | 0 | #endif |
8872 | |
|
8873 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_AVX2 |
8874 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8875 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_AVX2; |
8876 | 0 | } |
8877 | 0 | #endif |
8878 | |
|
8879 | | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_NEON |
8880 | | if (TestCpuFlag(kCpuHasNEON)) { |
8881 | | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_NEON; |
8882 | | } |
8883 | | #endif |
8884 | |
|
8885 | 0 | const int row_size = (2 * width + 31) & ~31; |
8886 | 0 | align_buffer_64(row, row_size * sizeof(uint16_t)); |
8887 | 0 | uint16_t* temp_uv = (uint16_t*)(row); |
8888 | 0 | if (!row) |
8889 | 0 | return 1; |
8890 | | |
8891 | 0 | for (y = 0; y < height; ++y) { |
8892 | 0 | ScaleRowUp2_Linear(src_uv, temp_uv, width); |
8893 | 0 | P410ToAR30Row(src_y, temp_uv, dst_ar30, yuvconstants, width); |
8894 | 0 | dst_ar30 += dst_stride_ar30; |
8895 | 0 | src_y += src_stride_y; |
8896 | 0 | src_uv += src_stride_uv; |
8897 | 0 | } |
8898 | |
|
8899 | 0 | free_aligned_buffer_64(row); |
8900 | 0 | return 0; |
8901 | 0 | } |
8902 | | |
8903 | | static int I422ToRGB24MatrixLinear(const uint8_t* src_y, |
8904 | | int src_stride_y, |
8905 | | const uint8_t* src_u, |
8906 | | int src_stride_u, |
8907 | | const uint8_t* src_v, |
8908 | | int src_stride_v, |
8909 | | uint8_t* dst_rgb24, |
8910 | | int dst_stride_rgb24, |
8911 | | const struct YuvConstants* yuvconstants, |
8912 | | int width, |
8913 | 146 | int height) { |
8914 | 146 | int y; |
8915 | 146 | void (*I444ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
8916 | 146 | const uint8_t* v_buf, uint8_t* rgb_buf, |
8917 | 146 | const struct YuvConstants* yuvconstants, int width) = |
8918 | 146 | I444ToRGB24Row_C; |
8919 | 146 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
8920 | 146 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
8921 | 146 | assert(yuvconstants); |
8922 | 146 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
8923 | 0 | return -1; |
8924 | 0 | } |
8925 | | // Negative height means invert the image. |
8926 | 146 | if (height < 0) { |
8927 | 0 | height = -height; |
8928 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
8929 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
8930 | 0 | } |
8931 | 146 | #if defined(HAS_I444TORGB24ROW_SSSE3) |
8932 | 146 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8933 | 146 | I444ToRGB24Row = I444ToRGB24Row_Any_SSSE3; |
8934 | 146 | if (IS_ALIGNED(width, 16)) { |
8935 | 42 | I444ToRGB24Row = I444ToRGB24Row_SSSE3; |
8936 | 42 | } |
8937 | 146 | } |
8938 | 146 | #endif |
8939 | 146 | #if defined(HAS_I444TORGB24ROW_AVX2) |
8940 | 146 | if (TestCpuFlag(kCpuHasAVX2)) { |
8941 | 146 | I444ToRGB24Row = I444ToRGB24Row_Any_AVX2; |
8942 | 146 | if (IS_ALIGNED(width, 32)) { |
8943 | 17 | I444ToRGB24Row = I444ToRGB24Row_AVX2; |
8944 | 17 | } |
8945 | 146 | } |
8946 | 146 | #endif |
8947 | | #if defined(HAS_I444TORGB24ROW_NEON) |
8948 | | if (TestCpuFlag(kCpuHasNEON)) { |
8949 | | I444ToRGB24Row = I444ToRGB24Row_Any_NEON; |
8950 | | if (IS_ALIGNED(width, 8)) { |
8951 | | I444ToRGB24Row = I444ToRGB24Row_NEON; |
8952 | | } |
8953 | | } |
8954 | | #endif |
8955 | | #if defined(HAS_I444TORGB24ROW_RVV) |
8956 | | if (TestCpuFlag(kCpuHasRVV)) { |
8957 | | I444ToRGB24Row = I444ToRGB24Row_RVV; |
8958 | | } |
8959 | | #endif |
8960 | 146 | #if defined(HAS_SCALEROWUP2_LINEAR_SSE2) |
8961 | 146 | if (TestCpuFlag(kCpuHasSSE2)) { |
8962 | 146 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
8963 | 146 | } |
8964 | 146 | #endif |
8965 | 146 | #if defined(HAS_SCALEROWUP2_LINEAR_SSSE3) |
8966 | 146 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8967 | 146 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
8968 | 146 | } |
8969 | 146 | #endif |
8970 | 146 | #if defined(HAS_SCALEROWUP2_LINEAR_AVX2) |
8971 | 146 | if (TestCpuFlag(kCpuHasAVX2)) { |
8972 | 146 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
8973 | 146 | } |
8974 | 146 | #endif |
8975 | | #if defined(HAS_SCALEROWUP2_LINEAR_NEON) |
8976 | | if (TestCpuFlag(kCpuHasNEON)) { |
8977 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
8978 | | } |
8979 | | #endif |
8980 | | #if defined(HAS_SCALEROWUP2_LINEAR_RVV) |
8981 | | if (TestCpuFlag(kCpuHasRVV)) { |
8982 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
8983 | | } |
8984 | | #endif |
8985 | | |
8986 | | // alloc 2 lines temp |
8987 | 146 | const int row_size = (width + 31) & ~31; |
8988 | 146 | align_buffer_64(row, row_size * 2); |
8989 | 146 | uint8_t* temp_u = row; |
8990 | 146 | uint8_t* temp_v = row + row_size; |
8991 | 146 | if (!row) |
8992 | 0 | return 1; |
8993 | | |
8994 | 2.23k | for (y = 0; y < height; ++y) { |
8995 | 2.09k | ScaleRowUp2_Linear(src_u, temp_u, width); |
8996 | 2.09k | ScaleRowUp2_Linear(src_v, temp_v, width); |
8997 | 2.09k | I444ToRGB24Row(src_y, temp_u, temp_v, dst_rgb24, yuvconstants, width); |
8998 | 2.09k | dst_rgb24 += dst_stride_rgb24; |
8999 | 2.09k | src_y += src_stride_y; |
9000 | 2.09k | src_u += src_stride_u; |
9001 | 2.09k | src_v += src_stride_v; |
9002 | 2.09k | } |
9003 | | |
9004 | 146 | free_aligned_buffer_64(row); |
9005 | 146 | return 0; |
9006 | 146 | } |
9007 | | |
9008 | | LIBYUV_API |
9009 | | int I422ToRGB24MatrixFilter(const uint8_t* src_y, |
9010 | | int src_stride_y, |
9011 | | const uint8_t* src_u, |
9012 | | int src_stride_u, |
9013 | | const uint8_t* src_v, |
9014 | | int src_stride_v, |
9015 | | uint8_t* dst_rgb24, |
9016 | | int dst_stride_rgb24, |
9017 | | const struct YuvConstants* yuvconstants, |
9018 | | int width, |
9019 | | int height, |
9020 | 226 | enum FilterMode filter) { |
9021 | 226 | switch (filter) { |
9022 | 80 | case kFilterNone: |
9023 | 80 | return I422ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9024 | 80 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
9025 | 80 | yuvconstants, width, height); |
9026 | 146 | case kFilterBilinear: |
9027 | 146 | case kFilterBox: |
9028 | 146 | case kFilterLinear: |
9029 | 146 | return I422ToRGB24MatrixLinear( |
9030 | 146 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9031 | 146 | dst_rgb24, dst_stride_rgb24, yuvconstants, width, height); |
9032 | 226 | } |
9033 | | |
9034 | 0 | return -1; |
9035 | 226 | } |
9036 | | |
9037 | | LIBYUV_API |
9038 | | int I420ToARGBMatrixFilter(const uint8_t* src_y, |
9039 | | int src_stride_y, |
9040 | | const uint8_t* src_u, |
9041 | | int src_stride_u, |
9042 | | const uint8_t* src_v, |
9043 | | int src_stride_v, |
9044 | | uint8_t* dst_argb, |
9045 | | int dst_stride_argb, |
9046 | | const struct YuvConstants* yuvconstants, |
9047 | | int width, |
9048 | | int height, |
9049 | 307 | enum FilterMode filter) { |
9050 | 307 | switch (filter) { |
9051 | 152 | case kFilterNone: |
9052 | 152 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9053 | 152 | src_stride_v, dst_argb, dst_stride_argb, |
9054 | 152 | yuvconstants, width, height); |
9055 | 155 | case kFilterBilinear: |
9056 | 155 | case kFilterBox: |
9057 | 155 | return I420ToARGBMatrixBilinear( |
9058 | 155 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9059 | 155 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
9060 | 0 | case kFilterLinear: |
9061 | | // Actually we can do this, but probably there's no usage. |
9062 | 0 | return -1; |
9063 | 307 | } |
9064 | | |
9065 | 0 | return -1; |
9066 | 307 | } |
9067 | | |
9068 | | LIBYUV_API |
9069 | | int I422ToARGBMatrixFilter(const uint8_t* src_y, |
9070 | | int src_stride_y, |
9071 | | const uint8_t* src_u, |
9072 | | int src_stride_u, |
9073 | | const uint8_t* src_v, |
9074 | | int src_stride_v, |
9075 | | uint8_t* dst_argb, |
9076 | | int dst_stride_argb, |
9077 | | const struct YuvConstants* yuvconstants, |
9078 | | int width, |
9079 | | int height, |
9080 | 264 | enum FilterMode filter) { |
9081 | 264 | switch (filter) { |
9082 | 108 | case kFilterNone: |
9083 | 108 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9084 | 108 | src_stride_v, dst_argb, dst_stride_argb, |
9085 | 108 | yuvconstants, width, height); |
9086 | 156 | case kFilterBilinear: |
9087 | 156 | case kFilterBox: |
9088 | 156 | case kFilterLinear: |
9089 | 156 | return I422ToARGBMatrixLinear( |
9090 | 156 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9091 | 156 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
9092 | 264 | } |
9093 | | |
9094 | 0 | return -1; |
9095 | 264 | } |
9096 | | |
9097 | | LIBYUV_API |
9098 | | int I420ToRGB24MatrixFilter(const uint8_t* src_y, |
9099 | | int src_stride_y, |
9100 | | const uint8_t* src_u, |
9101 | | int src_stride_u, |
9102 | | const uint8_t* src_v, |
9103 | | int src_stride_v, |
9104 | | uint8_t* dst_rgb24, |
9105 | | int dst_stride_rgb24, |
9106 | | const struct YuvConstants* yuvconstants, |
9107 | | int width, |
9108 | | int height, |
9109 | 272 | enum FilterMode filter) { |
9110 | 272 | switch (filter) { |
9111 | 183 | case kFilterNone: |
9112 | 183 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9113 | 183 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
9114 | 183 | yuvconstants, width, height); |
9115 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9116 | 88 | case kFilterBilinear: |
9117 | 88 | case kFilterBox: |
9118 | 88 | return I420ToRGB24MatrixBilinear( |
9119 | 88 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9120 | 88 | dst_rgb24, dst_stride_rgb24, yuvconstants, width, height); |
9121 | 272 | } |
9122 | | |
9123 | 0 | return -1; |
9124 | 272 | } |
9125 | | |
9126 | | LIBYUV_API |
9127 | | int I010ToAR30MatrixFilter(const uint16_t* src_y, |
9128 | | int src_stride_y, |
9129 | | const uint16_t* src_u, |
9130 | | int src_stride_u, |
9131 | | const uint16_t* src_v, |
9132 | | int src_stride_v, |
9133 | | uint8_t* dst_ar30, |
9134 | | int dst_stride_ar30, |
9135 | | const struct YuvConstants* yuvconstants, |
9136 | | int width, |
9137 | | int height, |
9138 | 0 | enum FilterMode filter) { |
9139 | 0 | switch (filter) { |
9140 | 0 | case kFilterNone: |
9141 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9142 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
9143 | 0 | yuvconstants, width, height); |
9144 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9145 | 0 | case kFilterBilinear: |
9146 | 0 | case kFilterBox: |
9147 | 0 | return I010ToAR30MatrixBilinear( |
9148 | 0 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9149 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, height); |
9150 | 0 | } |
9151 | | |
9152 | 0 | return -1; |
9153 | 0 | } |
9154 | | |
9155 | | LIBYUV_API |
9156 | | int I210ToAR30MatrixFilter(const uint16_t* src_y, |
9157 | | int src_stride_y, |
9158 | | const uint16_t* src_u, |
9159 | | int src_stride_u, |
9160 | | const uint16_t* src_v, |
9161 | | int src_stride_v, |
9162 | | uint8_t* dst_ar30, |
9163 | | int dst_stride_ar30, |
9164 | | const struct YuvConstants* yuvconstants, |
9165 | | int width, |
9166 | | int height, |
9167 | 0 | enum FilterMode filter) { |
9168 | 0 | switch (filter) { |
9169 | 0 | case kFilterNone: |
9170 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9171 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
9172 | 0 | yuvconstants, width, height); |
9173 | 0 | case kFilterBilinear: |
9174 | 0 | case kFilterBox: |
9175 | 0 | case kFilterLinear: |
9176 | 0 | return I210ToAR30MatrixLinear( |
9177 | 0 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9178 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, height); |
9179 | 0 | } |
9180 | | |
9181 | 0 | return -1; |
9182 | 0 | } |
9183 | | |
9184 | | LIBYUV_API |
9185 | | int I010ToARGBMatrixFilter(const uint16_t* src_y, |
9186 | | int src_stride_y, |
9187 | | const uint16_t* src_u, |
9188 | | int src_stride_u, |
9189 | | const uint16_t* src_v, |
9190 | | int src_stride_v, |
9191 | | uint8_t* dst_argb, |
9192 | | int dst_stride_argb, |
9193 | | const struct YuvConstants* yuvconstants, |
9194 | | int width, |
9195 | | int height, |
9196 | 217 | enum FilterMode filter) { |
9197 | 217 | switch (filter) { |
9198 | 128 | case kFilterNone: |
9199 | 128 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9200 | 128 | src_stride_v, dst_argb, dst_stride_argb, |
9201 | 128 | yuvconstants, width, height); |
9202 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9203 | 89 | case kFilterBilinear: |
9204 | 89 | case kFilterBox: |
9205 | 89 | return I010ToARGBMatrixBilinear( |
9206 | 89 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9207 | 89 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
9208 | 217 | } |
9209 | | |
9210 | 0 | return -1; |
9211 | 217 | } |
9212 | | |
9213 | | LIBYUV_API |
9214 | | int I210ToARGBMatrixFilter(const uint16_t* src_y, |
9215 | | int src_stride_y, |
9216 | | const uint16_t* src_u, |
9217 | | int src_stride_u, |
9218 | | const uint16_t* src_v, |
9219 | | int src_stride_v, |
9220 | | uint8_t* dst_argb, |
9221 | | int dst_stride_argb, |
9222 | | const struct YuvConstants* yuvconstants, |
9223 | | int width, |
9224 | | int height, |
9225 | 240 | enum FilterMode filter) { |
9226 | 240 | switch (filter) { |
9227 | 65 | case kFilterNone: |
9228 | 65 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
9229 | 65 | src_stride_v, dst_argb, dst_stride_argb, |
9230 | 65 | yuvconstants, width, height); |
9231 | 175 | case kFilterBilinear: |
9232 | 175 | case kFilterBox: |
9233 | 175 | case kFilterLinear: |
9234 | 175 | return I210ToARGBMatrixLinear( |
9235 | 175 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
9236 | 175 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
9237 | 240 | } |
9238 | | |
9239 | 0 | return -1; |
9240 | 240 | } |
9241 | | |
9242 | | LIBYUV_API |
9243 | | int I420AlphaToARGBMatrixFilter(const uint8_t* src_y, |
9244 | | int src_stride_y, |
9245 | | const uint8_t* src_u, |
9246 | | int src_stride_u, |
9247 | | const uint8_t* src_v, |
9248 | | int src_stride_v, |
9249 | | const uint8_t* src_a, |
9250 | | int src_stride_a, |
9251 | | uint8_t* dst_argb, |
9252 | | int dst_stride_argb, |
9253 | | const struct YuvConstants* yuvconstants, |
9254 | | int width, |
9255 | | int height, |
9256 | | int attenuate, |
9257 | 179 | enum FilterMode filter) { |
9258 | 179 | switch (filter) { |
9259 | 131 | case kFilterNone: |
9260 | 131 | return I420AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
9261 | 131 | src_v, src_stride_v, src_a, src_stride_a, |
9262 | 131 | dst_argb, dst_stride_argb, yuvconstants, |
9263 | 131 | width, height, attenuate); |
9264 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9265 | 48 | case kFilterBilinear: |
9266 | 48 | case kFilterBox: |
9267 | 48 | return I420AlphaToARGBMatrixBilinear( |
9268 | 48 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
9269 | 48 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
9270 | 48 | attenuate); |
9271 | 179 | } |
9272 | | |
9273 | 0 | return -1; |
9274 | 179 | } |
9275 | | |
9276 | | LIBYUV_API |
9277 | | int I422AlphaToARGBMatrixFilter(const uint8_t* src_y, |
9278 | | int src_stride_y, |
9279 | | const uint8_t* src_u, |
9280 | | int src_stride_u, |
9281 | | const uint8_t* src_v, |
9282 | | int src_stride_v, |
9283 | | const uint8_t* src_a, |
9284 | | int src_stride_a, |
9285 | | uint8_t* dst_argb, |
9286 | | int dst_stride_argb, |
9287 | | const struct YuvConstants* yuvconstants, |
9288 | | int width, |
9289 | | int height, |
9290 | | int attenuate, |
9291 | 216 | enum FilterMode filter) { |
9292 | 216 | switch (filter) { |
9293 | 93 | case kFilterNone: |
9294 | 93 | return I422AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
9295 | 93 | src_v, src_stride_v, src_a, src_stride_a, |
9296 | 93 | dst_argb, dst_stride_argb, yuvconstants, |
9297 | 93 | width, height, attenuate); |
9298 | 123 | case kFilterBilinear: |
9299 | 123 | case kFilterBox: |
9300 | 123 | case kFilterLinear: |
9301 | 123 | return I422AlphaToARGBMatrixLinear( |
9302 | 123 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
9303 | 123 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
9304 | 123 | attenuate); |
9305 | 216 | } |
9306 | | |
9307 | 0 | return -1; |
9308 | 216 | } |
9309 | | |
9310 | | LIBYUV_API |
9311 | | int I010AlphaToARGBMatrixFilter(const uint16_t* src_y, |
9312 | | int src_stride_y, |
9313 | | const uint16_t* src_u, |
9314 | | int src_stride_u, |
9315 | | const uint16_t* src_v, |
9316 | | int src_stride_v, |
9317 | | const uint16_t* src_a, |
9318 | | int src_stride_a, |
9319 | | uint8_t* dst_argb, |
9320 | | int dst_stride_argb, |
9321 | | const struct YuvConstants* yuvconstants, |
9322 | | int width, |
9323 | | int height, |
9324 | | int attenuate, |
9325 | 162 | enum FilterMode filter) { |
9326 | 162 | switch (filter) { |
9327 | 125 | case kFilterNone: |
9328 | 125 | return I010AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
9329 | 125 | src_v, src_stride_v, src_a, src_stride_a, |
9330 | 125 | dst_argb, dst_stride_argb, yuvconstants, |
9331 | 125 | width, height, attenuate); |
9332 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9333 | 37 | case kFilterBilinear: |
9334 | 37 | case kFilterBox: |
9335 | 37 | return I010AlphaToARGBMatrixBilinear( |
9336 | 37 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
9337 | 37 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
9338 | 37 | attenuate); |
9339 | 162 | } |
9340 | | |
9341 | 0 | return -1; |
9342 | 162 | } |
9343 | | |
9344 | | LIBYUV_API |
9345 | | int I210AlphaToARGBMatrixFilter(const uint16_t* src_y, |
9346 | | int src_stride_y, |
9347 | | const uint16_t* src_u, |
9348 | | int src_stride_u, |
9349 | | const uint16_t* src_v, |
9350 | | int src_stride_v, |
9351 | | const uint16_t* src_a, |
9352 | | int src_stride_a, |
9353 | | uint8_t* dst_argb, |
9354 | | int dst_stride_argb, |
9355 | | const struct YuvConstants* yuvconstants, |
9356 | | int width, |
9357 | | int height, |
9358 | | int attenuate, |
9359 | 180 | enum FilterMode filter) { |
9360 | 180 | switch (filter) { |
9361 | 84 | case kFilterNone: |
9362 | 84 | return I210AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
9363 | 84 | src_v, src_stride_v, src_a, src_stride_a, |
9364 | 84 | dst_argb, dst_stride_argb, yuvconstants, |
9365 | 84 | width, height, attenuate); |
9366 | 96 | case kFilterBilinear: |
9367 | 96 | case kFilterBox: |
9368 | 96 | case kFilterLinear: |
9369 | 96 | return I210AlphaToARGBMatrixLinear( |
9370 | 96 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
9371 | 96 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
9372 | 96 | attenuate); |
9373 | 180 | } |
9374 | | |
9375 | 0 | return -1; |
9376 | 180 | } |
9377 | | |
9378 | | // TODO(fb): Verify this function works correctly. P010 is like NV12 but 10 bit |
9379 | | // UV is biplanar. |
9380 | | LIBYUV_API |
9381 | | int P010ToARGBMatrixFilter(const uint16_t* src_y, |
9382 | | int src_stride_y, |
9383 | | const uint16_t* src_uv, |
9384 | | int src_stride_uv, |
9385 | | uint8_t* dst_argb, |
9386 | | int dst_stride_argb, |
9387 | | const struct YuvConstants* yuvconstants, |
9388 | | int width, |
9389 | | int height, |
9390 | 0 | enum FilterMode filter) { |
9391 | 0 | switch (filter) { |
9392 | 0 | case kFilterNone: |
9393 | 0 | return P010ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9394 | 0 | dst_argb, dst_stride_argb, yuvconstants, width, |
9395 | 0 | height); |
9396 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9397 | 0 | case kFilterBilinear: |
9398 | 0 | case kFilterBox: |
9399 | 0 | return P010ToARGBMatrixBilinear(src_y, src_stride_y, src_uv, |
9400 | 0 | src_stride_uv, dst_argb, dst_stride_argb, |
9401 | 0 | yuvconstants, width, height); |
9402 | 0 | } |
9403 | | |
9404 | 0 | return -1; |
9405 | 0 | } |
9406 | | |
9407 | | LIBYUV_API |
9408 | | int P210ToARGBMatrixFilter(const uint16_t* src_y, |
9409 | | int src_stride_y, |
9410 | | const uint16_t* src_uv, |
9411 | | int src_stride_uv, |
9412 | | uint8_t* dst_argb, |
9413 | | int dst_stride_argb, |
9414 | | const struct YuvConstants* yuvconstants, |
9415 | | int width, |
9416 | | int height, |
9417 | 0 | enum FilterMode filter) { |
9418 | 0 | switch (filter) { |
9419 | 0 | case kFilterNone: |
9420 | 0 | return P210ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9421 | 0 | dst_argb, dst_stride_argb, yuvconstants, width, |
9422 | 0 | height); |
9423 | 0 | case kFilterBilinear: |
9424 | 0 | case kFilterBox: |
9425 | 0 | case kFilterLinear: |
9426 | 0 | return P210ToARGBMatrixLinear(src_y, src_stride_y, src_uv, src_stride_uv, |
9427 | 0 | dst_argb, dst_stride_argb, yuvconstants, |
9428 | 0 | width, height); |
9429 | 0 | } |
9430 | | |
9431 | 0 | return -1; |
9432 | 0 | } |
9433 | | |
9434 | | LIBYUV_API |
9435 | | int P010ToAR30MatrixFilter(const uint16_t* src_y, |
9436 | | int src_stride_y, |
9437 | | const uint16_t* src_uv, |
9438 | | int src_stride_uv, |
9439 | | uint8_t* dst_ar30, |
9440 | | int dst_stride_ar30, |
9441 | | const struct YuvConstants* yuvconstants, |
9442 | | int width, |
9443 | | int height, |
9444 | 0 | enum FilterMode filter) { |
9445 | 0 | switch (filter) { |
9446 | 0 | case kFilterNone: |
9447 | 0 | return P010ToAR30Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9448 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, |
9449 | 0 | height); |
9450 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9451 | 0 | case kFilterBilinear: |
9452 | 0 | case kFilterBox: |
9453 | 0 | return P010ToAR30MatrixBilinear(src_y, src_stride_y, src_uv, |
9454 | 0 | src_stride_uv, dst_ar30, dst_stride_ar30, |
9455 | 0 | yuvconstants, width, height); |
9456 | 0 | } |
9457 | | |
9458 | 0 | return -1; |
9459 | 0 | } |
9460 | | |
9461 | | LIBYUV_API |
9462 | | int P210ToAR30MatrixFilter(const uint16_t* src_y, |
9463 | | int src_stride_y, |
9464 | | const uint16_t* src_uv, |
9465 | | int src_stride_uv, |
9466 | | uint8_t* dst_ar30, |
9467 | | int dst_stride_ar30, |
9468 | | const struct YuvConstants* yuvconstants, |
9469 | | int width, |
9470 | | int height, |
9471 | 0 | enum FilterMode filter) { |
9472 | 0 | switch (filter) { |
9473 | 0 | case kFilterNone: |
9474 | 0 | return P210ToAR30Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9475 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, |
9476 | 0 | height); |
9477 | 0 | case kFilterBilinear: |
9478 | 0 | case kFilterBox: |
9479 | 0 | case kFilterLinear: |
9480 | 0 | return P210ToAR30MatrixLinear(src_y, src_stride_y, src_uv, src_stride_uv, |
9481 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, |
9482 | 0 | width, height); |
9483 | 0 | } |
9484 | | |
9485 | 0 | return -1; |
9486 | 0 | } |
9487 | | |
9488 | | #ifdef __cplusplus |
9489 | | } // extern "C" |
9490 | | } // namespace libyuv |
9491 | | #endif |