/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 | 141 | int height) { |
64 | 141 | int y; |
65 | 141 | void (*I422ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
66 | 141 | const uint8_t* v_buf, uint8_t* rgb_buf, |
67 | 141 | const struct YuvConstants* yuvconstants, int width) = |
68 | 141 | I422ToARGBRow_C; |
69 | 141 | assert(yuvconstants); |
70 | 141 | 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 | 141 | 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 | 141 | #if defined(HAS_I422TOARGBROW_SSSE3) |
80 | 141 | if (TestCpuFlag(kCpuHasSSSE3)) { |
81 | 141 | I422ToARGBRow = I422ToARGBRow_Any_SSSE3; |
82 | 141 | if (IS_ALIGNED(width, 8)) { |
83 | 37 | I422ToARGBRow = I422ToARGBRow_SSSE3; |
84 | 37 | } |
85 | 141 | } |
86 | 141 | #endif |
87 | 141 | #if defined(HAS_I422TOARGBROW_AVX2) |
88 | 141 | if (TestCpuFlag(kCpuHasAVX2)) { |
89 | 141 | I422ToARGBRow = I422ToARGBRow_Any_AVX2; |
90 | 141 | if (IS_ALIGNED(width, 16)) { |
91 | 20 | I422ToARGBRow = I422ToARGBRow_AVX2; |
92 | 20 | } |
93 | 141 | } |
94 | 141 | #endif |
95 | 141 | #if defined(HAS_I422TOARGBROW_AVX512BW) |
96 | 141 | if (TestCpuFlag(kCpuHasAVX512BW | kCpuHasAVX512VL) == |
97 | 141 | (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 | 141 | #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_LSX) |
123 | | if (TestCpuFlag(kCpuHasLSX)) { |
124 | | I422ToARGBRow = I422ToARGBRow_Any_LSX; |
125 | | if (IS_ALIGNED(width, 16)) { |
126 | | I422ToARGBRow = I422ToARGBRow_LSX; |
127 | | } |
128 | | } |
129 | | #endif |
130 | | #if defined(HAS_I422TOARGBROW_LASX) |
131 | | if (TestCpuFlag(kCpuHasLASX)) { |
132 | | I422ToARGBRow = I422ToARGBRow_Any_LASX; |
133 | | if (IS_ALIGNED(width, 32)) { |
134 | | I422ToARGBRow = I422ToARGBRow_LASX; |
135 | | } |
136 | | } |
137 | | #endif |
138 | | #if defined(HAS_I422TOARGBROW_RVV) |
139 | | if (TestCpuFlag(kCpuHasRVV)) { |
140 | | I422ToARGBRow = I422ToARGBRow_RVV; |
141 | | } |
142 | | #endif |
143 | | |
144 | 2.13k | for (y = 0; y < height; ++y) { |
145 | 1.99k | I422ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
146 | 1.99k | dst_argb += dst_stride_argb; |
147 | 1.99k | src_y += src_stride_y; |
148 | 1.99k | if (y & 1) { |
149 | 977 | src_u += src_stride_u; |
150 | 977 | src_v += src_stride_v; |
151 | 977 | } |
152 | 1.99k | } |
153 | 141 | return 0; |
154 | 141 | } |
155 | | |
156 | | // Convert I420 to ARGB. |
157 | | LIBYUV_API |
158 | | int I420ToARGB(const uint8_t* src_y, |
159 | | int src_stride_y, |
160 | | const uint8_t* src_u, |
161 | | int src_stride_u, |
162 | | const uint8_t* src_v, |
163 | | int src_stride_v, |
164 | | uint8_t* dst_argb, |
165 | | int dst_stride_argb, |
166 | | int width, |
167 | 0 | int height) { |
168 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
169 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
170 | 0 | &kYuvI601Constants, width, height); |
171 | 0 | } |
172 | | |
173 | | // Convert I420 to ABGR. |
174 | | LIBYUV_API |
175 | | int I420ToABGR(const uint8_t* src_y, |
176 | | int src_stride_y, |
177 | | const uint8_t* src_u, |
178 | | int src_stride_u, |
179 | | const uint8_t* src_v, |
180 | | int src_stride_v, |
181 | | uint8_t* dst_abgr, |
182 | | int dst_stride_abgr, |
183 | | int width, |
184 | 0 | int height) { |
185 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
186 | 0 | src_stride_v, // Swap U and V |
187 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
188 | 0 | &kYvuI601Constants, // Use Yvu matrix |
189 | 0 | width, height); |
190 | 0 | } |
191 | | |
192 | | // Convert J420 to ARGB. |
193 | | LIBYUV_API |
194 | | int J420ToARGB(const uint8_t* src_y, |
195 | | int src_stride_y, |
196 | | const uint8_t* src_u, |
197 | | int src_stride_u, |
198 | | const uint8_t* src_v, |
199 | | int src_stride_v, |
200 | | uint8_t* dst_argb, |
201 | | int dst_stride_argb, |
202 | | int width, |
203 | 0 | int height) { |
204 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
205 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
206 | 0 | &kYuvJPEGConstants, width, height); |
207 | 0 | } |
208 | | |
209 | | // Convert J420 to ABGR. |
210 | | LIBYUV_API |
211 | | int J420ToABGR(const uint8_t* src_y, |
212 | | int src_stride_y, |
213 | | const uint8_t* src_u, |
214 | | int src_stride_u, |
215 | | const uint8_t* src_v, |
216 | | int src_stride_v, |
217 | | uint8_t* dst_abgr, |
218 | | int dst_stride_abgr, |
219 | | int width, |
220 | 0 | int height) { |
221 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
222 | 0 | src_stride_v, // Swap U and V |
223 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
224 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
225 | 0 | width, height); |
226 | 0 | } |
227 | | |
228 | | // Convert H420 to ARGB. |
229 | | LIBYUV_API |
230 | | int H420ToARGB(const uint8_t* src_y, |
231 | | int src_stride_y, |
232 | | const uint8_t* src_u, |
233 | | int src_stride_u, |
234 | | const uint8_t* src_v, |
235 | | int src_stride_v, |
236 | | uint8_t* dst_argb, |
237 | | int dst_stride_argb, |
238 | | int width, |
239 | 0 | int height) { |
240 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
241 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
242 | 0 | &kYuvH709Constants, width, height); |
243 | 0 | } |
244 | | |
245 | | // Convert H420 to ABGR. |
246 | | LIBYUV_API |
247 | | int H420ToABGR(const uint8_t* src_y, |
248 | | int src_stride_y, |
249 | | const uint8_t* src_u, |
250 | | int src_stride_u, |
251 | | const uint8_t* src_v, |
252 | | int src_stride_v, |
253 | | uint8_t* dst_abgr, |
254 | | int dst_stride_abgr, |
255 | | int width, |
256 | 0 | int height) { |
257 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
258 | 0 | src_stride_v, // Swap U and V |
259 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
260 | 0 | &kYvuH709Constants, // Use Yvu matrix |
261 | 0 | width, height); |
262 | 0 | } |
263 | | |
264 | | // Convert U420 to ARGB. |
265 | | LIBYUV_API |
266 | | int U420ToARGB(const uint8_t* src_y, |
267 | | int src_stride_y, |
268 | | const uint8_t* src_u, |
269 | | int src_stride_u, |
270 | | const uint8_t* src_v, |
271 | | int src_stride_v, |
272 | | uint8_t* dst_argb, |
273 | | int dst_stride_argb, |
274 | | int width, |
275 | 0 | int height) { |
276 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
277 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
278 | 0 | &kYuv2020Constants, width, height); |
279 | 0 | } |
280 | | |
281 | | // Convert U420 to ABGR. |
282 | | LIBYUV_API |
283 | | int U420ToABGR(const uint8_t* src_y, |
284 | | int src_stride_y, |
285 | | const uint8_t* src_u, |
286 | | int src_stride_u, |
287 | | const uint8_t* src_v, |
288 | | int src_stride_v, |
289 | | uint8_t* dst_abgr, |
290 | | int dst_stride_abgr, |
291 | | int width, |
292 | 0 | int height) { |
293 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_v, |
294 | 0 | src_stride_v, // Swap U and V |
295 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
296 | 0 | &kYvu2020Constants, // Use Yvu matrix |
297 | 0 | width, height); |
298 | 0 | } |
299 | | |
300 | | // Convert I422 to ARGB with matrix. |
301 | | LIBYUV_API |
302 | | int I422ToARGBMatrix(const uint8_t* src_y, |
303 | | int src_stride_y, |
304 | | const uint8_t* src_u, |
305 | | int src_stride_u, |
306 | | const uint8_t* src_v, |
307 | | int src_stride_v, |
308 | | uint8_t* dst_argb, |
309 | | int dst_stride_argb, |
310 | | const struct YuvConstants* yuvconstants, |
311 | | int width, |
312 | 95 | int height) { |
313 | 95 | int y; |
314 | 95 | void (*I422ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
315 | 95 | const uint8_t* v_buf, uint8_t* rgb_buf, |
316 | 95 | const struct YuvConstants* yuvconstants, int width) = |
317 | 95 | I422ToARGBRow_C; |
318 | 95 | assert(yuvconstants); |
319 | 95 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
320 | 0 | return -1; |
321 | 0 | } |
322 | | // Negative height means invert the image. |
323 | 95 | if (height < 0) { |
324 | 0 | height = -height; |
325 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
326 | 0 | dst_stride_argb = -dst_stride_argb; |
327 | 0 | } |
328 | | // Coalesce rows. |
329 | 95 | if (src_stride_y == width && src_stride_u * 2 == width && |
330 | 49 | src_stride_v * 2 == width && dst_stride_argb == width * 4) { |
331 | 49 | width *= height; |
332 | 49 | height = 1; |
333 | 49 | src_stride_y = src_stride_u = src_stride_v = dst_stride_argb = 0; |
334 | 49 | } |
335 | 95 | #if defined(HAS_I422TOARGBROW_SSSE3) |
336 | 95 | if (TestCpuFlag(kCpuHasSSSE3)) { |
337 | 95 | I422ToARGBRow = I422ToARGBRow_Any_SSSE3; |
338 | 95 | if (IS_ALIGNED(width, 8)) { |
339 | 37 | I422ToARGBRow = I422ToARGBRow_SSSE3; |
340 | 37 | } |
341 | 95 | } |
342 | 95 | #endif |
343 | 95 | #if defined(HAS_I422TOARGBROW_AVX2) |
344 | 95 | if (TestCpuFlag(kCpuHasAVX2)) { |
345 | 95 | I422ToARGBRow = I422ToARGBRow_Any_AVX2; |
346 | 95 | if (IS_ALIGNED(width, 16)) { |
347 | 19 | I422ToARGBRow = I422ToARGBRow_AVX2; |
348 | 19 | } |
349 | 95 | } |
350 | 95 | #endif |
351 | 95 | #if defined(HAS_I422TOARGBROW_AVX512BW) |
352 | 95 | if (TestCpuFlag(kCpuHasAVX512BW | kCpuHasAVX512VL) == |
353 | 95 | (kCpuHasAVX512BW | kCpuHasAVX512VL)) { |
354 | 0 | I422ToARGBRow = I422ToARGBRow_Any_AVX512BW; |
355 | 0 | if (IS_ALIGNED(width, 32)) { |
356 | 0 | I422ToARGBRow = I422ToARGBRow_AVX512BW; |
357 | 0 | } |
358 | 0 | } |
359 | 95 | #endif |
360 | | #if defined(HAS_I422TOARGBROW_NEON) |
361 | | if (TestCpuFlag(kCpuHasNEON)) { |
362 | | I422ToARGBRow = I422ToARGBRow_Any_NEON; |
363 | | if (IS_ALIGNED(width, 8)) { |
364 | | I422ToARGBRow = I422ToARGBRow_NEON; |
365 | | } |
366 | | } |
367 | | #endif |
368 | | #if defined(HAS_I422TOARGBROW_SVE2) |
369 | | if (TestCpuFlag(kCpuHasSVE2)) { |
370 | | I422ToARGBRow = I422ToARGBRow_SVE2; |
371 | | } |
372 | | #endif |
373 | | #if defined(HAS_I422TOARGBROW_SME) |
374 | | if (TestCpuFlag(kCpuHasSME)) { |
375 | | I422ToARGBRow = I422ToARGBRow_SME; |
376 | | } |
377 | | #endif |
378 | | #if defined(HAS_I422TOARGBROW_LSX) |
379 | | if (TestCpuFlag(kCpuHasLSX)) { |
380 | | I422ToARGBRow = I422ToARGBRow_Any_LSX; |
381 | | if (IS_ALIGNED(width, 16)) { |
382 | | I422ToARGBRow = I422ToARGBRow_LSX; |
383 | | } |
384 | | } |
385 | | #endif |
386 | | #if defined(HAS_I422TOARGBROW_LASX) |
387 | | if (TestCpuFlag(kCpuHasLASX)) { |
388 | | I422ToARGBRow = I422ToARGBRow_Any_LASX; |
389 | | if (IS_ALIGNED(width, 32)) { |
390 | | I422ToARGBRow = I422ToARGBRow_LASX; |
391 | | } |
392 | | } |
393 | | #endif |
394 | | #if defined(HAS_I422TOARGBROW_RVV) |
395 | | if (TestCpuFlag(kCpuHasRVV)) { |
396 | | I422ToARGBRow = I422ToARGBRow_RVV; |
397 | | } |
398 | | #endif |
399 | | |
400 | 591 | for (y = 0; y < height; ++y) { |
401 | 496 | I422ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
402 | 496 | dst_argb += dst_stride_argb; |
403 | 496 | src_y += src_stride_y; |
404 | 496 | src_u += src_stride_u; |
405 | 496 | src_v += src_stride_v; |
406 | 496 | } |
407 | 95 | return 0; |
408 | 95 | } |
409 | | |
410 | | // Convert I422 to ARGB. |
411 | | LIBYUV_API |
412 | | int I422ToARGB(const uint8_t* src_y, |
413 | | int src_stride_y, |
414 | | const uint8_t* src_u, |
415 | | int src_stride_u, |
416 | | const uint8_t* src_v, |
417 | | int src_stride_v, |
418 | | uint8_t* dst_argb, |
419 | | int dst_stride_argb, |
420 | | int width, |
421 | 0 | int height) { |
422 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
423 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
424 | 0 | &kYuvI601Constants, width, height); |
425 | 0 | } |
426 | | |
427 | | // Convert I422 to ABGR. |
428 | | LIBYUV_API |
429 | | int I422ToABGR(const uint8_t* src_y, |
430 | | int src_stride_y, |
431 | | const uint8_t* src_u, |
432 | | int src_stride_u, |
433 | | const uint8_t* src_v, |
434 | | int src_stride_v, |
435 | | uint8_t* dst_abgr, |
436 | | int dst_stride_abgr, |
437 | | int width, |
438 | 0 | int height) { |
439 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
440 | 0 | src_stride_v, // Swap U and V |
441 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
442 | 0 | &kYvuI601Constants, // Use Yvu matrix |
443 | 0 | width, height); |
444 | 0 | } |
445 | | |
446 | | // Convert J422 to ARGB. |
447 | | LIBYUV_API |
448 | | int J422ToARGB(const uint8_t* src_y, |
449 | | int src_stride_y, |
450 | | const uint8_t* src_u, |
451 | | int src_stride_u, |
452 | | const uint8_t* src_v, |
453 | | int src_stride_v, |
454 | | uint8_t* dst_argb, |
455 | | int dst_stride_argb, |
456 | | int width, |
457 | 0 | int height) { |
458 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
459 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
460 | 0 | &kYuvJPEGConstants, width, height); |
461 | 0 | } |
462 | | |
463 | | // Convert J422 to ABGR. |
464 | | LIBYUV_API |
465 | | int J422ToABGR(const uint8_t* src_y, |
466 | | int src_stride_y, |
467 | | const uint8_t* src_u, |
468 | | int src_stride_u, |
469 | | const uint8_t* src_v, |
470 | | int src_stride_v, |
471 | | uint8_t* dst_abgr, |
472 | | int dst_stride_abgr, |
473 | | int width, |
474 | 0 | int height) { |
475 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
476 | 0 | src_stride_v, // Swap U and V |
477 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
478 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
479 | 0 | width, height); |
480 | 0 | } |
481 | | |
482 | | // Convert H422 to ARGB. |
483 | | LIBYUV_API |
484 | | int H422ToARGB(const uint8_t* src_y, |
485 | | int src_stride_y, |
486 | | const uint8_t* src_u, |
487 | | int src_stride_u, |
488 | | const uint8_t* src_v, |
489 | | int src_stride_v, |
490 | | uint8_t* dst_argb, |
491 | | int dst_stride_argb, |
492 | | int width, |
493 | 0 | int height) { |
494 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
495 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
496 | 0 | &kYuvH709Constants, width, height); |
497 | 0 | } |
498 | | |
499 | | // Convert H422 to ABGR. |
500 | | LIBYUV_API |
501 | | int H422ToABGR(const uint8_t* src_y, |
502 | | int src_stride_y, |
503 | | const uint8_t* src_u, |
504 | | int src_stride_u, |
505 | | const uint8_t* src_v, |
506 | | int src_stride_v, |
507 | | uint8_t* dst_abgr, |
508 | | int dst_stride_abgr, |
509 | | int width, |
510 | 0 | int height) { |
511 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
512 | 0 | src_stride_v, // Swap U and V |
513 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
514 | 0 | &kYvuH709Constants, // Use Yvu matrix |
515 | 0 | width, height); |
516 | 0 | } |
517 | | |
518 | | // Convert U422 to ARGB. |
519 | | LIBYUV_API |
520 | | int U422ToARGB(const uint8_t* src_y, |
521 | | int src_stride_y, |
522 | | const uint8_t* src_u, |
523 | | int src_stride_u, |
524 | | const uint8_t* src_v, |
525 | | int src_stride_v, |
526 | | uint8_t* dst_argb, |
527 | | int dst_stride_argb, |
528 | | int width, |
529 | 0 | int height) { |
530 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
531 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
532 | 0 | &kYuv2020Constants, width, height); |
533 | 0 | } |
534 | | |
535 | | // Convert U422 to ABGR. |
536 | | LIBYUV_API |
537 | | int U422ToABGR(const uint8_t* src_y, |
538 | | int src_stride_y, |
539 | | const uint8_t* src_u, |
540 | | int src_stride_u, |
541 | | const uint8_t* src_v, |
542 | | int src_stride_v, |
543 | | uint8_t* dst_abgr, |
544 | | int dst_stride_abgr, |
545 | | int width, |
546 | 0 | int height) { |
547 | 0 | return I422ToARGBMatrix(src_y, src_stride_y, src_v, |
548 | 0 | src_stride_v, // Swap U and V |
549 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
550 | 0 | &kYvu2020Constants, // Use Yvu matrix |
551 | 0 | width, height); |
552 | 0 | } |
553 | | |
554 | | // Convert I444 to ARGB with matrix. |
555 | | LIBYUV_API |
556 | | int I444ToARGBMatrix(const uint8_t* src_y, |
557 | | int src_stride_y, |
558 | | const uint8_t* src_u, |
559 | | int src_stride_u, |
560 | | const uint8_t* src_v, |
561 | | int src_stride_v, |
562 | | uint8_t* dst_argb, |
563 | | int dst_stride_argb, |
564 | | const struct YuvConstants* yuvconstants, |
565 | | int width, |
566 | 90 | int height) { |
567 | 90 | int y; |
568 | 90 | void (*I444ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
569 | 90 | const uint8_t* v_buf, uint8_t* rgb_buf, |
570 | 90 | const struct YuvConstants* yuvconstants, int width) = |
571 | 90 | I444ToARGBRow_C; |
572 | 90 | assert(yuvconstants); |
573 | 90 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
574 | 0 | return -1; |
575 | 0 | } |
576 | | // Negative height means invert the image. |
577 | 90 | if (height < 0) { |
578 | 0 | height = -height; |
579 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
580 | 0 | dst_stride_argb = -dst_stride_argb; |
581 | 0 | } |
582 | | // Coalesce rows. |
583 | 90 | if (src_stride_y == width && src_stride_u == width && src_stride_v == width && |
584 | 90 | dst_stride_argb == width * 4) { |
585 | 90 | width *= height; |
586 | 90 | height = 1; |
587 | 90 | src_stride_y = src_stride_u = src_stride_v = dst_stride_argb = 0; |
588 | 90 | } |
589 | 90 | #if defined(HAS_I444TOARGBROW_SSSE3) |
590 | 90 | if (TestCpuFlag(kCpuHasSSSE3)) { |
591 | 90 | I444ToARGBRow = I444ToARGBRow_Any_SSSE3; |
592 | 90 | if (IS_ALIGNED(width, 8)) { |
593 | 39 | I444ToARGBRow = I444ToARGBRow_SSSE3; |
594 | 39 | } |
595 | 90 | } |
596 | 90 | #endif |
597 | 90 | #if defined(HAS_I444TOARGBROW_AVX2) |
598 | 90 | if (TestCpuFlag(kCpuHasAVX2)) { |
599 | 90 | I444ToARGBRow = I444ToARGBRow_Any_AVX2; |
600 | 90 | if (IS_ALIGNED(width, 16)) { |
601 | 19 | I444ToARGBRow = I444ToARGBRow_AVX2; |
602 | 19 | } |
603 | 90 | } |
604 | 90 | #endif |
605 | | #if defined(HAS_I444TOARGBROW_NEON) |
606 | | if (TestCpuFlag(kCpuHasNEON)) { |
607 | | I444ToARGBRow = I444ToARGBRow_Any_NEON; |
608 | | if (IS_ALIGNED(width, 8)) { |
609 | | I444ToARGBRow = I444ToARGBRow_NEON; |
610 | | } |
611 | | } |
612 | | #endif |
613 | | #if defined(HAS_I444TOARGBROW_SVE2) |
614 | | if (TestCpuFlag(kCpuHasSVE2)) { |
615 | | I444ToARGBRow = I444ToARGBRow_SVE2; |
616 | | } |
617 | | #endif |
618 | | #if defined(HAS_I444TOARGBROW_SME) |
619 | | if (TestCpuFlag(kCpuHasSME)) { |
620 | | I444ToARGBRow = I444ToARGBRow_SME; |
621 | | } |
622 | | #endif |
623 | | #if defined(HAS_I444TOARGBROW_LSX) |
624 | | if (TestCpuFlag(kCpuHasLSX)) { |
625 | | I444ToARGBRow = I444ToARGBRow_Any_LSX; |
626 | | if (IS_ALIGNED(width, 16)) { |
627 | | I444ToARGBRow = I444ToARGBRow_LSX; |
628 | | } |
629 | | } |
630 | | #endif |
631 | | #if defined(HAS_I444TOARGBROW_RVV) |
632 | | if (TestCpuFlag(kCpuHasRVV)) { |
633 | | I444ToARGBRow = I444ToARGBRow_RVV; |
634 | | } |
635 | | #endif |
636 | | |
637 | 180 | for (y = 0; y < height; ++y) { |
638 | 90 | I444ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
639 | 90 | dst_argb += dst_stride_argb; |
640 | 90 | src_y += src_stride_y; |
641 | 90 | src_u += src_stride_u; |
642 | 90 | src_v += src_stride_v; |
643 | 90 | } |
644 | 90 | return 0; |
645 | 90 | } |
646 | | |
647 | | // Convert I444 to ARGB. |
648 | | LIBYUV_API |
649 | | int I444ToARGB(const uint8_t* src_y, |
650 | | int src_stride_y, |
651 | | const uint8_t* src_u, |
652 | | int src_stride_u, |
653 | | const uint8_t* src_v, |
654 | | int src_stride_v, |
655 | | uint8_t* dst_argb, |
656 | | int dst_stride_argb, |
657 | | int width, |
658 | 0 | int height) { |
659 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
660 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
661 | 0 | &kYuvI601Constants, width, height); |
662 | 0 | } |
663 | | |
664 | | // Convert I444 to ABGR. |
665 | | LIBYUV_API |
666 | | int I444ToABGR(const uint8_t* src_y, |
667 | | int src_stride_y, |
668 | | const uint8_t* src_u, |
669 | | int src_stride_u, |
670 | | const uint8_t* src_v, |
671 | | int src_stride_v, |
672 | | uint8_t* dst_abgr, |
673 | | int dst_stride_abgr, |
674 | | int width, |
675 | 0 | int height) { |
676 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
677 | 0 | src_stride_v, // Swap U and V |
678 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
679 | 0 | &kYvuI601Constants, // Use Yvu matrix |
680 | 0 | width, height); |
681 | 0 | } |
682 | | |
683 | | // Convert J444 to ARGB. |
684 | | LIBYUV_API |
685 | | int J444ToARGB(const uint8_t* src_y, |
686 | | int src_stride_y, |
687 | | const uint8_t* src_u, |
688 | | int src_stride_u, |
689 | | const uint8_t* src_v, |
690 | | int src_stride_v, |
691 | | uint8_t* dst_argb, |
692 | | int dst_stride_argb, |
693 | | int width, |
694 | 0 | int height) { |
695 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
696 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
697 | 0 | &kYuvJPEGConstants, width, height); |
698 | 0 | } |
699 | | |
700 | | // Convert J444 to ABGR. |
701 | | LIBYUV_API |
702 | | int J444ToABGR(const uint8_t* src_y, |
703 | | int src_stride_y, |
704 | | const uint8_t* src_u, |
705 | | int src_stride_u, |
706 | | const uint8_t* src_v, |
707 | | int src_stride_v, |
708 | | uint8_t* dst_abgr, |
709 | | int dst_stride_abgr, |
710 | | int width, |
711 | 0 | int height) { |
712 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
713 | 0 | src_stride_v, // Swap U and V |
714 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
715 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
716 | 0 | width, height); |
717 | 0 | } |
718 | | |
719 | | // Convert H444 to ARGB. |
720 | | LIBYUV_API |
721 | | int H444ToARGB(const uint8_t* src_y, |
722 | | int src_stride_y, |
723 | | const uint8_t* src_u, |
724 | | int src_stride_u, |
725 | | const uint8_t* src_v, |
726 | | int src_stride_v, |
727 | | uint8_t* dst_argb, |
728 | | int dst_stride_argb, |
729 | | int width, |
730 | 0 | int height) { |
731 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
732 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
733 | 0 | &kYuvH709Constants, width, height); |
734 | 0 | } |
735 | | |
736 | | // Convert H444 to ABGR. |
737 | | LIBYUV_API |
738 | | int H444ToABGR(const uint8_t* src_y, |
739 | | int src_stride_y, |
740 | | const uint8_t* src_u, |
741 | | int src_stride_u, |
742 | | const uint8_t* src_v, |
743 | | int src_stride_v, |
744 | | uint8_t* dst_abgr, |
745 | | int dst_stride_abgr, |
746 | | int width, |
747 | 0 | int height) { |
748 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
749 | 0 | src_stride_v, // Swap U and V |
750 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
751 | 0 | &kYvuH709Constants, // Use Yvu matrix |
752 | 0 | width, height); |
753 | 0 | } |
754 | | |
755 | | // Convert U444 to ARGB. |
756 | | LIBYUV_API |
757 | | int U444ToARGB(const uint8_t* src_y, |
758 | | int src_stride_y, |
759 | | const uint8_t* src_u, |
760 | | int src_stride_u, |
761 | | const uint8_t* src_v, |
762 | | int src_stride_v, |
763 | | uint8_t* dst_argb, |
764 | | int dst_stride_argb, |
765 | | int width, |
766 | 0 | int height) { |
767 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
768 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
769 | 0 | &kYuv2020Constants, width, height); |
770 | 0 | } |
771 | | |
772 | | // Convert U444 to ABGR. |
773 | | LIBYUV_API |
774 | | int U444ToABGR(const uint8_t* src_y, |
775 | | int src_stride_y, |
776 | | const uint8_t* src_u, |
777 | | int src_stride_u, |
778 | | const uint8_t* src_v, |
779 | | int src_stride_v, |
780 | | uint8_t* dst_abgr, |
781 | | int dst_stride_abgr, |
782 | | int width, |
783 | 0 | int height) { |
784 | 0 | return I444ToARGBMatrix(src_y, src_stride_y, src_v, |
785 | 0 | src_stride_v, // Swap U and V |
786 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
787 | 0 | &kYvu2020Constants, // Use Yvu matrix |
788 | 0 | width, height); |
789 | 0 | } |
790 | | |
791 | | // Convert I444 to RGB24 with matrix. |
792 | | LIBYUV_API |
793 | | int I444ToRGB24Matrix(const uint8_t* src_y, |
794 | | int src_stride_y, |
795 | | const uint8_t* src_u, |
796 | | int src_stride_u, |
797 | | const uint8_t* src_v, |
798 | | int src_stride_v, |
799 | | uint8_t* dst_rgb24, |
800 | | int dst_stride_rgb24, |
801 | | const struct YuvConstants* yuvconstants, |
802 | | int width, |
803 | 123 | int height) { |
804 | 123 | int y; |
805 | 123 | void (*I444ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
806 | 123 | const uint8_t* v_buf, uint8_t* rgb_buf, |
807 | 123 | const struct YuvConstants* yuvconstants, int width) = |
808 | 123 | I444ToRGB24Row_C; |
809 | 123 | assert(yuvconstants); |
810 | 123 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
811 | 0 | return -1; |
812 | 0 | } |
813 | | // Negative height means invert the image. |
814 | 123 | if (height < 0) { |
815 | 0 | height = -height; |
816 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
817 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
818 | 0 | } |
819 | | // Coalesce rows. |
820 | 123 | if (src_stride_y == width && src_stride_u == width && src_stride_v == width && |
821 | 123 | dst_stride_rgb24 == width * 3) { |
822 | 123 | width *= height; |
823 | 123 | height = 1; |
824 | 123 | src_stride_y = src_stride_u = src_stride_v = dst_stride_rgb24 = 0; |
825 | 123 | } |
826 | 123 | #if defined(HAS_I444TORGB24ROW_SSSE3) |
827 | 123 | if (TestCpuFlag(kCpuHasSSSE3)) { |
828 | 123 | I444ToRGB24Row = I444ToRGB24Row_Any_SSSE3; |
829 | 123 | if (IS_ALIGNED(width, 16)) { |
830 | 52 | I444ToRGB24Row = I444ToRGB24Row_SSSE3; |
831 | 52 | } |
832 | 123 | } |
833 | 123 | #endif |
834 | 123 | #if defined(HAS_I444TORGB24ROW_AVX2) |
835 | 123 | if (TestCpuFlag(kCpuHasAVX2)) { |
836 | 123 | I444ToRGB24Row = I444ToRGB24Row_Any_AVX2; |
837 | 123 | if (IS_ALIGNED(width, 32)) { |
838 | 28 | I444ToRGB24Row = I444ToRGB24Row_AVX2; |
839 | 28 | } |
840 | 123 | } |
841 | 123 | #endif |
842 | | #if defined(HAS_I444TORGB24ROW_NEON) |
843 | | if (TestCpuFlag(kCpuHasNEON)) { |
844 | | I444ToRGB24Row = I444ToRGB24Row_Any_NEON; |
845 | | if (IS_ALIGNED(width, 8)) { |
846 | | I444ToRGB24Row = I444ToRGB24Row_NEON; |
847 | | } |
848 | | } |
849 | | #endif |
850 | | #if defined(HAS_I444TORGB24ROW_SVE2) |
851 | | if (TestCpuFlag(kCpuHasSVE2)) { |
852 | | I444ToRGB24Row = I444ToRGB24Row_SVE2; |
853 | | } |
854 | | #endif |
855 | | #if defined(HAS_I444TORGB24ROW_SME) |
856 | | if (TestCpuFlag(kCpuHasSME)) { |
857 | | I444ToRGB24Row = I444ToRGB24Row_SME; |
858 | | } |
859 | | #endif |
860 | | #if defined(HAS_I444TORGB24ROW_LSX) |
861 | | if (TestCpuFlag(kCpuHasLSX)) { |
862 | | I444ToRGB24Row = I444ToRGB24Row_Any_LSX; |
863 | | if (IS_ALIGNED(width, 16)) { |
864 | | I444ToRGB24Row = I444ToRGB24Row_LSX; |
865 | | } |
866 | | } |
867 | | #endif |
868 | | #if defined(HAS_I444TORGB24ROW_RVV) |
869 | | if (TestCpuFlag(kCpuHasRVV)) { |
870 | | I444ToRGB24Row = I444ToRGB24Row_RVV; |
871 | | } |
872 | | #endif |
873 | | |
874 | 246 | for (y = 0; y < height; ++y) { |
875 | 123 | I444ToRGB24Row(src_y, src_u, src_v, dst_rgb24, yuvconstants, width); |
876 | 123 | dst_rgb24 += dst_stride_rgb24; |
877 | 123 | src_y += src_stride_y; |
878 | 123 | src_u += src_stride_u; |
879 | 123 | src_v += src_stride_v; |
880 | 123 | } |
881 | 123 | return 0; |
882 | 123 | } |
883 | | |
884 | | // Convert I444 to RGB24. |
885 | | LIBYUV_API |
886 | | int I444ToRGB24(const uint8_t* src_y, |
887 | | int src_stride_y, |
888 | | const uint8_t* src_u, |
889 | | int src_stride_u, |
890 | | const uint8_t* src_v, |
891 | | int src_stride_v, |
892 | | uint8_t* dst_rgb24, |
893 | | int dst_stride_rgb24, |
894 | | int width, |
895 | 0 | int height) { |
896 | 0 | return I444ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
897 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
898 | 0 | &kYuvI601Constants, width, height); |
899 | 0 | } |
900 | | |
901 | | // Convert I444 to RAW. |
902 | | LIBYUV_API |
903 | | int I444ToRAW(const uint8_t* src_y, |
904 | | int src_stride_y, |
905 | | const uint8_t* src_u, |
906 | | int src_stride_u, |
907 | | const uint8_t* src_v, |
908 | | int src_stride_v, |
909 | | uint8_t* dst_raw, |
910 | | int dst_stride_raw, |
911 | | int width, |
912 | 0 | int height) { |
913 | 0 | return I444ToRGB24Matrix(src_y, src_stride_y, src_v, |
914 | 0 | src_stride_v, // Swap U and V |
915 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
916 | 0 | &kYvuI601Constants, // Use Yvu matrix |
917 | 0 | width, height); |
918 | 0 | } |
919 | | |
920 | | // Convert 10 bit YUV to ARGB with matrix. |
921 | | // TODO(fbarchard): Consider passing scale multiplier to I210ToARGB to |
922 | | // multiply 10 bit yuv into high bits to allow any number of bits. |
923 | | LIBYUV_API |
924 | | int I010ToAR30Matrix(const uint16_t* src_y, |
925 | | int src_stride_y, |
926 | | const uint16_t* src_u, |
927 | | int src_stride_u, |
928 | | const uint16_t* src_v, |
929 | | int src_stride_v, |
930 | | uint8_t* dst_ar30, |
931 | | int dst_stride_ar30, |
932 | | const struct YuvConstants* yuvconstants, |
933 | | int width, |
934 | 0 | int height) { |
935 | 0 | int y; |
936 | 0 | void (*I210ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
937 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
938 | 0 | const struct YuvConstants* yuvconstants, int width) = |
939 | 0 | I210ToAR30Row_C; |
940 | 0 | assert(yuvconstants); |
941 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
942 | 0 | return -1; |
943 | 0 | } |
944 | | // Negative height means invert the image. |
945 | 0 | if (height < 0) { |
946 | 0 | height = -height; |
947 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
948 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
949 | 0 | } |
950 | | #if defined(HAS_I210TOAR30ROW_NEON) |
951 | | if (TestCpuFlag(kCpuHasNEON)) { |
952 | | I210ToAR30Row = I210ToAR30Row_Any_NEON; |
953 | | if (IS_ALIGNED(width, 8)) { |
954 | | I210ToAR30Row = I210ToAR30Row_NEON; |
955 | | } |
956 | | } |
957 | | #endif |
958 | | #if defined(HAS_I210TOAR30ROW_SVE2) |
959 | | if (TestCpuFlag(kCpuHasSVE2)) { |
960 | | I210ToAR30Row = I210ToAR30Row_SVE2; |
961 | | } |
962 | | #endif |
963 | | #if defined(HAS_I210TOAR30ROW_SME) |
964 | | if (TestCpuFlag(kCpuHasSME)) { |
965 | | I210ToAR30Row = I210ToAR30Row_SME; |
966 | | } |
967 | | #endif |
968 | 0 | #if defined(HAS_I210TOAR30ROW_SSSE3) |
969 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
970 | 0 | I210ToAR30Row = I210ToAR30Row_Any_SSSE3; |
971 | 0 | if (IS_ALIGNED(width, 8)) { |
972 | 0 | I210ToAR30Row = I210ToAR30Row_SSSE3; |
973 | 0 | } |
974 | 0 | } |
975 | 0 | #endif |
976 | 0 | #if defined(HAS_I210TOAR30ROW_AVX2) |
977 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
978 | 0 | I210ToAR30Row = I210ToAR30Row_Any_AVX2; |
979 | 0 | if (IS_ALIGNED(width, 16)) { |
980 | 0 | I210ToAR30Row = I210ToAR30Row_AVX2; |
981 | 0 | } |
982 | 0 | } |
983 | 0 | #endif |
984 | 0 | for (y = 0; y < height; ++y) { |
985 | 0 | I210ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
986 | 0 | dst_ar30 += dst_stride_ar30; |
987 | 0 | src_y += src_stride_y; |
988 | 0 | if (y & 1) { |
989 | 0 | src_u += src_stride_u; |
990 | 0 | src_v += src_stride_v; |
991 | 0 | } |
992 | 0 | } |
993 | 0 | return 0; |
994 | 0 | } |
995 | | |
996 | | // Convert I010 to AR30. |
997 | | LIBYUV_API |
998 | | int I010ToAR30(const uint16_t* src_y, |
999 | | int src_stride_y, |
1000 | | const uint16_t* src_u, |
1001 | | int src_stride_u, |
1002 | | const uint16_t* src_v, |
1003 | | int src_stride_v, |
1004 | | uint8_t* dst_ar30, |
1005 | | int dst_stride_ar30, |
1006 | | int width, |
1007 | 0 | int height) { |
1008 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1009 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1010 | 0 | &kYuvI601Constants, width, height); |
1011 | 0 | } |
1012 | | |
1013 | | // Convert H010 to AR30. |
1014 | | LIBYUV_API |
1015 | | int H010ToAR30(const uint16_t* src_y, |
1016 | | int src_stride_y, |
1017 | | const uint16_t* src_u, |
1018 | | int src_stride_u, |
1019 | | const uint16_t* src_v, |
1020 | | int src_stride_v, |
1021 | | uint8_t* dst_ar30, |
1022 | | int dst_stride_ar30, |
1023 | | int width, |
1024 | 0 | int height) { |
1025 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1026 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1027 | 0 | &kYuvH709Constants, width, height); |
1028 | 0 | } |
1029 | | |
1030 | | // Convert U010 to AR30. |
1031 | | LIBYUV_API |
1032 | | int U010ToAR30(const uint16_t* src_y, |
1033 | | int src_stride_y, |
1034 | | const uint16_t* src_u, |
1035 | | int src_stride_u, |
1036 | | const uint16_t* src_v, |
1037 | | int src_stride_v, |
1038 | | uint8_t* dst_ar30, |
1039 | | int dst_stride_ar30, |
1040 | | int width, |
1041 | 0 | int height) { |
1042 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1043 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1044 | 0 | &kYuv2020Constants, width, height); |
1045 | 0 | } |
1046 | | |
1047 | | // Convert I010 to AB30. |
1048 | | LIBYUV_API |
1049 | | int I010ToAB30(const uint16_t* src_y, |
1050 | | int src_stride_y, |
1051 | | const uint16_t* src_u, |
1052 | | int src_stride_u, |
1053 | | const uint16_t* src_v, |
1054 | | int src_stride_v, |
1055 | | uint8_t* dst_ab30, |
1056 | | int dst_stride_ab30, |
1057 | | int width, |
1058 | 0 | int height) { |
1059 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1060 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1061 | 0 | &kYvuI601Constants, width, height); |
1062 | 0 | } |
1063 | | |
1064 | | // Convert H010 to AB30. |
1065 | | LIBYUV_API |
1066 | | int H010ToAB30(const uint16_t* src_y, |
1067 | | int src_stride_y, |
1068 | | const uint16_t* src_u, |
1069 | | int src_stride_u, |
1070 | | const uint16_t* src_v, |
1071 | | int src_stride_v, |
1072 | | uint8_t* dst_ab30, |
1073 | | int dst_stride_ab30, |
1074 | | int width, |
1075 | 0 | int height) { |
1076 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1077 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1078 | 0 | &kYvuH709Constants, width, height); |
1079 | 0 | } |
1080 | | |
1081 | | // Convert U010 to AB30. |
1082 | | LIBYUV_API |
1083 | | int U010ToAB30(const uint16_t* src_y, |
1084 | | int src_stride_y, |
1085 | | const uint16_t* src_u, |
1086 | | int src_stride_u, |
1087 | | const uint16_t* src_v, |
1088 | | int src_stride_v, |
1089 | | uint8_t* dst_ab30, |
1090 | | int dst_stride_ab30, |
1091 | | int width, |
1092 | 0 | int height) { |
1093 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1094 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1095 | 0 | &kYuv2020Constants, width, height); |
1096 | 0 | } |
1097 | | |
1098 | | // Convert 12 bit YUV to ARGB with matrix. |
1099 | | // TODO(fbarchard): Consider passing scale multiplier to I212ToARGB to |
1100 | | // multiply 12 bit yuv into high bits to allow any number of bits. |
1101 | | LIBYUV_API |
1102 | | int I012ToAR30Matrix(const uint16_t* src_y, |
1103 | | int src_stride_y, |
1104 | | const uint16_t* src_u, |
1105 | | int src_stride_u, |
1106 | | const uint16_t* src_v, |
1107 | | int src_stride_v, |
1108 | | uint8_t* dst_ar30, |
1109 | | int dst_stride_ar30, |
1110 | | const struct YuvConstants* yuvconstants, |
1111 | | int width, |
1112 | 0 | int height) { |
1113 | 0 | int y; |
1114 | 0 | void (*I212ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
1115 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1116 | 0 | const struct YuvConstants* yuvconstants, int width) = |
1117 | 0 | I212ToAR30Row_C; |
1118 | 0 | assert(yuvconstants); |
1119 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
1120 | 0 | return -1; |
1121 | 0 | } |
1122 | | // Negative height means invert the image. |
1123 | 0 | if (height < 0) { |
1124 | 0 | height = -height; |
1125 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
1126 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
1127 | 0 | } |
1128 | 0 | #if defined(HAS_I212TOAR30ROW_SSSE3) |
1129 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1130 | 0 | I212ToAR30Row = I212ToAR30Row_Any_SSSE3; |
1131 | 0 | if (IS_ALIGNED(width, 8)) { |
1132 | 0 | I212ToAR30Row = I212ToAR30Row_SSSE3; |
1133 | 0 | } |
1134 | 0 | } |
1135 | 0 | #endif |
1136 | 0 | #if defined(HAS_I212TOAR30ROW_AVX2) |
1137 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1138 | 0 | I212ToAR30Row = I212ToAR30Row_Any_AVX2; |
1139 | 0 | if (IS_ALIGNED(width, 16)) { |
1140 | 0 | I212ToAR30Row = I212ToAR30Row_AVX2; |
1141 | 0 | } |
1142 | 0 | } |
1143 | 0 | #endif |
1144 | | #if defined(HAS_I212TOAR30ROW_NEON) |
1145 | | if (TestCpuFlag(kCpuHasNEON)) { |
1146 | | I212ToAR30Row = I212ToAR30Row_Any_NEON; |
1147 | | if (IS_ALIGNED(width, 8)) { |
1148 | | I212ToAR30Row = I212ToAR30Row_NEON; |
1149 | | } |
1150 | | } |
1151 | | #endif |
1152 | | #if defined(HAS_I212TOAR30ROW_SVE2) |
1153 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1154 | | I212ToAR30Row = I212ToAR30Row_SVE2; |
1155 | | } |
1156 | | #endif |
1157 | | #if defined(HAS_I212TOAR30ROW_SME) |
1158 | | if (TestCpuFlag(kCpuHasSME)) { |
1159 | | I212ToAR30Row = I212ToAR30Row_SME; |
1160 | | } |
1161 | | #endif |
1162 | 0 | for (y = 0; y < height; ++y) { |
1163 | 0 | I212ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
1164 | 0 | dst_ar30 += dst_stride_ar30; |
1165 | 0 | src_y += src_stride_y; |
1166 | 0 | if (y & 1) { |
1167 | 0 | src_u += src_stride_u; |
1168 | 0 | src_v += src_stride_v; |
1169 | 0 | } |
1170 | 0 | } |
1171 | 0 | return 0; |
1172 | 0 | } |
1173 | | |
1174 | | // Convert 10 bit YUV to ARGB with matrix. |
1175 | | // TODO(fbarchard): Consider passing scale multiplier to I210ToARGB to |
1176 | | // multiply 10 bit yuv into high bits to allow any number of bits. |
1177 | | LIBYUV_API |
1178 | | int I210ToAR30Matrix(const uint16_t* src_y, |
1179 | | int src_stride_y, |
1180 | | const uint16_t* src_u, |
1181 | | int src_stride_u, |
1182 | | const uint16_t* src_v, |
1183 | | int src_stride_v, |
1184 | | uint8_t* dst_ar30, |
1185 | | int dst_stride_ar30, |
1186 | | const struct YuvConstants* yuvconstants, |
1187 | | int width, |
1188 | 0 | int height) { |
1189 | 0 | int y; |
1190 | 0 | void (*I210ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
1191 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1192 | 0 | const struct YuvConstants* yuvconstants, int width) = |
1193 | 0 | I210ToAR30Row_C; |
1194 | 0 | assert(yuvconstants); |
1195 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
1196 | 0 | return -1; |
1197 | 0 | } |
1198 | | // Negative height means invert the image. |
1199 | 0 | if (height < 0) { |
1200 | 0 | height = -height; |
1201 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
1202 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
1203 | 0 | } |
1204 | | #if defined(HAS_I210TOAR30ROW_NEON) |
1205 | | if (TestCpuFlag(kCpuHasNEON)) { |
1206 | | I210ToAR30Row = I210ToAR30Row_Any_NEON; |
1207 | | if (IS_ALIGNED(width, 8)) { |
1208 | | I210ToAR30Row = I210ToAR30Row_NEON; |
1209 | | } |
1210 | | } |
1211 | | #endif |
1212 | | #if defined(HAS_I210TOAR30ROW_SVE2) |
1213 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1214 | | I210ToAR30Row = I210ToAR30Row_SVE2; |
1215 | | } |
1216 | | #endif |
1217 | | #if defined(HAS_I210TOAR30ROW_SME) |
1218 | | if (TestCpuFlag(kCpuHasSME)) { |
1219 | | I210ToAR30Row = I210ToAR30Row_SME; |
1220 | | } |
1221 | | #endif |
1222 | 0 | #if defined(HAS_I210TOAR30ROW_SSSE3) |
1223 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1224 | 0 | I210ToAR30Row = I210ToAR30Row_Any_SSSE3; |
1225 | 0 | if (IS_ALIGNED(width, 8)) { |
1226 | 0 | I210ToAR30Row = I210ToAR30Row_SSSE3; |
1227 | 0 | } |
1228 | 0 | } |
1229 | 0 | #endif |
1230 | 0 | #if defined(HAS_I210TOAR30ROW_AVX2) |
1231 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1232 | 0 | I210ToAR30Row = I210ToAR30Row_Any_AVX2; |
1233 | 0 | if (IS_ALIGNED(width, 16)) { |
1234 | 0 | I210ToAR30Row = I210ToAR30Row_AVX2; |
1235 | 0 | } |
1236 | 0 | } |
1237 | 0 | #endif |
1238 | 0 | for (y = 0; y < height; ++y) { |
1239 | 0 | I210ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
1240 | 0 | dst_ar30 += dst_stride_ar30; |
1241 | 0 | src_y += src_stride_y; |
1242 | 0 | src_u += src_stride_u; |
1243 | 0 | src_v += src_stride_v; |
1244 | 0 | } |
1245 | 0 | return 0; |
1246 | 0 | } |
1247 | | |
1248 | | // Convert I210 to AR30. |
1249 | | LIBYUV_API |
1250 | | int I210ToAR30(const uint16_t* src_y, |
1251 | | int src_stride_y, |
1252 | | const uint16_t* src_u, |
1253 | | int src_stride_u, |
1254 | | const uint16_t* src_v, |
1255 | | int src_stride_v, |
1256 | | uint8_t* dst_ar30, |
1257 | | int dst_stride_ar30, |
1258 | | int width, |
1259 | 0 | int height) { |
1260 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1261 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1262 | 0 | &kYuvI601Constants, width, height); |
1263 | 0 | } |
1264 | | |
1265 | | // Convert H210 to AR30. |
1266 | | LIBYUV_API |
1267 | | int H210ToAR30(const uint16_t* src_y, |
1268 | | int src_stride_y, |
1269 | | const uint16_t* src_u, |
1270 | | int src_stride_u, |
1271 | | const uint16_t* src_v, |
1272 | | int src_stride_v, |
1273 | | uint8_t* dst_ar30, |
1274 | | int dst_stride_ar30, |
1275 | | int width, |
1276 | 0 | int height) { |
1277 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1278 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1279 | 0 | &kYuvH709Constants, width, height); |
1280 | 0 | } |
1281 | | |
1282 | | // Convert U210 to AR30. |
1283 | | LIBYUV_API |
1284 | | int U210ToAR30(const uint16_t* src_y, |
1285 | | int src_stride_y, |
1286 | | const uint16_t* src_u, |
1287 | | int src_stride_u, |
1288 | | const uint16_t* src_v, |
1289 | | int src_stride_v, |
1290 | | uint8_t* dst_ar30, |
1291 | | int dst_stride_ar30, |
1292 | | int width, |
1293 | 0 | int height) { |
1294 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1295 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
1296 | 0 | &kYuv2020Constants, width, height); |
1297 | 0 | } |
1298 | | |
1299 | | // Convert I210 to AB30. |
1300 | | LIBYUV_API |
1301 | | int I210ToAB30(const uint16_t* src_y, |
1302 | | int src_stride_y, |
1303 | | const uint16_t* src_u, |
1304 | | int src_stride_u, |
1305 | | const uint16_t* src_v, |
1306 | | int src_stride_v, |
1307 | | uint8_t* dst_ab30, |
1308 | | int dst_stride_ab30, |
1309 | | int width, |
1310 | 0 | int height) { |
1311 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1312 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1313 | 0 | &kYvuI601Constants, width, height); |
1314 | 0 | } |
1315 | | |
1316 | | // Convert H210 to AB30. |
1317 | | LIBYUV_API |
1318 | | int H210ToAB30(const uint16_t* src_y, |
1319 | | int src_stride_y, |
1320 | | const uint16_t* src_u, |
1321 | | int src_stride_u, |
1322 | | const uint16_t* src_v, |
1323 | | int src_stride_v, |
1324 | | uint8_t* dst_ab30, |
1325 | | int dst_stride_ab30, |
1326 | | int width, |
1327 | 0 | int height) { |
1328 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1329 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1330 | 0 | &kYvuH709Constants, width, height); |
1331 | 0 | } |
1332 | | |
1333 | | // Convert U210 to AB30. |
1334 | | LIBYUV_API |
1335 | | int U210ToAB30(const uint16_t* src_y, |
1336 | | int src_stride_y, |
1337 | | const uint16_t* src_u, |
1338 | | int src_stride_u, |
1339 | | const uint16_t* src_v, |
1340 | | int src_stride_v, |
1341 | | uint8_t* dst_ab30, |
1342 | | int dst_stride_ab30, |
1343 | | int width, |
1344 | 0 | int height) { |
1345 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
1346 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
1347 | 0 | &kYuv2020Constants, width, height); |
1348 | 0 | } |
1349 | | |
1350 | | LIBYUV_API |
1351 | | int I410ToAR30Matrix(const uint16_t* src_y, |
1352 | | int src_stride_y, |
1353 | | const uint16_t* src_u, |
1354 | | int src_stride_u, |
1355 | | const uint16_t* src_v, |
1356 | | int src_stride_v, |
1357 | | uint8_t* dst_ar30, |
1358 | | int dst_stride_ar30, |
1359 | | const struct YuvConstants* yuvconstants, |
1360 | | int width, |
1361 | 0 | int height) { |
1362 | 0 | int y; |
1363 | 0 | void (*I410ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
1364 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1365 | 0 | const struct YuvConstants* yuvconstants, int width) = |
1366 | 0 | I410ToAR30Row_C; |
1367 | 0 | assert(yuvconstants); |
1368 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
1369 | 0 | return -1; |
1370 | 0 | } |
1371 | | // Negative height means invert the image. |
1372 | 0 | if (height < 0) { |
1373 | 0 | height = -height; |
1374 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
1375 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
1376 | 0 | } |
1377 | | #if defined(HAS_I410TOAR30ROW_NEON) |
1378 | | if (TestCpuFlag(kCpuHasNEON)) { |
1379 | | I410ToAR30Row = I410ToAR30Row_Any_NEON; |
1380 | | if (IS_ALIGNED(width, 8)) { |
1381 | | I410ToAR30Row = I410ToAR30Row_NEON; |
1382 | | } |
1383 | | } |
1384 | | #endif |
1385 | | #if defined(HAS_I410TOAR30ROW_SVE2) |
1386 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1387 | | I410ToAR30Row = I410ToAR30Row_SVE2; |
1388 | | } |
1389 | | #endif |
1390 | | #if defined(HAS_I410TOAR30ROW_SME) |
1391 | | if (TestCpuFlag(kCpuHasSME)) { |
1392 | | I410ToAR30Row = I410ToAR30Row_SME; |
1393 | | } |
1394 | | #endif |
1395 | 0 | #if defined(HAS_I410TOAR30ROW_SSSE3) |
1396 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1397 | 0 | I410ToAR30Row = I410ToAR30Row_Any_SSSE3; |
1398 | 0 | if (IS_ALIGNED(width, 8)) { |
1399 | 0 | I410ToAR30Row = I410ToAR30Row_SSSE3; |
1400 | 0 | } |
1401 | 0 | } |
1402 | 0 | #endif |
1403 | 0 | #if defined(HAS_I410TOAR30ROW_AVX2) |
1404 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1405 | 0 | I410ToAR30Row = I410ToAR30Row_Any_AVX2; |
1406 | 0 | if (IS_ALIGNED(width, 16)) { |
1407 | 0 | I410ToAR30Row = I410ToAR30Row_AVX2; |
1408 | 0 | } |
1409 | 0 | } |
1410 | 0 | #endif |
1411 | 0 | for (y = 0; y < height; ++y) { |
1412 | 0 | I410ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
1413 | 0 | dst_ar30 += dst_stride_ar30; |
1414 | 0 | src_y += src_stride_y; |
1415 | 0 | src_u += src_stride_u; |
1416 | 0 | src_v += src_stride_v; |
1417 | 0 | } |
1418 | 0 | return 0; |
1419 | 0 | } |
1420 | | |
1421 | | // Convert 10 bit YUV to ARGB with matrix. |
1422 | | LIBYUV_API |
1423 | | int I010ToARGBMatrix(const uint16_t* src_y, |
1424 | | int src_stride_y, |
1425 | | const uint16_t* src_u, |
1426 | | int src_stride_u, |
1427 | | const uint16_t* src_v, |
1428 | | int src_stride_v, |
1429 | | uint8_t* dst_argb, |
1430 | | int dst_stride_argb, |
1431 | | const struct YuvConstants* yuvconstants, |
1432 | | int width, |
1433 | 97 | int height) { |
1434 | 97 | int y; |
1435 | 97 | void (*I210ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1436 | 97 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1437 | 97 | const struct YuvConstants* yuvconstants, int width) = |
1438 | 97 | I210ToARGBRow_C; |
1439 | 97 | assert(yuvconstants); |
1440 | 97 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1441 | 0 | return -1; |
1442 | 0 | } |
1443 | | // Negative height means invert the image. |
1444 | 97 | if (height < 0) { |
1445 | 0 | height = -height; |
1446 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1447 | 0 | dst_stride_argb = -dst_stride_argb; |
1448 | 0 | } |
1449 | 97 | #if defined(HAS_I210TOARGBROW_SSSE3) |
1450 | 97 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1451 | 97 | I210ToARGBRow = I210ToARGBRow_Any_SSSE3; |
1452 | 97 | if (IS_ALIGNED(width, 8)) { |
1453 | 33 | I210ToARGBRow = I210ToARGBRow_SSSE3; |
1454 | 33 | } |
1455 | 97 | } |
1456 | 97 | #endif |
1457 | | #if defined(HAS_I210TOARGBROW_NEON) |
1458 | | if (TestCpuFlag(kCpuHasNEON)) { |
1459 | | I210ToARGBRow = I210ToARGBRow_Any_NEON; |
1460 | | if (IS_ALIGNED(width, 8)) { |
1461 | | I210ToARGBRow = I210ToARGBRow_NEON; |
1462 | | } |
1463 | | } |
1464 | | #endif |
1465 | | #if defined(HAS_I210TOARGBROW_SVE2) |
1466 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1467 | | I210ToARGBRow = I210ToARGBRow_SVE2; |
1468 | | } |
1469 | | #endif |
1470 | | #if defined(HAS_I210TOARGBROW_SME) |
1471 | | if (TestCpuFlag(kCpuHasSME)) { |
1472 | | I210ToARGBRow = I210ToARGBRow_SME; |
1473 | | } |
1474 | | #endif |
1475 | 97 | #if defined(HAS_I210TOARGBROW_AVX2) |
1476 | 97 | if (TestCpuFlag(kCpuHasAVX2)) { |
1477 | 97 | I210ToARGBRow = I210ToARGBRow_Any_AVX2; |
1478 | 97 | if (IS_ALIGNED(width, 16)) { |
1479 | 16 | I210ToARGBRow = I210ToARGBRow_AVX2; |
1480 | 16 | } |
1481 | 97 | } |
1482 | 97 | #endif |
1483 | 1.49k | for (y = 0; y < height; ++y) { |
1484 | 1.40k | I210ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1485 | 1.40k | dst_argb += dst_stride_argb; |
1486 | 1.40k | src_y += src_stride_y; |
1487 | 1.40k | if (y & 1) { |
1488 | 691 | src_u += src_stride_u; |
1489 | 691 | src_v += src_stride_v; |
1490 | 691 | } |
1491 | 1.40k | } |
1492 | 97 | return 0; |
1493 | 97 | } |
1494 | | |
1495 | | // Convert I010 to ARGB. |
1496 | | LIBYUV_API |
1497 | | int I010ToARGB(const uint16_t* src_y, |
1498 | | int src_stride_y, |
1499 | | const uint16_t* src_u, |
1500 | | int src_stride_u, |
1501 | | const uint16_t* src_v, |
1502 | | int src_stride_v, |
1503 | | uint8_t* dst_argb, |
1504 | | int dst_stride_argb, |
1505 | | int width, |
1506 | 0 | int height) { |
1507 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1508 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1509 | 0 | &kYuvI601Constants, width, height); |
1510 | 0 | } |
1511 | | |
1512 | | // Convert I010 to ABGR. |
1513 | | LIBYUV_API |
1514 | | int I010ToABGR(const uint16_t* src_y, |
1515 | | int src_stride_y, |
1516 | | const uint16_t* src_u, |
1517 | | int src_stride_u, |
1518 | | const uint16_t* src_v, |
1519 | | int src_stride_v, |
1520 | | uint8_t* dst_abgr, |
1521 | | int dst_stride_abgr, |
1522 | | int width, |
1523 | 0 | int height) { |
1524 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_v, |
1525 | 0 | src_stride_v, // Swap U and V |
1526 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1527 | 0 | &kYvuI601Constants, // Use Yvu matrix |
1528 | 0 | width, height); |
1529 | 0 | } |
1530 | | |
1531 | | // Convert H010 to ARGB. |
1532 | | LIBYUV_API |
1533 | | int H010ToARGB(const uint16_t* src_y, |
1534 | | int src_stride_y, |
1535 | | const uint16_t* src_u, |
1536 | | int src_stride_u, |
1537 | | const uint16_t* src_v, |
1538 | | int src_stride_v, |
1539 | | uint8_t* dst_argb, |
1540 | | int dst_stride_argb, |
1541 | | int width, |
1542 | 0 | int height) { |
1543 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1544 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1545 | 0 | &kYuvH709Constants, width, height); |
1546 | 0 | } |
1547 | | |
1548 | | // Convert H010 to ABGR. |
1549 | | LIBYUV_API |
1550 | | int H010ToABGR(const uint16_t* src_y, |
1551 | | int src_stride_y, |
1552 | | const uint16_t* src_u, |
1553 | | int src_stride_u, |
1554 | | const uint16_t* src_v, |
1555 | | int src_stride_v, |
1556 | | uint8_t* dst_abgr, |
1557 | | int dst_stride_abgr, |
1558 | | int width, |
1559 | 0 | int height) { |
1560 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_v, |
1561 | 0 | src_stride_v, // Swap U and V |
1562 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1563 | 0 | &kYvuH709Constants, // Use Yvu matrix |
1564 | 0 | width, height); |
1565 | 0 | } |
1566 | | |
1567 | | // Convert U010 to ARGB. |
1568 | | LIBYUV_API |
1569 | | int U010ToARGB(const uint16_t* src_y, |
1570 | | int src_stride_y, |
1571 | | const uint16_t* src_u, |
1572 | | int src_stride_u, |
1573 | | const uint16_t* src_v, |
1574 | | int src_stride_v, |
1575 | | uint8_t* dst_argb, |
1576 | | int dst_stride_argb, |
1577 | | int width, |
1578 | 0 | int height) { |
1579 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1580 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1581 | 0 | &kYuv2020Constants, width, height); |
1582 | 0 | } |
1583 | | |
1584 | | // Convert U010 to ABGR. |
1585 | | LIBYUV_API |
1586 | | int U010ToABGR(const uint16_t* src_y, |
1587 | | int src_stride_y, |
1588 | | const uint16_t* src_u, |
1589 | | int src_stride_u, |
1590 | | const uint16_t* src_v, |
1591 | | int src_stride_v, |
1592 | | uint8_t* dst_abgr, |
1593 | | int dst_stride_abgr, |
1594 | | int width, |
1595 | 0 | int height) { |
1596 | 0 | return I010ToARGBMatrix(src_y, src_stride_y, src_v, |
1597 | 0 | src_stride_v, // Swap U and V |
1598 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1599 | 0 | &kYvu2020Constants, // Use Yvu matrix |
1600 | 0 | width, height); |
1601 | 0 | } |
1602 | | |
1603 | | // Convert 12 bit YUV to ARGB with matrix. |
1604 | | LIBYUV_API |
1605 | | int I012ToARGBMatrix(const uint16_t* src_y, |
1606 | | int src_stride_y, |
1607 | | const uint16_t* src_u, |
1608 | | int src_stride_u, |
1609 | | const uint16_t* src_v, |
1610 | | int src_stride_v, |
1611 | | uint8_t* dst_argb, |
1612 | | int dst_stride_argb, |
1613 | | const struct YuvConstants* yuvconstants, |
1614 | | int width, |
1615 | 118 | int height) { |
1616 | 118 | int y; |
1617 | 118 | void (*I212ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1618 | 118 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1619 | 118 | const struct YuvConstants* yuvconstants, int width) = |
1620 | 118 | I212ToARGBRow_C; |
1621 | 118 | assert(yuvconstants); |
1622 | 118 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1623 | 0 | return -1; |
1624 | 0 | } |
1625 | | // Negative height means invert the image. |
1626 | 118 | if (height < 0) { |
1627 | 0 | height = -height; |
1628 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1629 | 0 | dst_stride_argb = -dst_stride_argb; |
1630 | 0 | } |
1631 | 118 | #if defined(HAS_I212TOARGBROW_SSSE3) |
1632 | 118 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1633 | 118 | I212ToARGBRow = I212ToARGBRow_Any_SSSE3; |
1634 | 118 | if (IS_ALIGNED(width, 8)) { |
1635 | 38 | I212ToARGBRow = I212ToARGBRow_SSSE3; |
1636 | 38 | } |
1637 | 118 | } |
1638 | 118 | #endif |
1639 | 118 | #if defined(HAS_I212TOARGBROW_AVX2) |
1640 | 118 | if (TestCpuFlag(kCpuHasAVX2)) { |
1641 | 118 | I212ToARGBRow = I212ToARGBRow_Any_AVX2; |
1642 | 118 | if (IS_ALIGNED(width, 16)) { |
1643 | 20 | I212ToARGBRow = I212ToARGBRow_AVX2; |
1644 | 20 | } |
1645 | 118 | } |
1646 | 118 | #endif |
1647 | | #if defined(HAS_I212TOARGBROW_NEON) |
1648 | | if (TestCpuFlag(kCpuHasNEON)) { |
1649 | | I212ToARGBRow = I212ToARGBRow_Any_NEON; |
1650 | | if (IS_ALIGNED(width, 8)) { |
1651 | | I212ToARGBRow = I212ToARGBRow_NEON; |
1652 | | } |
1653 | | } |
1654 | | #endif |
1655 | | #if defined(HAS_I212TOARGBROW_SVE2) |
1656 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1657 | | I212ToARGBRow = I212ToARGBRow_SVE2; |
1658 | | } |
1659 | | #endif |
1660 | | #if defined(HAS_I212TOARGBROW_SME) |
1661 | | if (TestCpuFlag(kCpuHasSME)) { |
1662 | | I212ToARGBRow = I212ToARGBRow_SME; |
1663 | | } |
1664 | | #endif |
1665 | 4.53k | for (y = 0; y < height; ++y) { |
1666 | 4.41k | I212ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1667 | 4.41k | dst_argb += dst_stride_argb; |
1668 | 4.41k | src_y += src_stride_y; |
1669 | 4.41k | if (y & 1) { |
1670 | 2.18k | src_u += src_stride_u; |
1671 | 2.18k | src_v += src_stride_v; |
1672 | 2.18k | } |
1673 | 4.41k | } |
1674 | 118 | return 0; |
1675 | 118 | } |
1676 | | |
1677 | | // Convert 10 bit 422 YUV to ARGB with matrix. |
1678 | | LIBYUV_API |
1679 | | int I210ToARGBMatrix(const uint16_t* src_y, |
1680 | | int src_stride_y, |
1681 | | const uint16_t* src_u, |
1682 | | int src_stride_u, |
1683 | | const uint16_t* src_v, |
1684 | | int src_stride_v, |
1685 | | uint8_t* dst_argb, |
1686 | | int dst_stride_argb, |
1687 | | const struct YuvConstants* yuvconstants, |
1688 | | int width, |
1689 | 87 | int height) { |
1690 | 87 | int y; |
1691 | 87 | void (*I210ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1692 | 87 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1693 | 87 | const struct YuvConstants* yuvconstants, int width) = |
1694 | 87 | I210ToARGBRow_C; |
1695 | 87 | assert(yuvconstants); |
1696 | 87 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1697 | 0 | return -1; |
1698 | 0 | } |
1699 | | // Negative height means invert the image. |
1700 | 87 | if (height < 0) { |
1701 | 0 | height = -height; |
1702 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1703 | 0 | dst_stride_argb = -dst_stride_argb; |
1704 | 0 | } |
1705 | 87 | #if defined(HAS_I210TOARGBROW_SSSE3) |
1706 | 87 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1707 | 87 | I210ToARGBRow = I210ToARGBRow_Any_SSSE3; |
1708 | 87 | if (IS_ALIGNED(width, 8)) { |
1709 | 34 | I210ToARGBRow = I210ToARGBRow_SSSE3; |
1710 | 34 | } |
1711 | 87 | } |
1712 | 87 | #endif |
1713 | | #if defined(HAS_I210TOARGBROW_NEON) |
1714 | | if (TestCpuFlag(kCpuHasNEON)) { |
1715 | | I210ToARGBRow = I210ToARGBRow_Any_NEON; |
1716 | | if (IS_ALIGNED(width, 8)) { |
1717 | | I210ToARGBRow = I210ToARGBRow_NEON; |
1718 | | } |
1719 | | } |
1720 | | #endif |
1721 | | #if defined(HAS_I210TOARGBROW_SVE2) |
1722 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1723 | | I210ToARGBRow = I210ToARGBRow_SVE2; |
1724 | | } |
1725 | | #endif |
1726 | | #if defined(HAS_I210TOARGBROW_SME) |
1727 | | if (TestCpuFlag(kCpuHasSME)) { |
1728 | | I210ToARGBRow = I210ToARGBRow_SME; |
1729 | | } |
1730 | | #endif |
1731 | 87 | #if defined(HAS_I210TOARGBROW_AVX2) |
1732 | 87 | if (TestCpuFlag(kCpuHasAVX2)) { |
1733 | 87 | I210ToARGBRow = I210ToARGBRow_Any_AVX2; |
1734 | 87 | if (IS_ALIGNED(width, 16)) { |
1735 | 17 | I210ToARGBRow = I210ToARGBRow_AVX2; |
1736 | 17 | } |
1737 | 87 | } |
1738 | 87 | #endif |
1739 | 2.29k | for (y = 0; y < height; ++y) { |
1740 | 2.21k | I210ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1741 | 2.21k | dst_argb += dst_stride_argb; |
1742 | 2.21k | src_y += src_stride_y; |
1743 | 2.21k | src_u += src_stride_u; |
1744 | 2.21k | src_v += src_stride_v; |
1745 | 2.21k | } |
1746 | 87 | return 0; |
1747 | 87 | } |
1748 | | |
1749 | | // Convert I210 to ARGB. |
1750 | | LIBYUV_API |
1751 | | int I210ToARGB(const uint16_t* src_y, |
1752 | | int src_stride_y, |
1753 | | const uint16_t* src_u, |
1754 | | int src_stride_u, |
1755 | | const uint16_t* src_v, |
1756 | | int src_stride_v, |
1757 | | uint8_t* dst_argb, |
1758 | | int dst_stride_argb, |
1759 | | int width, |
1760 | 0 | int height) { |
1761 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1762 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1763 | 0 | &kYuvI601Constants, width, height); |
1764 | 0 | } |
1765 | | |
1766 | | // Convert I210 to ABGR. |
1767 | | LIBYUV_API |
1768 | | int I210ToABGR(const uint16_t* src_y, |
1769 | | int src_stride_y, |
1770 | | const uint16_t* src_u, |
1771 | | int src_stride_u, |
1772 | | const uint16_t* src_v, |
1773 | | int src_stride_v, |
1774 | | uint8_t* dst_abgr, |
1775 | | int dst_stride_abgr, |
1776 | | int width, |
1777 | 0 | int height) { |
1778 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_v, |
1779 | 0 | src_stride_v, // Swap U and V |
1780 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1781 | 0 | &kYvuI601Constants, // Use Yvu matrix |
1782 | 0 | width, height); |
1783 | 0 | } |
1784 | | |
1785 | | // Convert H210 to ARGB. |
1786 | | LIBYUV_API |
1787 | | int H210ToARGB(const uint16_t* src_y, |
1788 | | int src_stride_y, |
1789 | | const uint16_t* src_u, |
1790 | | int src_stride_u, |
1791 | | const uint16_t* src_v, |
1792 | | int src_stride_v, |
1793 | | uint8_t* dst_argb, |
1794 | | int dst_stride_argb, |
1795 | | int width, |
1796 | 0 | int height) { |
1797 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1798 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1799 | 0 | &kYuvH709Constants, width, height); |
1800 | 0 | } |
1801 | | |
1802 | | // Convert H210 to ABGR. |
1803 | | LIBYUV_API |
1804 | | int H210ToABGR(const uint16_t* src_y, |
1805 | | int src_stride_y, |
1806 | | const uint16_t* src_u, |
1807 | | int src_stride_u, |
1808 | | const uint16_t* src_v, |
1809 | | int src_stride_v, |
1810 | | uint8_t* dst_abgr, |
1811 | | int dst_stride_abgr, |
1812 | | int width, |
1813 | 0 | int height) { |
1814 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_v, |
1815 | 0 | src_stride_v, // Swap U and V |
1816 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1817 | 0 | &kYvuH709Constants, // Use Yvu matrix |
1818 | 0 | width, height); |
1819 | 0 | } |
1820 | | |
1821 | | // Convert U210 to ARGB. |
1822 | | LIBYUV_API |
1823 | | int U210ToARGB(const uint16_t* src_y, |
1824 | | int src_stride_y, |
1825 | | const uint16_t* src_u, |
1826 | | int src_stride_u, |
1827 | | const uint16_t* src_v, |
1828 | | int src_stride_v, |
1829 | | uint8_t* dst_argb, |
1830 | | int dst_stride_argb, |
1831 | | int width, |
1832 | 0 | int height) { |
1833 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
1834 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
1835 | 0 | &kYuv2020Constants, width, height); |
1836 | 0 | } |
1837 | | |
1838 | | // Convert U210 to ABGR. |
1839 | | LIBYUV_API |
1840 | | int U210ToABGR(const uint16_t* src_y, |
1841 | | int src_stride_y, |
1842 | | const uint16_t* src_u, |
1843 | | int src_stride_u, |
1844 | | const uint16_t* src_v, |
1845 | | int src_stride_v, |
1846 | | uint8_t* dst_abgr, |
1847 | | int dst_stride_abgr, |
1848 | | int width, |
1849 | 0 | int height) { |
1850 | 0 | return I210ToARGBMatrix(src_y, src_stride_y, src_v, |
1851 | 0 | src_stride_v, // Swap U and V |
1852 | 0 | src_u, src_stride_u, dst_abgr, dst_stride_abgr, |
1853 | 0 | &kYvu2020Constants, // Use Yvu matrix |
1854 | 0 | width, height); |
1855 | 0 | } |
1856 | | |
1857 | | LIBYUV_API |
1858 | | int I410ToARGBMatrix(const uint16_t* src_y, |
1859 | | int src_stride_y, |
1860 | | const uint16_t* src_u, |
1861 | | int src_stride_u, |
1862 | | const uint16_t* src_v, |
1863 | | int src_stride_v, |
1864 | | uint8_t* dst_argb, |
1865 | | int dst_stride_argb, |
1866 | | const struct YuvConstants* yuvconstants, |
1867 | | int width, |
1868 | 74 | int height) { |
1869 | 74 | int y; |
1870 | 74 | void (*I410ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
1871 | 74 | const uint16_t* v_buf, uint8_t* rgb_buf, |
1872 | 74 | const struct YuvConstants* yuvconstants, int width) = |
1873 | 74 | I410ToARGBRow_C; |
1874 | 74 | assert(yuvconstants); |
1875 | 74 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
1876 | 0 | return -1; |
1877 | 0 | } |
1878 | | // Negative height means invert the image. |
1879 | 74 | if (height < 0) { |
1880 | 0 | height = -height; |
1881 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1882 | 0 | dst_stride_argb = -dst_stride_argb; |
1883 | 0 | } |
1884 | 74 | #if defined(HAS_I410TOARGBROW_SSSE3) |
1885 | 74 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1886 | 74 | I410ToARGBRow = I410ToARGBRow_Any_SSSE3; |
1887 | 74 | if (IS_ALIGNED(width, 8)) { |
1888 | 35 | I410ToARGBRow = I410ToARGBRow_SSSE3; |
1889 | 35 | } |
1890 | 74 | } |
1891 | 74 | #endif |
1892 | | #if defined(HAS_I410TOARGBROW_NEON) |
1893 | | if (TestCpuFlag(kCpuHasNEON)) { |
1894 | | I410ToARGBRow = I410ToARGBRow_Any_NEON; |
1895 | | if (IS_ALIGNED(width, 8)) { |
1896 | | I410ToARGBRow = I410ToARGBRow_NEON; |
1897 | | } |
1898 | | } |
1899 | | #endif |
1900 | | #if defined(HAS_I410TOARGBROW_SVE2) |
1901 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1902 | | I410ToARGBRow = I410ToARGBRow_SVE2; |
1903 | | } |
1904 | | #endif |
1905 | | #if defined(HAS_I410TOARGBROW_SME) |
1906 | | if (TestCpuFlag(kCpuHasSME)) { |
1907 | | I410ToARGBRow = I410ToARGBRow_SME; |
1908 | | } |
1909 | | #endif |
1910 | 74 | #if defined(HAS_I410TOARGBROW_AVX2) |
1911 | 74 | if (TestCpuFlag(kCpuHasAVX2)) { |
1912 | 74 | I410ToARGBRow = I410ToARGBRow_Any_AVX2; |
1913 | 74 | if (IS_ALIGNED(width, 16)) { |
1914 | 18 | I410ToARGBRow = I410ToARGBRow_AVX2; |
1915 | 18 | } |
1916 | 74 | } |
1917 | 74 | #endif |
1918 | 1.09k | for (y = 0; y < height; ++y) { |
1919 | 1.01k | I410ToARGBRow(src_y, src_u, src_v, dst_argb, yuvconstants, width); |
1920 | 1.01k | dst_argb += dst_stride_argb; |
1921 | 1.01k | src_y += src_stride_y; |
1922 | 1.01k | src_u += src_stride_u; |
1923 | 1.01k | src_v += src_stride_v; |
1924 | 1.01k | } |
1925 | 74 | return 0; |
1926 | 74 | } |
1927 | | |
1928 | | LIBYUV_API |
1929 | | int P010ToARGBMatrix(const uint16_t* src_y, |
1930 | | int src_stride_y, |
1931 | | const uint16_t* src_uv, |
1932 | | int src_stride_uv, |
1933 | | uint8_t* dst_argb, |
1934 | | int dst_stride_argb, |
1935 | | const struct YuvConstants* yuvconstants, |
1936 | | int width, |
1937 | 0 | int height) { |
1938 | 0 | int y; |
1939 | 0 | void (*P210ToARGBRow)( |
1940 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
1941 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToARGBRow_C; |
1942 | 0 | assert(yuvconstants); |
1943 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
1944 | 0 | return -1; |
1945 | 0 | } |
1946 | | // Negative height means invert the image. |
1947 | 0 | if (height < 0) { |
1948 | 0 | height = -height; |
1949 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
1950 | 0 | dst_stride_argb = -dst_stride_argb; |
1951 | 0 | } |
1952 | 0 | #if defined(HAS_P210TOARGBROW_SSSE3) |
1953 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1954 | 0 | P210ToARGBRow = P210ToARGBRow_Any_SSSE3; |
1955 | 0 | if (IS_ALIGNED(width, 8)) { |
1956 | 0 | P210ToARGBRow = P210ToARGBRow_SSSE3; |
1957 | 0 | } |
1958 | 0 | } |
1959 | 0 | #endif |
1960 | 0 | #if defined(HAS_P210TOARGBROW_AVX2) |
1961 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1962 | 0 | P210ToARGBRow = P210ToARGBRow_Any_AVX2; |
1963 | 0 | if (IS_ALIGNED(width, 16)) { |
1964 | 0 | P210ToARGBRow = P210ToARGBRow_AVX2; |
1965 | 0 | } |
1966 | 0 | } |
1967 | 0 | #endif |
1968 | | #if defined(HAS_P210TOARGBROW_NEON) |
1969 | | if (TestCpuFlag(kCpuHasNEON)) { |
1970 | | P210ToARGBRow = P210ToARGBRow_Any_NEON; |
1971 | | if (IS_ALIGNED(width, 8)) { |
1972 | | P210ToARGBRow = P210ToARGBRow_NEON; |
1973 | | } |
1974 | | } |
1975 | | #endif |
1976 | | #if defined(HAS_P210TOARGBROW_SVE2) |
1977 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1978 | | P210ToARGBRow = P210ToARGBRow_SVE2; |
1979 | | } |
1980 | | #endif |
1981 | | #if defined(HAS_P210TOARGBROW_SME) |
1982 | | if (TestCpuFlag(kCpuHasSME)) { |
1983 | | P210ToARGBRow = P210ToARGBRow_SME; |
1984 | | } |
1985 | | #endif |
1986 | 0 | for (y = 0; y < height; ++y) { |
1987 | 0 | P210ToARGBRow(src_y, src_uv, dst_argb, yuvconstants, width); |
1988 | 0 | dst_argb += dst_stride_argb; |
1989 | 0 | src_y += src_stride_y; |
1990 | 0 | if (y & 1) { |
1991 | 0 | src_uv += src_stride_uv; |
1992 | 0 | } |
1993 | 0 | } |
1994 | 0 | return 0; |
1995 | 0 | } |
1996 | | |
1997 | | LIBYUV_API |
1998 | | int P210ToARGBMatrix(const uint16_t* src_y, |
1999 | | int src_stride_y, |
2000 | | const uint16_t* src_uv, |
2001 | | int src_stride_uv, |
2002 | | uint8_t* dst_argb, |
2003 | | int dst_stride_argb, |
2004 | | const struct YuvConstants* yuvconstants, |
2005 | | int width, |
2006 | 0 | int height) { |
2007 | 0 | int y; |
2008 | 0 | void (*P210ToARGBRow)( |
2009 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
2010 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToARGBRow_C; |
2011 | 0 | assert(yuvconstants); |
2012 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
2013 | 0 | return -1; |
2014 | 0 | } |
2015 | | // Negative height means invert the image. |
2016 | 0 | if (height < 0) { |
2017 | 0 | height = -height; |
2018 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2019 | 0 | dst_stride_argb = -dst_stride_argb; |
2020 | 0 | } |
2021 | 0 | #if defined(HAS_P210TOARGBROW_SSSE3) |
2022 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2023 | 0 | P210ToARGBRow = P210ToARGBRow_Any_SSSE3; |
2024 | 0 | if (IS_ALIGNED(width, 8)) { |
2025 | 0 | P210ToARGBRow = P210ToARGBRow_SSSE3; |
2026 | 0 | } |
2027 | 0 | } |
2028 | 0 | #endif |
2029 | 0 | #if defined(HAS_P210TOARGBROW_AVX2) |
2030 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2031 | 0 | P210ToARGBRow = P210ToARGBRow_Any_AVX2; |
2032 | 0 | if (IS_ALIGNED(width, 16)) { |
2033 | 0 | P210ToARGBRow = P210ToARGBRow_AVX2; |
2034 | 0 | } |
2035 | 0 | } |
2036 | 0 | #endif |
2037 | | #if defined(HAS_P210TOARGBROW_NEON) |
2038 | | if (TestCpuFlag(kCpuHasNEON)) { |
2039 | | P210ToARGBRow = P210ToARGBRow_Any_NEON; |
2040 | | if (IS_ALIGNED(width, 8)) { |
2041 | | P210ToARGBRow = P210ToARGBRow_NEON; |
2042 | | } |
2043 | | } |
2044 | | #endif |
2045 | | #if defined(HAS_P210TOARGBROW_SVE2) |
2046 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2047 | | P210ToARGBRow = P210ToARGBRow_SVE2; |
2048 | | } |
2049 | | #endif |
2050 | | #if defined(HAS_P210TOARGBROW_SME) |
2051 | | if (TestCpuFlag(kCpuHasSME)) { |
2052 | | P210ToARGBRow = P210ToARGBRow_SME; |
2053 | | } |
2054 | | #endif |
2055 | 0 | for (y = 0; y < height; ++y) { |
2056 | 0 | P210ToARGBRow(src_y, src_uv, dst_argb, yuvconstants, width); |
2057 | 0 | dst_argb += dst_stride_argb; |
2058 | 0 | src_y += src_stride_y; |
2059 | 0 | src_uv += src_stride_uv; |
2060 | 0 | } |
2061 | 0 | return 0; |
2062 | 0 | } |
2063 | | |
2064 | | LIBYUV_API |
2065 | | int P010ToAR30Matrix(const uint16_t* src_y, |
2066 | | int src_stride_y, |
2067 | | const uint16_t* src_uv, |
2068 | | int src_stride_uv, |
2069 | | uint8_t* dst_ar30, |
2070 | | int dst_stride_ar30, |
2071 | | const struct YuvConstants* yuvconstants, |
2072 | | int width, |
2073 | 0 | int height) { |
2074 | 0 | int y; |
2075 | 0 | void (*P210ToAR30Row)( |
2076 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
2077 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToAR30Row_C; |
2078 | 0 | assert(yuvconstants); |
2079 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
2080 | 0 | return -1; |
2081 | 0 | } |
2082 | | // Negative height means invert the image. |
2083 | 0 | if (height < 0) { |
2084 | 0 | height = -height; |
2085 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
2086 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
2087 | 0 | } |
2088 | 0 | #if defined(HAS_P210TOAR30ROW_SSSE3) |
2089 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2090 | 0 | P210ToAR30Row = P210ToAR30Row_Any_SSSE3; |
2091 | 0 | if (IS_ALIGNED(width, 8)) { |
2092 | 0 | P210ToAR30Row = P210ToAR30Row_SSSE3; |
2093 | 0 | } |
2094 | 0 | } |
2095 | 0 | #endif |
2096 | 0 | #if defined(HAS_P210TOAR30ROW_AVX2) |
2097 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2098 | 0 | P210ToAR30Row = P210ToAR30Row_Any_AVX2; |
2099 | 0 | if (IS_ALIGNED(width, 16)) { |
2100 | 0 | P210ToAR30Row = P210ToAR30Row_AVX2; |
2101 | 0 | } |
2102 | 0 | } |
2103 | 0 | #endif |
2104 | | #if defined(HAS_P210TOAR30ROW_NEON) |
2105 | | if (TestCpuFlag(kCpuHasNEON)) { |
2106 | | P210ToAR30Row = P210ToAR30Row_Any_NEON; |
2107 | | if (IS_ALIGNED(width, 8)) { |
2108 | | P210ToAR30Row = P210ToAR30Row_NEON; |
2109 | | } |
2110 | | } |
2111 | | #endif |
2112 | | #if defined(HAS_P210TOAR30ROW_SVE2) |
2113 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2114 | | P210ToAR30Row = P210ToAR30Row_SVE2; |
2115 | | } |
2116 | | #endif |
2117 | | #if defined(HAS_P210TOAR30ROW_SME) |
2118 | | if (TestCpuFlag(kCpuHasSME)) { |
2119 | | P210ToAR30Row = P210ToAR30Row_SME; |
2120 | | } |
2121 | | #endif |
2122 | 0 | for (y = 0; y < height; ++y) { |
2123 | 0 | P210ToAR30Row(src_y, src_uv, dst_ar30, yuvconstants, width); |
2124 | 0 | dst_ar30 += dst_stride_ar30; |
2125 | 0 | src_y += src_stride_y; |
2126 | 0 | if (y & 1) { |
2127 | 0 | src_uv += src_stride_uv; |
2128 | 0 | } |
2129 | 0 | } |
2130 | 0 | return 0; |
2131 | 0 | } |
2132 | | |
2133 | | LIBYUV_API |
2134 | | int P210ToAR30Matrix(const uint16_t* src_y, |
2135 | | int src_stride_y, |
2136 | | const uint16_t* src_uv, |
2137 | | int src_stride_uv, |
2138 | | uint8_t* dst_ar30, |
2139 | | int dst_stride_ar30, |
2140 | | const struct YuvConstants* yuvconstants, |
2141 | | int width, |
2142 | 0 | int height) { |
2143 | 0 | int y; |
2144 | 0 | void (*P210ToAR30Row)( |
2145 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
2146 | 0 | const struct YuvConstants* yuvconstants, int width) = P210ToAR30Row_C; |
2147 | 0 | assert(yuvconstants); |
2148 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
2149 | 0 | return -1; |
2150 | 0 | } |
2151 | | // Negative height means invert the image. |
2152 | 0 | if (height < 0) { |
2153 | 0 | height = -height; |
2154 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
2155 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
2156 | 0 | } |
2157 | 0 | #if defined(HAS_P210TOAR30ROW_SSSE3) |
2158 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2159 | 0 | P210ToAR30Row = P210ToAR30Row_Any_SSSE3; |
2160 | 0 | if (IS_ALIGNED(width, 8)) { |
2161 | 0 | P210ToAR30Row = P210ToAR30Row_SSSE3; |
2162 | 0 | } |
2163 | 0 | } |
2164 | 0 | #endif |
2165 | 0 | #if defined(HAS_P210TOAR30ROW_AVX2) |
2166 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2167 | 0 | P210ToAR30Row = P210ToAR30Row_Any_AVX2; |
2168 | 0 | if (IS_ALIGNED(width, 16)) { |
2169 | 0 | P210ToAR30Row = P210ToAR30Row_AVX2; |
2170 | 0 | } |
2171 | 0 | } |
2172 | 0 | #endif |
2173 | | #if defined(HAS_P210TOAR30ROW_NEON) |
2174 | | if (TestCpuFlag(kCpuHasNEON)) { |
2175 | | P210ToAR30Row = P210ToAR30Row_Any_NEON; |
2176 | | if (IS_ALIGNED(width, 8)) { |
2177 | | P210ToAR30Row = P210ToAR30Row_NEON; |
2178 | | } |
2179 | | } |
2180 | | #endif |
2181 | | #if defined(HAS_P210TOAR30ROW_SVE2) |
2182 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2183 | | P210ToAR30Row = P210ToAR30Row_SVE2; |
2184 | | } |
2185 | | #endif |
2186 | | #if defined(HAS_P210TOAR30ROW_SME) |
2187 | | if (TestCpuFlag(kCpuHasSME)) { |
2188 | | P210ToAR30Row = P210ToAR30Row_SME; |
2189 | | } |
2190 | | #endif |
2191 | 0 | for (y = 0; y < height; ++y) { |
2192 | 0 | P210ToAR30Row(src_y, src_uv, dst_ar30, yuvconstants, width); |
2193 | 0 | dst_ar30 += dst_stride_ar30; |
2194 | 0 | src_y += src_stride_y; |
2195 | 0 | src_uv += src_stride_uv; |
2196 | 0 | } |
2197 | 0 | return 0; |
2198 | 0 | } |
2199 | | |
2200 | | // Convert I420 with Alpha to preattenuated ARGB with matrix. |
2201 | | LIBYUV_API |
2202 | | int I420AlphaToARGBMatrix(const uint8_t* src_y, |
2203 | | int src_stride_y, |
2204 | | const uint8_t* src_u, |
2205 | | int src_stride_u, |
2206 | | const uint8_t* src_v, |
2207 | | int src_stride_v, |
2208 | | const uint8_t* src_a, |
2209 | | int src_stride_a, |
2210 | | uint8_t* dst_argb, |
2211 | | int dst_stride_argb, |
2212 | | const struct YuvConstants* yuvconstants, |
2213 | | int width, |
2214 | | int height, |
2215 | 148 | int attenuate) { |
2216 | 148 | int y; |
2217 | 148 | void (*I422AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
2218 | 148 | const uint8_t* v_buf, const uint8_t* a_buf, |
2219 | 148 | uint8_t* dst_argb, |
2220 | 148 | const struct YuvConstants* yuvconstants, |
2221 | 148 | int width) = I422AlphaToARGBRow_C; |
2222 | 148 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2223 | 148 | int width) = ARGBAttenuateRow_C; |
2224 | 148 | assert(yuvconstants); |
2225 | 148 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2226 | 148 | height == 0) { |
2227 | 0 | return -1; |
2228 | 0 | } |
2229 | | // Negative height means invert the image. |
2230 | 148 | if (height < 0) { |
2231 | 0 | height = -height; |
2232 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2233 | 0 | dst_stride_argb = -dst_stride_argb; |
2234 | 0 | } |
2235 | 148 | #if defined(HAS_I422ALPHATOARGBROW_SSSE3) |
2236 | 148 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2237 | 148 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_SSSE3; |
2238 | 148 | if (IS_ALIGNED(width, 8)) { |
2239 | 36 | I422AlphaToARGBRow = I422AlphaToARGBRow_SSSE3; |
2240 | 36 | } |
2241 | 148 | } |
2242 | 148 | #endif |
2243 | 148 | #if defined(HAS_I422ALPHATOARGBROW_AVX2) |
2244 | 148 | if (TestCpuFlag(kCpuHasAVX2)) { |
2245 | 148 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_AVX2; |
2246 | 148 | if (IS_ALIGNED(width, 16)) { |
2247 | 19 | I422AlphaToARGBRow = I422AlphaToARGBRow_AVX2; |
2248 | 19 | } |
2249 | 148 | } |
2250 | 148 | #endif |
2251 | | #if defined(HAS_I422ALPHATOARGBROW_NEON) |
2252 | | if (TestCpuFlag(kCpuHasNEON)) { |
2253 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_NEON; |
2254 | | if (IS_ALIGNED(width, 8)) { |
2255 | | I422AlphaToARGBRow = I422AlphaToARGBRow_NEON; |
2256 | | } |
2257 | | } |
2258 | | #endif |
2259 | | #if defined(HAS_I422ALPHATOARGBROW_SVE2) |
2260 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2261 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SVE2; |
2262 | | } |
2263 | | #endif |
2264 | | #if defined(HAS_I422ALPHATOARGBROW_SME) |
2265 | | if (TestCpuFlag(kCpuHasSME)) { |
2266 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SME; |
2267 | | } |
2268 | | #endif |
2269 | | #if defined(HAS_I422ALPHATOARGBROW_LSX) |
2270 | | if (TestCpuFlag(kCpuHasLSX)) { |
2271 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LSX; |
2272 | | if (IS_ALIGNED(width, 16)) { |
2273 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LSX; |
2274 | | } |
2275 | | } |
2276 | | #endif |
2277 | | #if defined(HAS_I422ALPHATOARGBROW_LASX) |
2278 | | if (TestCpuFlag(kCpuHasLASX)) { |
2279 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LASX; |
2280 | | if (IS_ALIGNED(width, 16)) { |
2281 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LASX; |
2282 | | } |
2283 | | } |
2284 | | #endif |
2285 | | #if defined(HAS_I422ALPHATOARGBROW_RVV) |
2286 | | if (TestCpuFlag(kCpuHasRVV)) { |
2287 | | I422AlphaToARGBRow = I422AlphaToARGBRow_RVV; |
2288 | | } |
2289 | | #endif |
2290 | 148 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2291 | 148 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2292 | 148 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2293 | 148 | if (IS_ALIGNED(width, 4)) { |
2294 | 53 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2295 | 53 | } |
2296 | 148 | } |
2297 | 148 | #endif |
2298 | 148 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2299 | 148 | if (TestCpuFlag(kCpuHasAVX2)) { |
2300 | 148 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2301 | 148 | if (IS_ALIGNED(width, 8)) { |
2302 | 36 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2303 | 36 | } |
2304 | 148 | } |
2305 | 148 | #endif |
2306 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2307 | | if (TestCpuFlag(kCpuHasNEON)) { |
2308 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2309 | | if (IS_ALIGNED(width, 8)) { |
2310 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2311 | | } |
2312 | | } |
2313 | | #endif |
2314 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2315 | | if (TestCpuFlag(kCpuHasRVV)) { |
2316 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2317 | | } |
2318 | | #endif |
2319 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2320 | | if (TestCpuFlag(kCpuHasLSX)) { |
2321 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2322 | | if (IS_ALIGNED(width, 8)) { |
2323 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2324 | | } |
2325 | | } |
2326 | | #endif |
2327 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2328 | | if (TestCpuFlag(kCpuHasLASX)) { |
2329 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2330 | | if (IS_ALIGNED(width, 16)) { |
2331 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2332 | | } |
2333 | | } |
2334 | | #endif |
2335 | | |
2336 | 2.78k | for (y = 0; y < height; ++y) { |
2337 | 2.63k | I422AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2338 | 2.63k | width); |
2339 | 2.63k | if (attenuate) { |
2340 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2341 | 0 | } |
2342 | 2.63k | dst_argb += dst_stride_argb; |
2343 | 2.63k | src_a += src_stride_a; |
2344 | 2.63k | src_y += src_stride_y; |
2345 | 2.63k | if (y & 1) { |
2346 | 1.30k | src_u += src_stride_u; |
2347 | 1.30k | src_v += src_stride_v; |
2348 | 1.30k | } |
2349 | 2.63k | } |
2350 | 148 | return 0; |
2351 | 148 | } |
2352 | | |
2353 | | // Convert I422 with Alpha to preattenuated ARGB with matrix. |
2354 | | LIBYUV_API |
2355 | | int I422AlphaToARGBMatrix(const uint8_t* src_y, |
2356 | | int src_stride_y, |
2357 | | const uint8_t* src_u, |
2358 | | int src_stride_u, |
2359 | | const uint8_t* src_v, |
2360 | | int src_stride_v, |
2361 | | const uint8_t* src_a, |
2362 | | int src_stride_a, |
2363 | | uint8_t* dst_argb, |
2364 | | int dst_stride_argb, |
2365 | | const struct YuvConstants* yuvconstants, |
2366 | | int width, |
2367 | | int height, |
2368 | 103 | int attenuate) { |
2369 | 103 | int y; |
2370 | 103 | void (*I422AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
2371 | 103 | const uint8_t* v_buf, const uint8_t* a_buf, |
2372 | 103 | uint8_t* dst_argb, |
2373 | 103 | const struct YuvConstants* yuvconstants, |
2374 | 103 | int width) = I422AlphaToARGBRow_C; |
2375 | 103 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2376 | 103 | int width) = ARGBAttenuateRow_C; |
2377 | 103 | assert(yuvconstants); |
2378 | 103 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2379 | 103 | height == 0) { |
2380 | 0 | return -1; |
2381 | 0 | } |
2382 | | // Negative height means invert the image. |
2383 | 103 | if (height < 0) { |
2384 | 0 | height = -height; |
2385 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2386 | 0 | dst_stride_argb = -dst_stride_argb; |
2387 | 0 | } |
2388 | 103 | #if defined(HAS_I422ALPHATOARGBROW_SSSE3) |
2389 | 103 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2390 | 103 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_SSSE3; |
2391 | 103 | if (IS_ALIGNED(width, 8)) { |
2392 | 34 | I422AlphaToARGBRow = I422AlphaToARGBRow_SSSE3; |
2393 | 34 | } |
2394 | 103 | } |
2395 | 103 | #endif |
2396 | 103 | #if defined(HAS_I422ALPHATOARGBROW_AVX2) |
2397 | 103 | if (TestCpuFlag(kCpuHasAVX2)) { |
2398 | 103 | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_AVX2; |
2399 | 103 | if (IS_ALIGNED(width, 16)) { |
2400 | 17 | I422AlphaToARGBRow = I422AlphaToARGBRow_AVX2; |
2401 | 17 | } |
2402 | 103 | } |
2403 | 103 | #endif |
2404 | | #if defined(HAS_I422ALPHATOARGBROW_NEON) |
2405 | | if (TestCpuFlag(kCpuHasNEON)) { |
2406 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_NEON; |
2407 | | if (IS_ALIGNED(width, 8)) { |
2408 | | I422AlphaToARGBRow = I422AlphaToARGBRow_NEON; |
2409 | | } |
2410 | | } |
2411 | | #endif |
2412 | | #if defined(HAS_I422ALPHATOARGBROW_SVE2) |
2413 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2414 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SVE2; |
2415 | | } |
2416 | | #endif |
2417 | | #if defined(HAS_I422ALPHATOARGBROW_SME) |
2418 | | if (TestCpuFlag(kCpuHasSME)) { |
2419 | | I422AlphaToARGBRow = I422AlphaToARGBRow_SME; |
2420 | | } |
2421 | | #endif |
2422 | | #if defined(HAS_I422ALPHATOARGBROW_LSX) |
2423 | | if (TestCpuFlag(kCpuHasLSX)) { |
2424 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LSX; |
2425 | | if (IS_ALIGNED(width, 16)) { |
2426 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LSX; |
2427 | | } |
2428 | | } |
2429 | | #endif |
2430 | | #if defined(HAS_I422ALPHATOARGBROW_LASX) |
2431 | | if (TestCpuFlag(kCpuHasLASX)) { |
2432 | | I422AlphaToARGBRow = I422AlphaToARGBRow_Any_LASX; |
2433 | | if (IS_ALIGNED(width, 16)) { |
2434 | | I422AlphaToARGBRow = I422AlphaToARGBRow_LASX; |
2435 | | } |
2436 | | } |
2437 | | #endif |
2438 | | #if defined(HAS_I422ALPHATOARGBROW_RVV) |
2439 | | if (TestCpuFlag(kCpuHasRVV)) { |
2440 | | I422AlphaToARGBRow = I422AlphaToARGBRow_RVV; |
2441 | | } |
2442 | | #endif |
2443 | 103 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2444 | 103 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2445 | 103 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2446 | 103 | if (IS_ALIGNED(width, 4)) { |
2447 | 52 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2448 | 52 | } |
2449 | 103 | } |
2450 | 103 | #endif |
2451 | 103 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2452 | 103 | if (TestCpuFlag(kCpuHasAVX2)) { |
2453 | 103 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2454 | 103 | if (IS_ALIGNED(width, 8)) { |
2455 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2456 | 34 | } |
2457 | 103 | } |
2458 | 103 | #endif |
2459 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2460 | | if (TestCpuFlag(kCpuHasNEON)) { |
2461 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2462 | | if (IS_ALIGNED(width, 8)) { |
2463 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2464 | | } |
2465 | | } |
2466 | | #endif |
2467 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2468 | | if (TestCpuFlag(kCpuHasRVV)) { |
2469 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2470 | | } |
2471 | | #endif |
2472 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2473 | | if (TestCpuFlag(kCpuHasLSX)) { |
2474 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2475 | | if (IS_ALIGNED(width, 8)) { |
2476 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2477 | | } |
2478 | | } |
2479 | | #endif |
2480 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2481 | | if (TestCpuFlag(kCpuHasLASX)) { |
2482 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2483 | | if (IS_ALIGNED(width, 16)) { |
2484 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2485 | | } |
2486 | | } |
2487 | | #endif |
2488 | | |
2489 | 2.16k | for (y = 0; y < height; ++y) { |
2490 | 2.06k | I422AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2491 | 2.06k | width); |
2492 | 2.06k | if (attenuate) { |
2493 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2494 | 0 | } |
2495 | 2.06k | dst_argb += dst_stride_argb; |
2496 | 2.06k | src_a += src_stride_a; |
2497 | 2.06k | src_y += src_stride_y; |
2498 | 2.06k | src_u += src_stride_u; |
2499 | 2.06k | src_v += src_stride_v; |
2500 | 2.06k | } |
2501 | 103 | return 0; |
2502 | 103 | } |
2503 | | |
2504 | | // Convert I444 with Alpha to preattenuated ARGB with matrix. |
2505 | | LIBYUV_API |
2506 | | int I444AlphaToARGBMatrix(const uint8_t* src_y, |
2507 | | int src_stride_y, |
2508 | | const uint8_t* src_u, |
2509 | | int src_stride_u, |
2510 | | const uint8_t* src_v, |
2511 | | int src_stride_v, |
2512 | | const uint8_t* src_a, |
2513 | | int src_stride_a, |
2514 | | uint8_t* dst_argb, |
2515 | | int dst_stride_argb, |
2516 | | const struct YuvConstants* yuvconstants, |
2517 | | int width, |
2518 | | int height, |
2519 | 108 | int attenuate) { |
2520 | 108 | int y; |
2521 | 108 | void (*I444AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
2522 | 108 | const uint8_t* v_buf, const uint8_t* a_buf, |
2523 | 108 | uint8_t* dst_argb, |
2524 | 108 | const struct YuvConstants* yuvconstants, |
2525 | 108 | int width) = I444AlphaToARGBRow_C; |
2526 | 108 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2527 | 108 | int width) = ARGBAttenuateRow_C; |
2528 | 108 | assert(yuvconstants); |
2529 | 108 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2530 | 108 | height == 0) { |
2531 | 0 | return -1; |
2532 | 0 | } |
2533 | | // Negative height means invert the image. |
2534 | 108 | if (height < 0) { |
2535 | 0 | height = -height; |
2536 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2537 | 0 | dst_stride_argb = -dst_stride_argb; |
2538 | 0 | } |
2539 | 108 | #if defined(HAS_I444ALPHATOARGBROW_SSSE3) |
2540 | 108 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2541 | 108 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_SSSE3; |
2542 | 108 | if (IS_ALIGNED(width, 8)) { |
2543 | 43 | I444AlphaToARGBRow = I444AlphaToARGBRow_SSSE3; |
2544 | 43 | } |
2545 | 108 | } |
2546 | 108 | #endif |
2547 | 108 | #if defined(HAS_I444ALPHATOARGBROW_AVX2) |
2548 | 108 | if (TestCpuFlag(kCpuHasAVX2)) { |
2549 | 108 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_AVX2; |
2550 | 108 | if (IS_ALIGNED(width, 16)) { |
2551 | 25 | I444AlphaToARGBRow = I444AlphaToARGBRow_AVX2; |
2552 | 25 | } |
2553 | 108 | } |
2554 | 108 | #endif |
2555 | | #if defined(HAS_I444ALPHATOARGBROW_NEON) |
2556 | | if (TestCpuFlag(kCpuHasNEON)) { |
2557 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_NEON; |
2558 | | if (IS_ALIGNED(width, 8)) { |
2559 | | I444AlphaToARGBRow = I444AlphaToARGBRow_NEON; |
2560 | | } |
2561 | | } |
2562 | | #endif |
2563 | | #if defined(HAS_I444ALPHATOARGBROW_SVE2) |
2564 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2565 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SVE2; |
2566 | | } |
2567 | | #endif |
2568 | | #if defined(HAS_I444ALPHATOARGBROW_SME) |
2569 | | if (TestCpuFlag(kCpuHasSME)) { |
2570 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SME; |
2571 | | } |
2572 | | #endif |
2573 | | #if defined(HAS_I444ALPHATOARGBROW_RVV) |
2574 | | if (TestCpuFlag(kCpuHasRVV)) { |
2575 | | I444AlphaToARGBRow = I444AlphaToARGBRow_RVV; |
2576 | | } |
2577 | | #endif |
2578 | 108 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2579 | 108 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2580 | 108 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2581 | 108 | if (IS_ALIGNED(width, 4)) { |
2582 | 61 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2583 | 61 | } |
2584 | 108 | } |
2585 | 108 | #endif |
2586 | 108 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2587 | 108 | if (TestCpuFlag(kCpuHasAVX2)) { |
2588 | 108 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2589 | 108 | if (IS_ALIGNED(width, 8)) { |
2590 | 43 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2591 | 43 | } |
2592 | 108 | } |
2593 | 108 | #endif |
2594 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2595 | | if (TestCpuFlag(kCpuHasNEON)) { |
2596 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2597 | | if (IS_ALIGNED(width, 8)) { |
2598 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2599 | | } |
2600 | | } |
2601 | | #endif |
2602 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2603 | | if (TestCpuFlag(kCpuHasRVV)) { |
2604 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2605 | | } |
2606 | | #endif |
2607 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2608 | | if (TestCpuFlag(kCpuHasLSX)) { |
2609 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2610 | | if (IS_ALIGNED(width, 8)) { |
2611 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2612 | | } |
2613 | | } |
2614 | | #endif |
2615 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2616 | | if (TestCpuFlag(kCpuHasLASX)) { |
2617 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2618 | | if (IS_ALIGNED(width, 16)) { |
2619 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2620 | | } |
2621 | | } |
2622 | | #endif |
2623 | | |
2624 | 1.77k | for (y = 0; y < height; ++y) { |
2625 | 1.66k | I444AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2626 | 1.66k | width); |
2627 | 1.66k | if (attenuate) { |
2628 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2629 | 0 | } |
2630 | 1.66k | dst_argb += dst_stride_argb; |
2631 | 1.66k | src_a += src_stride_a; |
2632 | 1.66k | src_y += src_stride_y; |
2633 | 1.66k | src_u += src_stride_u; |
2634 | 1.66k | src_v += src_stride_v; |
2635 | 1.66k | } |
2636 | 108 | return 0; |
2637 | 108 | } |
2638 | | |
2639 | | // Convert I420 with Alpha to ARGB. |
2640 | | LIBYUV_API |
2641 | | int I420AlphaToARGB(const uint8_t* src_y, |
2642 | | int src_stride_y, |
2643 | | const uint8_t* src_u, |
2644 | | int src_stride_u, |
2645 | | const uint8_t* src_v, |
2646 | | int src_stride_v, |
2647 | | const uint8_t* src_a, |
2648 | | int src_stride_a, |
2649 | | uint8_t* dst_argb, |
2650 | | int dst_stride_argb, |
2651 | | int width, |
2652 | | int height, |
2653 | 0 | int attenuate) { |
2654 | 0 | return I420AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
2655 | 0 | src_stride_v, src_a, src_stride_a, dst_argb, |
2656 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
2657 | 0 | height, attenuate); |
2658 | 0 | } |
2659 | | |
2660 | | // Convert I420 with Alpha to ABGR. |
2661 | | LIBYUV_API |
2662 | | int I420AlphaToABGR(const uint8_t* src_y, |
2663 | | int src_stride_y, |
2664 | | const uint8_t* src_u, |
2665 | | int src_stride_u, |
2666 | | const uint8_t* src_v, |
2667 | | int src_stride_v, |
2668 | | const uint8_t* src_a, |
2669 | | int src_stride_a, |
2670 | | uint8_t* dst_abgr, |
2671 | | int dst_stride_abgr, |
2672 | | int width, |
2673 | | int height, |
2674 | 0 | int attenuate) { |
2675 | 0 | return I420AlphaToARGBMatrix( |
2676 | 0 | src_y, src_stride_y, src_v, src_stride_v, // Swap U and V |
2677 | 0 | src_u, src_stride_u, src_a, src_stride_a, dst_abgr, dst_stride_abgr, |
2678 | 0 | &kYvuI601Constants, // Use Yvu matrix |
2679 | 0 | width, height, attenuate); |
2680 | 0 | } |
2681 | | |
2682 | | // Convert I422 with Alpha to ARGB. |
2683 | | LIBYUV_API |
2684 | | int I422AlphaToARGB(const uint8_t* src_y, |
2685 | | int src_stride_y, |
2686 | | const uint8_t* src_u, |
2687 | | int src_stride_u, |
2688 | | const uint8_t* src_v, |
2689 | | int src_stride_v, |
2690 | | const uint8_t* src_a, |
2691 | | int src_stride_a, |
2692 | | uint8_t* dst_argb, |
2693 | | int dst_stride_argb, |
2694 | | int width, |
2695 | | int height, |
2696 | 0 | int attenuate) { |
2697 | 0 | return I422AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
2698 | 0 | src_stride_v, src_a, src_stride_a, dst_argb, |
2699 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
2700 | 0 | height, attenuate); |
2701 | 0 | } |
2702 | | |
2703 | | // Convert I422 with Alpha to ABGR. |
2704 | | LIBYUV_API |
2705 | | int I422AlphaToABGR(const uint8_t* src_y, |
2706 | | int src_stride_y, |
2707 | | const uint8_t* src_u, |
2708 | | int src_stride_u, |
2709 | | const uint8_t* src_v, |
2710 | | int src_stride_v, |
2711 | | const uint8_t* src_a, |
2712 | | int src_stride_a, |
2713 | | uint8_t* dst_abgr, |
2714 | | int dst_stride_abgr, |
2715 | | int width, |
2716 | | int height, |
2717 | 0 | int attenuate) { |
2718 | 0 | return I422AlphaToARGBMatrix( |
2719 | 0 | src_y, src_stride_y, src_v, src_stride_v, // Swap U and V |
2720 | 0 | src_u, src_stride_u, src_a, src_stride_a, dst_abgr, dst_stride_abgr, |
2721 | 0 | &kYvuI601Constants, // Use Yvu matrix |
2722 | 0 | width, height, attenuate); |
2723 | 0 | } |
2724 | | |
2725 | | // Convert I444 with Alpha to ARGB. |
2726 | | LIBYUV_API |
2727 | | int I444AlphaToARGB(const uint8_t* src_y, |
2728 | | int src_stride_y, |
2729 | | const uint8_t* src_u, |
2730 | | int src_stride_u, |
2731 | | const uint8_t* src_v, |
2732 | | int src_stride_v, |
2733 | | const uint8_t* src_a, |
2734 | | int src_stride_a, |
2735 | | uint8_t* dst_argb, |
2736 | | int dst_stride_argb, |
2737 | | int width, |
2738 | | int height, |
2739 | 0 | int attenuate) { |
2740 | 0 | return I444AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
2741 | 0 | src_stride_v, src_a, src_stride_a, dst_argb, |
2742 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
2743 | 0 | height, attenuate); |
2744 | 0 | } |
2745 | | |
2746 | | // Convert I444 with Alpha to ABGR. |
2747 | | LIBYUV_API |
2748 | | int I444AlphaToABGR(const uint8_t* src_y, |
2749 | | int src_stride_y, |
2750 | | const uint8_t* src_u, |
2751 | | int src_stride_u, |
2752 | | const uint8_t* src_v, |
2753 | | int src_stride_v, |
2754 | | const uint8_t* src_a, |
2755 | | int src_stride_a, |
2756 | | uint8_t* dst_abgr, |
2757 | | int dst_stride_abgr, |
2758 | | int width, |
2759 | | int height, |
2760 | 0 | int attenuate) { |
2761 | 0 | return I444AlphaToARGBMatrix( |
2762 | 0 | src_y, src_stride_y, src_v, src_stride_v, // Swap U and V |
2763 | 0 | src_u, src_stride_u, src_a, src_stride_a, dst_abgr, dst_stride_abgr, |
2764 | 0 | &kYvuI601Constants, // Use Yvu matrix |
2765 | 0 | width, height, attenuate); |
2766 | 0 | } |
2767 | | |
2768 | | // Convert I010 with Alpha to preattenuated ARGB with matrix. |
2769 | | LIBYUV_API |
2770 | | int I010AlphaToARGBMatrix(const uint16_t* src_y, |
2771 | | int src_stride_y, |
2772 | | const uint16_t* src_u, |
2773 | | int src_stride_u, |
2774 | | const uint16_t* src_v, |
2775 | | int src_stride_v, |
2776 | | const uint16_t* src_a, |
2777 | | int src_stride_a, |
2778 | | uint8_t* dst_argb, |
2779 | | int dst_stride_argb, |
2780 | | const struct YuvConstants* yuvconstants, |
2781 | | int width, |
2782 | | int height, |
2783 | 133 | int attenuate) { |
2784 | 133 | int y; |
2785 | 133 | void (*I210AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
2786 | 133 | const uint16_t* v_buf, const uint16_t* a_buf, |
2787 | 133 | uint8_t* dst_argb, |
2788 | 133 | const struct YuvConstants* yuvconstants, |
2789 | 133 | int width) = I210AlphaToARGBRow_C; |
2790 | 133 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2791 | 133 | int width) = ARGBAttenuateRow_C; |
2792 | 133 | assert(yuvconstants); |
2793 | 133 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2794 | 133 | height == 0) { |
2795 | 0 | return -1; |
2796 | 0 | } |
2797 | | // Negative height means invert the image. |
2798 | 133 | if (height < 0) { |
2799 | 0 | height = -height; |
2800 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2801 | 0 | dst_stride_argb = -dst_stride_argb; |
2802 | 0 | } |
2803 | | #if defined(HAS_I210ALPHATOARGBROW_NEON) |
2804 | | if (TestCpuFlag(kCpuHasNEON)) { |
2805 | | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_NEON; |
2806 | | if (IS_ALIGNED(width, 8)) { |
2807 | | I210AlphaToARGBRow = I210AlphaToARGBRow_NEON; |
2808 | | } |
2809 | | } |
2810 | | #endif |
2811 | | #if defined(HAS_I210ALPHATOARGBROW_SVE2) |
2812 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2813 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SVE2; |
2814 | | } |
2815 | | #endif |
2816 | | #if defined(HAS_I210ALPHATOARGBROW_SME) |
2817 | | if (TestCpuFlag(kCpuHasSME)) { |
2818 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SME; |
2819 | | } |
2820 | | #endif |
2821 | 133 | #if defined(HAS_I210ALPHATOARGBROW_SSSE3) |
2822 | 133 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2823 | 133 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_SSSE3; |
2824 | 133 | if (IS_ALIGNED(width, 8)) { |
2825 | 34 | I210AlphaToARGBRow = I210AlphaToARGBRow_SSSE3; |
2826 | 34 | } |
2827 | 133 | } |
2828 | 133 | #endif |
2829 | 133 | #if defined(HAS_I210ALPHATOARGBROW_AVX2) |
2830 | 133 | if (TestCpuFlag(kCpuHasAVX2)) { |
2831 | 133 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_AVX2; |
2832 | 133 | if (IS_ALIGNED(width, 16)) { |
2833 | 17 | I210AlphaToARGBRow = I210AlphaToARGBRow_AVX2; |
2834 | 17 | } |
2835 | 133 | } |
2836 | 133 | #endif |
2837 | 133 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2838 | 133 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2839 | 133 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2840 | 133 | if (IS_ALIGNED(width, 4)) { |
2841 | 52 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2842 | 52 | } |
2843 | 133 | } |
2844 | 133 | #endif |
2845 | 133 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2846 | 133 | if (TestCpuFlag(kCpuHasAVX2)) { |
2847 | 133 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2848 | 133 | if (IS_ALIGNED(width, 8)) { |
2849 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2850 | 34 | } |
2851 | 133 | } |
2852 | 133 | #endif |
2853 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2854 | | if (TestCpuFlag(kCpuHasNEON)) { |
2855 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2856 | | if (IS_ALIGNED(width, 8)) { |
2857 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2858 | | } |
2859 | | } |
2860 | | #endif |
2861 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2862 | | if (TestCpuFlag(kCpuHasRVV)) { |
2863 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2864 | | } |
2865 | | #endif |
2866 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2867 | | if (TestCpuFlag(kCpuHasLSX)) { |
2868 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
2869 | | if (IS_ALIGNED(width, 8)) { |
2870 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
2871 | | } |
2872 | | } |
2873 | | #endif |
2874 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
2875 | | if (TestCpuFlag(kCpuHasLASX)) { |
2876 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
2877 | | if (IS_ALIGNED(width, 16)) { |
2878 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
2879 | | } |
2880 | | } |
2881 | | #endif |
2882 | | |
2883 | 2.24k | for (y = 0; y < height; ++y) { |
2884 | 2.11k | I210AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
2885 | 2.11k | width); |
2886 | 2.11k | if (attenuate) { |
2887 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
2888 | 0 | } |
2889 | 2.11k | dst_argb += dst_stride_argb; |
2890 | 2.11k | src_a += src_stride_a; |
2891 | 2.11k | src_y += src_stride_y; |
2892 | 2.11k | if (y & 1) { |
2893 | 1.04k | src_u += src_stride_u; |
2894 | 1.04k | src_v += src_stride_v; |
2895 | 1.04k | } |
2896 | 2.11k | } |
2897 | 133 | return 0; |
2898 | 133 | } |
2899 | | |
2900 | | // Convert I210 with Alpha to preattenuated ARGB with matrix. |
2901 | | LIBYUV_API |
2902 | | int I210AlphaToARGBMatrix(const uint16_t* src_y, |
2903 | | int src_stride_y, |
2904 | | const uint16_t* src_u, |
2905 | | int src_stride_u, |
2906 | | const uint16_t* src_v, |
2907 | | int src_stride_v, |
2908 | | const uint16_t* src_a, |
2909 | | int src_stride_a, |
2910 | | uint8_t* dst_argb, |
2911 | | int dst_stride_argb, |
2912 | | const struct YuvConstants* yuvconstants, |
2913 | | int width, |
2914 | | int height, |
2915 | 74 | int attenuate) { |
2916 | 74 | int y; |
2917 | 74 | void (*I210AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
2918 | 74 | const uint16_t* v_buf, const uint16_t* a_buf, |
2919 | 74 | uint8_t* dst_argb, |
2920 | 74 | const struct YuvConstants* yuvconstants, |
2921 | 74 | int width) = I210AlphaToARGBRow_C; |
2922 | 74 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
2923 | 74 | int width) = ARGBAttenuateRow_C; |
2924 | 74 | assert(yuvconstants); |
2925 | 74 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
2926 | 74 | height == 0) { |
2927 | 0 | return -1; |
2928 | 0 | } |
2929 | | // Negative height means invert the image. |
2930 | 74 | if (height < 0) { |
2931 | 0 | height = -height; |
2932 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
2933 | 0 | dst_stride_argb = -dst_stride_argb; |
2934 | 0 | } |
2935 | | #if defined(HAS_I210ALPHATOARGBROW_NEON) |
2936 | | if (TestCpuFlag(kCpuHasNEON)) { |
2937 | | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_NEON; |
2938 | | if (IS_ALIGNED(width, 8)) { |
2939 | | I210AlphaToARGBRow = I210AlphaToARGBRow_NEON; |
2940 | | } |
2941 | | } |
2942 | | #endif |
2943 | | #if defined(HAS_I210ALPHATOARGBROW_SVE2) |
2944 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2945 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SVE2; |
2946 | | } |
2947 | | #endif |
2948 | | #if defined(HAS_I210ALPHATOARGBROW_SME) |
2949 | | if (TestCpuFlag(kCpuHasSME)) { |
2950 | | I210AlphaToARGBRow = I210AlphaToARGBRow_SME; |
2951 | | } |
2952 | | #endif |
2953 | 74 | #if defined(HAS_I210ALPHATOARGBROW_SSSE3) |
2954 | 74 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2955 | 74 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_SSSE3; |
2956 | 74 | if (IS_ALIGNED(width, 8)) { |
2957 | 28 | I210AlphaToARGBRow = I210AlphaToARGBRow_SSSE3; |
2958 | 28 | } |
2959 | 74 | } |
2960 | 74 | #endif |
2961 | 74 | #if defined(HAS_I210ALPHATOARGBROW_AVX2) |
2962 | 74 | if (TestCpuFlag(kCpuHasAVX2)) { |
2963 | 74 | I210AlphaToARGBRow = I210AlphaToARGBRow_Any_AVX2; |
2964 | 74 | if (IS_ALIGNED(width, 16)) { |
2965 | 11 | I210AlphaToARGBRow = I210AlphaToARGBRow_AVX2; |
2966 | 11 | } |
2967 | 74 | } |
2968 | 74 | #endif |
2969 | 74 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
2970 | 74 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2971 | 74 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
2972 | 74 | if (IS_ALIGNED(width, 4)) { |
2973 | 46 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
2974 | 46 | } |
2975 | 74 | } |
2976 | 74 | #endif |
2977 | 74 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
2978 | 74 | if (TestCpuFlag(kCpuHasAVX2)) { |
2979 | 74 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
2980 | 74 | if (IS_ALIGNED(width, 8)) { |
2981 | 28 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
2982 | 28 | } |
2983 | 74 | } |
2984 | 74 | #endif |
2985 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
2986 | | if (TestCpuFlag(kCpuHasNEON)) { |
2987 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
2988 | | if (IS_ALIGNED(width, 8)) { |
2989 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
2990 | | } |
2991 | | } |
2992 | | #endif |
2993 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
2994 | | if (TestCpuFlag(kCpuHasRVV)) { |
2995 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
2996 | | } |
2997 | | #endif |
2998 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
2999 | | if (TestCpuFlag(kCpuHasLSX)) { |
3000 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
3001 | | if (IS_ALIGNED(width, 8)) { |
3002 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
3003 | | } |
3004 | | } |
3005 | | #endif |
3006 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
3007 | | if (TestCpuFlag(kCpuHasLASX)) { |
3008 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
3009 | | if (IS_ALIGNED(width, 16)) { |
3010 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
3011 | | } |
3012 | | } |
3013 | | #endif |
3014 | | |
3015 | 1.19k | for (y = 0; y < height; ++y) { |
3016 | 1.12k | I210AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
3017 | 1.12k | width); |
3018 | 1.12k | if (attenuate) { |
3019 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
3020 | 0 | } |
3021 | 1.12k | dst_argb += dst_stride_argb; |
3022 | 1.12k | src_a += src_stride_a; |
3023 | 1.12k | src_y += src_stride_y; |
3024 | 1.12k | src_u += src_stride_u; |
3025 | 1.12k | src_v += src_stride_v; |
3026 | 1.12k | } |
3027 | 74 | return 0; |
3028 | 74 | } |
3029 | | |
3030 | | // Convert I410 with Alpha to preattenuated ARGB with matrix. |
3031 | | LIBYUV_API |
3032 | | int I410AlphaToARGBMatrix(const uint16_t* src_y, |
3033 | | int src_stride_y, |
3034 | | const uint16_t* src_u, |
3035 | | int src_stride_u, |
3036 | | const uint16_t* src_v, |
3037 | | int src_stride_v, |
3038 | | const uint16_t* src_a, |
3039 | | int src_stride_a, |
3040 | | uint8_t* dst_argb, |
3041 | | int dst_stride_argb, |
3042 | | const struct YuvConstants* yuvconstants, |
3043 | | int width, |
3044 | | int height, |
3045 | 86 | int attenuate) { |
3046 | 86 | int y; |
3047 | 86 | void (*I410AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
3048 | 86 | const uint16_t* v_buf, const uint16_t* a_buf, |
3049 | 86 | uint8_t* dst_argb, |
3050 | 86 | const struct YuvConstants* yuvconstants, |
3051 | 86 | int width) = I410AlphaToARGBRow_C; |
3052 | 86 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
3053 | 86 | int width) = ARGBAttenuateRow_C; |
3054 | 86 | assert(yuvconstants); |
3055 | 86 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
3056 | 86 | height == 0) { |
3057 | 0 | return -1; |
3058 | 0 | } |
3059 | | // Negative height means invert the image. |
3060 | 86 | if (height < 0) { |
3061 | 0 | height = -height; |
3062 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
3063 | 0 | dst_stride_argb = -dst_stride_argb; |
3064 | 0 | } |
3065 | | #if defined(HAS_I410ALPHATOARGBROW_NEON) |
3066 | | if (TestCpuFlag(kCpuHasNEON)) { |
3067 | | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_NEON; |
3068 | | if (IS_ALIGNED(width, 8)) { |
3069 | | I410AlphaToARGBRow = I410AlphaToARGBRow_NEON; |
3070 | | } |
3071 | | } |
3072 | | #endif |
3073 | | #if defined(HAS_I410ALPHATOARGBROW_SVE2) |
3074 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3075 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SVE2; |
3076 | | } |
3077 | | #endif |
3078 | | #if defined(HAS_I410ALPHATOARGBROW_SME) |
3079 | | if (TestCpuFlag(kCpuHasSME)) { |
3080 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SME; |
3081 | | } |
3082 | | #endif |
3083 | 86 | #if defined(HAS_I410ALPHATOARGBROW_SSSE3) |
3084 | 86 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3085 | 86 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_SSSE3; |
3086 | 86 | if (IS_ALIGNED(width, 8)) { |
3087 | 35 | I410AlphaToARGBRow = I410AlphaToARGBRow_SSSE3; |
3088 | 35 | } |
3089 | 86 | } |
3090 | 86 | #endif |
3091 | 86 | #if defined(HAS_I410ALPHATOARGBROW_AVX2) |
3092 | 86 | if (TestCpuFlag(kCpuHasAVX2)) { |
3093 | 86 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_AVX2; |
3094 | 86 | if (IS_ALIGNED(width, 16)) { |
3095 | 18 | I410AlphaToARGBRow = I410AlphaToARGBRow_AVX2; |
3096 | 18 | } |
3097 | 86 | } |
3098 | 86 | #endif |
3099 | 86 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
3100 | 86 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3101 | 86 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
3102 | 86 | if (IS_ALIGNED(width, 4)) { |
3103 | 52 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
3104 | 52 | } |
3105 | 86 | } |
3106 | 86 | #endif |
3107 | 86 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
3108 | 86 | if (TestCpuFlag(kCpuHasAVX2)) { |
3109 | 86 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
3110 | 86 | if (IS_ALIGNED(width, 8)) { |
3111 | 35 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
3112 | 35 | } |
3113 | 86 | } |
3114 | 86 | #endif |
3115 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
3116 | | if (TestCpuFlag(kCpuHasNEON)) { |
3117 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
3118 | | if (IS_ALIGNED(width, 8)) { |
3119 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
3120 | | } |
3121 | | } |
3122 | | #endif |
3123 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
3124 | | if (TestCpuFlag(kCpuHasRVV)) { |
3125 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
3126 | | } |
3127 | | #endif |
3128 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
3129 | | if (TestCpuFlag(kCpuHasLSX)) { |
3130 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
3131 | | if (IS_ALIGNED(width, 8)) { |
3132 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
3133 | | } |
3134 | | } |
3135 | | #endif |
3136 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
3137 | | if (TestCpuFlag(kCpuHasLASX)) { |
3138 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
3139 | | if (IS_ALIGNED(width, 16)) { |
3140 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
3141 | | } |
3142 | | } |
3143 | | #endif |
3144 | | |
3145 | 1.57k | for (y = 0; y < height; ++y) { |
3146 | 1.48k | I410AlphaToARGBRow(src_y, src_u, src_v, src_a, dst_argb, yuvconstants, |
3147 | 1.48k | width); |
3148 | 1.48k | if (attenuate) { |
3149 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
3150 | 0 | } |
3151 | 1.48k | dst_argb += dst_stride_argb; |
3152 | 1.48k | src_a += src_stride_a; |
3153 | 1.48k | src_y += src_stride_y; |
3154 | 1.48k | src_u += src_stride_u; |
3155 | 1.48k | src_v += src_stride_v; |
3156 | 1.48k | } |
3157 | 86 | return 0; |
3158 | 86 | } |
3159 | | |
3160 | | // Convert I400 to ARGB with matrix. |
3161 | | LIBYUV_API |
3162 | | int I400ToARGBMatrix(const uint8_t* src_y, |
3163 | | int src_stride_y, |
3164 | | uint8_t* dst_argb, |
3165 | | int dst_stride_argb, |
3166 | | const struct YuvConstants* yuvconstants, |
3167 | | int width, |
3168 | 296 | int height) { |
3169 | 296 | int y; |
3170 | 296 | void (*I400ToARGBRow)(const uint8_t* y_buf, uint8_t* rgb_buf, |
3171 | 296 | const struct YuvConstants* yuvconstants, int width) = |
3172 | 296 | I400ToARGBRow_C; |
3173 | 296 | assert(yuvconstants); |
3174 | 296 | if (!src_y || !dst_argb || width <= 0 || height == 0) { |
3175 | 0 | return -1; |
3176 | 0 | } |
3177 | | // Negative height means invert the image. |
3178 | 296 | if (height < 0) { |
3179 | 0 | height = -height; |
3180 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
3181 | 0 | dst_stride_argb = -dst_stride_argb; |
3182 | 0 | } |
3183 | | // Coalesce rows. |
3184 | 296 | if (src_stride_y == width && dst_stride_argb == width * 4) { |
3185 | 296 | width *= height; |
3186 | 296 | height = 1; |
3187 | 296 | src_stride_y = dst_stride_argb = 0; |
3188 | 296 | } |
3189 | 296 | #if defined(HAS_I400TOARGBROW_SSE2) |
3190 | 296 | if (TestCpuFlag(kCpuHasSSE2)) { |
3191 | 296 | I400ToARGBRow = I400ToARGBRow_Any_SSE2; |
3192 | 296 | if (IS_ALIGNED(width, 8)) { |
3193 | 82 | I400ToARGBRow = I400ToARGBRow_SSE2; |
3194 | 82 | } |
3195 | 296 | } |
3196 | 296 | #endif |
3197 | 296 | #if defined(HAS_I400TOARGBROW_AVX2) |
3198 | 296 | if (TestCpuFlag(kCpuHasAVX2)) { |
3199 | 296 | I400ToARGBRow = I400ToARGBRow_Any_AVX2; |
3200 | 296 | if (IS_ALIGNED(width, 16)) { |
3201 | 51 | I400ToARGBRow = I400ToARGBRow_AVX2; |
3202 | 51 | } |
3203 | 296 | } |
3204 | 296 | #endif |
3205 | | #if defined(HAS_I400TOARGBROW_NEON) |
3206 | | if (TestCpuFlag(kCpuHasNEON)) { |
3207 | | I400ToARGBRow = I400ToARGBRow_Any_NEON; |
3208 | | if (IS_ALIGNED(width, 8)) { |
3209 | | I400ToARGBRow = I400ToARGBRow_NEON; |
3210 | | } |
3211 | | } |
3212 | | #endif |
3213 | | #if defined(HAS_I400TOARGBROW_SVE2) |
3214 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3215 | | I400ToARGBRow = I400ToARGBRow_SVE2; |
3216 | | } |
3217 | | #endif |
3218 | | #if defined(HAS_I400TOARGBROW_SME) |
3219 | | if (TestCpuFlag(kCpuHasSME)) { |
3220 | | I400ToARGBRow = I400ToARGBRow_SME; |
3221 | | } |
3222 | | #endif |
3223 | | #if defined(HAS_I400TOARGBROW_LSX) |
3224 | | if (TestCpuFlag(kCpuHasLSX)) { |
3225 | | I400ToARGBRow = I400ToARGBRow_Any_LSX; |
3226 | | if (IS_ALIGNED(width, 16)) { |
3227 | | I400ToARGBRow = I400ToARGBRow_LSX; |
3228 | | } |
3229 | | } |
3230 | | #endif |
3231 | | #if defined(HAS_I400TOARGBROW_RVV) |
3232 | | if (TestCpuFlag(kCpuHasRVV)) { |
3233 | | I400ToARGBRow = I400ToARGBRow_RVV; |
3234 | | } |
3235 | | #endif |
3236 | | |
3237 | 592 | for (y = 0; y < height; ++y) { |
3238 | 296 | I400ToARGBRow(src_y, dst_argb, yuvconstants, width); |
3239 | 296 | dst_argb += dst_stride_argb; |
3240 | 296 | src_y += src_stride_y; |
3241 | 296 | } |
3242 | 296 | return 0; |
3243 | 296 | } |
3244 | | |
3245 | | // Convert I400 to ARGB. |
3246 | | LIBYUV_API |
3247 | | int I400ToARGB(const uint8_t* src_y, |
3248 | | int src_stride_y, |
3249 | | uint8_t* dst_argb, |
3250 | | int dst_stride_argb, |
3251 | | int width, |
3252 | 0 | int height) { |
3253 | 0 | return I400ToARGBMatrix(src_y, src_stride_y, dst_argb, dst_stride_argb, |
3254 | 0 | &kYuvI601Constants, width, height); |
3255 | 0 | } |
3256 | | |
3257 | | // Convert J400 to ARGB. |
3258 | | LIBYUV_API |
3259 | | int J400ToARGB(const uint8_t* src_y, |
3260 | | int src_stride_y, |
3261 | | uint8_t* dst_argb, |
3262 | | int dst_stride_argb, |
3263 | | int width, |
3264 | 0 | int height) { |
3265 | 0 | int y; |
3266 | 0 | void (*J400ToARGBRow)(const uint8_t* src_y, uint8_t* dst_argb, int width) = |
3267 | 0 | J400ToARGBRow_C; |
3268 | 0 | if (!src_y || !dst_argb || width <= 0 || height == 0) { |
3269 | 0 | return -1; |
3270 | 0 | } |
3271 | | // Negative height means invert the image. |
3272 | 0 | if (height < 0) { |
3273 | 0 | height = -height; |
3274 | 0 | src_y = src_y + (height - 1) * src_stride_y; |
3275 | 0 | src_stride_y = -src_stride_y; |
3276 | 0 | } |
3277 | | // Coalesce rows. |
3278 | 0 | if (src_stride_y == width && dst_stride_argb == width * 4) { |
3279 | 0 | width *= height; |
3280 | 0 | height = 1; |
3281 | 0 | src_stride_y = dst_stride_argb = 0; |
3282 | 0 | } |
3283 | 0 | #if defined(HAS_J400TOARGBROW_SSE2) |
3284 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
3285 | 0 | J400ToARGBRow = J400ToARGBRow_Any_SSE2; |
3286 | 0 | if (IS_ALIGNED(width, 8)) { |
3287 | 0 | J400ToARGBRow = J400ToARGBRow_SSE2; |
3288 | 0 | } |
3289 | 0 | } |
3290 | 0 | #endif |
3291 | | #if defined(HAS_J400TOARGBROW_AVX2) |
3292 | | if (TestCpuFlag(kCpuHasAVX2)) { |
3293 | | J400ToARGBRow = J400ToARGBRow_Any_AVX2; |
3294 | | if (IS_ALIGNED(width, 16)) { |
3295 | | J400ToARGBRow = J400ToARGBRow_AVX2; |
3296 | | } |
3297 | | } |
3298 | | #endif |
3299 | | #if defined(HAS_J400TOARGBROW_NEON) |
3300 | | if (TestCpuFlag(kCpuHasNEON)) { |
3301 | | J400ToARGBRow = J400ToARGBRow_Any_NEON; |
3302 | | if (IS_ALIGNED(width, 8)) { |
3303 | | J400ToARGBRow = J400ToARGBRow_NEON; |
3304 | | } |
3305 | | } |
3306 | | #endif |
3307 | | #if defined(HAS_J400TOARGBROW_LSX) |
3308 | | if (TestCpuFlag(kCpuHasLSX)) { |
3309 | | J400ToARGBRow = J400ToARGBRow_Any_LSX; |
3310 | | if (IS_ALIGNED(width, 16)) { |
3311 | | J400ToARGBRow = J400ToARGBRow_LSX; |
3312 | | } |
3313 | | } |
3314 | | #endif |
3315 | | #if defined(HAS_J400TOARGBROW_RVV) |
3316 | | if (TestCpuFlag(kCpuHasRVV)) { |
3317 | | J400ToARGBRow = J400ToARGBRow_RVV; |
3318 | | } |
3319 | | #endif |
3320 | |
|
3321 | 0 | for (y = 0; y < height; ++y) { |
3322 | 0 | J400ToARGBRow(src_y, dst_argb, width); |
3323 | 0 | src_y += src_stride_y; |
3324 | 0 | dst_argb += dst_stride_argb; |
3325 | 0 | } |
3326 | 0 | return 0; |
3327 | 0 | } |
3328 | | |
3329 | | #ifndef __riscv |
3330 | | // Shuffle table for converting BGRA to ARGB. |
3331 | | static const uvec8 kShuffleMaskBGRAToARGB = { |
3332 | | 3u, 2u, 1u, 0u, 7u, 6u, 5u, 4u, 11u, 10u, 9u, 8u, 15u, 14u, 13u, 12u}; |
3333 | | |
3334 | | // Shuffle table for converting ABGR to ARGB. |
3335 | | static const uvec8 kShuffleMaskABGRToARGB = { |
3336 | | 2u, 1u, 0u, 3u, 6u, 5u, 4u, 7u, 10u, 9u, 8u, 11u, 14u, 13u, 12u, 15u}; |
3337 | | |
3338 | | // Shuffle table for converting RGBA to ARGB. |
3339 | | static const uvec8 kShuffleMaskRGBAToARGB = { |
3340 | | 1u, 2u, 3u, 0u, 5u, 6u, 7u, 4u, 9u, 10u, 11u, 8u, 13u, 14u, 15u, 12u}; |
3341 | | |
3342 | | // Shuffle table for converting AR64 to AB64. |
3343 | | static const uvec8 kShuffleMaskAR64ToAB64 = { |
3344 | | 4u, 5u, 2u, 3u, 0u, 1u, 6u, 7u, 12u, 13u, 10u, 11u, 8u, 9u, 14u, 15u}; |
3345 | | |
3346 | | // Convert BGRA to ARGB. |
3347 | | LIBYUV_API |
3348 | | int BGRAToARGB(const uint8_t* src_bgra, |
3349 | | int src_stride_bgra, |
3350 | | uint8_t* dst_argb, |
3351 | | int dst_stride_argb, |
3352 | | int width, |
3353 | 0 | int height) { |
3354 | 0 | return ARGBShuffle(src_bgra, src_stride_bgra, dst_argb, dst_stride_argb, |
3355 | 0 | (const uint8_t*)&kShuffleMaskBGRAToARGB, width, height); |
3356 | 0 | } |
3357 | | |
3358 | | // Convert ARGB to BGRA (same as BGRAToARGB). |
3359 | | LIBYUV_API |
3360 | | int ARGBToBGRA(const uint8_t* src_bgra, |
3361 | | int src_stride_bgra, |
3362 | | uint8_t* dst_argb, |
3363 | | int dst_stride_argb, |
3364 | | int width, |
3365 | 0 | int height) { |
3366 | 0 | return ARGBShuffle(src_bgra, src_stride_bgra, dst_argb, dst_stride_argb, |
3367 | 0 | (const uint8_t*)&kShuffleMaskBGRAToARGB, width, height); |
3368 | 0 | } |
3369 | | |
3370 | | // Convert ABGR to ARGB. |
3371 | | LIBYUV_API |
3372 | | int ABGRToARGB(const uint8_t* src_abgr, |
3373 | | int src_stride_abgr, |
3374 | | uint8_t* dst_argb, |
3375 | | int dst_stride_argb, |
3376 | | int width, |
3377 | 0 | int height) { |
3378 | 0 | return ARGBShuffle(src_abgr, src_stride_abgr, dst_argb, dst_stride_argb, |
3379 | 0 | (const uint8_t*)&kShuffleMaskABGRToARGB, width, height); |
3380 | 0 | } |
3381 | | |
3382 | | // Convert ARGB to ABGR to (same as ABGRToARGB). |
3383 | | LIBYUV_API |
3384 | | int ARGBToABGR(const uint8_t* src_abgr, |
3385 | | int src_stride_abgr, |
3386 | | uint8_t* dst_argb, |
3387 | | int dst_stride_argb, |
3388 | | int width, |
3389 | 0 | int height) { |
3390 | 0 | return ARGBShuffle(src_abgr, src_stride_abgr, dst_argb, dst_stride_argb, |
3391 | 0 | (const uint8_t*)&kShuffleMaskABGRToARGB, width, height); |
3392 | 0 | } |
3393 | | |
3394 | | // Convert RGBA to ARGB. |
3395 | | LIBYUV_API |
3396 | | int RGBAToARGB(const uint8_t* src_rgba, |
3397 | | int src_stride_rgba, |
3398 | | uint8_t* dst_argb, |
3399 | | int dst_stride_argb, |
3400 | | int width, |
3401 | 0 | int height) { |
3402 | 0 | return ARGBShuffle(src_rgba, src_stride_rgba, dst_argb, dst_stride_argb, |
3403 | 0 | (const uint8_t*)&kShuffleMaskRGBAToARGB, width, height); |
3404 | 0 | } |
3405 | | |
3406 | | // Convert AR64 To AB64. |
3407 | | LIBYUV_API |
3408 | | int AR64ToAB64(const uint16_t* src_ar64, |
3409 | | int src_stride_ar64, |
3410 | | uint16_t* dst_ab64, |
3411 | | int dst_stride_ab64, |
3412 | | int width, |
3413 | 0 | int height) { |
3414 | 0 | return AR64Shuffle(src_ar64, src_stride_ar64, dst_ab64, dst_stride_ab64, |
3415 | 0 | (const uint8_t*)&kShuffleMaskAR64ToAB64, width, height); |
3416 | 0 | } |
3417 | | #else |
3418 | | // Convert BGRA to ARGB (same as ARGBToBGRA). |
3419 | | LIBYUV_API |
3420 | | int BGRAToARGB(const uint8_t* src_bgra, |
3421 | | int src_stride_bgra, |
3422 | | uint8_t* dst_argb, |
3423 | | int dst_stride_argb, |
3424 | | int width, |
3425 | | int height) { |
3426 | | return ARGBToBGRA(src_bgra, src_stride_bgra, dst_argb, dst_stride_argb, width, |
3427 | | height); |
3428 | | } |
3429 | | |
3430 | | // Convert ARGB to BGRA. |
3431 | | LIBYUV_API |
3432 | | int ARGBToBGRA(const uint8_t* src_argb, |
3433 | | int src_stride_argb, |
3434 | | uint8_t* dst_bgra, |
3435 | | int dst_stride_bgra, |
3436 | | int width, |
3437 | | int height) { |
3438 | | int y; |
3439 | | void (*ARGBToBGRARow)(const uint8_t* src_argb, uint8_t* dst_bgra, int width) = |
3440 | | ARGBToBGRARow_C; |
3441 | | if (!src_argb || !dst_bgra || width <= 0 || height == 0) { |
3442 | | return -1; |
3443 | | } |
3444 | | // Negative height means invert the image. |
3445 | | if (height < 0) { |
3446 | | height = -height; |
3447 | | src_argb = src_argb + (height - 1) * src_stride_argb; |
3448 | | src_stride_argb = -src_stride_argb; |
3449 | | } |
3450 | | // Coalesce rows. |
3451 | | if (src_stride_argb == width * 4 && dst_stride_bgra == width * 4) { |
3452 | | width *= height; |
3453 | | height = 1; |
3454 | | src_stride_argb = dst_stride_bgra = 0; |
3455 | | } |
3456 | | |
3457 | | #if defined(HAS_ARGBTOBGRAROW_RVV) |
3458 | | if (TestCpuFlag(kCpuHasRVV)) { |
3459 | | ARGBToBGRARow = ARGBToBGRARow_RVV; |
3460 | | } |
3461 | | #endif |
3462 | | |
3463 | | for (y = 0; y < height; ++y) { |
3464 | | ARGBToBGRARow(src_argb, dst_bgra, width); |
3465 | | src_argb += src_stride_argb; |
3466 | | dst_bgra += dst_stride_bgra; |
3467 | | } |
3468 | | return 0; |
3469 | | } |
3470 | | |
3471 | | // Convert ARGB to ABGR. |
3472 | | LIBYUV_API |
3473 | | int ARGBToABGR(const uint8_t* src_argb, |
3474 | | int src_stride_argb, |
3475 | | uint8_t* dst_abgr, |
3476 | | int dst_stride_abgr, |
3477 | | int width, |
3478 | | int height) { |
3479 | | int y; |
3480 | | void (*ARGBToABGRRow)(const uint8_t* src_argb, uint8_t* dst_abgr, int width) = |
3481 | | ARGBToABGRRow_C; |
3482 | | if (!src_argb || !dst_abgr || width <= 0 || height == 0) { |
3483 | | return -1; |
3484 | | } |
3485 | | // Negative height means invert the image. |
3486 | | if (height < 0) { |
3487 | | height = -height; |
3488 | | src_argb = src_argb + (height - 1) * src_stride_argb; |
3489 | | src_stride_argb = -src_stride_argb; |
3490 | | } |
3491 | | // Coalesce rows. |
3492 | | if (src_stride_argb == width * 4 && dst_stride_abgr == width * 4) { |
3493 | | width *= height; |
3494 | | height = 1; |
3495 | | src_stride_argb = dst_stride_abgr = 0; |
3496 | | } |
3497 | | |
3498 | | #if defined(HAS_ARGBTOABGRROW_RVV) |
3499 | | if (TestCpuFlag(kCpuHasRVV)) { |
3500 | | ARGBToABGRRow = ARGBToABGRRow_RVV; |
3501 | | } |
3502 | | #endif |
3503 | | |
3504 | | for (y = 0; y < height; ++y) { |
3505 | | ARGBToABGRRow(src_argb, dst_abgr, width); |
3506 | | src_argb += src_stride_argb; |
3507 | | dst_abgr += dst_stride_abgr; |
3508 | | } |
3509 | | return 0; |
3510 | | } |
3511 | | |
3512 | | // Convert ABGR to ARGB (same as ARGBToABGR). |
3513 | | LIBYUV_API |
3514 | | int ABGRToARGB(const uint8_t* src_abgr, |
3515 | | int src_stride_abgr, |
3516 | | uint8_t* dst_argb, |
3517 | | int dst_stride_argb, |
3518 | | int width, |
3519 | | int height) { |
3520 | | return ARGBToABGR(src_abgr, src_stride_abgr, dst_argb, dst_stride_argb, width, |
3521 | | height); |
3522 | | } |
3523 | | |
3524 | | // Convert RGBA to ARGB. |
3525 | | LIBYUV_API |
3526 | | int RGBAToARGB(const uint8_t* src_rgba, |
3527 | | int src_stride_rgba, |
3528 | | uint8_t* dst_argb, |
3529 | | int dst_stride_argb, |
3530 | | int width, |
3531 | | int height) { |
3532 | | int y; |
3533 | | void (*RGBAToARGBRow)(const uint8_t* src_rgba, uint8_t* dst_argb, int width) = |
3534 | | RGBAToARGBRow_C; |
3535 | | if (!src_rgba || !dst_argb || width <= 0 || height == 0) { |
3536 | | return -1; |
3537 | | } |
3538 | | // Negative height means invert the image. |
3539 | | if (height < 0) { |
3540 | | height = -height; |
3541 | | src_rgba = src_rgba + (height - 1) * src_stride_rgba; |
3542 | | src_stride_rgba = -src_stride_rgba; |
3543 | | } |
3544 | | // Coalesce rows. |
3545 | | if (src_stride_rgba == width * 4 && dst_stride_argb == width * 4) { |
3546 | | width *= height; |
3547 | | height = 1; |
3548 | | src_stride_rgba = dst_stride_argb = 0; |
3549 | | } |
3550 | | |
3551 | | #if defined(HAS_RGBATOARGBROW_RVV) |
3552 | | if (TestCpuFlag(kCpuHasRVV)) { |
3553 | | RGBAToARGBRow = RGBAToARGBRow_RVV; |
3554 | | } |
3555 | | #endif |
3556 | | |
3557 | | for (y = 0; y < height; ++y) { |
3558 | | RGBAToARGBRow(src_rgba, dst_argb, width); |
3559 | | src_rgba += src_stride_rgba; |
3560 | | dst_argb += dst_stride_argb; |
3561 | | } |
3562 | | return 0; |
3563 | | } |
3564 | | |
3565 | | // Convert AR64 To AB64. |
3566 | | LIBYUV_API |
3567 | | int AR64ToAB64(const uint16_t* src_ar64, |
3568 | | int src_stride_ar64, |
3569 | | uint16_t* dst_ab64, |
3570 | | int dst_stride_ab64, |
3571 | | int width, |
3572 | | int height) { |
3573 | | int y; |
3574 | | void (*AR64ToAB64Row)(const uint16_t* src_ar64, uint16_t* dst_ab64, |
3575 | | int width) = AR64ToAB64Row_C; |
3576 | | if (!src_ar64 || !dst_ab64 || width <= 0 || height == 0) { |
3577 | | return -1; |
3578 | | } |
3579 | | // Negative height means invert the image. |
3580 | | if (height < 0) { |
3581 | | height = -height; |
3582 | | src_ar64 = src_ar64 + (height - 1) * src_stride_ar64; |
3583 | | src_stride_ar64 = -src_stride_ar64; |
3584 | | } |
3585 | | // Coalesce rows. |
3586 | | if (src_stride_ar64 == width * 4 && dst_stride_ab64 == width * 4) { |
3587 | | width *= height; |
3588 | | height = 1; |
3589 | | src_stride_ar64 = dst_stride_ab64 = 0; |
3590 | | } |
3591 | | |
3592 | | #if defined(HAS_AR64TOAB64ROW_RVV) |
3593 | | if (TestCpuFlag(kCpuHasRVV)) { |
3594 | | AR64ToAB64Row = AR64ToAB64Row_RVV; |
3595 | | } |
3596 | | #endif |
3597 | | |
3598 | | for (y = 0; y < height; ++y) { |
3599 | | AR64ToAB64Row(src_ar64, dst_ab64, width); |
3600 | | src_ar64 += src_stride_ar64; |
3601 | | dst_ab64 += dst_stride_ab64; |
3602 | | } |
3603 | | return 0; |
3604 | | } |
3605 | | #endif |
3606 | | |
3607 | | // Convert RGB24 to ARGB. |
3608 | | LIBYUV_API |
3609 | | int RGB24ToARGB(const uint8_t* src_rgb24, |
3610 | | int src_stride_rgb24, |
3611 | | uint8_t* dst_argb, |
3612 | | int dst_stride_argb, |
3613 | | int width, |
3614 | 0 | int height) { |
3615 | 0 | int y; |
3616 | 0 | void (*RGB24ToARGBRow)(const uint8_t* src_rgb, uint8_t* dst_argb, int width) = |
3617 | 0 | RGB24ToARGBRow_C; |
3618 | 0 | if (!src_rgb24 || !dst_argb || width <= 0 || height == 0) { |
3619 | 0 | return -1; |
3620 | 0 | } |
3621 | | // Negative height means invert the image. |
3622 | 0 | if (height < 0) { |
3623 | 0 | height = -height; |
3624 | 0 | src_rgb24 = src_rgb24 + (height - 1) * src_stride_rgb24; |
3625 | 0 | src_stride_rgb24 = -src_stride_rgb24; |
3626 | 0 | } |
3627 | | // Coalesce rows. |
3628 | 0 | if (src_stride_rgb24 == width * 3 && dst_stride_argb == width * 4) { |
3629 | 0 | width *= height; |
3630 | 0 | height = 1; |
3631 | 0 | src_stride_rgb24 = dst_stride_argb = 0; |
3632 | 0 | } |
3633 | 0 | #if defined(HAS_RGB24TOARGBROW_SSSE3) |
3634 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3635 | 0 | RGB24ToARGBRow = RGB24ToARGBRow_Any_SSSE3; |
3636 | 0 | if (IS_ALIGNED(width, 16)) { |
3637 | 0 | RGB24ToARGBRow = RGB24ToARGBRow_SSSE3; |
3638 | 0 | } |
3639 | 0 | } |
3640 | 0 | #endif |
3641 | | #if defined(HAS_RGB24TOARGBROW_NEON) |
3642 | | if (TestCpuFlag(kCpuHasNEON)) { |
3643 | | RGB24ToARGBRow = RGB24ToARGBRow_Any_NEON; |
3644 | | if (IS_ALIGNED(width, 8)) { |
3645 | | RGB24ToARGBRow = RGB24ToARGBRow_NEON; |
3646 | | } |
3647 | | } |
3648 | | #endif |
3649 | | #if defined(HAS_RGB24TOARGBROW_SVE2) |
3650 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3651 | | RGB24ToARGBRow = RGB24ToARGBRow_SVE2; |
3652 | | } |
3653 | | #endif |
3654 | | #if defined(HAS_RGB24TOARGBROW_LSX) |
3655 | | if (TestCpuFlag(kCpuHasLSX)) { |
3656 | | RGB24ToARGBRow = RGB24ToARGBRow_Any_LSX; |
3657 | | if (IS_ALIGNED(width, 16)) { |
3658 | | RGB24ToARGBRow = RGB24ToARGBRow_LSX; |
3659 | | } |
3660 | | } |
3661 | | #endif |
3662 | | #if defined(HAS_RGB24TOARGBROW_LASX) |
3663 | | if (TestCpuFlag(kCpuHasLASX)) { |
3664 | | RGB24ToARGBRow = RGB24ToARGBRow_Any_LASX; |
3665 | | if (IS_ALIGNED(width, 32)) { |
3666 | | RGB24ToARGBRow = RGB24ToARGBRow_LASX; |
3667 | | } |
3668 | | } |
3669 | | #endif |
3670 | | #if defined(HAS_RGB24TOARGBROW_RVV) |
3671 | | if (TestCpuFlag(kCpuHasRVV)) { |
3672 | | RGB24ToARGBRow = RGB24ToARGBRow_RVV; |
3673 | | } |
3674 | | #endif |
3675 | |
|
3676 | 0 | for (y = 0; y < height; ++y) { |
3677 | 0 | RGB24ToARGBRow(src_rgb24, dst_argb, width); |
3678 | 0 | src_rgb24 += src_stride_rgb24; |
3679 | 0 | dst_argb += dst_stride_argb; |
3680 | 0 | } |
3681 | 0 | return 0; |
3682 | 0 | } |
3683 | | |
3684 | | // Convert RAW to ARGB. |
3685 | | LIBYUV_API |
3686 | | int RAWToARGB(const uint8_t* src_raw, |
3687 | | int src_stride_raw, |
3688 | | uint8_t* dst_argb, |
3689 | | int dst_stride_argb, |
3690 | | int width, |
3691 | 118k | int height) { |
3692 | 118k | int y; |
3693 | 118k | void (*RAWToARGBRow)(const uint8_t* src_rgb, uint8_t* dst_argb, int width) = |
3694 | 118k | RAWToARGBRow_C; |
3695 | 118k | if (!src_raw || !dst_argb || width <= 0 || height == 0) { |
3696 | 0 | return -1; |
3697 | 0 | } |
3698 | | // Negative height means invert the image. |
3699 | 118k | if (height < 0) { |
3700 | 0 | height = -height; |
3701 | 0 | src_raw = src_raw + (height - 1) * src_stride_raw; |
3702 | 0 | src_stride_raw = -src_stride_raw; |
3703 | 0 | } |
3704 | | // Coalesce rows. |
3705 | 118k | if (src_stride_raw == width * 3 && dst_stride_argb == width * 4) { |
3706 | 118k | width *= height; |
3707 | 118k | height = 1; |
3708 | 118k | src_stride_raw = dst_stride_argb = 0; |
3709 | 118k | } |
3710 | 118k | #if defined(HAS_RAWTOARGBROW_SSSE3) |
3711 | 118k | if (TestCpuFlag(kCpuHasSSSE3)) { |
3712 | 118k | RAWToARGBRow = RAWToARGBRow_Any_SSSE3; |
3713 | 118k | if (IS_ALIGNED(width, 16)) { |
3714 | 12.6k | RAWToARGBRow = RAWToARGBRow_SSSE3; |
3715 | 12.6k | } |
3716 | 118k | } |
3717 | 118k | #endif |
3718 | 118k | #if defined(HAS_RAWTOARGBROW_AVX2) |
3719 | 118k | if (TestCpuFlag(kCpuHasAVX2)) { |
3720 | 118k | RAWToARGBRow = RAWToARGBRow_Any_AVX2; |
3721 | 118k | if (IS_ALIGNED(width, 32)) { |
3722 | 10.3k | RAWToARGBRow = RAWToARGBRow_AVX2; |
3723 | 10.3k | } |
3724 | 118k | } |
3725 | 118k | #endif |
3726 | | #if defined(HAS_RAWTOARGBROW_NEON) |
3727 | | if (TestCpuFlag(kCpuHasNEON)) { |
3728 | | RAWToARGBRow = RAWToARGBRow_Any_NEON; |
3729 | | if (IS_ALIGNED(width, 8)) { |
3730 | | RAWToARGBRow = RAWToARGBRow_NEON; |
3731 | | } |
3732 | | } |
3733 | | #endif |
3734 | | #if defined(HAS_RAWTOARGBROW_SVE2) |
3735 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3736 | | RAWToARGBRow = RAWToARGBRow_SVE2; |
3737 | | } |
3738 | | #endif |
3739 | | #if defined(HAS_RAWTOARGBROW_LSX) |
3740 | | if (TestCpuFlag(kCpuHasLSX)) { |
3741 | | RAWToARGBRow = RAWToARGBRow_Any_LSX; |
3742 | | if (IS_ALIGNED(width, 16)) { |
3743 | | RAWToARGBRow = RAWToARGBRow_LSX; |
3744 | | } |
3745 | | } |
3746 | | #endif |
3747 | | #if defined(HAS_RAWTOARGBROW_LASX) |
3748 | | if (TestCpuFlag(kCpuHasLASX)) { |
3749 | | RAWToARGBRow = RAWToARGBRow_Any_LASX; |
3750 | | if (IS_ALIGNED(width, 32)) { |
3751 | | RAWToARGBRow = RAWToARGBRow_LASX; |
3752 | | } |
3753 | | } |
3754 | | #endif |
3755 | | #if defined(HAS_RAWTOARGBROW_RVV) |
3756 | | if (TestCpuFlag(kCpuHasRVV)) { |
3757 | | RAWToARGBRow = RAWToARGBRow_RVV; |
3758 | | } |
3759 | | #endif |
3760 | | |
3761 | 236k | for (y = 0; y < height; ++y) { |
3762 | 118k | RAWToARGBRow(src_raw, dst_argb, width); |
3763 | 118k | src_raw += src_stride_raw; |
3764 | 118k | dst_argb += dst_stride_argb; |
3765 | 118k | } |
3766 | 118k | return 0; |
3767 | 118k | } |
3768 | | |
3769 | | // Convert RAW to RGBA. |
3770 | | LIBYUV_API |
3771 | | int RAWToRGBA(const uint8_t* src_raw, |
3772 | | int src_stride_raw, |
3773 | | uint8_t* dst_rgba, |
3774 | | int dst_stride_rgba, |
3775 | | int width, |
3776 | 0 | int height) { |
3777 | 0 | int y; |
3778 | 0 | void (*RAWToRGBARow)(const uint8_t* src_rgb, uint8_t* dst_rgba, int width) = |
3779 | 0 | RAWToRGBARow_C; |
3780 | 0 | if (!src_raw || !dst_rgba || width <= 0 || height == 0) { |
3781 | 0 | return -1; |
3782 | 0 | } |
3783 | | // Negative height means invert the image. |
3784 | 0 | if (height < 0) { |
3785 | 0 | height = -height; |
3786 | 0 | src_raw = src_raw + (height - 1) * src_stride_raw; |
3787 | 0 | src_stride_raw = -src_stride_raw; |
3788 | 0 | } |
3789 | | // Coalesce rows. |
3790 | 0 | if (src_stride_raw == width * 3 && dst_stride_rgba == width * 4) { |
3791 | 0 | width *= height; |
3792 | 0 | height = 1; |
3793 | 0 | src_stride_raw = dst_stride_rgba = 0; |
3794 | 0 | } |
3795 | 0 | #if defined(HAS_RAWTORGBAROW_SSSE3) |
3796 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
3797 | 0 | RAWToRGBARow = RAWToRGBARow_Any_SSSE3; |
3798 | 0 | if (IS_ALIGNED(width, 16)) { |
3799 | 0 | RAWToRGBARow = RAWToRGBARow_SSSE3; |
3800 | 0 | } |
3801 | 0 | } |
3802 | 0 | #endif |
3803 | | #if defined(HAS_RAWTORGBAROW_NEON) |
3804 | | if (TestCpuFlag(kCpuHasNEON)) { |
3805 | | RAWToRGBARow = RAWToRGBARow_Any_NEON; |
3806 | | if (IS_ALIGNED(width, 8)) { |
3807 | | RAWToRGBARow = RAWToRGBARow_NEON; |
3808 | | } |
3809 | | } |
3810 | | #endif |
3811 | | #if defined(HAS_RAWTORGBAROW_SVE2) |
3812 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3813 | | RAWToRGBARow = RAWToRGBARow_SVE2; |
3814 | | } |
3815 | | #endif |
3816 | | #if defined(HAS_RAWTORGBAROW_RVV) |
3817 | | if (TestCpuFlag(kCpuHasRVV)) { |
3818 | | RAWToRGBARow = RAWToRGBARow_RVV; |
3819 | | } |
3820 | | #endif |
3821 | |
|
3822 | 0 | for (y = 0; y < height; ++y) { |
3823 | 0 | RAWToRGBARow(src_raw, dst_rgba, width); |
3824 | 0 | src_raw += src_stride_raw; |
3825 | 0 | dst_rgba += dst_stride_rgba; |
3826 | 0 | } |
3827 | 0 | return 0; |
3828 | 0 | } |
3829 | | |
3830 | | // Convert RGB565 to ARGB. |
3831 | | LIBYUV_API |
3832 | | int RGB565ToARGB(const uint8_t* src_rgb565, |
3833 | | int src_stride_rgb565, |
3834 | | uint8_t* dst_argb, |
3835 | | int dst_stride_argb, |
3836 | | int width, |
3837 | 0 | int height) { |
3838 | 0 | int y; |
3839 | 0 | void (*RGB565ToARGBRow)(const uint8_t* src_rgb565, uint8_t* dst_argb, |
3840 | 0 | int width) = RGB565ToARGBRow_C; |
3841 | 0 | if (!src_rgb565 || !dst_argb || width <= 0 || height == 0) { |
3842 | 0 | return -1; |
3843 | 0 | } |
3844 | | // Negative height means invert the image. |
3845 | 0 | if (height < 0) { |
3846 | 0 | height = -height; |
3847 | 0 | src_rgb565 = src_rgb565 + (height - 1) * src_stride_rgb565; |
3848 | 0 | src_stride_rgb565 = -src_stride_rgb565; |
3849 | 0 | } |
3850 | | // Coalesce rows. |
3851 | 0 | if (src_stride_rgb565 == width * 2 && dst_stride_argb == width * 4) { |
3852 | 0 | width *= height; |
3853 | 0 | height = 1; |
3854 | 0 | src_stride_rgb565 = dst_stride_argb = 0; |
3855 | 0 | } |
3856 | 0 | #if defined(HAS_RGB565TOARGBROW_SSE2) |
3857 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
3858 | 0 | RGB565ToARGBRow = RGB565ToARGBRow_Any_SSE2; |
3859 | 0 | if (IS_ALIGNED(width, 8)) { |
3860 | 0 | RGB565ToARGBRow = RGB565ToARGBRow_SSE2; |
3861 | 0 | } |
3862 | 0 | } |
3863 | 0 | #endif |
3864 | | #if defined(HAS_RGB565TOARGBROW_AVX2) |
3865 | | if (TestCpuFlag(kCpuHasAVX2)) { |
3866 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_AVX2; |
3867 | | if (IS_ALIGNED(width, 16)) { |
3868 | | RGB565ToARGBRow = RGB565ToARGBRow_AVX2; |
3869 | | } |
3870 | | } |
3871 | | #endif |
3872 | | #if defined(HAS_RGB565TOARGBROW_NEON) |
3873 | | if (TestCpuFlag(kCpuHasNEON)) { |
3874 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_NEON; |
3875 | | if (IS_ALIGNED(width, 16)) { |
3876 | | RGB565ToARGBRow = RGB565ToARGBRow_NEON; |
3877 | | } |
3878 | | } |
3879 | | #endif |
3880 | | #if defined(HAS_RGB565TOARGBROW_LSX) |
3881 | | if (TestCpuFlag(kCpuHasLSX)) { |
3882 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_LSX; |
3883 | | if (IS_ALIGNED(width, 16)) { |
3884 | | RGB565ToARGBRow = RGB565ToARGBRow_LSX; |
3885 | | } |
3886 | | } |
3887 | | #endif |
3888 | | #if defined(HAS_RGB565TOARGBROW_LASX) |
3889 | | if (TestCpuFlag(kCpuHasLASX)) { |
3890 | | RGB565ToARGBRow = RGB565ToARGBRow_Any_LASX; |
3891 | | if (IS_ALIGNED(width, 32)) { |
3892 | | RGB565ToARGBRow = RGB565ToARGBRow_LASX; |
3893 | | } |
3894 | | } |
3895 | | #endif |
3896 | |
|
3897 | 0 | for (y = 0; y < height; ++y) { |
3898 | 0 | RGB565ToARGBRow(src_rgb565, dst_argb, width); |
3899 | 0 | src_rgb565 += src_stride_rgb565; |
3900 | 0 | dst_argb += dst_stride_argb; |
3901 | 0 | } |
3902 | 0 | return 0; |
3903 | 0 | } |
3904 | | |
3905 | | // Convert ARGB1555 to ARGB. |
3906 | | LIBYUV_API |
3907 | | int ARGB1555ToARGB(const uint8_t* src_argb1555, |
3908 | | int src_stride_argb1555, |
3909 | | uint8_t* dst_argb, |
3910 | | int dst_stride_argb, |
3911 | | int width, |
3912 | 0 | int height) { |
3913 | 0 | int y; |
3914 | 0 | void (*ARGB1555ToARGBRow)(const uint8_t* src_argb1555, uint8_t* dst_argb, |
3915 | 0 | int width) = ARGB1555ToARGBRow_C; |
3916 | 0 | if (!src_argb1555 || !dst_argb || width <= 0 || height == 0) { |
3917 | 0 | return -1; |
3918 | 0 | } |
3919 | | // Negative height means invert the image. |
3920 | 0 | if (height < 0) { |
3921 | 0 | height = -height; |
3922 | 0 | src_argb1555 = src_argb1555 + (height - 1) * src_stride_argb1555; |
3923 | 0 | src_stride_argb1555 = -src_stride_argb1555; |
3924 | 0 | } |
3925 | | // Coalesce rows. |
3926 | 0 | if (src_stride_argb1555 == width * 2 && dst_stride_argb == width * 4) { |
3927 | 0 | width *= height; |
3928 | 0 | height = 1; |
3929 | 0 | src_stride_argb1555 = dst_stride_argb = 0; |
3930 | 0 | } |
3931 | 0 | #if defined(HAS_ARGB1555TOARGBROW_SSE2) |
3932 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
3933 | 0 | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_SSE2; |
3934 | 0 | if (IS_ALIGNED(width, 8)) { |
3935 | 0 | ARGB1555ToARGBRow = ARGB1555ToARGBRow_SSE2; |
3936 | 0 | } |
3937 | 0 | } |
3938 | 0 | #endif |
3939 | | #if defined(HAS_ARGB1555TOARGBROW_AVX2) |
3940 | | if (TestCpuFlag(kCpuHasAVX2)) { |
3941 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_AVX2; |
3942 | | if (IS_ALIGNED(width, 16)) { |
3943 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_AVX2; |
3944 | | } |
3945 | | } |
3946 | | #endif |
3947 | | #if defined(HAS_ARGB1555TOARGBROW_NEON) |
3948 | | if (TestCpuFlag(kCpuHasNEON)) { |
3949 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_NEON; |
3950 | | if (IS_ALIGNED(width, 16)) { |
3951 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_NEON; |
3952 | | } |
3953 | | } |
3954 | | #endif |
3955 | | #if defined(HAS_ARGB1555TOARGBROW_SVE2) |
3956 | | if (TestCpuFlag(kCpuHasSVE2)) { |
3957 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_SVE2; |
3958 | | } |
3959 | | #endif |
3960 | | #if defined(HAS_ARGB1555TOARGBROW_LSX) |
3961 | | if (TestCpuFlag(kCpuHasLSX)) { |
3962 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_LSX; |
3963 | | if (IS_ALIGNED(width, 16)) { |
3964 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_LSX; |
3965 | | } |
3966 | | } |
3967 | | #endif |
3968 | | #if defined(HAS_ARGB1555TOARGBROW_LASX) |
3969 | | if (TestCpuFlag(kCpuHasLASX)) { |
3970 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_Any_LASX; |
3971 | | if (IS_ALIGNED(width, 32)) { |
3972 | | ARGB1555ToARGBRow = ARGB1555ToARGBRow_LASX; |
3973 | | } |
3974 | | } |
3975 | | #endif |
3976 | |
|
3977 | 0 | for (y = 0; y < height; ++y) { |
3978 | 0 | ARGB1555ToARGBRow(src_argb1555, dst_argb, width); |
3979 | 0 | src_argb1555 += src_stride_argb1555; |
3980 | 0 | dst_argb += dst_stride_argb; |
3981 | 0 | } |
3982 | 0 | return 0; |
3983 | 0 | } |
3984 | | |
3985 | | // Convert ARGB4444 to ARGB. |
3986 | | LIBYUV_API |
3987 | | int ARGB4444ToARGB(const uint8_t* src_argb4444, |
3988 | | int src_stride_argb4444, |
3989 | | uint8_t* dst_argb, |
3990 | | int dst_stride_argb, |
3991 | | int width, |
3992 | 0 | int height) { |
3993 | 0 | int y; |
3994 | 0 | void (*ARGB4444ToARGBRow)(const uint8_t* src_argb4444, uint8_t* dst_argb, |
3995 | 0 | int width) = ARGB4444ToARGBRow_C; |
3996 | 0 | if (!src_argb4444 || !dst_argb || width <= 0 || height == 0) { |
3997 | 0 | return -1; |
3998 | 0 | } |
3999 | | // Negative height means invert the image. |
4000 | 0 | if (height < 0) { |
4001 | 0 | height = -height; |
4002 | 0 | src_argb4444 = src_argb4444 + (height - 1) * src_stride_argb4444; |
4003 | 0 | src_stride_argb4444 = -src_stride_argb4444; |
4004 | 0 | } |
4005 | | // Coalesce rows. |
4006 | 0 | if (src_stride_argb4444 == width * 2 && dst_stride_argb == width * 4) { |
4007 | 0 | width *= height; |
4008 | 0 | height = 1; |
4009 | 0 | src_stride_argb4444 = dst_stride_argb = 0; |
4010 | 0 | } |
4011 | 0 | #if defined(HAS_ARGB4444TOARGBROW_SSE2) |
4012 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
4013 | 0 | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_SSE2; |
4014 | 0 | if (IS_ALIGNED(width, 8)) { |
4015 | 0 | ARGB4444ToARGBRow = ARGB4444ToARGBRow_SSE2; |
4016 | 0 | } |
4017 | 0 | } |
4018 | 0 | #endif |
4019 | | #if defined(HAS_ARGB4444TOARGBROW_AVX2) |
4020 | | if (TestCpuFlag(kCpuHasAVX2)) { |
4021 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_AVX2; |
4022 | | if (IS_ALIGNED(width, 16)) { |
4023 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_AVX2; |
4024 | | } |
4025 | | } |
4026 | | #endif |
4027 | | #if defined(HAS_ARGB4444TOARGBROW_NEON) |
4028 | | if (TestCpuFlag(kCpuHasNEON)) { |
4029 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_NEON; |
4030 | | if (IS_ALIGNED(width, 8)) { |
4031 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_NEON; |
4032 | | } |
4033 | | } |
4034 | | #endif |
4035 | | #if defined(HAS_ARGB4444TOARGBROW_LSX) |
4036 | | if (TestCpuFlag(kCpuHasLSX)) { |
4037 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_LSX; |
4038 | | if (IS_ALIGNED(width, 16)) { |
4039 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_LSX; |
4040 | | } |
4041 | | } |
4042 | | #endif |
4043 | | #if defined(HAS_ARGB4444TOARGBROW_LASX) |
4044 | | if (TestCpuFlag(kCpuHasLASX)) { |
4045 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_Any_LASX; |
4046 | | if (IS_ALIGNED(width, 32)) { |
4047 | | ARGB4444ToARGBRow = ARGB4444ToARGBRow_LASX; |
4048 | | } |
4049 | | } |
4050 | | #endif |
4051 | |
|
4052 | 0 | for (y = 0; y < height; ++y) { |
4053 | 0 | ARGB4444ToARGBRow(src_argb4444, dst_argb, width); |
4054 | 0 | src_argb4444 += src_stride_argb4444; |
4055 | 0 | dst_argb += dst_stride_argb; |
4056 | 0 | } |
4057 | 0 | return 0; |
4058 | 0 | } |
4059 | | |
4060 | | // Convert AR30 to ARGB. |
4061 | | LIBYUV_API |
4062 | | int AR30ToARGB(const uint8_t* src_ar30, |
4063 | | int src_stride_ar30, |
4064 | | uint8_t* dst_argb, |
4065 | | int dst_stride_argb, |
4066 | | int width, |
4067 | 0 | int height) { |
4068 | 0 | int y; |
4069 | 0 | if (!src_ar30 || !dst_argb || width <= 0 || height == 0) { |
4070 | 0 | return -1; |
4071 | 0 | } |
4072 | | // Negative height means invert the image. |
4073 | 0 | if (height < 0) { |
4074 | 0 | height = -height; |
4075 | 0 | src_ar30 = src_ar30 + (height - 1) * src_stride_ar30; |
4076 | 0 | src_stride_ar30 = -src_stride_ar30; |
4077 | 0 | } |
4078 | | // Coalesce rows. |
4079 | 0 | if (src_stride_ar30 == width * 4 && dst_stride_argb == width * 4) { |
4080 | 0 | width *= height; |
4081 | 0 | height = 1; |
4082 | 0 | src_stride_ar30 = dst_stride_argb = 0; |
4083 | 0 | } |
4084 | 0 | for (y = 0; y < height; ++y) { |
4085 | 0 | AR30ToARGBRow_C(src_ar30, dst_argb, width); |
4086 | 0 | src_ar30 += src_stride_ar30; |
4087 | 0 | dst_argb += dst_stride_argb; |
4088 | 0 | } |
4089 | 0 | return 0; |
4090 | 0 | } |
4091 | | |
4092 | | // Convert AR30 to ABGR. |
4093 | | LIBYUV_API |
4094 | | int AR30ToABGR(const uint8_t* src_ar30, |
4095 | | int src_stride_ar30, |
4096 | | uint8_t* dst_abgr, |
4097 | | int dst_stride_abgr, |
4098 | | int width, |
4099 | 0 | int height) { |
4100 | 0 | int y; |
4101 | 0 | if (!src_ar30 || !dst_abgr || width <= 0 || height == 0) { |
4102 | 0 | return -1; |
4103 | 0 | } |
4104 | | // Negative height means invert the image. |
4105 | 0 | if (height < 0) { |
4106 | 0 | height = -height; |
4107 | 0 | src_ar30 = src_ar30 + (height - 1) * src_stride_ar30; |
4108 | 0 | src_stride_ar30 = -src_stride_ar30; |
4109 | 0 | } |
4110 | | // Coalesce rows. |
4111 | 0 | if (src_stride_ar30 == width * 4 && dst_stride_abgr == width * 4) { |
4112 | 0 | width *= height; |
4113 | 0 | height = 1; |
4114 | 0 | src_stride_ar30 = dst_stride_abgr = 0; |
4115 | 0 | } |
4116 | 0 | for (y = 0; y < height; ++y) { |
4117 | 0 | AR30ToABGRRow_C(src_ar30, dst_abgr, width); |
4118 | 0 | src_ar30 += src_stride_ar30; |
4119 | 0 | dst_abgr += dst_stride_abgr; |
4120 | 0 | } |
4121 | 0 | return 0; |
4122 | 0 | } |
4123 | | |
4124 | | // Convert AR30 to AB30. |
4125 | | LIBYUV_API |
4126 | | int AR30ToAB30(const uint8_t* src_ar30, |
4127 | | int src_stride_ar30, |
4128 | | uint8_t* dst_ab30, |
4129 | | int dst_stride_ab30, |
4130 | | int width, |
4131 | 0 | int height) { |
4132 | 0 | int y; |
4133 | 0 | if (!src_ar30 || !dst_ab30 || width <= 0 || height == 0) { |
4134 | 0 | return -1; |
4135 | 0 | } |
4136 | | // Negative height means invert the image. |
4137 | 0 | if (height < 0) { |
4138 | 0 | height = -height; |
4139 | 0 | src_ar30 = src_ar30 + (height - 1) * src_stride_ar30; |
4140 | 0 | src_stride_ar30 = -src_stride_ar30; |
4141 | 0 | } |
4142 | | // Coalesce rows. |
4143 | 0 | if (src_stride_ar30 == width * 4 && dst_stride_ab30 == width * 4) { |
4144 | 0 | width *= height; |
4145 | 0 | height = 1; |
4146 | 0 | src_stride_ar30 = dst_stride_ab30 = 0; |
4147 | 0 | } |
4148 | 0 | for (y = 0; y < height; ++y) { |
4149 | 0 | AR30ToAB30Row_C(src_ar30, dst_ab30, width); |
4150 | 0 | src_ar30 += src_stride_ar30; |
4151 | 0 | dst_ab30 += dst_stride_ab30; |
4152 | 0 | } |
4153 | 0 | return 0; |
4154 | 0 | } |
4155 | | |
4156 | | // Convert AR64 to ARGB. |
4157 | | LIBYUV_API |
4158 | | int AR64ToARGB(const uint16_t* src_ar64, |
4159 | | int src_stride_ar64, |
4160 | | uint8_t* dst_argb, |
4161 | | int dst_stride_argb, |
4162 | | int width, |
4163 | 0 | int height) { |
4164 | 0 | int y; |
4165 | 0 | void (*AR64ToARGBRow)(const uint16_t* src_ar64, uint8_t* dst_argb, |
4166 | 0 | int width) = AR64ToARGBRow_C; |
4167 | 0 | if (!src_ar64 || !dst_argb || width <= 0 || height == 0) { |
4168 | 0 | return -1; |
4169 | 0 | } |
4170 | | // Negative height means invert the image. |
4171 | 0 | if (height < 0) { |
4172 | 0 | height = -height; |
4173 | 0 | src_ar64 = src_ar64 + (height - 1) * src_stride_ar64; |
4174 | 0 | src_stride_ar64 = -src_stride_ar64; |
4175 | 0 | } |
4176 | | // Coalesce rows. |
4177 | 0 | if (src_stride_ar64 == width * 4 && dst_stride_argb == width * 4) { |
4178 | 0 | width *= height; |
4179 | 0 | height = 1; |
4180 | 0 | src_stride_ar64 = dst_stride_argb = 0; |
4181 | 0 | } |
4182 | 0 | #if defined(HAS_AR64TOARGBROW_SSSE3) |
4183 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4184 | 0 | AR64ToARGBRow = AR64ToARGBRow_Any_SSSE3; |
4185 | 0 | if (IS_ALIGNED(width, 4)) { |
4186 | 0 | AR64ToARGBRow = AR64ToARGBRow_SSSE3; |
4187 | 0 | } |
4188 | 0 | } |
4189 | 0 | #endif |
4190 | 0 | #if defined(HAS_AR64TOARGBROW_AVX2) |
4191 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4192 | 0 | AR64ToARGBRow = AR64ToARGBRow_Any_AVX2; |
4193 | 0 | if (IS_ALIGNED(width, 8)) { |
4194 | 0 | AR64ToARGBRow = AR64ToARGBRow_AVX2; |
4195 | 0 | } |
4196 | 0 | } |
4197 | 0 | #endif |
4198 | | #if defined(HAS_AR64TOARGBROW_NEON) |
4199 | | if (TestCpuFlag(kCpuHasNEON)) { |
4200 | | AR64ToARGBRow = AR64ToARGBRow_Any_NEON; |
4201 | | if (IS_ALIGNED(width, 8)) { |
4202 | | AR64ToARGBRow = AR64ToARGBRow_NEON; |
4203 | | } |
4204 | | } |
4205 | | #endif |
4206 | | #if defined(HAS_AR64TOARGBROW_RVV) |
4207 | | if (TestCpuFlag(kCpuHasRVV)) { |
4208 | | AR64ToARGBRow = AR64ToARGBRow_RVV; |
4209 | | } |
4210 | | #endif |
4211 | |
|
4212 | 0 | for (y = 0; y < height; ++y) { |
4213 | 0 | AR64ToARGBRow(src_ar64, dst_argb, width); |
4214 | 0 | src_ar64 += src_stride_ar64; |
4215 | 0 | dst_argb += dst_stride_argb; |
4216 | 0 | } |
4217 | 0 | return 0; |
4218 | 0 | } |
4219 | | |
4220 | | // Convert AB64 to ARGB. |
4221 | | LIBYUV_API |
4222 | | int AB64ToARGB(const uint16_t* src_ab64, |
4223 | | int src_stride_ab64, |
4224 | | uint8_t* dst_argb, |
4225 | | int dst_stride_argb, |
4226 | | int width, |
4227 | 0 | int height) { |
4228 | 0 | int y; |
4229 | 0 | void (*AB64ToARGBRow)(const uint16_t* src_ar64, uint8_t* dst_argb, |
4230 | 0 | int width) = AB64ToARGBRow_C; |
4231 | 0 | if (!src_ab64 || !dst_argb || width <= 0 || height == 0) { |
4232 | 0 | return -1; |
4233 | 0 | } |
4234 | | // Negative height means invert the image. |
4235 | 0 | if (height < 0) { |
4236 | 0 | height = -height; |
4237 | 0 | src_ab64 = src_ab64 + (height - 1) * src_stride_ab64; |
4238 | 0 | src_stride_ab64 = -src_stride_ab64; |
4239 | 0 | } |
4240 | | // Coalesce rows. |
4241 | 0 | if (src_stride_ab64 == width * 4 && dst_stride_argb == width * 4) { |
4242 | 0 | width *= height; |
4243 | 0 | height = 1; |
4244 | 0 | src_stride_ab64 = dst_stride_argb = 0; |
4245 | 0 | } |
4246 | 0 | #if defined(HAS_AB64TOARGBROW_SSSE3) |
4247 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4248 | 0 | AB64ToARGBRow = AB64ToARGBRow_Any_SSSE3; |
4249 | 0 | if (IS_ALIGNED(width, 4)) { |
4250 | 0 | AB64ToARGBRow = AB64ToARGBRow_SSSE3; |
4251 | 0 | } |
4252 | 0 | } |
4253 | 0 | #endif |
4254 | 0 | #if defined(HAS_AB64TOARGBROW_AVX2) |
4255 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4256 | 0 | AB64ToARGBRow = AB64ToARGBRow_Any_AVX2; |
4257 | 0 | if (IS_ALIGNED(width, 8)) { |
4258 | 0 | AB64ToARGBRow = AB64ToARGBRow_AVX2; |
4259 | 0 | } |
4260 | 0 | } |
4261 | 0 | #endif |
4262 | | #if defined(HAS_AB64TOARGBROW_NEON) |
4263 | | if (TestCpuFlag(kCpuHasNEON)) { |
4264 | | AB64ToARGBRow = AB64ToARGBRow_Any_NEON; |
4265 | | if (IS_ALIGNED(width, 8)) { |
4266 | | AB64ToARGBRow = AB64ToARGBRow_NEON; |
4267 | | } |
4268 | | } |
4269 | | #endif |
4270 | | #if defined(HAS_AB64TOARGBROW_RVV) |
4271 | | if (TestCpuFlag(kCpuHasRVV)) { |
4272 | | AB64ToARGBRow = AB64ToARGBRow_RVV; |
4273 | | } |
4274 | | #endif |
4275 | |
|
4276 | 0 | for (y = 0; y < height; ++y) { |
4277 | 0 | AB64ToARGBRow(src_ab64, dst_argb, width); |
4278 | 0 | src_ab64 += src_stride_ab64; |
4279 | 0 | dst_argb += dst_stride_argb; |
4280 | 0 | } |
4281 | 0 | return 0; |
4282 | 0 | } |
4283 | | |
4284 | | // Convert NV12 to ARGB with matrix. |
4285 | | LIBYUV_API |
4286 | | int NV12ToARGBMatrix(const uint8_t* src_y, |
4287 | | int src_stride_y, |
4288 | | const uint8_t* src_uv, |
4289 | | int src_stride_uv, |
4290 | | uint8_t* dst_argb, |
4291 | | int dst_stride_argb, |
4292 | | const struct YuvConstants* yuvconstants, |
4293 | | int width, |
4294 | 0 | int height) { |
4295 | 0 | int y; |
4296 | 0 | void (*NV12ToARGBRow)( |
4297 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4298 | 0 | const struct YuvConstants* yuvconstants, int width) = NV12ToARGBRow_C; |
4299 | 0 | assert(yuvconstants); |
4300 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
4301 | 0 | return -1; |
4302 | 0 | } |
4303 | | // Negative height means invert the image. |
4304 | 0 | if (height < 0) { |
4305 | 0 | height = -height; |
4306 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
4307 | 0 | dst_stride_argb = -dst_stride_argb; |
4308 | 0 | } |
4309 | 0 | #if defined(HAS_NV12TOARGBROW_SSSE3) |
4310 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4311 | 0 | NV12ToARGBRow = NV12ToARGBRow_Any_SSSE3; |
4312 | 0 | if (IS_ALIGNED(width, 8)) { |
4313 | 0 | NV12ToARGBRow = NV12ToARGBRow_SSSE3; |
4314 | 0 | } |
4315 | 0 | } |
4316 | 0 | #endif |
4317 | 0 | #if defined(HAS_NV12TOARGBROW_AVX2) |
4318 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4319 | 0 | NV12ToARGBRow = NV12ToARGBRow_Any_AVX2; |
4320 | 0 | if (IS_ALIGNED(width, 16)) { |
4321 | 0 | NV12ToARGBRow = NV12ToARGBRow_AVX2; |
4322 | 0 | } |
4323 | 0 | } |
4324 | 0 | #endif |
4325 | | #if defined(HAS_NV12TOARGBROW_NEON) |
4326 | | if (TestCpuFlag(kCpuHasNEON)) { |
4327 | | NV12ToARGBRow = NV12ToARGBRow_Any_NEON; |
4328 | | if (IS_ALIGNED(width, 8)) { |
4329 | | NV12ToARGBRow = NV12ToARGBRow_NEON; |
4330 | | } |
4331 | | } |
4332 | | #endif |
4333 | | #if defined(HAS_NV12TOARGBROW_SVE2) |
4334 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4335 | | NV12ToARGBRow = NV12ToARGBRow_SVE2; |
4336 | | } |
4337 | | #endif |
4338 | | #if defined(HAS_NV12TOARGBROW_SME) |
4339 | | if (TestCpuFlag(kCpuHasSME)) { |
4340 | | NV12ToARGBRow = NV12ToARGBRow_SME; |
4341 | | } |
4342 | | #endif |
4343 | | #if defined(HAS_NV12TOARGBROW_LSX) |
4344 | | if (TestCpuFlag(kCpuHasLSX)) { |
4345 | | NV12ToARGBRow = NV12ToARGBRow_Any_LSX; |
4346 | | if (IS_ALIGNED(width, 8)) { |
4347 | | NV12ToARGBRow = NV12ToARGBRow_LSX; |
4348 | | } |
4349 | | } |
4350 | | #endif |
4351 | | #if defined(HAS_NV12TOARGBROW_LASX) |
4352 | | if (TestCpuFlag(kCpuHasLASX)) { |
4353 | | NV12ToARGBRow = NV12ToARGBRow_Any_LASX; |
4354 | | if (IS_ALIGNED(width, 16)) { |
4355 | | NV12ToARGBRow = NV12ToARGBRow_LASX; |
4356 | | } |
4357 | | } |
4358 | | #endif |
4359 | | #if defined(HAS_NV12TOARGBROW_RVV) |
4360 | | if (TestCpuFlag(kCpuHasRVV)) { |
4361 | | NV12ToARGBRow = NV12ToARGBRow_RVV; |
4362 | | } |
4363 | | #endif |
4364 | |
|
4365 | 0 | for (y = 0; y < height; ++y) { |
4366 | 0 | NV12ToARGBRow(src_y, src_uv, dst_argb, yuvconstants, width); |
4367 | 0 | dst_argb += dst_stride_argb; |
4368 | 0 | src_y += src_stride_y; |
4369 | 0 | if (y & 1) { |
4370 | 0 | src_uv += src_stride_uv; |
4371 | 0 | } |
4372 | 0 | } |
4373 | 0 | return 0; |
4374 | 0 | } |
4375 | | |
4376 | | // Convert NV21 to ARGB with matrix. |
4377 | | LIBYUV_API |
4378 | | int NV21ToARGBMatrix(const uint8_t* src_y, |
4379 | | int src_stride_y, |
4380 | | const uint8_t* src_vu, |
4381 | | int src_stride_vu, |
4382 | | uint8_t* dst_argb, |
4383 | | int dst_stride_argb, |
4384 | | const struct YuvConstants* yuvconstants, |
4385 | | int width, |
4386 | 0 | int height) { |
4387 | 0 | int y; |
4388 | 0 | void (*NV21ToARGBRow)( |
4389 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4390 | 0 | const struct YuvConstants* yuvconstants, int width) = NV21ToARGBRow_C; |
4391 | 0 | assert(yuvconstants); |
4392 | 0 | if (!src_y || !src_vu || !dst_argb || width <= 0 || height == 0) { |
4393 | 0 | return -1; |
4394 | 0 | } |
4395 | | // Negative height means invert the image. |
4396 | 0 | if (height < 0) { |
4397 | 0 | height = -height; |
4398 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
4399 | 0 | dst_stride_argb = -dst_stride_argb; |
4400 | 0 | } |
4401 | 0 | #if defined(HAS_NV21TOARGBROW_SSSE3) |
4402 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4403 | 0 | NV21ToARGBRow = NV21ToARGBRow_Any_SSSE3; |
4404 | 0 | if (IS_ALIGNED(width, 8)) { |
4405 | 0 | NV21ToARGBRow = NV21ToARGBRow_SSSE3; |
4406 | 0 | } |
4407 | 0 | } |
4408 | 0 | #endif |
4409 | 0 | #if defined(HAS_NV21TOARGBROW_AVX2) |
4410 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4411 | 0 | NV21ToARGBRow = NV21ToARGBRow_Any_AVX2; |
4412 | 0 | if (IS_ALIGNED(width, 16)) { |
4413 | 0 | NV21ToARGBRow = NV21ToARGBRow_AVX2; |
4414 | 0 | } |
4415 | 0 | } |
4416 | 0 | #endif |
4417 | | #if defined(HAS_NV21TOARGBROW_NEON) |
4418 | | if (TestCpuFlag(kCpuHasNEON)) { |
4419 | | NV21ToARGBRow = NV21ToARGBRow_Any_NEON; |
4420 | | if (IS_ALIGNED(width, 8)) { |
4421 | | NV21ToARGBRow = NV21ToARGBRow_NEON; |
4422 | | } |
4423 | | } |
4424 | | #endif |
4425 | | #if defined(HAS_NV21TOARGBROW_SVE2) |
4426 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4427 | | NV21ToARGBRow = NV21ToARGBRow_SVE2; |
4428 | | } |
4429 | | #endif |
4430 | | #if defined(HAS_NV21TOARGBROW_SME) |
4431 | | if (TestCpuFlag(kCpuHasSME)) { |
4432 | | NV21ToARGBRow = NV21ToARGBRow_SME; |
4433 | | } |
4434 | | #endif |
4435 | | #if defined(HAS_NV21TOARGBROW_LSX) |
4436 | | if (TestCpuFlag(kCpuHasLSX)) { |
4437 | | NV21ToARGBRow = NV21ToARGBRow_Any_LSX; |
4438 | | if (IS_ALIGNED(width, 8)) { |
4439 | | NV21ToARGBRow = NV21ToARGBRow_LSX; |
4440 | | } |
4441 | | } |
4442 | | #endif |
4443 | | #if defined(HAS_NV21TOARGBROW_LASX) |
4444 | | if (TestCpuFlag(kCpuHasLASX)) { |
4445 | | NV21ToARGBRow = NV21ToARGBRow_Any_LASX; |
4446 | | if (IS_ALIGNED(width, 16)) { |
4447 | | NV21ToARGBRow = NV21ToARGBRow_LASX; |
4448 | | } |
4449 | | } |
4450 | | #endif |
4451 | | #if defined(HAS_NV21TOARGBROW_RVV) |
4452 | | if (TestCpuFlag(kCpuHasRVV)) { |
4453 | | NV21ToARGBRow = NV21ToARGBRow_RVV; |
4454 | | } |
4455 | | #endif |
4456 | |
|
4457 | 0 | for (y = 0; y < height; ++y) { |
4458 | 0 | NV21ToARGBRow(src_y, src_vu, dst_argb, yuvconstants, width); |
4459 | 0 | dst_argb += dst_stride_argb; |
4460 | 0 | src_y += src_stride_y; |
4461 | 0 | if (y & 1) { |
4462 | 0 | src_vu += src_stride_vu; |
4463 | 0 | } |
4464 | 0 | } |
4465 | 0 | return 0; |
4466 | 0 | } |
4467 | | |
4468 | | // Convert NV12 to ARGB. |
4469 | | LIBYUV_API |
4470 | | int NV12ToARGB(const uint8_t* src_y, |
4471 | | int src_stride_y, |
4472 | | const uint8_t* src_uv, |
4473 | | int src_stride_uv, |
4474 | | uint8_t* dst_argb, |
4475 | | int dst_stride_argb, |
4476 | | int width, |
4477 | 0 | int height) { |
4478 | 0 | return NV12ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, dst_argb, |
4479 | 0 | dst_stride_argb, &kYuvI601Constants, width, height); |
4480 | 0 | } |
4481 | | |
4482 | | // Convert NV21 to ARGB. |
4483 | | LIBYUV_API |
4484 | | int NV21ToARGB(const uint8_t* src_y, |
4485 | | int src_stride_y, |
4486 | | const uint8_t* src_vu, |
4487 | | int src_stride_vu, |
4488 | | uint8_t* dst_argb, |
4489 | | int dst_stride_argb, |
4490 | | int width, |
4491 | 0 | int height) { |
4492 | 0 | return NV21ToARGBMatrix(src_y, src_stride_y, src_vu, src_stride_vu, dst_argb, |
4493 | 0 | dst_stride_argb, &kYuvI601Constants, width, height); |
4494 | 0 | } |
4495 | | |
4496 | | // Convert NV12 to ABGR. |
4497 | | // To output ABGR instead of ARGB swap the UV and use a mirrored yuv matrix. |
4498 | | // To swap the UV use NV12 instead of NV21.LIBYUV_API |
4499 | | LIBYUV_API |
4500 | | int NV12ToABGR(const uint8_t* src_y, |
4501 | | int src_stride_y, |
4502 | | const uint8_t* src_uv, |
4503 | | int src_stride_uv, |
4504 | | uint8_t* dst_abgr, |
4505 | | int dst_stride_abgr, |
4506 | | int width, |
4507 | 0 | int height) { |
4508 | 0 | return NV21ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, dst_abgr, |
4509 | 0 | dst_stride_abgr, &kYvuI601Constants, width, height); |
4510 | 0 | } |
4511 | | |
4512 | | // Convert NV21 to ABGR. |
4513 | | LIBYUV_API |
4514 | | int NV21ToABGR(const uint8_t* src_y, |
4515 | | int src_stride_y, |
4516 | | const uint8_t* src_vu, |
4517 | | int src_stride_vu, |
4518 | | uint8_t* dst_abgr, |
4519 | | int dst_stride_abgr, |
4520 | | int width, |
4521 | 0 | int height) { |
4522 | 0 | return NV12ToARGBMatrix(src_y, src_stride_y, src_vu, src_stride_vu, dst_abgr, |
4523 | 0 | dst_stride_abgr, &kYvuI601Constants, width, height); |
4524 | 0 | } |
4525 | | |
4526 | | // TODO(fbarchard): Consider SSSE3 2 step conversion. |
4527 | | // Convert NV12 to RGB24 with matrix. |
4528 | | LIBYUV_API |
4529 | | int NV12ToRGB24Matrix(const uint8_t* src_y, |
4530 | | int src_stride_y, |
4531 | | const uint8_t* src_uv, |
4532 | | int src_stride_uv, |
4533 | | uint8_t* dst_rgb24, |
4534 | | int dst_stride_rgb24, |
4535 | | const struct YuvConstants* yuvconstants, |
4536 | | int width, |
4537 | 0 | int height) { |
4538 | 0 | int y; |
4539 | 0 | void (*NV12ToRGB24Row)( |
4540 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4541 | 0 | const struct YuvConstants* yuvconstants, int width) = NV12ToRGB24Row_C; |
4542 | 0 | assert(yuvconstants); |
4543 | 0 | if (!src_y || !src_uv || !dst_rgb24 || width <= 0 || height == 0) { |
4544 | 0 | return -1; |
4545 | 0 | } |
4546 | | // Negative height means invert the image. |
4547 | 0 | if (height < 0) { |
4548 | 0 | height = -height; |
4549 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
4550 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
4551 | 0 | } |
4552 | | #if defined(HAS_NV12TORGB24ROW_NEON) |
4553 | | if (TestCpuFlag(kCpuHasNEON)) { |
4554 | | NV12ToRGB24Row = NV12ToRGB24Row_Any_NEON; |
4555 | | if (IS_ALIGNED(width, 8)) { |
4556 | | NV12ToRGB24Row = NV12ToRGB24Row_NEON; |
4557 | | } |
4558 | | } |
4559 | | #endif |
4560 | | #if defined(HAS_NV12TORGB24ROW_SVE2) |
4561 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4562 | | NV12ToRGB24Row = NV12ToRGB24Row_SVE2; |
4563 | | } |
4564 | | #endif |
4565 | | #if defined(HAS_NV12TORGB24ROW_SME) |
4566 | | if (TestCpuFlag(kCpuHasSME)) { |
4567 | | NV12ToRGB24Row = NV12ToRGB24Row_SME; |
4568 | | } |
4569 | | #endif |
4570 | 0 | #if defined(HAS_NV12TORGB24ROW_SSSE3) |
4571 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4572 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_Any_SSSE3; |
4573 | 0 | if (IS_ALIGNED(width, 16)) { |
4574 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_SSSE3; |
4575 | 0 | } |
4576 | 0 | } |
4577 | 0 | #endif |
4578 | 0 | #if defined(HAS_NV12TORGB24ROW_AVX2) |
4579 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4580 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_Any_AVX2; |
4581 | 0 | if (IS_ALIGNED(width, 32)) { |
4582 | 0 | NV12ToRGB24Row = NV12ToRGB24Row_AVX2; |
4583 | 0 | } |
4584 | 0 | } |
4585 | 0 | #endif |
4586 | | #if defined(HAS_NV12TORGB24ROW_RVV) |
4587 | | if (TestCpuFlag(kCpuHasRVV)) { |
4588 | | NV12ToRGB24Row = NV12ToRGB24Row_RVV; |
4589 | | } |
4590 | | #endif |
4591 | |
|
4592 | 0 | for (y = 0; y < height; ++y) { |
4593 | 0 | NV12ToRGB24Row(src_y, src_uv, dst_rgb24, yuvconstants, width); |
4594 | 0 | dst_rgb24 += dst_stride_rgb24; |
4595 | 0 | src_y += src_stride_y; |
4596 | 0 | if (y & 1) { |
4597 | 0 | src_uv += src_stride_uv; |
4598 | 0 | } |
4599 | 0 | } |
4600 | 0 | return 0; |
4601 | 0 | } |
4602 | | |
4603 | | // Convert NV21 to RGB24 with matrix. |
4604 | | LIBYUV_API |
4605 | | int NV21ToRGB24Matrix(const uint8_t* src_y, |
4606 | | int src_stride_y, |
4607 | | const uint8_t* src_vu, |
4608 | | int src_stride_vu, |
4609 | | uint8_t* dst_rgb24, |
4610 | | int dst_stride_rgb24, |
4611 | | const struct YuvConstants* yuvconstants, |
4612 | | int width, |
4613 | 0 | int height) { |
4614 | 0 | int y; |
4615 | 0 | void (*NV21ToRGB24Row)( |
4616 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
4617 | 0 | const struct YuvConstants* yuvconstants, int width) = NV21ToRGB24Row_C; |
4618 | 0 | assert(yuvconstants); |
4619 | 0 | if (!src_y || !src_vu || !dst_rgb24 || width <= 0 || height == 0) { |
4620 | 0 | return -1; |
4621 | 0 | } |
4622 | | // Negative height means invert the image. |
4623 | 0 | if (height < 0) { |
4624 | 0 | height = -height; |
4625 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
4626 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
4627 | 0 | } |
4628 | | #if defined(HAS_NV21TORGB24ROW_NEON) |
4629 | | if (TestCpuFlag(kCpuHasNEON)) { |
4630 | | NV21ToRGB24Row = NV21ToRGB24Row_Any_NEON; |
4631 | | if (IS_ALIGNED(width, 8)) { |
4632 | | NV21ToRGB24Row = NV21ToRGB24Row_NEON; |
4633 | | } |
4634 | | } |
4635 | | #endif |
4636 | | #if defined(HAS_NV21TORGB24ROW_SVE2) |
4637 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4638 | | NV21ToRGB24Row = NV21ToRGB24Row_SVE2; |
4639 | | } |
4640 | | #endif |
4641 | | #if defined(HAS_NV21TORGB24ROW_SME) |
4642 | | if (TestCpuFlag(kCpuHasSME)) { |
4643 | | NV21ToRGB24Row = NV21ToRGB24Row_SME; |
4644 | | } |
4645 | | #endif |
4646 | 0 | #if defined(HAS_NV21TORGB24ROW_SSSE3) |
4647 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4648 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_Any_SSSE3; |
4649 | 0 | if (IS_ALIGNED(width, 16)) { |
4650 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_SSSE3; |
4651 | 0 | } |
4652 | 0 | } |
4653 | 0 | #endif |
4654 | 0 | #if defined(HAS_NV21TORGB24ROW_AVX2) |
4655 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4656 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_Any_AVX2; |
4657 | 0 | if (IS_ALIGNED(width, 32)) { |
4658 | 0 | NV21ToRGB24Row = NV21ToRGB24Row_AVX2; |
4659 | 0 | } |
4660 | 0 | } |
4661 | 0 | #endif |
4662 | | #if defined(HAS_NV21TORGB24ROW_RVV) |
4663 | | if (TestCpuFlag(kCpuHasRVV)) { |
4664 | | NV21ToRGB24Row = NV21ToRGB24Row_RVV; |
4665 | | } |
4666 | | #endif |
4667 | |
|
4668 | 0 | for (y = 0; y < height; ++y) { |
4669 | 0 | NV21ToRGB24Row(src_y, src_vu, dst_rgb24, yuvconstants, width); |
4670 | 0 | dst_rgb24 += dst_stride_rgb24; |
4671 | 0 | src_y += src_stride_y; |
4672 | 0 | if (y & 1) { |
4673 | 0 | src_vu += src_stride_vu; |
4674 | 0 | } |
4675 | 0 | } |
4676 | 0 | return 0; |
4677 | 0 | } |
4678 | | |
4679 | | // Convert NV12 to RGB24. |
4680 | | LIBYUV_API |
4681 | | int NV12ToRGB24(const uint8_t* src_y, |
4682 | | int src_stride_y, |
4683 | | const uint8_t* src_uv, |
4684 | | int src_stride_uv, |
4685 | | uint8_t* dst_rgb24, |
4686 | | int dst_stride_rgb24, |
4687 | | int width, |
4688 | 0 | int height) { |
4689 | 0 | return NV12ToRGB24Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
4690 | 0 | dst_rgb24, dst_stride_rgb24, &kYuvI601Constants, |
4691 | 0 | width, height); |
4692 | 0 | } |
4693 | | |
4694 | | // Convert NV21 to RGB24. |
4695 | | LIBYUV_API |
4696 | | int NV21ToRGB24(const uint8_t* src_y, |
4697 | | int src_stride_y, |
4698 | | const uint8_t* src_vu, |
4699 | | int src_stride_vu, |
4700 | | uint8_t* dst_rgb24, |
4701 | | int dst_stride_rgb24, |
4702 | | int width, |
4703 | 0 | int height) { |
4704 | 0 | return NV21ToRGB24Matrix(src_y, src_stride_y, src_vu, src_stride_vu, |
4705 | 0 | dst_rgb24, dst_stride_rgb24, &kYuvI601Constants, |
4706 | 0 | width, height); |
4707 | 0 | } |
4708 | | |
4709 | | // Convert NV12 to RAW. |
4710 | | LIBYUV_API |
4711 | | int NV12ToRAW(const uint8_t* src_y, |
4712 | | int src_stride_y, |
4713 | | const uint8_t* src_uv, |
4714 | | int src_stride_uv, |
4715 | | uint8_t* dst_raw, |
4716 | | int dst_stride_raw, |
4717 | | int width, |
4718 | 0 | int height) { |
4719 | 0 | return NV21ToRGB24Matrix(src_y, src_stride_y, src_uv, src_stride_uv, dst_raw, |
4720 | 0 | dst_stride_raw, &kYvuI601Constants, width, height); |
4721 | 0 | } |
4722 | | |
4723 | | // Convert NV21 to RAW. |
4724 | | LIBYUV_API |
4725 | | int NV21ToRAW(const uint8_t* src_y, |
4726 | | int src_stride_y, |
4727 | | const uint8_t* src_vu, |
4728 | | int src_stride_vu, |
4729 | | uint8_t* dst_raw, |
4730 | | int dst_stride_raw, |
4731 | | int width, |
4732 | 0 | int height) { |
4733 | 0 | return NV12ToRGB24Matrix(src_y, src_stride_y, src_vu, src_stride_vu, dst_raw, |
4734 | 0 | dst_stride_raw, &kYvuI601Constants, width, height); |
4735 | 0 | } |
4736 | | |
4737 | | // Convert NV21 to YUV24 |
4738 | | int NV21ToYUV24(const uint8_t* src_y, |
4739 | | int src_stride_y, |
4740 | | const uint8_t* src_vu, |
4741 | | int src_stride_vu, |
4742 | | uint8_t* dst_yuv24, |
4743 | | int dst_stride_yuv24, |
4744 | | int width, |
4745 | 0 | int height) { |
4746 | 0 | int y; |
4747 | 0 | void (*NV21ToYUV24Row)(const uint8_t* src_y, const uint8_t* src_vu, |
4748 | 0 | uint8_t* dst_yuv24, int width) = NV21ToYUV24Row_C; |
4749 | 0 | if (!src_y || !src_vu || !dst_yuv24 || width <= 0 || height == 0) { |
4750 | 0 | return -1; |
4751 | 0 | } |
4752 | | // Negative height means invert the image. |
4753 | 0 | if (height < 0) { |
4754 | 0 | height = -height; |
4755 | 0 | dst_yuv24 = dst_yuv24 + (height - 1) * dst_stride_yuv24; |
4756 | 0 | dst_stride_yuv24 = -dst_stride_yuv24; |
4757 | 0 | } |
4758 | | #if defined(HAS_NV21TOYUV24ROW_NEON) |
4759 | | if (TestCpuFlag(kCpuHasNEON)) { |
4760 | | NV21ToYUV24Row = NV21ToYUV24Row_Any_NEON; |
4761 | | if (IS_ALIGNED(width, 16)) { |
4762 | | NV21ToYUV24Row = NV21ToYUV24Row_NEON; |
4763 | | } |
4764 | | } |
4765 | | #endif |
4766 | 0 | #if defined(HAS_NV21TOYUV24ROW_SSSE3) |
4767 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4768 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_Any_SSSE3; |
4769 | 0 | if (IS_ALIGNED(width, 16)) { |
4770 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_SSSE3; |
4771 | 0 | } |
4772 | 0 | } |
4773 | 0 | #endif |
4774 | 0 | #if defined(HAS_NV21TOYUV24ROW_AVX2) |
4775 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4776 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_Any_AVX2; |
4777 | 0 | if (IS_ALIGNED(width, 32)) { |
4778 | 0 | NV21ToYUV24Row = NV21ToYUV24Row_AVX2; |
4779 | 0 | } |
4780 | 0 | } |
4781 | 0 | #endif |
4782 | 0 | for (y = 0; y < height; ++y) { |
4783 | 0 | NV21ToYUV24Row(src_y, src_vu, dst_yuv24, width); |
4784 | 0 | dst_yuv24 += dst_stride_yuv24; |
4785 | 0 | src_y += src_stride_y; |
4786 | 0 | if (y & 1) { |
4787 | 0 | src_vu += src_stride_vu; |
4788 | 0 | } |
4789 | 0 | } |
4790 | 0 | return 0; |
4791 | 0 | } |
4792 | | |
4793 | | // Convert YUY2 to ARGB with matrix. |
4794 | | LIBYUV_API |
4795 | | int YUY2ToARGBMatrix(const uint8_t* src_yuy2, |
4796 | | int src_stride_yuy2, |
4797 | | uint8_t* dst_argb, |
4798 | | int dst_stride_argb, |
4799 | | const struct YuvConstants* yuvconstants, |
4800 | | int width, |
4801 | 0 | int height) { |
4802 | 0 | int y; |
4803 | 0 | void (*YUY2ToARGBRow)(const uint8_t* src_yuy2, uint8_t* dst_argb, |
4804 | 0 | const struct YuvConstants* yuvconstants, int width) = |
4805 | 0 | YUY2ToARGBRow_C; |
4806 | 0 | if (!src_yuy2 || !dst_argb || width <= 0 || height == 0) { |
4807 | 0 | return -1; |
4808 | 0 | } |
4809 | | // Negative height means invert the image. |
4810 | 0 | if (height < 0) { |
4811 | 0 | height = -height; |
4812 | 0 | src_yuy2 = src_yuy2 + (height - 1) * src_stride_yuy2; |
4813 | 0 | src_stride_yuy2 = -src_stride_yuy2; |
4814 | 0 | } |
4815 | | // Coalesce rows. |
4816 | 0 | if (src_stride_yuy2 == width * 2 && dst_stride_argb == width * 4) { |
4817 | 0 | width *= height; |
4818 | 0 | height = 1; |
4819 | 0 | src_stride_yuy2 = dst_stride_argb = 0; |
4820 | 0 | } |
4821 | 0 | #if defined(HAS_YUY2TOARGBROW_SSSE3) |
4822 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4823 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_Any_SSSE3; |
4824 | 0 | if (IS_ALIGNED(width, 16)) { |
4825 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_SSSE3; |
4826 | 0 | } |
4827 | 0 | } |
4828 | 0 | #endif |
4829 | 0 | #if defined(HAS_YUY2TOARGBROW_AVX2) |
4830 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4831 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_Any_AVX2; |
4832 | 0 | if (IS_ALIGNED(width, 32)) { |
4833 | 0 | YUY2ToARGBRow = YUY2ToARGBRow_AVX2; |
4834 | 0 | } |
4835 | 0 | } |
4836 | 0 | #endif |
4837 | | #if defined(HAS_YUY2TOARGBROW_NEON) |
4838 | | if (TestCpuFlag(kCpuHasNEON)) { |
4839 | | YUY2ToARGBRow = YUY2ToARGBRow_Any_NEON; |
4840 | | if (IS_ALIGNED(width, 8)) { |
4841 | | YUY2ToARGBRow = YUY2ToARGBRow_NEON; |
4842 | | } |
4843 | | } |
4844 | | #endif |
4845 | | #if defined(HAS_YUY2TOARGBROW_SVE2) |
4846 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4847 | | YUY2ToARGBRow = YUY2ToARGBRow_SVE2; |
4848 | | } |
4849 | | #endif |
4850 | | #if defined(HAS_YUY2TOARGBROW_SME) |
4851 | | if (TestCpuFlag(kCpuHasSME)) { |
4852 | | YUY2ToARGBRow = YUY2ToARGBRow_SME; |
4853 | | } |
4854 | | #endif |
4855 | | #if defined(HAS_YUY2TOARGBROW_LSX) |
4856 | | if (TestCpuFlag(kCpuHasLSX)) { |
4857 | | YUY2ToARGBRow = YUY2ToARGBRow_Any_LSX; |
4858 | | if (IS_ALIGNED(width, 8)) { |
4859 | | YUY2ToARGBRow = YUY2ToARGBRow_LSX; |
4860 | | } |
4861 | | } |
4862 | | #endif |
4863 | 0 | for (y = 0; y < height; ++y) { |
4864 | 0 | YUY2ToARGBRow(src_yuy2, dst_argb, yuvconstants, width); |
4865 | 0 | src_yuy2 += src_stride_yuy2; |
4866 | 0 | dst_argb += dst_stride_argb; |
4867 | 0 | } |
4868 | 0 | return 0; |
4869 | 0 | } |
4870 | | |
4871 | | // Convert YUY2 to ARGB. |
4872 | | LIBYUV_API |
4873 | | int YUY2ToARGB(const uint8_t* src_yuy2, |
4874 | | int src_stride_yuy2, |
4875 | | uint8_t* dst_argb, |
4876 | | int dst_stride_argb, |
4877 | | int width, |
4878 | 0 | int height) { |
4879 | 0 | return YUY2ToARGBMatrix(src_yuy2, src_stride_yuy2, dst_argb, dst_stride_argb, |
4880 | 0 | &kYuvI601Constants, width, height); |
4881 | 0 | } |
4882 | | |
4883 | | // Convert UYVY to ARGB with matrix. |
4884 | | LIBYUV_API |
4885 | | int UYVYToARGBMatrix(const uint8_t* src_uyvy, |
4886 | | int src_stride_uyvy, |
4887 | | uint8_t* dst_argb, |
4888 | | int dst_stride_argb, |
4889 | | const struct YuvConstants* yuvconstants, |
4890 | | int width, |
4891 | 0 | int height) { |
4892 | 0 | int y; |
4893 | 0 | void (*UYVYToARGBRow)(const uint8_t* src_uyvy, uint8_t* dst_argb, |
4894 | 0 | const struct YuvConstants* yuvconstants, int width) = |
4895 | 0 | UYVYToARGBRow_C; |
4896 | 0 | if (!src_uyvy || !dst_argb || width <= 0 || height == 0) { |
4897 | 0 | return -1; |
4898 | 0 | } |
4899 | | // Negative height means invert the image. |
4900 | 0 | if (height < 0) { |
4901 | 0 | height = -height; |
4902 | 0 | src_uyvy = src_uyvy + (height - 1) * src_stride_uyvy; |
4903 | 0 | src_stride_uyvy = -src_stride_uyvy; |
4904 | 0 | } |
4905 | | // Coalesce rows. |
4906 | 0 | if (src_stride_uyvy == width * 2 && dst_stride_argb == width * 4) { |
4907 | 0 | width *= height; |
4908 | 0 | height = 1; |
4909 | 0 | src_stride_uyvy = dst_stride_argb = 0; |
4910 | 0 | } |
4911 | 0 | #if defined(HAS_UYVYTOARGBROW_SSSE3) |
4912 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
4913 | 0 | UYVYToARGBRow = UYVYToARGBRow_Any_SSSE3; |
4914 | 0 | if (IS_ALIGNED(width, 16)) { |
4915 | 0 | UYVYToARGBRow = UYVYToARGBRow_SSSE3; |
4916 | 0 | } |
4917 | 0 | } |
4918 | 0 | #endif |
4919 | 0 | #if defined(HAS_UYVYTOARGBROW_AVX2) |
4920 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
4921 | 0 | UYVYToARGBRow = UYVYToARGBRow_Any_AVX2; |
4922 | 0 | if (IS_ALIGNED(width, 32)) { |
4923 | 0 | UYVYToARGBRow = UYVYToARGBRow_AVX2; |
4924 | 0 | } |
4925 | 0 | } |
4926 | 0 | #endif |
4927 | | #if defined(HAS_UYVYTOARGBROW_NEON) |
4928 | | if (TestCpuFlag(kCpuHasNEON)) { |
4929 | | UYVYToARGBRow = UYVYToARGBRow_Any_NEON; |
4930 | | if (IS_ALIGNED(width, 8)) { |
4931 | | UYVYToARGBRow = UYVYToARGBRow_NEON; |
4932 | | } |
4933 | | } |
4934 | | #endif |
4935 | | #if defined(HAS_UYVYTOARGBROW_SVE2) |
4936 | | if (TestCpuFlag(kCpuHasSVE2)) { |
4937 | | UYVYToARGBRow = UYVYToARGBRow_SVE2; |
4938 | | } |
4939 | | #endif |
4940 | | #if defined(HAS_UYVYTOARGBROW_SME) |
4941 | | if (TestCpuFlag(kCpuHasSME)) { |
4942 | | UYVYToARGBRow = UYVYToARGBRow_SME; |
4943 | | } |
4944 | | #endif |
4945 | | #if defined(HAS_UYVYTOARGBROW_LSX) |
4946 | | if (TestCpuFlag(kCpuHasLSX)) { |
4947 | | UYVYToARGBRow = UYVYToARGBRow_Any_LSX; |
4948 | | if (IS_ALIGNED(width, 8)) { |
4949 | | UYVYToARGBRow = UYVYToARGBRow_LSX; |
4950 | | } |
4951 | | } |
4952 | | #endif |
4953 | 0 | for (y = 0; y < height; ++y) { |
4954 | 0 | UYVYToARGBRow(src_uyvy, dst_argb, yuvconstants, width); |
4955 | 0 | src_uyvy += src_stride_uyvy; |
4956 | 0 | dst_argb += dst_stride_argb; |
4957 | 0 | } |
4958 | 0 | return 0; |
4959 | 0 | } |
4960 | | |
4961 | | // Convert UYVY to ARGB. |
4962 | | LIBYUV_API |
4963 | | int UYVYToARGB(const uint8_t* src_uyvy, |
4964 | | int src_stride_uyvy, |
4965 | | uint8_t* dst_argb, |
4966 | | int dst_stride_argb, |
4967 | | int width, |
4968 | 0 | int height) { |
4969 | 0 | return UYVYToARGBMatrix(src_uyvy, src_stride_uyvy, dst_argb, dst_stride_argb, |
4970 | 0 | &kYuvI601Constants, width, height); |
4971 | 0 | } |
4972 | | |
4973 | | static void WeavePixels(const uint8_t* src_u, |
4974 | | const uint8_t* src_v, |
4975 | | int src_pixel_stride_uv, |
4976 | | uint8_t* dst_uv, |
4977 | 0 | int width) { |
4978 | 0 | int i; |
4979 | 0 | for (i = 0; i < width; ++i) { |
4980 | 0 | dst_uv[0] = *src_u; |
4981 | 0 | dst_uv[1] = *src_v; |
4982 | 0 | dst_uv += 2; |
4983 | 0 | src_u += src_pixel_stride_uv; |
4984 | 0 | src_v += src_pixel_stride_uv; |
4985 | 0 | } |
4986 | 0 | } |
4987 | | |
4988 | | // Convert Android420 to ARGB with matrix. |
4989 | | LIBYUV_API |
4990 | | int Android420ToARGBMatrix(const uint8_t* src_y, |
4991 | | int src_stride_y, |
4992 | | const uint8_t* src_u, |
4993 | | int src_stride_u, |
4994 | | const uint8_t* src_v, |
4995 | | int src_stride_v, |
4996 | | int src_pixel_stride_uv, |
4997 | | uint8_t* dst_argb, |
4998 | | int dst_stride_argb, |
4999 | | const struct YuvConstants* yuvconstants, |
5000 | | int width, |
5001 | 0 | int height) { |
5002 | 0 | int y; |
5003 | 0 | uint8_t* dst_uv; |
5004 | 0 | const ptrdiff_t vu_off = src_v - src_u; |
5005 | 0 | int halfwidth = (width + 1) >> 1; |
5006 | 0 | int halfheight = (height + 1) >> 1; |
5007 | 0 | assert(yuvconstants); |
5008 | 0 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
5009 | 0 | return -1; |
5010 | 0 | } |
5011 | | // Negative height means invert the image. |
5012 | 0 | if (height < 0) { |
5013 | 0 | height = -height; |
5014 | 0 | halfheight = (height + 1) >> 1; |
5015 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
5016 | 0 | dst_stride_argb = -dst_stride_argb; |
5017 | 0 | } |
5018 | | |
5019 | | // I420 |
5020 | 0 | if (src_pixel_stride_uv == 1) { |
5021 | 0 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5022 | 0 | src_stride_v, dst_argb, dst_stride_argb, |
5023 | 0 | yuvconstants, width, height); |
5024 | | // NV21 |
5025 | 0 | } |
5026 | 0 | if (src_pixel_stride_uv == 2 && vu_off == -1 && |
5027 | 0 | src_stride_u == src_stride_v) { |
5028 | 0 | return NV21ToARGBMatrix(src_y, src_stride_y, src_v, src_stride_v, dst_argb, |
5029 | 0 | dst_stride_argb, yuvconstants, width, height); |
5030 | | // NV12 |
5031 | 0 | } |
5032 | 0 | if (src_pixel_stride_uv == 2 && vu_off == 1 && src_stride_u == src_stride_v) { |
5033 | 0 | return NV12ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, dst_argb, |
5034 | 0 | dst_stride_argb, yuvconstants, width, height); |
5035 | 0 | } |
5036 | | |
5037 | | // General case fallback creates NV12 |
5038 | 0 | align_buffer_64(plane_uv, halfwidth * 2 * halfheight); |
5039 | 0 | if (!plane_uv) |
5040 | 0 | return 1; |
5041 | 0 | dst_uv = plane_uv; |
5042 | 0 | for (y = 0; y < halfheight; ++y) { |
5043 | 0 | WeavePixels(src_u, src_v, src_pixel_stride_uv, dst_uv, halfwidth); |
5044 | 0 | src_u += src_stride_u; |
5045 | 0 | src_v += src_stride_v; |
5046 | 0 | dst_uv += halfwidth * 2; |
5047 | 0 | } |
5048 | 0 | NV12ToARGBMatrix(src_y, src_stride_y, plane_uv, halfwidth * 2, dst_argb, |
5049 | 0 | dst_stride_argb, yuvconstants, width, height); |
5050 | 0 | free_aligned_buffer_64(plane_uv); |
5051 | 0 | return 0; |
5052 | 0 | } |
5053 | | |
5054 | | // Convert Android420 to ARGB. |
5055 | | LIBYUV_API |
5056 | | int Android420ToARGB(const uint8_t* src_y, |
5057 | | int src_stride_y, |
5058 | | const uint8_t* src_u, |
5059 | | int src_stride_u, |
5060 | | const uint8_t* src_v, |
5061 | | int src_stride_v, |
5062 | | int src_pixel_stride_uv, |
5063 | | uint8_t* dst_argb, |
5064 | | int dst_stride_argb, |
5065 | | int width, |
5066 | 0 | int height) { |
5067 | 0 | return Android420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5068 | 0 | src_stride_v, src_pixel_stride_uv, dst_argb, |
5069 | 0 | dst_stride_argb, &kYuvI601Constants, width, |
5070 | 0 | height); |
5071 | 0 | } |
5072 | | |
5073 | | // Convert Android420 to ABGR. |
5074 | | LIBYUV_API |
5075 | | int Android420ToABGR(const uint8_t* src_y, |
5076 | | int src_stride_y, |
5077 | | const uint8_t* src_u, |
5078 | | int src_stride_u, |
5079 | | const uint8_t* src_v, |
5080 | | int src_stride_v, |
5081 | | int src_pixel_stride_uv, |
5082 | | uint8_t* dst_abgr, |
5083 | | int dst_stride_abgr, |
5084 | | int width, |
5085 | 0 | int height) { |
5086 | 0 | return Android420ToARGBMatrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
5087 | 0 | src_stride_u, src_pixel_stride_uv, dst_abgr, |
5088 | 0 | dst_stride_abgr, &kYvuI601Constants, width, |
5089 | 0 | height); |
5090 | 0 | } |
5091 | | |
5092 | | // Convert I422 to RGBA with matrix. |
5093 | | LIBYUV_API |
5094 | | int I422ToRGBAMatrix(const uint8_t* src_y, |
5095 | | int src_stride_y, |
5096 | | const uint8_t* src_u, |
5097 | | int src_stride_u, |
5098 | | const uint8_t* src_v, |
5099 | | int src_stride_v, |
5100 | | uint8_t* dst_rgba, |
5101 | | int dst_stride_rgba, |
5102 | | const struct YuvConstants* yuvconstants, |
5103 | | int width, |
5104 | 89 | int height) { |
5105 | 89 | int y; |
5106 | 89 | void (*I422ToRGBARow)(const uint8_t* y_buf, const uint8_t* u_buf, |
5107 | 89 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5108 | 89 | const struct YuvConstants* yuvconstants, int width) = |
5109 | 89 | I422ToRGBARow_C; |
5110 | 89 | assert(yuvconstants); |
5111 | 89 | if (!src_y || !src_u || !src_v || !dst_rgba || width <= 0 || height == 0) { |
5112 | 0 | return -1; |
5113 | 0 | } |
5114 | | // Negative height means invert the image. |
5115 | 89 | if (height < 0) { |
5116 | 0 | height = -height; |
5117 | 0 | dst_rgba = dst_rgba + (height - 1) * dst_stride_rgba; |
5118 | 0 | dst_stride_rgba = -dst_stride_rgba; |
5119 | 0 | } |
5120 | 89 | #if defined(HAS_I422TORGBAROW_SSSE3) |
5121 | 89 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5122 | 89 | I422ToRGBARow = I422ToRGBARow_Any_SSSE3; |
5123 | 89 | if (IS_ALIGNED(width, 8)) { |
5124 | 34 | I422ToRGBARow = I422ToRGBARow_SSSE3; |
5125 | 34 | } |
5126 | 89 | } |
5127 | 89 | #endif |
5128 | 89 | #if defined(HAS_I422TORGBAROW_AVX2) |
5129 | 89 | if (TestCpuFlag(kCpuHasAVX2)) { |
5130 | 89 | I422ToRGBARow = I422ToRGBARow_Any_AVX2; |
5131 | 89 | if (IS_ALIGNED(width, 16)) { |
5132 | 17 | I422ToRGBARow = I422ToRGBARow_AVX2; |
5133 | 17 | } |
5134 | 89 | } |
5135 | 89 | #endif |
5136 | | #if defined(HAS_I422TORGBAROW_NEON) |
5137 | | if (TestCpuFlag(kCpuHasNEON)) { |
5138 | | I422ToRGBARow = I422ToRGBARow_Any_NEON; |
5139 | | if (IS_ALIGNED(width, 8)) { |
5140 | | I422ToRGBARow = I422ToRGBARow_NEON; |
5141 | | } |
5142 | | } |
5143 | | #endif |
5144 | | #if defined(HAS_I422TORGBAROW_SVE2) |
5145 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5146 | | I422ToRGBARow = I422ToRGBARow_SVE2; |
5147 | | } |
5148 | | #endif |
5149 | | #if defined(HAS_I422TORGBAROW_SME) |
5150 | | if (TestCpuFlag(kCpuHasSME)) { |
5151 | | I422ToRGBARow = I422ToRGBARow_SME; |
5152 | | } |
5153 | | #endif |
5154 | | #if defined(HAS_I422TORGBAROW_LSX) |
5155 | | if (TestCpuFlag(kCpuHasLSX)) { |
5156 | | I422ToRGBARow = I422ToRGBARow_Any_LSX; |
5157 | | if (IS_ALIGNED(width, 16)) { |
5158 | | I422ToRGBARow = I422ToRGBARow_LSX; |
5159 | | } |
5160 | | } |
5161 | | #endif |
5162 | | #if defined(HAS_I422TORGBAROW_LASX) |
5163 | | if (TestCpuFlag(kCpuHasLASX)) { |
5164 | | I422ToRGBARow = I422ToRGBARow_Any_LASX; |
5165 | | if (IS_ALIGNED(width, 32)) { |
5166 | | I422ToRGBARow = I422ToRGBARow_LASX; |
5167 | | } |
5168 | | } |
5169 | | #endif |
5170 | | #if defined(HAS_I422TORGBAROW_RVV) |
5171 | | if (TestCpuFlag(kCpuHasRVV)) { |
5172 | | I422ToRGBARow = I422ToRGBARow_RVV; |
5173 | | } |
5174 | | #endif |
5175 | | |
5176 | 1.38k | for (y = 0; y < height; ++y) { |
5177 | 1.29k | I422ToRGBARow(src_y, src_u, src_v, dst_rgba, yuvconstants, width); |
5178 | 1.29k | dst_rgba += dst_stride_rgba; |
5179 | 1.29k | src_y += src_stride_y; |
5180 | 1.29k | src_u += src_stride_u; |
5181 | 1.29k | src_v += src_stride_v; |
5182 | 1.29k | } |
5183 | 89 | return 0; |
5184 | 89 | } |
5185 | | |
5186 | | // Convert I422 to RGBA. |
5187 | | LIBYUV_API |
5188 | | int I422ToRGBA(const uint8_t* src_y, |
5189 | | int src_stride_y, |
5190 | | const uint8_t* src_u, |
5191 | | int src_stride_u, |
5192 | | const uint8_t* src_v, |
5193 | | int src_stride_v, |
5194 | | uint8_t* dst_rgba, |
5195 | | int dst_stride_rgba, |
5196 | | int width, |
5197 | 0 | int height) { |
5198 | 0 | return I422ToRGBAMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5199 | 0 | src_stride_v, dst_rgba, dst_stride_rgba, |
5200 | 0 | &kYuvI601Constants, width, height); |
5201 | 0 | } |
5202 | | |
5203 | | // Convert I422 to BGRA. |
5204 | | LIBYUV_API |
5205 | | int I422ToBGRA(const uint8_t* src_y, |
5206 | | int src_stride_y, |
5207 | | const uint8_t* src_u, |
5208 | | int src_stride_u, |
5209 | | const uint8_t* src_v, |
5210 | | int src_stride_v, |
5211 | | uint8_t* dst_bgra, |
5212 | | int dst_stride_bgra, |
5213 | | int width, |
5214 | 0 | int height) { |
5215 | 0 | return I422ToRGBAMatrix(src_y, src_stride_y, src_v, |
5216 | 0 | src_stride_v, // Swap U and V |
5217 | 0 | src_u, src_stride_u, dst_bgra, dst_stride_bgra, |
5218 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5219 | 0 | width, height); |
5220 | 0 | } |
5221 | | |
5222 | | // Convert NV12 to RGB565 with matrix. |
5223 | | LIBYUV_API |
5224 | | int NV12ToRGB565Matrix(const uint8_t* src_y, |
5225 | | int src_stride_y, |
5226 | | const uint8_t* src_uv, |
5227 | | int src_stride_uv, |
5228 | | uint8_t* dst_rgb565, |
5229 | | int dst_stride_rgb565, |
5230 | | const struct YuvConstants* yuvconstants, |
5231 | | int width, |
5232 | 0 | int height) { |
5233 | 0 | int y; |
5234 | 0 | void (*NV12ToRGB565Row)( |
5235 | 0 | const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* rgb_buf, |
5236 | 0 | const struct YuvConstants* yuvconstants, int width) = NV12ToRGB565Row_C; |
5237 | 0 | assert(yuvconstants); |
5238 | 0 | if (!src_y || !src_uv || !dst_rgb565 || width <= 0 || height == 0) { |
5239 | 0 | return -1; |
5240 | 0 | } |
5241 | | // Negative height means invert the image. |
5242 | 0 | if (height < 0) { |
5243 | 0 | height = -height; |
5244 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
5245 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
5246 | 0 | } |
5247 | 0 | #if defined(HAS_NV12TORGB565ROW_SSSE3) |
5248 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5249 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_Any_SSSE3; |
5250 | 0 | if (IS_ALIGNED(width, 8)) { |
5251 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_SSSE3; |
5252 | 0 | } |
5253 | 0 | } |
5254 | 0 | #endif |
5255 | 0 | #if defined(HAS_NV12TORGB565ROW_AVX2) |
5256 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
5257 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_Any_AVX2; |
5258 | 0 | if (IS_ALIGNED(width, 16)) { |
5259 | 0 | NV12ToRGB565Row = NV12ToRGB565Row_AVX2; |
5260 | 0 | } |
5261 | 0 | } |
5262 | 0 | #endif |
5263 | | #if defined(HAS_NV12TORGB565ROW_NEON) |
5264 | | if (TestCpuFlag(kCpuHasNEON)) { |
5265 | | NV12ToRGB565Row = NV12ToRGB565Row_Any_NEON; |
5266 | | if (IS_ALIGNED(width, 8)) { |
5267 | | NV12ToRGB565Row = NV12ToRGB565Row_NEON; |
5268 | | } |
5269 | | } |
5270 | | #endif |
5271 | | #if defined(HAS_NV12TORGB565ROW_LSX) |
5272 | | if (TestCpuFlag(kCpuHasLSX)) { |
5273 | | NV12ToRGB565Row = NV12ToRGB565Row_Any_LSX; |
5274 | | if (IS_ALIGNED(width, 8)) { |
5275 | | NV12ToRGB565Row = NV12ToRGB565Row_LSX; |
5276 | | } |
5277 | | } |
5278 | | #endif |
5279 | | #if defined(HAS_NV12TORGB565ROW_LASX) |
5280 | | if (TestCpuFlag(kCpuHasLASX)) { |
5281 | | NV12ToRGB565Row = NV12ToRGB565Row_Any_LASX; |
5282 | | if (IS_ALIGNED(width, 16)) { |
5283 | | NV12ToRGB565Row = NV12ToRGB565Row_LASX; |
5284 | | } |
5285 | | } |
5286 | | #endif |
5287 | |
|
5288 | 0 | for (y = 0; y < height; ++y) { |
5289 | 0 | NV12ToRGB565Row(src_y, src_uv, dst_rgb565, yuvconstants, width); |
5290 | 0 | dst_rgb565 += dst_stride_rgb565; |
5291 | 0 | src_y += src_stride_y; |
5292 | 0 | if (y & 1) { |
5293 | 0 | src_uv += src_stride_uv; |
5294 | 0 | } |
5295 | 0 | } |
5296 | 0 | return 0; |
5297 | 0 | } |
5298 | | |
5299 | | // Convert NV12 to RGB565. |
5300 | | LIBYUV_API |
5301 | | int NV12ToRGB565(const uint8_t* src_y, |
5302 | | int src_stride_y, |
5303 | | const uint8_t* src_uv, |
5304 | | int src_stride_uv, |
5305 | | uint8_t* dst_rgb565, |
5306 | | int dst_stride_rgb565, |
5307 | | int width, |
5308 | 0 | int height) { |
5309 | 0 | return NV12ToRGB565Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
5310 | 0 | dst_rgb565, dst_stride_rgb565, &kYuvI601Constants, |
5311 | 0 | width, height); |
5312 | 0 | } |
5313 | | |
5314 | | // Convert I422 to RGBA with matrix. |
5315 | | LIBYUV_API |
5316 | | int I420ToRGBAMatrix(const uint8_t* src_y, |
5317 | | int src_stride_y, |
5318 | | const uint8_t* src_u, |
5319 | | int src_stride_u, |
5320 | | const uint8_t* src_v, |
5321 | | int src_stride_v, |
5322 | | uint8_t* dst_rgba, |
5323 | | int dst_stride_rgba, |
5324 | | const struct YuvConstants* yuvconstants, |
5325 | | int width, |
5326 | 177 | int height) { |
5327 | 177 | int y; |
5328 | 177 | void (*I422ToRGBARow)(const uint8_t* y_buf, const uint8_t* u_buf, |
5329 | 177 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5330 | 177 | const struct YuvConstants* yuvconstants, int width) = |
5331 | 177 | I422ToRGBARow_C; |
5332 | 177 | assert(yuvconstants); |
5333 | 177 | if (!src_y || !src_u || !src_v || !dst_rgba || width <= 0 || height == 0) { |
5334 | 0 | return -1; |
5335 | 0 | } |
5336 | | // Negative height means invert the image. |
5337 | 177 | if (height < 0) { |
5338 | 0 | height = -height; |
5339 | 0 | dst_rgba = dst_rgba + (height - 1) * dst_stride_rgba; |
5340 | 0 | dst_stride_rgba = -dst_stride_rgba; |
5341 | 0 | } |
5342 | 177 | #if defined(HAS_I422TORGBAROW_SSSE3) |
5343 | 177 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5344 | 177 | I422ToRGBARow = I422ToRGBARow_Any_SSSE3; |
5345 | 177 | if (IS_ALIGNED(width, 8)) { |
5346 | 34 | I422ToRGBARow = I422ToRGBARow_SSSE3; |
5347 | 34 | } |
5348 | 177 | } |
5349 | 177 | #endif |
5350 | 177 | #if defined(HAS_I422TORGBAROW_AVX2) |
5351 | 177 | if (TestCpuFlag(kCpuHasAVX2)) { |
5352 | 177 | I422ToRGBARow = I422ToRGBARow_Any_AVX2; |
5353 | 177 | if (IS_ALIGNED(width, 16)) { |
5354 | 17 | I422ToRGBARow = I422ToRGBARow_AVX2; |
5355 | 17 | } |
5356 | 177 | } |
5357 | 177 | #endif |
5358 | | #if defined(HAS_I422TORGBAROW_NEON) |
5359 | | if (TestCpuFlag(kCpuHasNEON)) { |
5360 | | I422ToRGBARow = I422ToRGBARow_Any_NEON; |
5361 | | if (IS_ALIGNED(width, 8)) { |
5362 | | I422ToRGBARow = I422ToRGBARow_NEON; |
5363 | | } |
5364 | | } |
5365 | | #endif |
5366 | | #if defined(HAS_I422TORGBAROW_SVE2) |
5367 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5368 | | I422ToRGBARow = I422ToRGBARow_SVE2; |
5369 | | } |
5370 | | #endif |
5371 | | #if defined(HAS_I422TORGBAROW_SME) |
5372 | | if (TestCpuFlag(kCpuHasSME)) { |
5373 | | I422ToRGBARow = I422ToRGBARow_SME; |
5374 | | } |
5375 | | #endif |
5376 | | #if defined(HAS_I422TORGBAROW_LSX) |
5377 | | if (TestCpuFlag(kCpuHasLSX)) { |
5378 | | I422ToRGBARow = I422ToRGBARow_Any_LSX; |
5379 | | if (IS_ALIGNED(width, 16)) { |
5380 | | I422ToRGBARow = I422ToRGBARow_LSX; |
5381 | | } |
5382 | | } |
5383 | | #endif |
5384 | | #if defined(HAS_I422TORGBAROW_LASX) |
5385 | | if (TestCpuFlag(kCpuHasLASX)) { |
5386 | | I422ToRGBARow = I422ToRGBARow_Any_LASX; |
5387 | | if (IS_ALIGNED(width, 32)) { |
5388 | | I422ToRGBARow = I422ToRGBARow_LASX; |
5389 | | } |
5390 | | } |
5391 | | #endif |
5392 | | #if defined(HAS_I422TORGBAROW_RVV) |
5393 | | if (TestCpuFlag(kCpuHasRVV)) { |
5394 | | I422ToRGBARow = I422ToRGBARow_RVV; |
5395 | | } |
5396 | | #endif |
5397 | | |
5398 | 2.87k | for (y = 0; y < height; ++y) { |
5399 | 2.70k | I422ToRGBARow(src_y, src_u, src_v, dst_rgba, yuvconstants, width); |
5400 | 2.70k | dst_rgba += dst_stride_rgba; |
5401 | 2.70k | src_y += src_stride_y; |
5402 | 2.70k | if (y & 1) { |
5403 | 1.32k | src_u += src_stride_u; |
5404 | 1.32k | src_v += src_stride_v; |
5405 | 1.32k | } |
5406 | 2.70k | } |
5407 | 177 | return 0; |
5408 | 177 | } |
5409 | | |
5410 | | // Convert I420 to RGBA. |
5411 | | LIBYUV_API |
5412 | | int I420ToRGBA(const uint8_t* src_y, |
5413 | | int src_stride_y, |
5414 | | const uint8_t* src_u, |
5415 | | int src_stride_u, |
5416 | | const uint8_t* src_v, |
5417 | | int src_stride_v, |
5418 | | uint8_t* dst_rgba, |
5419 | | int dst_stride_rgba, |
5420 | | int width, |
5421 | 0 | int height) { |
5422 | 0 | return I420ToRGBAMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5423 | 0 | src_stride_v, dst_rgba, dst_stride_rgba, |
5424 | 0 | &kYuvI601Constants, width, height); |
5425 | 0 | } |
5426 | | |
5427 | | // Convert I420 to BGRA. |
5428 | | LIBYUV_API |
5429 | | int I420ToBGRA(const uint8_t* src_y, |
5430 | | int src_stride_y, |
5431 | | const uint8_t* src_u, |
5432 | | int src_stride_u, |
5433 | | const uint8_t* src_v, |
5434 | | int src_stride_v, |
5435 | | uint8_t* dst_bgra, |
5436 | | int dst_stride_bgra, |
5437 | | int width, |
5438 | 0 | int height) { |
5439 | 0 | return I420ToRGBAMatrix(src_y, src_stride_y, src_v, |
5440 | 0 | src_stride_v, // Swap U and V |
5441 | 0 | src_u, src_stride_u, dst_bgra, dst_stride_bgra, |
5442 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5443 | 0 | width, height); |
5444 | 0 | } |
5445 | | |
5446 | | // Convert I420 to RGB24 with matrix. |
5447 | | LIBYUV_API |
5448 | | int I420ToRGB24Matrix(const uint8_t* src_y, |
5449 | | int src_stride_y, |
5450 | | const uint8_t* src_u, |
5451 | | int src_stride_u, |
5452 | | const uint8_t* src_v, |
5453 | | int src_stride_v, |
5454 | | uint8_t* dst_rgb24, |
5455 | | int dst_stride_rgb24, |
5456 | | const struct YuvConstants* yuvconstants, |
5457 | | int width, |
5458 | 217 | int height) { |
5459 | 217 | int y; |
5460 | 217 | void (*I422ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
5461 | 217 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5462 | 217 | const struct YuvConstants* yuvconstants, int width) = |
5463 | 217 | I422ToRGB24Row_C; |
5464 | 217 | assert(yuvconstants); |
5465 | 217 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
5466 | 0 | return -1; |
5467 | 0 | } |
5468 | | // Negative height means invert the image. |
5469 | 217 | if (height < 0) { |
5470 | 0 | height = -height; |
5471 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
5472 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
5473 | 0 | } |
5474 | 217 | #if defined(HAS_I422TORGB24ROW_SSSE3) |
5475 | 217 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5476 | 217 | I422ToRGB24Row = I422ToRGB24Row_Any_SSSE3; |
5477 | 217 | if (IS_ALIGNED(width, 16)) { |
5478 | 35 | I422ToRGB24Row = I422ToRGB24Row_SSSE3; |
5479 | 35 | } |
5480 | 217 | } |
5481 | 217 | #endif |
5482 | 217 | #if defined(HAS_I422TORGB24ROW_AVX2) |
5483 | 217 | if (TestCpuFlag(kCpuHasAVX2)) { |
5484 | 217 | I422ToRGB24Row = I422ToRGB24Row_Any_AVX2; |
5485 | 217 | if (IS_ALIGNED(width, 32)) { |
5486 | 18 | I422ToRGB24Row = I422ToRGB24Row_AVX2; |
5487 | 18 | } |
5488 | 217 | } |
5489 | 217 | #endif |
5490 | | #if defined(HAS_I422TORGB24ROW_NEON) |
5491 | | if (TestCpuFlag(kCpuHasNEON)) { |
5492 | | I422ToRGB24Row = I422ToRGB24Row_Any_NEON; |
5493 | | if (IS_ALIGNED(width, 8)) { |
5494 | | I422ToRGB24Row = I422ToRGB24Row_NEON; |
5495 | | } |
5496 | | } |
5497 | | #endif |
5498 | | #if defined(HAS_I422TORGB24ROW_SVE2) |
5499 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5500 | | I422ToRGB24Row = I422ToRGB24Row_SVE2; |
5501 | | } |
5502 | | #endif |
5503 | | #if defined(HAS_I422TORGB24ROW_SME) |
5504 | | if (TestCpuFlag(kCpuHasSME)) { |
5505 | | I422ToRGB24Row = I422ToRGB24Row_SME; |
5506 | | } |
5507 | | #endif |
5508 | | #if defined(HAS_I422TORGB24ROW_LSX) |
5509 | | if (TestCpuFlag(kCpuHasLSX)) { |
5510 | | I422ToRGB24Row = I422ToRGB24Row_Any_LSX; |
5511 | | if (IS_ALIGNED(width, 16)) { |
5512 | | I422ToRGB24Row = I422ToRGB24Row_LSX; |
5513 | | } |
5514 | | } |
5515 | | #endif |
5516 | | #if defined(HAS_I422TORGB24ROW_LASX) |
5517 | | if (TestCpuFlag(kCpuHasLASX)) { |
5518 | | I422ToRGB24Row = I422ToRGB24Row_Any_LASX; |
5519 | | if (IS_ALIGNED(width, 32)) { |
5520 | | I422ToRGB24Row = I422ToRGB24Row_LASX; |
5521 | | } |
5522 | | } |
5523 | | #endif |
5524 | | #if defined(HAS_I422TORGB24ROW_RVV) |
5525 | | if (TestCpuFlag(kCpuHasRVV)) { |
5526 | | I422ToRGB24Row = I422ToRGB24Row_RVV; |
5527 | | } |
5528 | | #endif |
5529 | | |
5530 | 3.94k | for (y = 0; y < height; ++y) { |
5531 | 3.73k | I422ToRGB24Row(src_y, src_u, src_v, dst_rgb24, yuvconstants, width); |
5532 | 3.73k | dst_rgb24 += dst_stride_rgb24; |
5533 | 3.73k | src_y += src_stride_y; |
5534 | 3.73k | if (y & 1) { |
5535 | 1.83k | src_u += src_stride_u; |
5536 | 1.83k | src_v += src_stride_v; |
5537 | 1.83k | } |
5538 | 3.73k | } |
5539 | 217 | return 0; |
5540 | 217 | } |
5541 | | |
5542 | | // Convert I420 to RGB24. |
5543 | | LIBYUV_API |
5544 | | int I420ToRGB24(const uint8_t* src_y, |
5545 | | int src_stride_y, |
5546 | | const uint8_t* src_u, |
5547 | | int src_stride_u, |
5548 | | const uint8_t* src_v, |
5549 | | int src_stride_v, |
5550 | | uint8_t* dst_rgb24, |
5551 | | int dst_stride_rgb24, |
5552 | | int width, |
5553 | 0 | int height) { |
5554 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5555 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5556 | 0 | &kYuvI601Constants, width, height); |
5557 | 0 | } |
5558 | | |
5559 | | // Convert I420 to RAW. |
5560 | | LIBYUV_API |
5561 | | int I420ToRAW(const uint8_t* src_y, |
5562 | | int src_stride_y, |
5563 | | const uint8_t* src_u, |
5564 | | int src_stride_u, |
5565 | | const uint8_t* src_v, |
5566 | | int src_stride_v, |
5567 | | uint8_t* dst_raw, |
5568 | | int dst_stride_raw, |
5569 | | int width, |
5570 | 0 | int height) { |
5571 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_v, |
5572 | 0 | src_stride_v, // Swap U and V |
5573 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5574 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5575 | 0 | width, height); |
5576 | 0 | } |
5577 | | |
5578 | | // Convert J420 to RGB24. |
5579 | | LIBYUV_API |
5580 | | int J420ToRGB24(const uint8_t* src_y, |
5581 | | int src_stride_y, |
5582 | | const uint8_t* src_u, |
5583 | | int src_stride_u, |
5584 | | const uint8_t* src_v, |
5585 | | int src_stride_v, |
5586 | | uint8_t* dst_rgb24, |
5587 | | int dst_stride_rgb24, |
5588 | | int width, |
5589 | 0 | int height) { |
5590 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5591 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5592 | 0 | &kYuvJPEGConstants, width, height); |
5593 | 0 | } |
5594 | | |
5595 | | // Convert J420 to RAW. |
5596 | | LIBYUV_API |
5597 | | int J420ToRAW(const uint8_t* src_y, |
5598 | | int src_stride_y, |
5599 | | const uint8_t* src_u, |
5600 | | int src_stride_u, |
5601 | | const uint8_t* src_v, |
5602 | | int src_stride_v, |
5603 | | uint8_t* dst_raw, |
5604 | | int dst_stride_raw, |
5605 | | int width, |
5606 | 0 | int height) { |
5607 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_v, |
5608 | 0 | src_stride_v, // Swap U and V |
5609 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5610 | 0 | &kYvuJPEGConstants, // Use Yvu matrix |
5611 | 0 | width, height); |
5612 | 0 | } |
5613 | | |
5614 | | // Convert H420 to RGB24. |
5615 | | LIBYUV_API |
5616 | | int H420ToRGB24(const uint8_t* src_y, |
5617 | | int src_stride_y, |
5618 | | const uint8_t* src_u, |
5619 | | int src_stride_u, |
5620 | | const uint8_t* src_v, |
5621 | | int src_stride_v, |
5622 | | uint8_t* dst_rgb24, |
5623 | | int dst_stride_rgb24, |
5624 | | int width, |
5625 | 0 | int height) { |
5626 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5627 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5628 | 0 | &kYuvH709Constants, width, height); |
5629 | 0 | } |
5630 | | |
5631 | | // Convert H420 to RAW. |
5632 | | LIBYUV_API |
5633 | | int H420ToRAW(const uint8_t* src_y, |
5634 | | int src_stride_y, |
5635 | | const uint8_t* src_u, |
5636 | | int src_stride_u, |
5637 | | const uint8_t* src_v, |
5638 | | int src_stride_v, |
5639 | | uint8_t* dst_raw, |
5640 | | int dst_stride_raw, |
5641 | | int width, |
5642 | 0 | int height) { |
5643 | 0 | return I420ToRGB24Matrix(src_y, src_stride_y, src_v, |
5644 | 0 | src_stride_v, // Swap U and V |
5645 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5646 | 0 | &kYvuH709Constants, // Use Yvu matrix |
5647 | 0 | width, height); |
5648 | 0 | } |
5649 | | |
5650 | | // Convert I422 to RGB24 with matrix. |
5651 | | LIBYUV_API |
5652 | | int I422ToRGB24Matrix(const uint8_t* src_y, |
5653 | | int src_stride_y, |
5654 | | const uint8_t* src_u, |
5655 | | int src_stride_u, |
5656 | | const uint8_t* src_v, |
5657 | | int src_stride_v, |
5658 | | uint8_t* dst_rgb24, |
5659 | | int dst_stride_rgb24, |
5660 | | const struct YuvConstants* yuvconstants, |
5661 | | int width, |
5662 | 83 | int height) { |
5663 | 83 | int y; |
5664 | 83 | void (*I422ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
5665 | 83 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5666 | 83 | const struct YuvConstants* yuvconstants, int width) = |
5667 | 83 | I422ToRGB24Row_C; |
5668 | 83 | assert(yuvconstants); |
5669 | 83 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
5670 | 0 | return -1; |
5671 | 0 | } |
5672 | | // Negative height means invert the image. |
5673 | 83 | if (height < 0) { |
5674 | 0 | height = -height; |
5675 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
5676 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
5677 | 0 | } |
5678 | 83 | #if defined(HAS_I422TORGB24ROW_SSSE3) |
5679 | 83 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5680 | 83 | I422ToRGB24Row = I422ToRGB24Row_Any_SSSE3; |
5681 | 83 | if (IS_ALIGNED(width, 16)) { |
5682 | 34 | I422ToRGB24Row = I422ToRGB24Row_SSSE3; |
5683 | 34 | } |
5684 | 83 | } |
5685 | 83 | #endif |
5686 | 83 | #if defined(HAS_I422TORGB24ROW_AVX2) |
5687 | 83 | if (TestCpuFlag(kCpuHasAVX2)) { |
5688 | 83 | I422ToRGB24Row = I422ToRGB24Row_Any_AVX2; |
5689 | 83 | if (IS_ALIGNED(width, 32)) { |
5690 | 17 | I422ToRGB24Row = I422ToRGB24Row_AVX2; |
5691 | 17 | } |
5692 | 83 | } |
5693 | 83 | #endif |
5694 | | #if defined(HAS_I422TORGB24ROW_NEON) |
5695 | | if (TestCpuFlag(kCpuHasNEON)) { |
5696 | | I422ToRGB24Row = I422ToRGB24Row_Any_NEON; |
5697 | | if (IS_ALIGNED(width, 8)) { |
5698 | | I422ToRGB24Row = I422ToRGB24Row_NEON; |
5699 | | } |
5700 | | } |
5701 | | #endif |
5702 | | #if defined(HAS_I422TORGB24ROW_SVE2) |
5703 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5704 | | I422ToRGB24Row = I422ToRGB24Row_SVE2; |
5705 | | } |
5706 | | #endif |
5707 | | #if defined(HAS_I422TORGB24ROW_SME) |
5708 | | if (TestCpuFlag(kCpuHasSME)) { |
5709 | | I422ToRGB24Row = I422ToRGB24Row_SME; |
5710 | | } |
5711 | | #endif |
5712 | | #if defined(HAS_I422TORGB24ROW_LSX) |
5713 | | if (TestCpuFlag(kCpuHasLSX)) { |
5714 | | I422ToRGB24Row = I422ToRGB24Row_Any_LSX; |
5715 | | if (IS_ALIGNED(width, 16)) { |
5716 | | I422ToRGB24Row = I422ToRGB24Row_LSX; |
5717 | | } |
5718 | | } |
5719 | | #endif |
5720 | | #if defined(HAS_I422TORGB24ROW_LASX) |
5721 | | if (TestCpuFlag(kCpuHasLASX)) { |
5722 | | I422ToRGB24Row = I422ToRGB24Row_Any_LASX; |
5723 | | if (IS_ALIGNED(width, 32)) { |
5724 | | I422ToRGB24Row = I422ToRGB24Row_LASX; |
5725 | | } |
5726 | | } |
5727 | | #endif |
5728 | | #if defined(HAS_I422TORGB24ROW_RVV) |
5729 | | if (TestCpuFlag(kCpuHasRVV)) { |
5730 | | I422ToRGB24Row = I422ToRGB24Row_RVV; |
5731 | | } |
5732 | | #endif |
5733 | | |
5734 | 1.56k | for (y = 0; y < height; ++y) { |
5735 | 1.48k | I422ToRGB24Row(src_y, src_u, src_v, dst_rgb24, yuvconstants, width); |
5736 | 1.48k | dst_rgb24 += dst_stride_rgb24; |
5737 | 1.48k | src_y += src_stride_y; |
5738 | 1.48k | src_u += src_stride_u; |
5739 | 1.48k | src_v += src_stride_v; |
5740 | 1.48k | } |
5741 | 83 | return 0; |
5742 | 83 | } |
5743 | | |
5744 | | // Convert I422 to RGB24. |
5745 | | LIBYUV_API |
5746 | | int I422ToRGB24(const uint8_t* src_y, |
5747 | | int src_stride_y, |
5748 | | const uint8_t* src_u, |
5749 | | int src_stride_u, |
5750 | | const uint8_t* src_v, |
5751 | | int src_stride_v, |
5752 | | uint8_t* dst_rgb24, |
5753 | | int dst_stride_rgb24, |
5754 | | int width, |
5755 | 0 | int height) { |
5756 | 0 | return I422ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
5757 | 0 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
5758 | 0 | &kYuvI601Constants, width, height); |
5759 | 0 | } |
5760 | | |
5761 | | // Convert I422 to RAW. |
5762 | | LIBYUV_API |
5763 | | int I422ToRAW(const uint8_t* src_y, |
5764 | | int src_stride_y, |
5765 | | const uint8_t* src_u, |
5766 | | int src_stride_u, |
5767 | | const uint8_t* src_v, |
5768 | | int src_stride_v, |
5769 | | uint8_t* dst_raw, |
5770 | | int dst_stride_raw, |
5771 | | int width, |
5772 | 0 | int height) { |
5773 | 0 | return I422ToRGB24Matrix(src_y, src_stride_y, src_v, |
5774 | 0 | src_stride_v, // Swap U and V |
5775 | 0 | src_u, src_stride_u, dst_raw, dst_stride_raw, |
5776 | 0 | &kYvuI601Constants, // Use Yvu matrix |
5777 | 0 | width, height); |
5778 | 0 | } |
5779 | | |
5780 | | // Convert I420 to ARGB1555. |
5781 | | LIBYUV_API |
5782 | | int I420ToARGB1555(const uint8_t* src_y, |
5783 | | int src_stride_y, |
5784 | | const uint8_t* src_u, |
5785 | | int src_stride_u, |
5786 | | const uint8_t* src_v, |
5787 | | int src_stride_v, |
5788 | | uint8_t* dst_argb1555, |
5789 | | int dst_stride_argb1555, |
5790 | | int width, |
5791 | 0 | int height) { |
5792 | 0 | int y; |
5793 | 0 | void (*I422ToARGB1555Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
5794 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5795 | 0 | const struct YuvConstants* yuvconstants, |
5796 | 0 | int width) = I422ToARGB1555Row_C; |
5797 | 0 | if (!src_y || !src_u || !src_v || !dst_argb1555 || width <= 0 || |
5798 | 0 | height == 0) { |
5799 | 0 | return -1; |
5800 | 0 | } |
5801 | | // Negative height means invert the image. |
5802 | 0 | if (height < 0) { |
5803 | 0 | height = -height; |
5804 | 0 | dst_argb1555 = dst_argb1555 + (height - 1) * dst_stride_argb1555; |
5805 | 0 | dst_stride_argb1555 = -dst_stride_argb1555; |
5806 | 0 | } |
5807 | 0 | #if defined(HAS_I422TOARGB1555ROW_SSSE3) |
5808 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5809 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_Any_SSSE3; |
5810 | 0 | if (IS_ALIGNED(width, 8)) { |
5811 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_SSSE3; |
5812 | 0 | } |
5813 | 0 | } |
5814 | 0 | #endif |
5815 | 0 | #if defined(HAS_I422TOARGB1555ROW_AVX2) |
5816 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
5817 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_Any_AVX2; |
5818 | 0 | if (IS_ALIGNED(width, 16)) { |
5819 | 0 | I422ToARGB1555Row = I422ToARGB1555Row_AVX2; |
5820 | 0 | } |
5821 | 0 | } |
5822 | 0 | #endif |
5823 | | #if defined(HAS_I422TOARGB1555ROW_NEON) |
5824 | | if (TestCpuFlag(kCpuHasNEON)) { |
5825 | | I422ToARGB1555Row = I422ToARGB1555Row_Any_NEON; |
5826 | | if (IS_ALIGNED(width, 8)) { |
5827 | | I422ToARGB1555Row = I422ToARGB1555Row_NEON; |
5828 | | } |
5829 | | } |
5830 | | #endif |
5831 | | #if defined(HAS_I422TOARGB1555ROW_SVE2) |
5832 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5833 | | I422ToARGB1555Row = I422ToARGB1555Row_SVE2; |
5834 | | } |
5835 | | #endif |
5836 | | #if defined(HAS_I422TOARGB1555ROW_SME) |
5837 | | if (TestCpuFlag(kCpuHasSME)) { |
5838 | | I422ToARGB1555Row = I422ToARGB1555Row_SME; |
5839 | | } |
5840 | | #endif |
5841 | | #if defined(HAS_I422TOARGB1555ROW_LSX) |
5842 | | if (TestCpuFlag(kCpuHasLSX)) { |
5843 | | I422ToARGB1555Row = I422ToARGB1555Row_Any_LSX; |
5844 | | if (IS_ALIGNED(width, 16)) { |
5845 | | I422ToARGB1555Row = I422ToARGB1555Row_LSX; |
5846 | | } |
5847 | | } |
5848 | | #endif |
5849 | | #if defined(HAS_I422TOARGB1555ROW_LASX) |
5850 | | if (TestCpuFlag(kCpuHasLASX)) { |
5851 | | I422ToARGB1555Row = I422ToARGB1555Row_Any_LASX; |
5852 | | if (IS_ALIGNED(width, 8)) { |
5853 | | I422ToARGB1555Row = I422ToARGB1555Row_LASX; |
5854 | | } |
5855 | | } |
5856 | | #endif |
5857 | |
|
5858 | 0 | for (y = 0; y < height; ++y) { |
5859 | 0 | I422ToARGB1555Row(src_y, src_u, src_v, dst_argb1555, &kYuvI601Constants, |
5860 | 0 | width); |
5861 | 0 | dst_argb1555 += dst_stride_argb1555; |
5862 | 0 | src_y += src_stride_y; |
5863 | 0 | if (y & 1) { |
5864 | 0 | src_u += src_stride_u; |
5865 | 0 | src_v += src_stride_v; |
5866 | 0 | } |
5867 | 0 | } |
5868 | 0 | return 0; |
5869 | 0 | } |
5870 | | |
5871 | | // Convert I420 to ARGB4444. |
5872 | | LIBYUV_API |
5873 | | int I420ToARGB4444(const uint8_t* src_y, |
5874 | | int src_stride_y, |
5875 | | const uint8_t* src_u, |
5876 | | int src_stride_u, |
5877 | | const uint8_t* src_v, |
5878 | | int src_stride_v, |
5879 | | uint8_t* dst_argb4444, |
5880 | | int dst_stride_argb4444, |
5881 | | int width, |
5882 | 0 | int height) { |
5883 | 0 | int y; |
5884 | 0 | void (*I422ToARGB4444Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
5885 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5886 | 0 | const struct YuvConstants* yuvconstants, |
5887 | 0 | int width) = I422ToARGB4444Row_C; |
5888 | 0 | if (!src_y || !src_u || !src_v || !dst_argb4444 || width <= 0 || |
5889 | 0 | height == 0) { |
5890 | 0 | return -1; |
5891 | 0 | } |
5892 | | // Negative height means invert the image. |
5893 | 0 | if (height < 0) { |
5894 | 0 | height = -height; |
5895 | 0 | dst_argb4444 = dst_argb4444 + (height - 1) * dst_stride_argb4444; |
5896 | 0 | dst_stride_argb4444 = -dst_stride_argb4444; |
5897 | 0 | } |
5898 | 0 | #if defined(HAS_I422TOARGB4444ROW_SSSE3) |
5899 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5900 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_Any_SSSE3; |
5901 | 0 | if (IS_ALIGNED(width, 8)) { |
5902 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_SSSE3; |
5903 | 0 | } |
5904 | 0 | } |
5905 | 0 | #endif |
5906 | 0 | #if defined(HAS_I422TOARGB4444ROW_AVX2) |
5907 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
5908 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_Any_AVX2; |
5909 | 0 | if (IS_ALIGNED(width, 16)) { |
5910 | 0 | I422ToARGB4444Row = I422ToARGB4444Row_AVX2; |
5911 | 0 | } |
5912 | 0 | } |
5913 | 0 | #endif |
5914 | | #if defined(HAS_I422TOARGB4444ROW_NEON) |
5915 | | if (TestCpuFlag(kCpuHasNEON)) { |
5916 | | I422ToARGB4444Row = I422ToARGB4444Row_Any_NEON; |
5917 | | if (IS_ALIGNED(width, 8)) { |
5918 | | I422ToARGB4444Row = I422ToARGB4444Row_NEON; |
5919 | | } |
5920 | | } |
5921 | | #endif |
5922 | | #if defined(HAS_I422TOARGB4444ROW_SVE2) |
5923 | | if (TestCpuFlag(kCpuHasSVE2)) { |
5924 | | I422ToARGB4444Row = I422ToARGB4444Row_SVE2; |
5925 | | } |
5926 | | #endif |
5927 | | #if defined(HAS_I422TOARGB4444ROW_SME) |
5928 | | if (TestCpuFlag(kCpuHasSME)) { |
5929 | | I422ToARGB4444Row = I422ToARGB4444Row_SME; |
5930 | | } |
5931 | | #endif |
5932 | | #if defined(HAS_I422TOARGB4444ROW_LSX) |
5933 | | if (TestCpuFlag(kCpuHasLSX)) { |
5934 | | I422ToARGB4444Row = I422ToARGB4444Row_Any_LSX; |
5935 | | if (IS_ALIGNED(width, 16)) { |
5936 | | I422ToARGB4444Row = I422ToARGB4444Row_LSX; |
5937 | | } |
5938 | | } |
5939 | | #endif |
5940 | | #if defined(HAS_I422TOARGB4444ROW_LASX) |
5941 | | if (TestCpuFlag(kCpuHasLASX)) { |
5942 | | I422ToARGB4444Row = I422ToARGB4444Row_Any_LASX; |
5943 | | if (IS_ALIGNED(width, 8)) { |
5944 | | I422ToARGB4444Row = I422ToARGB4444Row_LASX; |
5945 | | } |
5946 | | } |
5947 | | #endif |
5948 | |
|
5949 | 0 | for (y = 0; y < height; ++y) { |
5950 | 0 | I422ToARGB4444Row(src_y, src_u, src_v, dst_argb4444, &kYuvI601Constants, |
5951 | 0 | width); |
5952 | 0 | dst_argb4444 += dst_stride_argb4444; |
5953 | 0 | src_y += src_stride_y; |
5954 | 0 | if (y & 1) { |
5955 | 0 | src_u += src_stride_u; |
5956 | 0 | src_v += src_stride_v; |
5957 | 0 | } |
5958 | 0 | } |
5959 | 0 | return 0; |
5960 | 0 | } |
5961 | | |
5962 | | // Convert I420 to RGB565 with specified color matrix. |
5963 | | LIBYUV_API |
5964 | | int I420ToRGB565Matrix(const uint8_t* src_y, |
5965 | | int src_stride_y, |
5966 | | const uint8_t* src_u, |
5967 | | int src_stride_u, |
5968 | | const uint8_t* src_v, |
5969 | | int src_stride_v, |
5970 | | uint8_t* dst_rgb565, |
5971 | | int dst_stride_rgb565, |
5972 | | const struct YuvConstants* yuvconstants, |
5973 | | int width, |
5974 | 125 | int height) { |
5975 | 125 | int y; |
5976 | 125 | void (*I422ToRGB565Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
5977 | 125 | const uint8_t* v_buf, uint8_t* rgb_buf, |
5978 | 125 | const struct YuvConstants* yuvconstants, int width) = |
5979 | 125 | I422ToRGB565Row_C; |
5980 | 125 | assert(yuvconstants); |
5981 | 125 | if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) { |
5982 | 0 | return -1; |
5983 | 0 | } |
5984 | | // Negative height means invert the image. |
5985 | 125 | if (height < 0) { |
5986 | 0 | height = -height; |
5987 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
5988 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
5989 | 0 | } |
5990 | 125 | #if defined(HAS_I422TORGB565ROW_SSSE3) |
5991 | 125 | if (TestCpuFlag(kCpuHasSSSE3)) { |
5992 | 125 | I422ToRGB565Row = I422ToRGB565Row_Any_SSSE3; |
5993 | 125 | if (IS_ALIGNED(width, 8)) { |
5994 | 36 | I422ToRGB565Row = I422ToRGB565Row_SSSE3; |
5995 | 36 | } |
5996 | 125 | } |
5997 | 125 | #endif |
5998 | 125 | #if defined(HAS_I422TORGB565ROW_AVX2) |
5999 | 125 | if (TestCpuFlag(kCpuHasAVX2)) { |
6000 | 125 | I422ToRGB565Row = I422ToRGB565Row_Any_AVX2; |
6001 | 125 | if (IS_ALIGNED(width, 16)) { |
6002 | 17 | I422ToRGB565Row = I422ToRGB565Row_AVX2; |
6003 | 17 | } |
6004 | 125 | } |
6005 | 125 | #endif |
6006 | | #if defined(HAS_I422TORGB565ROW_NEON) |
6007 | | if (TestCpuFlag(kCpuHasNEON)) { |
6008 | | I422ToRGB565Row = I422ToRGB565Row_Any_NEON; |
6009 | | if (IS_ALIGNED(width, 8)) { |
6010 | | I422ToRGB565Row = I422ToRGB565Row_NEON; |
6011 | | } |
6012 | | } |
6013 | | #endif |
6014 | | #if defined(HAS_I422TORGB565ROW_SVE2) |
6015 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6016 | | I422ToRGB565Row = I422ToRGB565Row_SVE2; |
6017 | | } |
6018 | | #endif |
6019 | | #if defined(HAS_I422TORGB565ROW_SME) |
6020 | | if (TestCpuFlag(kCpuHasSME)) { |
6021 | | I422ToRGB565Row = I422ToRGB565Row_SME; |
6022 | | } |
6023 | | #endif |
6024 | | #if defined(HAS_I422TORGB565ROW_LSX) |
6025 | | if (TestCpuFlag(kCpuHasLSX)) { |
6026 | | I422ToRGB565Row = I422ToRGB565Row_Any_LSX; |
6027 | | if (IS_ALIGNED(width, 16)) { |
6028 | | I422ToRGB565Row = I422ToRGB565Row_LSX; |
6029 | | } |
6030 | | } |
6031 | | #endif |
6032 | | #if defined(HAS_I422TORGB565ROW_LASX) |
6033 | | if (TestCpuFlag(kCpuHasLASX)) { |
6034 | | I422ToRGB565Row = I422ToRGB565Row_Any_LASX; |
6035 | | if (IS_ALIGNED(width, 32)) { |
6036 | | I422ToRGB565Row = I422ToRGB565Row_LASX; |
6037 | | } |
6038 | | } |
6039 | | #endif |
6040 | | |
6041 | 2.25k | for (y = 0; y < height; ++y) { |
6042 | 2.13k | I422ToRGB565Row(src_y, src_u, src_v, dst_rgb565, yuvconstants, width); |
6043 | 2.13k | dst_rgb565 += dst_stride_rgb565; |
6044 | 2.13k | src_y += src_stride_y; |
6045 | 2.13k | if (y & 1) { |
6046 | 1.05k | src_u += src_stride_u; |
6047 | 1.05k | src_v += src_stride_v; |
6048 | 1.05k | } |
6049 | 2.13k | } |
6050 | 125 | return 0; |
6051 | 125 | } |
6052 | | |
6053 | | // Convert I420 to RGB565. |
6054 | | LIBYUV_API |
6055 | | int I420ToRGB565(const uint8_t* src_y, |
6056 | | int src_stride_y, |
6057 | | const uint8_t* src_u, |
6058 | | int src_stride_u, |
6059 | | const uint8_t* src_v, |
6060 | | int src_stride_v, |
6061 | | uint8_t* dst_rgb565, |
6062 | | int dst_stride_rgb565, |
6063 | | int width, |
6064 | 0 | int height) { |
6065 | 0 | return I420ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6066 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6067 | 0 | &kYuvI601Constants, width, height); |
6068 | 0 | } |
6069 | | |
6070 | | // Convert J420 to RGB565. |
6071 | | LIBYUV_API |
6072 | | int J420ToRGB565(const uint8_t* src_y, |
6073 | | int src_stride_y, |
6074 | | const uint8_t* src_u, |
6075 | | int src_stride_u, |
6076 | | const uint8_t* src_v, |
6077 | | int src_stride_v, |
6078 | | uint8_t* dst_rgb565, |
6079 | | int dst_stride_rgb565, |
6080 | | int width, |
6081 | 0 | int height) { |
6082 | 0 | return I420ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6083 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6084 | 0 | &kYuvJPEGConstants, width, height); |
6085 | 0 | } |
6086 | | |
6087 | | // Convert H420 to RGB565. |
6088 | | LIBYUV_API |
6089 | | int H420ToRGB565(const uint8_t* src_y, |
6090 | | int src_stride_y, |
6091 | | const uint8_t* src_u, |
6092 | | int src_stride_u, |
6093 | | const uint8_t* src_v, |
6094 | | int src_stride_v, |
6095 | | uint8_t* dst_rgb565, |
6096 | | int dst_stride_rgb565, |
6097 | | int width, |
6098 | 0 | int height) { |
6099 | 0 | return I420ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6100 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6101 | 0 | &kYuvH709Constants, width, height); |
6102 | 0 | } |
6103 | | |
6104 | | // Convert I422 to RGB565 with specified color matrix. |
6105 | | LIBYUV_API |
6106 | | int I422ToRGB565Matrix(const uint8_t* src_y, |
6107 | | int src_stride_y, |
6108 | | const uint8_t* src_u, |
6109 | | int src_stride_u, |
6110 | | const uint8_t* src_v, |
6111 | | int src_stride_v, |
6112 | | uint8_t* dst_rgb565, |
6113 | | int dst_stride_rgb565, |
6114 | | const struct YuvConstants* yuvconstants, |
6115 | | int width, |
6116 | 71 | int height) { |
6117 | 71 | int y; |
6118 | 71 | void (*I422ToRGB565Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6119 | 71 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6120 | 71 | const struct YuvConstants* yuvconstants, int width) = |
6121 | 71 | I422ToRGB565Row_C; |
6122 | 71 | assert(yuvconstants); |
6123 | 71 | if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) { |
6124 | 0 | return -1; |
6125 | 0 | } |
6126 | | // Negative height means invert the image. |
6127 | 71 | if (height < 0) { |
6128 | 0 | height = -height; |
6129 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
6130 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
6131 | 0 | } |
6132 | 71 | #if defined(HAS_I422TORGB565ROW_SSSE3) |
6133 | 71 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6134 | 71 | I422ToRGB565Row = I422ToRGB565Row_Any_SSSE3; |
6135 | 71 | if (IS_ALIGNED(width, 8)) { |
6136 | 34 | I422ToRGB565Row = I422ToRGB565Row_SSSE3; |
6137 | 34 | } |
6138 | 71 | } |
6139 | 71 | #endif |
6140 | 71 | #if defined(HAS_I422TORGB565ROW_AVX2) |
6141 | 71 | if (TestCpuFlag(kCpuHasAVX2)) { |
6142 | 71 | I422ToRGB565Row = I422ToRGB565Row_Any_AVX2; |
6143 | 71 | if (IS_ALIGNED(width, 16)) { |
6144 | 17 | I422ToRGB565Row = I422ToRGB565Row_AVX2; |
6145 | 17 | } |
6146 | 71 | } |
6147 | 71 | #endif |
6148 | | #if defined(HAS_I422TORGB565ROW_NEON) |
6149 | | if (TestCpuFlag(kCpuHasNEON)) { |
6150 | | I422ToRGB565Row = I422ToRGB565Row_Any_NEON; |
6151 | | if (IS_ALIGNED(width, 8)) { |
6152 | | I422ToRGB565Row = I422ToRGB565Row_NEON; |
6153 | | } |
6154 | | } |
6155 | | #endif |
6156 | | #if defined(HAS_I422TORGB565ROW_SVE2) |
6157 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6158 | | I422ToRGB565Row = I422ToRGB565Row_SVE2; |
6159 | | } |
6160 | | #endif |
6161 | | #if defined(HAS_I422TORGB565ROW_SME) |
6162 | | if (TestCpuFlag(kCpuHasSME)) { |
6163 | | I422ToRGB565Row = I422ToRGB565Row_SME; |
6164 | | } |
6165 | | #endif |
6166 | | #if defined(HAS_I422TORGB565ROW_LSX) |
6167 | | if (TestCpuFlag(kCpuHasLSX)) { |
6168 | | I422ToRGB565Row = I422ToRGB565Row_Any_LSX; |
6169 | | if (IS_ALIGNED(width, 16)) { |
6170 | | I422ToRGB565Row = I422ToRGB565Row_LSX; |
6171 | | } |
6172 | | } |
6173 | | #endif |
6174 | | #if defined(HAS_I422TORGB565ROW_LASX) |
6175 | | if (TestCpuFlag(kCpuHasLASX)) { |
6176 | | I422ToRGB565Row = I422ToRGB565Row_Any_LASX; |
6177 | | if (IS_ALIGNED(width, 32)) { |
6178 | | I422ToRGB565Row = I422ToRGB565Row_LASX; |
6179 | | } |
6180 | | } |
6181 | | #endif |
6182 | | |
6183 | 1.20k | for (y = 0; y < height; ++y) { |
6184 | 1.13k | I422ToRGB565Row(src_y, src_u, src_v, dst_rgb565, yuvconstants, width); |
6185 | 1.13k | dst_rgb565 += dst_stride_rgb565; |
6186 | 1.13k | src_y += src_stride_y; |
6187 | 1.13k | src_u += src_stride_u; |
6188 | 1.13k | src_v += src_stride_v; |
6189 | 1.13k | } |
6190 | 71 | return 0; |
6191 | 71 | } |
6192 | | |
6193 | | // Convert I422 to RGB565. |
6194 | | LIBYUV_API |
6195 | | int I422ToRGB565(const uint8_t* src_y, |
6196 | | int src_stride_y, |
6197 | | const uint8_t* src_u, |
6198 | | int src_stride_u, |
6199 | | const uint8_t* src_v, |
6200 | | int src_stride_v, |
6201 | | uint8_t* dst_rgb565, |
6202 | | int dst_stride_rgb565, |
6203 | | int width, |
6204 | 0 | int height) { |
6205 | 0 | return I422ToRGB565Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6206 | 0 | src_stride_v, dst_rgb565, dst_stride_rgb565, |
6207 | 0 | &kYuvI601Constants, width, height); |
6208 | 0 | } |
6209 | | |
6210 | | // Ordered 8x8 dither for 888 to 565. Values from 0 to 7. |
6211 | | static const uint8_t kDither565_4x4[16] = { |
6212 | | 0, 4, 1, 5, 6, 2, 7, 3, 1, 5, 0, 4, 7, 3, 6, 2, |
6213 | | }; |
6214 | | |
6215 | | // Convert I420 to RGB565 with dithering. |
6216 | | LIBYUV_API |
6217 | | int I420ToRGB565Dither(const uint8_t* src_y, |
6218 | | int src_stride_y, |
6219 | | const uint8_t* src_u, |
6220 | | int src_stride_u, |
6221 | | const uint8_t* src_v, |
6222 | | int src_stride_v, |
6223 | | uint8_t* dst_rgb565, |
6224 | | int dst_stride_rgb565, |
6225 | | const uint8_t* dither4x4, |
6226 | | int width, |
6227 | 0 | int height) { |
6228 | 0 | int y; |
6229 | 0 | void (*I422ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
6230 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6231 | 0 | const struct YuvConstants* yuvconstants, int width) = |
6232 | 0 | I422ToARGBRow_C; |
6233 | 0 | void (*ARGBToRGB565DitherRow)(const uint8_t* src_argb, uint8_t* dst_rgb, |
6234 | 0 | uint32_t dither4, int width) = |
6235 | 0 | ARGBToRGB565DitherRow_C; |
6236 | 0 | if (!src_y || !src_u || !src_v || !dst_rgb565 || width <= 0 || height == 0) { |
6237 | 0 | return -1; |
6238 | 0 | } |
6239 | | // Negative height means invert the image. |
6240 | 0 | if (height < 0) { |
6241 | 0 | height = -height; |
6242 | 0 | dst_rgb565 = dst_rgb565 + (height - 1) * dst_stride_rgb565; |
6243 | 0 | dst_stride_rgb565 = -dst_stride_rgb565; |
6244 | 0 | } |
6245 | 0 | if (!dither4x4) { |
6246 | 0 | dither4x4 = kDither565_4x4; |
6247 | 0 | } |
6248 | 0 | #if defined(HAS_I422TOARGBROW_SSSE3) |
6249 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6250 | 0 | I422ToARGBRow = I422ToARGBRow_Any_SSSE3; |
6251 | 0 | if (IS_ALIGNED(width, 8)) { |
6252 | 0 | I422ToARGBRow = I422ToARGBRow_SSSE3; |
6253 | 0 | } |
6254 | 0 | } |
6255 | 0 | #endif |
6256 | 0 | #if defined(HAS_I422TOARGBROW_AVX2) |
6257 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6258 | 0 | I422ToARGBRow = I422ToARGBRow_Any_AVX2; |
6259 | 0 | if (IS_ALIGNED(width, 16)) { |
6260 | 0 | I422ToARGBRow = I422ToARGBRow_AVX2; |
6261 | 0 | } |
6262 | 0 | } |
6263 | 0 | #endif |
6264 | 0 | #if defined(HAS_I422TOARGBROW_AVX512BW) |
6265 | 0 | if (TestCpuFlag(kCpuHasAVX512BW | kCpuHasAVX512VL) == |
6266 | 0 | (kCpuHasAVX512BW | kCpuHasAVX512VL)) { |
6267 | 0 | I422ToARGBRow = I422ToARGBRow_Any_AVX512BW; |
6268 | 0 | if (IS_ALIGNED(width, 32)) { |
6269 | 0 | I422ToARGBRow = I422ToARGBRow_AVX512BW; |
6270 | 0 | } |
6271 | 0 | } |
6272 | 0 | #endif |
6273 | | #if defined(HAS_I422TOARGBROW_NEON) |
6274 | | if (TestCpuFlag(kCpuHasNEON)) { |
6275 | | I422ToARGBRow = I422ToARGBRow_Any_NEON; |
6276 | | if (IS_ALIGNED(width, 8)) { |
6277 | | I422ToARGBRow = I422ToARGBRow_NEON; |
6278 | | } |
6279 | | } |
6280 | | #endif |
6281 | | #if defined(HAS_I422TOARGBROW_SVE2) |
6282 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6283 | | I422ToARGBRow = I422ToARGBRow_SVE2; |
6284 | | } |
6285 | | #endif |
6286 | | #if defined(HAS_I422TOARGBROW_SME) |
6287 | | if (TestCpuFlag(kCpuHasSME)) { |
6288 | | I422ToARGBRow = I422ToARGBRow_SME; |
6289 | | } |
6290 | | #endif |
6291 | | #if defined(HAS_I422TOARGBROW_LSX) |
6292 | | if (TestCpuFlag(kCpuHasLSX)) { |
6293 | | I422ToARGBRow = I422ToARGBRow_Any_LSX; |
6294 | | if (IS_ALIGNED(width, 16)) { |
6295 | | I422ToARGBRow = I422ToARGBRow_LSX; |
6296 | | } |
6297 | | } |
6298 | | #endif |
6299 | | #if defined(HAS_I422TOARGBROW_LASX) |
6300 | | if (TestCpuFlag(kCpuHasLASX)) { |
6301 | | I422ToARGBRow = I422ToARGBRow_Any_LASX; |
6302 | | if (IS_ALIGNED(width, 32)) { |
6303 | | I422ToARGBRow = I422ToARGBRow_LASX; |
6304 | | } |
6305 | | } |
6306 | | #endif |
6307 | | #if defined(HAS_I422TOARGBROW_RVV) |
6308 | | if (TestCpuFlag(kCpuHasRVV)) { |
6309 | | I422ToARGBRow = I422ToARGBRow_RVV; |
6310 | | } |
6311 | | #endif |
6312 | 0 | #if defined(HAS_ARGBTORGB565DITHERROW_SSE2) |
6313 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
6314 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_SSE2; |
6315 | 0 | if (IS_ALIGNED(width, 4)) { |
6316 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_SSE2; |
6317 | 0 | } |
6318 | 0 | } |
6319 | 0 | #endif |
6320 | 0 | #if defined(HAS_ARGBTORGB565DITHERROW_AVX2) |
6321 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6322 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_AVX2; |
6323 | 0 | if (IS_ALIGNED(width, 8)) { |
6324 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_AVX2; |
6325 | 0 | } |
6326 | 0 | } |
6327 | 0 | #endif |
6328 | | #if defined(HAS_ARGBTORGB565DITHERROW_NEON) |
6329 | | if (TestCpuFlag(kCpuHasNEON)) { |
6330 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_NEON; |
6331 | | if (IS_ALIGNED(width, 8)) { |
6332 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_NEON; |
6333 | | } |
6334 | | } |
6335 | | #endif |
6336 | | #if defined(HAS_ARGBTORGB565DITHERROW_SVE2) |
6337 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6338 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_SVE2; |
6339 | | } |
6340 | | #endif |
6341 | | #if defined(HAS_ARGBTORGB565DITHERROW_LSX) |
6342 | | if (TestCpuFlag(kCpuHasLSX)) { |
6343 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LSX; |
6344 | | if (IS_ALIGNED(width, 8)) { |
6345 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LSX; |
6346 | | } |
6347 | | } |
6348 | | #endif |
6349 | | #if defined(HAS_ARGBTORGB565DITHERROW_LASX) |
6350 | | if (TestCpuFlag(kCpuHasLASX)) { |
6351 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LASX; |
6352 | | if (IS_ALIGNED(width, 16)) { |
6353 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LASX; |
6354 | | } |
6355 | | } |
6356 | | #endif |
6357 | 0 | { |
6358 | | // Allocate a row of argb. |
6359 | 0 | align_buffer_64(row_argb, width * 4); |
6360 | 0 | if (!row_argb) |
6361 | 0 | return 1; |
6362 | 0 | for (y = 0; y < height; ++y) { |
6363 | 0 | I422ToARGBRow(src_y, src_u, src_v, row_argb, &kYuvI601Constants, width); |
6364 | 0 | ARGBToRGB565DitherRow(row_argb, dst_rgb565, |
6365 | 0 | *(const uint32_t*)(dither4x4 + ((y & 3) << 2)), |
6366 | 0 | width); |
6367 | 0 | dst_rgb565 += dst_stride_rgb565; |
6368 | 0 | src_y += src_stride_y; |
6369 | 0 | if (y & 1) { |
6370 | 0 | src_u += src_stride_u; |
6371 | 0 | src_v += src_stride_v; |
6372 | 0 | } |
6373 | 0 | } |
6374 | 0 | free_aligned_buffer_64(row_argb); |
6375 | 0 | } |
6376 | 0 | return 0; |
6377 | 0 | } |
6378 | | |
6379 | | // Convert I420 to AR30 with matrix. |
6380 | | LIBYUV_API |
6381 | | int I420ToAR30Matrix(const uint8_t* src_y, |
6382 | | int src_stride_y, |
6383 | | const uint8_t* src_u, |
6384 | | int src_stride_u, |
6385 | | const uint8_t* src_v, |
6386 | | int src_stride_v, |
6387 | | uint8_t* dst_ar30, |
6388 | | int dst_stride_ar30, |
6389 | | const struct YuvConstants* yuvconstants, |
6390 | | int width, |
6391 | 0 | int height) { |
6392 | 0 | int y; |
6393 | 0 | void (*I422ToAR30Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6394 | 0 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6395 | 0 | const struct YuvConstants* yuvconstants, int width) = |
6396 | 0 | I422ToAR30Row_C; |
6397 | |
|
6398 | 0 | assert(yuvconstants); |
6399 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
6400 | 0 | return -1; |
6401 | 0 | } |
6402 | | // Negative height means invert the image. |
6403 | 0 | if (height < 0) { |
6404 | 0 | height = -height; |
6405 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
6406 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
6407 | 0 | } |
6408 | |
|
6409 | 0 | #if defined(HAS_I422TOAR30ROW_SSSE3) |
6410 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6411 | 0 | I422ToAR30Row = I422ToAR30Row_Any_SSSE3; |
6412 | 0 | if (IS_ALIGNED(width, 8)) { |
6413 | 0 | I422ToAR30Row = I422ToAR30Row_SSSE3; |
6414 | 0 | } |
6415 | 0 | } |
6416 | 0 | #endif |
6417 | 0 | #if defined(HAS_I422TOAR30ROW_AVX2) |
6418 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
6419 | 0 | I422ToAR30Row = I422ToAR30Row_Any_AVX2; |
6420 | 0 | if (IS_ALIGNED(width, 16)) { |
6421 | 0 | I422ToAR30Row = I422ToAR30Row_AVX2; |
6422 | 0 | } |
6423 | 0 | } |
6424 | 0 | #endif |
6425 | | #if defined(HAS_I422TOAR30ROW_NEON) |
6426 | | if (TestCpuFlag(kCpuHasNEON)) { |
6427 | | I422ToAR30Row = I422ToAR30Row_Any_NEON; |
6428 | | if (IS_ALIGNED(width, 8)) { |
6429 | | I422ToAR30Row = I422ToAR30Row_NEON; |
6430 | | } |
6431 | | } |
6432 | | #endif |
6433 | | #if defined(HAS_I422TOAR30ROW_SVE2) |
6434 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6435 | | I422ToAR30Row = I422ToAR30Row_SVE2; |
6436 | | } |
6437 | | #endif |
6438 | | #if defined(HAS_I422TOAR30ROW_SME) |
6439 | | if (TestCpuFlag(kCpuHasSME)) { |
6440 | | I422ToAR30Row = I422ToAR30Row_SME; |
6441 | | } |
6442 | | #endif |
6443 | |
|
6444 | 0 | for (y = 0; y < height; ++y) { |
6445 | 0 | I422ToAR30Row(src_y, src_u, src_v, dst_ar30, yuvconstants, width); |
6446 | 0 | dst_ar30 += dst_stride_ar30; |
6447 | 0 | src_y += src_stride_y; |
6448 | 0 | if (y & 1) { |
6449 | 0 | src_u += src_stride_u; |
6450 | 0 | src_v += src_stride_v; |
6451 | 0 | } |
6452 | 0 | } |
6453 | 0 | return 0; |
6454 | 0 | } |
6455 | | |
6456 | | // Convert I420 to AR30. |
6457 | | LIBYUV_API |
6458 | | int I420ToAR30(const uint8_t* src_y, |
6459 | | int src_stride_y, |
6460 | | const uint8_t* src_u, |
6461 | | int src_stride_u, |
6462 | | const uint8_t* src_v, |
6463 | | int src_stride_v, |
6464 | | uint8_t* dst_ar30, |
6465 | | int dst_stride_ar30, |
6466 | | int width, |
6467 | 0 | int height) { |
6468 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6469 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
6470 | 0 | &kYuvI601Constants, width, height); |
6471 | 0 | } |
6472 | | |
6473 | | // Convert H420 to AR30. |
6474 | | LIBYUV_API |
6475 | | int H420ToAR30(const uint8_t* src_y, |
6476 | | int src_stride_y, |
6477 | | const uint8_t* src_u, |
6478 | | int src_stride_u, |
6479 | | const uint8_t* src_v, |
6480 | | int src_stride_v, |
6481 | | uint8_t* dst_ar30, |
6482 | | int dst_stride_ar30, |
6483 | | int width, |
6484 | 0 | int height) { |
6485 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
6486 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
6487 | 0 | &kYvuH709Constants, width, height); |
6488 | 0 | } |
6489 | | |
6490 | | // Convert I420 to AB30. |
6491 | | LIBYUV_API |
6492 | | int I420ToAB30(const uint8_t* src_y, |
6493 | | int src_stride_y, |
6494 | | const uint8_t* src_u, |
6495 | | int src_stride_u, |
6496 | | const uint8_t* src_v, |
6497 | | int src_stride_v, |
6498 | | uint8_t* dst_ab30, |
6499 | | int dst_stride_ab30, |
6500 | | int width, |
6501 | 0 | int height) { |
6502 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
6503 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
6504 | 0 | &kYvuI601Constants, width, height); |
6505 | 0 | } |
6506 | | |
6507 | | // Convert H420 to AB30. |
6508 | | LIBYUV_API |
6509 | | int H420ToAB30(const uint8_t* src_y, |
6510 | | int src_stride_y, |
6511 | | const uint8_t* src_u, |
6512 | | int src_stride_u, |
6513 | | const uint8_t* src_v, |
6514 | | int src_stride_v, |
6515 | | uint8_t* dst_ab30, |
6516 | | int dst_stride_ab30, |
6517 | | int width, |
6518 | 0 | int height) { |
6519 | 0 | return I420ToAR30Matrix(src_y, src_stride_y, src_v, src_stride_v, src_u, |
6520 | 0 | src_stride_u, dst_ab30, dst_stride_ab30, |
6521 | 0 | &kYvuH709Constants, width, height); |
6522 | 0 | } |
6523 | | |
6524 | | static int I420ToARGBMatrixBilinear(const uint8_t* src_y, |
6525 | | int src_stride_y, |
6526 | | const uint8_t* src_u, |
6527 | | int src_stride_u, |
6528 | | const uint8_t* src_v, |
6529 | | int src_stride_v, |
6530 | | uint8_t* dst_argb, |
6531 | | int dst_stride_argb, |
6532 | | const struct YuvConstants* yuvconstants, |
6533 | | int width, |
6534 | 179 | int height) { |
6535 | 179 | int y; |
6536 | 179 | void (*I444ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
6537 | 179 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6538 | 179 | const struct YuvConstants* yuvconstants, int width) = |
6539 | 179 | I444ToARGBRow_C; |
6540 | 179 | void (*Scale2RowUp_Bilinear)(const uint8_t* src_ptr, ptrdiff_t src_stride, |
6541 | 179 | uint8_t* dst_ptr, ptrdiff_t dst_stride, |
6542 | 179 | int dst_width) = ScaleRowUp2_Bilinear_Any_C; |
6543 | 179 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
6544 | 179 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
6545 | 179 | assert(yuvconstants); |
6546 | 179 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
6547 | 0 | return -1; |
6548 | 0 | } |
6549 | | // Negative height means invert the image. |
6550 | 179 | if (height < 0) { |
6551 | 0 | height = -height; |
6552 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
6553 | 0 | dst_stride_argb = -dst_stride_argb; |
6554 | 0 | } |
6555 | 179 | #if defined(HAS_I444TOARGBROW_SSSE3) |
6556 | 179 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6557 | 179 | I444ToARGBRow = I444ToARGBRow_Any_SSSE3; |
6558 | 179 | if (IS_ALIGNED(width, 8)) { |
6559 | 8 | I444ToARGBRow = I444ToARGBRow_SSSE3; |
6560 | 8 | } |
6561 | 179 | } |
6562 | 179 | #endif |
6563 | 179 | #if defined(HAS_I444TOARGBROW_AVX2) |
6564 | 179 | if (TestCpuFlag(kCpuHasAVX2)) { |
6565 | 179 | I444ToARGBRow = I444ToARGBRow_Any_AVX2; |
6566 | 179 | if (IS_ALIGNED(width, 16)) { |
6567 | 3 | I444ToARGBRow = I444ToARGBRow_AVX2; |
6568 | 3 | } |
6569 | 179 | } |
6570 | 179 | #endif |
6571 | | #if defined(HAS_I444TOARGBROW_NEON) |
6572 | | if (TestCpuFlag(kCpuHasNEON)) { |
6573 | | I444ToARGBRow = I444ToARGBRow_Any_NEON; |
6574 | | if (IS_ALIGNED(width, 8)) { |
6575 | | I444ToARGBRow = I444ToARGBRow_NEON; |
6576 | | } |
6577 | | } |
6578 | | #endif |
6579 | | #if defined(HAS_I444TOARGBROW_SVE2) |
6580 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6581 | | I444ToARGBRow = I444ToARGBRow_SVE2; |
6582 | | } |
6583 | | #endif |
6584 | | #if defined(HAS_I444TOARGBROW_SME) |
6585 | | if (TestCpuFlag(kCpuHasSME)) { |
6586 | | I444ToARGBRow = I444ToARGBRow_SME; |
6587 | | } |
6588 | | #endif |
6589 | | #if defined(HAS_I444TOARGBROW_LASX) |
6590 | | if (TestCpuFlag(kCpuHasLASX)) { |
6591 | | I444ToARGBRow = I444ToARGBRow_Any_LASX; |
6592 | | if (IS_ALIGNED(width, 32)) { |
6593 | | I444ToARGBRow = I444ToARGBRow_LASX; |
6594 | | } |
6595 | | } |
6596 | | #endif |
6597 | | #if defined(HAS_I444TOARGBROW_RVV) |
6598 | | if (TestCpuFlag(kCpuHasRVV)) { |
6599 | | I444ToARGBRow = I444ToARGBRow_RVV; |
6600 | | } |
6601 | | #endif |
6602 | | |
6603 | 179 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSE2) |
6604 | 179 | if (TestCpuFlag(kCpuHasSSE2)) { |
6605 | 179 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSE2; |
6606 | 179 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
6607 | 179 | } |
6608 | 179 | #endif |
6609 | | |
6610 | 179 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSSE3) |
6611 | 179 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6612 | 179 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSSE3; |
6613 | 179 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
6614 | 179 | } |
6615 | 179 | #endif |
6616 | | |
6617 | 179 | #if defined(HAS_SCALEROWUP2_BILINEAR_AVX2) |
6618 | 179 | if (TestCpuFlag(kCpuHasAVX2)) { |
6619 | 179 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_AVX2; |
6620 | 179 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
6621 | 179 | } |
6622 | 179 | #endif |
6623 | | |
6624 | | #if defined(HAS_SCALEROWUP2_BILINEAR_NEON) |
6625 | | if (TestCpuFlag(kCpuHasNEON)) { |
6626 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_NEON; |
6627 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
6628 | | } |
6629 | | #endif |
6630 | | #if defined(HAS_SCALEROWUP2_BILINEAR_RVV) |
6631 | | if (TestCpuFlag(kCpuHasRVV)) { |
6632 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_RVV; |
6633 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
6634 | | } |
6635 | | #endif |
6636 | | |
6637 | | // alloc 4 lines temp |
6638 | 179 | const int row_size = (width + 31) & ~31; |
6639 | 179 | align_buffer_64(row, row_size * 4); |
6640 | 179 | uint8_t* temp_u_1 = row; |
6641 | 179 | uint8_t* temp_u_2 = row + row_size; |
6642 | 179 | uint8_t* temp_v_1 = row + row_size * 2; |
6643 | 179 | uint8_t* temp_v_2 = row + row_size * 3; |
6644 | 179 | if (!row) |
6645 | 0 | return 1; |
6646 | | |
6647 | 179 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
6648 | 179 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
6649 | 179 | I444ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
6650 | 179 | dst_argb += dst_stride_argb; |
6651 | 179 | src_y += src_stride_y; |
6652 | | |
6653 | 2.84k | for (y = 0; y < height - 2; y += 2) { |
6654 | 2.66k | Scale2RowUp_Bilinear(src_u, src_stride_u, temp_u_1, row_size, width); |
6655 | 2.66k | Scale2RowUp_Bilinear(src_v, src_stride_v, temp_v_1, row_size, width); |
6656 | 2.66k | I444ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
6657 | 2.66k | dst_argb += dst_stride_argb; |
6658 | 2.66k | src_y += src_stride_y; |
6659 | 2.66k | I444ToARGBRow(src_y, temp_u_2, temp_v_2, dst_argb, yuvconstants, width); |
6660 | 2.66k | dst_argb += dst_stride_argb; |
6661 | 2.66k | src_y += src_stride_y; |
6662 | 2.66k | src_u += src_stride_u; |
6663 | 2.66k | src_v += src_stride_v; |
6664 | 2.66k | } |
6665 | | |
6666 | 179 | if (!(height & 1)) { |
6667 | 46 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
6668 | 46 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
6669 | 46 | I444ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
6670 | 46 | } |
6671 | | |
6672 | 179 | free_aligned_buffer_64(row); |
6673 | 179 | return 0; |
6674 | 179 | } |
6675 | | |
6676 | | static int I422ToARGBMatrixLinear(const uint8_t* src_y, |
6677 | | int src_stride_y, |
6678 | | const uint8_t* src_u, |
6679 | | int src_stride_u, |
6680 | | const uint8_t* src_v, |
6681 | | int src_stride_v, |
6682 | | uint8_t* dst_argb, |
6683 | | int dst_stride_argb, |
6684 | | const struct YuvConstants* yuvconstants, |
6685 | | int width, |
6686 | 194 | int height) { |
6687 | 194 | int y; |
6688 | 194 | void (*I444ToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
6689 | 194 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6690 | 194 | const struct YuvConstants* yuvconstants, int width) = |
6691 | 194 | I444ToARGBRow_C; |
6692 | 194 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
6693 | 194 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
6694 | 194 | assert(yuvconstants); |
6695 | 194 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
6696 | 0 | return -1; |
6697 | 0 | } |
6698 | | // Negative height means invert the image. |
6699 | 194 | if (height < 0) { |
6700 | 0 | height = -height; |
6701 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
6702 | 0 | dst_stride_argb = -dst_stride_argb; |
6703 | 0 | } |
6704 | 194 | #if defined(HAS_I444TOARGBROW_SSSE3) |
6705 | 194 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6706 | 194 | I444ToARGBRow = I444ToARGBRow_Any_SSSE3; |
6707 | 194 | if (IS_ALIGNED(width, 8)) { |
6708 | 63 | I444ToARGBRow = I444ToARGBRow_SSSE3; |
6709 | 63 | } |
6710 | 194 | } |
6711 | 194 | #endif |
6712 | 194 | #if defined(HAS_I444TOARGBROW_AVX2) |
6713 | 194 | if (TestCpuFlag(kCpuHasAVX2)) { |
6714 | 194 | I444ToARGBRow = I444ToARGBRow_Any_AVX2; |
6715 | 194 | if (IS_ALIGNED(width, 16)) { |
6716 | 21 | I444ToARGBRow = I444ToARGBRow_AVX2; |
6717 | 21 | } |
6718 | 194 | } |
6719 | 194 | #endif |
6720 | | #if defined(HAS_I444TOARGBROW_NEON) |
6721 | | if (TestCpuFlag(kCpuHasNEON)) { |
6722 | | I444ToARGBRow = I444ToARGBRow_Any_NEON; |
6723 | | if (IS_ALIGNED(width, 8)) { |
6724 | | I444ToARGBRow = I444ToARGBRow_NEON; |
6725 | | } |
6726 | | } |
6727 | | #endif |
6728 | | #if defined(HAS_I444TOARGBROW_SVE2) |
6729 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6730 | | I444ToARGBRow = I444ToARGBRow_SVE2; |
6731 | | } |
6732 | | #endif |
6733 | | #if defined(HAS_I444TOARGBROW_SME) |
6734 | | if (TestCpuFlag(kCpuHasSME)) { |
6735 | | I444ToARGBRow = I444ToARGBRow_SME; |
6736 | | } |
6737 | | #endif |
6738 | | #if defined(HAS_I444TOARGBROW_LASX) |
6739 | | if (TestCpuFlag(kCpuHasLASX)) { |
6740 | | I444ToARGBRow = I444ToARGBRow_Any_LASX; |
6741 | | if (IS_ALIGNED(width, 32)) { |
6742 | | I444ToARGBRow = I444ToARGBRow_LASX; |
6743 | | } |
6744 | | } |
6745 | | #endif |
6746 | | #if defined(HAS_I444TOARGBROW_RVV) |
6747 | | if (TestCpuFlag(kCpuHasRVV)) { |
6748 | | I444ToARGBRow = I444ToARGBRow_RVV; |
6749 | | } |
6750 | | #endif |
6751 | 194 | #if defined(HAS_SCALEROWUP2_LINEAR_SSE2) |
6752 | 194 | if (TestCpuFlag(kCpuHasSSE2)) { |
6753 | 194 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
6754 | 194 | } |
6755 | 194 | #endif |
6756 | 194 | #if defined(HAS_SCALEROWUP2_LINEAR_SSSE3) |
6757 | 194 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6758 | 194 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
6759 | 194 | } |
6760 | 194 | #endif |
6761 | 194 | #if defined(HAS_SCALEROWUP2_LINEAR_AVX2) |
6762 | 194 | if (TestCpuFlag(kCpuHasAVX2)) { |
6763 | 194 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
6764 | 194 | } |
6765 | 194 | #endif |
6766 | | #if defined(HAS_SCALEROWUP2_LINEAR_NEON) |
6767 | | if (TestCpuFlag(kCpuHasNEON)) { |
6768 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
6769 | | } |
6770 | | #endif |
6771 | | #if defined(HAS_SCALEROWUP2_LINEAR_RVV) |
6772 | | if (TestCpuFlag(kCpuHasRVV)) { |
6773 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
6774 | | } |
6775 | | #endif |
6776 | | |
6777 | | // alloc 2 lines temp |
6778 | 194 | const int row_size = (width + 31) & ~31; |
6779 | 194 | align_buffer_64(row, row_size * 2); |
6780 | 194 | uint8_t* temp_u = row; |
6781 | 194 | uint8_t* temp_v = row + row_size; |
6782 | 194 | if (!row) |
6783 | 0 | return 1; |
6784 | | |
6785 | 7.50k | for (y = 0; y < height; ++y) { |
6786 | 7.31k | ScaleRowUp2_Linear(src_u, temp_u, width); |
6787 | 7.31k | ScaleRowUp2_Linear(src_v, temp_v, width); |
6788 | 7.31k | I444ToARGBRow(src_y, temp_u, temp_v, dst_argb, yuvconstants, width); |
6789 | 7.31k | dst_argb += dst_stride_argb; |
6790 | 7.31k | src_y += src_stride_y; |
6791 | 7.31k | src_u += src_stride_u; |
6792 | 7.31k | src_v += src_stride_v; |
6793 | 7.31k | } |
6794 | | |
6795 | 194 | free_aligned_buffer_64(row); |
6796 | 194 | return 0; |
6797 | 194 | } |
6798 | | |
6799 | | static int I420ToRGB24MatrixBilinear(const uint8_t* src_y, |
6800 | | int src_stride_y, |
6801 | | const uint8_t* src_u, |
6802 | | int src_stride_u, |
6803 | | const uint8_t* src_v, |
6804 | | int src_stride_v, |
6805 | | uint8_t* dst_rgb24, |
6806 | | int dst_stride_rgb24, |
6807 | | const struct YuvConstants* yuvconstants, |
6808 | | int width, |
6809 | 79 | int height) { |
6810 | 79 | int y; |
6811 | 79 | void (*I444ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
6812 | 79 | const uint8_t* v_buf, uint8_t* rgb_buf, |
6813 | 79 | const struct YuvConstants* yuvconstants, int width) = |
6814 | 79 | I444ToRGB24Row_C; |
6815 | 79 | void (*Scale2RowUp_Bilinear)(const uint8_t* src_ptr, ptrdiff_t src_stride, |
6816 | 79 | uint8_t* dst_ptr, ptrdiff_t dst_stride, |
6817 | 79 | int dst_width) = ScaleRowUp2_Bilinear_Any_C; |
6818 | 79 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
6819 | 79 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
6820 | 79 | assert(yuvconstants); |
6821 | 79 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
6822 | 0 | return -1; |
6823 | 0 | } |
6824 | | // Negative height means invert the image. |
6825 | 79 | if (height < 0) { |
6826 | 0 | height = -height; |
6827 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
6828 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
6829 | 0 | } |
6830 | 79 | #if defined(HAS_I444TORGB24ROW_SSSE3) |
6831 | 79 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6832 | 79 | I444ToRGB24Row = I444ToRGB24Row_Any_SSSE3; |
6833 | 79 | if (IS_ALIGNED(width, 16)) { |
6834 | 4 | I444ToRGB24Row = I444ToRGB24Row_SSSE3; |
6835 | 4 | } |
6836 | 79 | } |
6837 | 79 | #endif |
6838 | 79 | #if defined(HAS_I444TORGB24ROW_AVX2) |
6839 | 79 | if (TestCpuFlag(kCpuHasAVX2)) { |
6840 | 79 | I444ToRGB24Row = I444ToRGB24Row_Any_AVX2; |
6841 | 79 | if (IS_ALIGNED(width, 32)) { |
6842 | 3 | I444ToRGB24Row = I444ToRGB24Row_AVX2; |
6843 | 3 | } |
6844 | 79 | } |
6845 | 79 | #endif |
6846 | | #if defined(HAS_I444TORGB24ROW_NEON) |
6847 | | if (TestCpuFlag(kCpuHasNEON)) { |
6848 | | I444ToRGB24Row = I444ToRGB24Row_Any_NEON; |
6849 | | if (IS_ALIGNED(width, 8)) { |
6850 | | I444ToRGB24Row = I444ToRGB24Row_NEON; |
6851 | | } |
6852 | | } |
6853 | | #endif |
6854 | | #if defined(HAS_I444TORGB24ROW_SVE2) |
6855 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6856 | | I444ToRGB24Row = I444ToRGB24Row_SVE2; |
6857 | | } |
6858 | | #endif |
6859 | | #if defined(HAS_I444TORGB24ROW_SME) |
6860 | | if (TestCpuFlag(kCpuHasSME)) { |
6861 | | I444ToRGB24Row = I444ToRGB24Row_SME; |
6862 | | } |
6863 | | #endif |
6864 | | #if defined(HAS_I444TORGB24ROW_LASX) |
6865 | | if (TestCpuFlag(kCpuHasLASX)) { |
6866 | | I444ToRGB24Row = I444ToRGB24Row_Any_LASX; |
6867 | | if (IS_ALIGNED(width, 32)) { |
6868 | | I444ToRGB24Row = I444ToRGB24Row_LASX; |
6869 | | } |
6870 | | } |
6871 | | #endif |
6872 | | #if defined(HAS_I444TORGB24ROW_RVV) |
6873 | | if (TestCpuFlag(kCpuHasRVV)) { |
6874 | | I444ToRGB24Row = I444ToRGB24Row_RVV; |
6875 | | } |
6876 | | #endif |
6877 | | |
6878 | 79 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSE2) |
6879 | 79 | if (TestCpuFlag(kCpuHasSSE2)) { |
6880 | 79 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSE2; |
6881 | 79 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
6882 | 79 | } |
6883 | 79 | #endif |
6884 | | |
6885 | 79 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSSE3) |
6886 | 79 | if (TestCpuFlag(kCpuHasSSSE3)) { |
6887 | 79 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSSE3; |
6888 | 79 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
6889 | 79 | } |
6890 | 79 | #endif |
6891 | | |
6892 | 79 | #if defined(HAS_SCALEROWUP2_BILINEAR_AVX2) |
6893 | 79 | if (TestCpuFlag(kCpuHasAVX2)) { |
6894 | 79 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_AVX2; |
6895 | 79 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
6896 | 79 | } |
6897 | 79 | #endif |
6898 | | |
6899 | | #if defined(HAS_SCALEROWUP2_BILINEAR_NEON) |
6900 | | if (TestCpuFlag(kCpuHasNEON)) { |
6901 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_NEON; |
6902 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
6903 | | } |
6904 | | #endif |
6905 | | #if defined(HAS_SCALEROWUP2_BILINEAR_RVV) |
6906 | | if (TestCpuFlag(kCpuHasRVV)) { |
6907 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_RVV; |
6908 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
6909 | | } |
6910 | | #endif |
6911 | | |
6912 | | // alloc 4 lines temp |
6913 | 79 | const int row_size = (width + 31) & ~31; |
6914 | 79 | align_buffer_64(row, row_size * 4); |
6915 | 79 | uint8_t* temp_u_1 = row; |
6916 | 79 | uint8_t* temp_u_2 = row + row_size; |
6917 | 79 | uint8_t* temp_v_1 = row + row_size * 2; |
6918 | 79 | uint8_t* temp_v_2 = row + row_size * 3; |
6919 | 79 | if (!row) |
6920 | 0 | return 1; |
6921 | | |
6922 | 79 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
6923 | 79 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
6924 | 79 | I444ToRGB24Row(src_y, temp_u_1, temp_v_1, dst_rgb24, yuvconstants, width); |
6925 | 79 | dst_rgb24 += dst_stride_rgb24; |
6926 | 79 | src_y += src_stride_y; |
6927 | | |
6928 | 1.47k | for (y = 0; y < height - 2; y += 2) { |
6929 | 1.39k | Scale2RowUp_Bilinear(src_u, src_stride_u, temp_u_1, row_size, width); |
6930 | 1.39k | Scale2RowUp_Bilinear(src_v, src_stride_v, temp_v_1, row_size, width); |
6931 | 1.39k | I444ToRGB24Row(src_y, temp_u_1, temp_v_1, dst_rgb24, yuvconstants, width); |
6932 | 1.39k | dst_rgb24 += dst_stride_rgb24; |
6933 | 1.39k | src_y += src_stride_y; |
6934 | 1.39k | I444ToRGB24Row(src_y, temp_u_2, temp_v_2, dst_rgb24, yuvconstants, width); |
6935 | 1.39k | dst_rgb24 += dst_stride_rgb24; |
6936 | 1.39k | src_y += src_stride_y; |
6937 | 1.39k | src_u += src_stride_u; |
6938 | 1.39k | src_v += src_stride_v; |
6939 | 1.39k | } |
6940 | | |
6941 | 79 | if (!(height & 1)) { |
6942 | 26 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
6943 | 26 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
6944 | 26 | I444ToRGB24Row(src_y, temp_u_1, temp_v_1, dst_rgb24, yuvconstants, width); |
6945 | 26 | } |
6946 | | |
6947 | 79 | free_aligned_buffer_64(row); |
6948 | 79 | return 0; |
6949 | 79 | } |
6950 | | |
6951 | | static int I010ToAR30MatrixBilinear(const uint16_t* src_y, |
6952 | | int src_stride_y, |
6953 | | const uint16_t* src_u, |
6954 | | int src_stride_u, |
6955 | | const uint16_t* src_v, |
6956 | | int src_stride_v, |
6957 | | uint8_t* dst_ar30, |
6958 | | int dst_stride_ar30, |
6959 | | const struct YuvConstants* yuvconstants, |
6960 | | int width, |
6961 | 0 | int height) { |
6962 | 0 | int y; |
6963 | 0 | void (*I410ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
6964 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
6965 | 0 | const struct YuvConstants* yuvconstants, int width) = |
6966 | 0 | I410ToAR30Row_C; |
6967 | 0 | void (*Scale2RowUp_Bilinear_12)( |
6968 | 0 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
6969 | 0 | ptrdiff_t dst_stride, int dst_width) = ScaleRowUp2_Bilinear_16_Any_C; |
6970 | 0 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
6971 | 0 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
6972 | 0 | assert(yuvconstants); |
6973 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
6974 | 0 | return -1; |
6975 | 0 | } |
6976 | | // Negative height means invert the image. |
6977 | 0 | if (height < 0) { |
6978 | 0 | height = -height; |
6979 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
6980 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
6981 | 0 | } |
6982 | | #if defined(HAS_I410TOAR30ROW_NEON) |
6983 | | if (TestCpuFlag(kCpuHasNEON)) { |
6984 | | I410ToAR30Row = I410ToAR30Row_Any_NEON; |
6985 | | if (IS_ALIGNED(width, 8)) { |
6986 | | I410ToAR30Row = I410ToAR30Row_NEON; |
6987 | | } |
6988 | | } |
6989 | | #endif |
6990 | | #if defined(HAS_I410TOAR30ROW_SVE2) |
6991 | | if (TestCpuFlag(kCpuHasSVE2)) { |
6992 | | I410ToAR30Row = I410ToAR30Row_SVE2; |
6993 | | } |
6994 | | #endif |
6995 | | #if defined(HAS_I410TOAR30ROW_SME) |
6996 | | if (TestCpuFlag(kCpuHasSME)) { |
6997 | | I410ToAR30Row = I410ToAR30Row_SME; |
6998 | | } |
6999 | | #endif |
7000 | 0 | #if defined(HAS_I410TOAR30ROW_SSSE3) |
7001 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7002 | 0 | I410ToAR30Row = I410ToAR30Row_Any_SSSE3; |
7003 | 0 | if (IS_ALIGNED(width, 8)) { |
7004 | 0 | I410ToAR30Row = I410ToAR30Row_SSSE3; |
7005 | 0 | } |
7006 | 0 | } |
7007 | 0 | #endif |
7008 | 0 | #if defined(HAS_I410TOAR30ROW_AVX2) |
7009 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7010 | 0 | I410ToAR30Row = I410ToAR30Row_Any_AVX2; |
7011 | 0 | if (IS_ALIGNED(width, 16)) { |
7012 | 0 | I410ToAR30Row = I410ToAR30Row_AVX2; |
7013 | 0 | } |
7014 | 0 | } |
7015 | 0 | #endif |
7016 | |
|
7017 | 0 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_SSSE3) |
7018 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7019 | 0 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_SSSE3; |
7020 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7021 | 0 | } |
7022 | 0 | #endif |
7023 | |
|
7024 | 0 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_AVX2) |
7025 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7026 | 0 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_AVX2; |
7027 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7028 | 0 | } |
7029 | 0 | #endif |
7030 | |
|
7031 | | #if defined(HAS_SCALEROWUP2_BILINEAR_12_NEON) |
7032 | | if (TestCpuFlag(kCpuHasNEON)) { |
7033 | | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_NEON; |
7034 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7035 | | } |
7036 | | #endif |
7037 | | |
7038 | | // alloc 4 lines temp |
7039 | 0 | const int row_size = (width + 31) & ~31; |
7040 | 0 | align_buffer_64(row, row_size * 4 * sizeof(uint16_t)); |
7041 | 0 | uint16_t* temp_u_1 = (uint16_t*)(row); |
7042 | 0 | uint16_t* temp_u_2 = (uint16_t*)(row) + row_size; |
7043 | 0 | uint16_t* temp_v_1 = (uint16_t*)(row) + row_size * 2; |
7044 | 0 | uint16_t* temp_v_2 = (uint16_t*)(row) + row_size * 3; |
7045 | 0 | if (!row) |
7046 | 0 | return 1; |
7047 | | |
7048 | 0 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7049 | 0 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7050 | 0 | I410ToAR30Row(src_y, temp_u_1, temp_v_1, dst_ar30, yuvconstants, width); |
7051 | 0 | dst_ar30 += dst_stride_ar30; |
7052 | 0 | src_y += src_stride_y; |
7053 | |
|
7054 | 0 | for (y = 0; y < height - 2; y += 2) { |
7055 | 0 | Scale2RowUp_Bilinear_12(src_u, src_stride_u, temp_u_1, row_size, width); |
7056 | 0 | Scale2RowUp_Bilinear_12(src_v, src_stride_v, temp_v_1, row_size, width); |
7057 | 0 | I410ToAR30Row(src_y, temp_u_1, temp_v_1, dst_ar30, yuvconstants, width); |
7058 | 0 | dst_ar30 += dst_stride_ar30; |
7059 | 0 | src_y += src_stride_y; |
7060 | 0 | I410ToAR30Row(src_y, temp_u_2, temp_v_2, dst_ar30, yuvconstants, width); |
7061 | 0 | dst_ar30 += dst_stride_ar30; |
7062 | 0 | src_y += src_stride_y; |
7063 | 0 | src_u += src_stride_u; |
7064 | 0 | src_v += src_stride_v; |
7065 | 0 | } |
7066 | |
|
7067 | 0 | if (!(height & 1)) { |
7068 | 0 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7069 | 0 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7070 | 0 | I410ToAR30Row(src_y, temp_u_1, temp_v_1, dst_ar30, yuvconstants, width); |
7071 | 0 | } |
7072 | |
|
7073 | 0 | free_aligned_buffer_64(row); |
7074 | |
|
7075 | 0 | return 0; |
7076 | 0 | } |
7077 | | |
7078 | | static int I210ToAR30MatrixLinear(const uint16_t* src_y, |
7079 | | int src_stride_y, |
7080 | | const uint16_t* src_u, |
7081 | | int src_stride_u, |
7082 | | const uint16_t* src_v, |
7083 | | int src_stride_v, |
7084 | | uint8_t* dst_ar30, |
7085 | | int dst_stride_ar30, |
7086 | | const struct YuvConstants* yuvconstants, |
7087 | | int width, |
7088 | 0 | int height) { |
7089 | 0 | int y; |
7090 | 0 | void (*I410ToAR30Row)(const uint16_t* y_buf, const uint16_t* u_buf, |
7091 | 0 | const uint16_t* v_buf, uint8_t* rgb_buf, |
7092 | 0 | const struct YuvConstants* yuvconstants, int width) = |
7093 | 0 | I410ToAR30Row_C; |
7094 | 0 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7095 | 0 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7096 | 0 | assert(yuvconstants); |
7097 | 0 | if (!src_y || !src_u || !src_v || !dst_ar30 || width <= 0 || height == 0) { |
7098 | 0 | return -1; |
7099 | 0 | } |
7100 | | // Negative height means invert the image. |
7101 | 0 | if (height < 0) { |
7102 | 0 | height = -height; |
7103 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
7104 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
7105 | 0 | } |
7106 | | #if defined(HAS_I410TOAR30ROW_NEON) |
7107 | | if (TestCpuFlag(kCpuHasNEON)) { |
7108 | | I410ToAR30Row = I410ToAR30Row_Any_NEON; |
7109 | | if (IS_ALIGNED(width, 8)) { |
7110 | | I410ToAR30Row = I410ToAR30Row_NEON; |
7111 | | } |
7112 | | } |
7113 | | #endif |
7114 | | #if defined(HAS_I410TOAR30ROW_SVE2) |
7115 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7116 | | I410ToAR30Row = I410ToAR30Row_SVE2; |
7117 | | } |
7118 | | #endif |
7119 | | #if defined(HAS_I410TOAR30ROW_SME) |
7120 | | if (TestCpuFlag(kCpuHasSME)) { |
7121 | | I410ToAR30Row = I410ToAR30Row_SME; |
7122 | | } |
7123 | | #endif |
7124 | 0 | #if defined(HAS_I410TOAR30ROW_SSSE3) |
7125 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7126 | 0 | I410ToAR30Row = I410ToAR30Row_Any_SSSE3; |
7127 | 0 | if (IS_ALIGNED(width, 8)) { |
7128 | 0 | I410ToAR30Row = I410ToAR30Row_SSSE3; |
7129 | 0 | } |
7130 | 0 | } |
7131 | 0 | #endif |
7132 | 0 | #if defined(HAS_I410TOAR30ROW_AVX2) |
7133 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7134 | 0 | I410ToAR30Row = I410ToAR30Row_Any_AVX2; |
7135 | 0 | if (IS_ALIGNED(width, 16)) { |
7136 | 0 | I410ToAR30Row = I410ToAR30Row_AVX2; |
7137 | 0 | } |
7138 | 0 | } |
7139 | 0 | #endif |
7140 | |
|
7141 | 0 | #if defined(HAS_SCALEROWUP2_LINEAR_12_SSSE3) |
7142 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7143 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7144 | 0 | } |
7145 | 0 | #endif |
7146 | 0 | #if defined(HAS_SCALEROWUP2_LINEAR_12_AVX2) |
7147 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
7148 | 0 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7149 | 0 | } |
7150 | 0 | #endif |
7151 | | #if defined(HAS_SCALEROWUP2_LINEAR_12_NEON) |
7152 | | if (TestCpuFlag(kCpuHasNEON)) { |
7153 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7154 | | } |
7155 | | #endif |
7156 | | |
7157 | | // alloc 2 lines temp |
7158 | 0 | const int row_size = (width + 31) & ~31; |
7159 | 0 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
7160 | 0 | uint16_t* temp_u = (uint16_t*)(row); |
7161 | 0 | uint16_t* temp_v = (uint16_t*)(row) + row_size; |
7162 | 0 | if (!row) |
7163 | 0 | return 1; |
7164 | | |
7165 | 0 | for (y = 0; y < height; ++y) { |
7166 | 0 | ScaleRowUp2_Linear_12(src_u, temp_u, width); |
7167 | 0 | ScaleRowUp2_Linear_12(src_v, temp_v, width); |
7168 | 0 | I410ToAR30Row(src_y, temp_u, temp_v, dst_ar30, yuvconstants, width); |
7169 | 0 | dst_ar30 += dst_stride_ar30; |
7170 | 0 | src_y += src_stride_y; |
7171 | 0 | src_u += src_stride_u; |
7172 | 0 | src_v += src_stride_v; |
7173 | 0 | } |
7174 | 0 | free_aligned_buffer_64(row); |
7175 | 0 | return 0; |
7176 | 0 | } |
7177 | | |
7178 | | static int I010ToARGBMatrixBilinear(const uint16_t* src_y, |
7179 | | int src_stride_y, |
7180 | | const uint16_t* src_u, |
7181 | | int src_stride_u, |
7182 | | const uint16_t* src_v, |
7183 | | int src_stride_v, |
7184 | | uint8_t* dst_argb, |
7185 | | int dst_stride_argb, |
7186 | | const struct YuvConstants* yuvconstants, |
7187 | | int width, |
7188 | 94 | int height) { |
7189 | 94 | int y; |
7190 | 94 | void (*I410ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
7191 | 94 | const uint16_t* v_buf, uint8_t* rgb_buf, |
7192 | 94 | const struct YuvConstants* yuvconstants, int width) = |
7193 | 94 | I410ToARGBRow_C; |
7194 | 94 | void (*Scale2RowUp_Bilinear_12)( |
7195 | 94 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
7196 | 94 | ptrdiff_t dst_stride, int dst_width) = ScaleRowUp2_Bilinear_16_Any_C; |
7197 | 94 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7198 | 94 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7199 | 94 | assert(yuvconstants); |
7200 | 94 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
7201 | 0 | return -1; |
7202 | 0 | } |
7203 | | // Negative height means invert the image. |
7204 | 94 | if (height < 0) { |
7205 | 0 | height = -height; |
7206 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7207 | 0 | dst_stride_argb = -dst_stride_argb; |
7208 | 0 | } |
7209 | 94 | #if defined(HAS_I410TOARGBROW_SSSE3) |
7210 | 94 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7211 | 94 | I410ToARGBRow = I410ToARGBRow_Any_SSSE3; |
7212 | 94 | if (IS_ALIGNED(width, 8)) { |
7213 | 5 | I410ToARGBRow = I410ToARGBRow_SSSE3; |
7214 | 5 | } |
7215 | 94 | } |
7216 | 94 | #endif |
7217 | | #if defined(HAS_I410TOARGBROW_NEON) |
7218 | | if (TestCpuFlag(kCpuHasNEON)) { |
7219 | | I410ToARGBRow = I410ToARGBRow_Any_NEON; |
7220 | | if (IS_ALIGNED(width, 8)) { |
7221 | | I410ToARGBRow = I410ToARGBRow_NEON; |
7222 | | } |
7223 | | } |
7224 | | #endif |
7225 | | #if defined(HAS_I410TOARGBROW_SVE2) |
7226 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7227 | | I410ToARGBRow = I410ToARGBRow_SVE2; |
7228 | | } |
7229 | | #endif |
7230 | | #if defined(HAS_I410TOARGBROW_SME) |
7231 | | if (TestCpuFlag(kCpuHasSME)) { |
7232 | | I410ToARGBRow = I410ToARGBRow_SME; |
7233 | | } |
7234 | | #endif |
7235 | 94 | #if defined(HAS_I410TOARGBROW_AVX2) |
7236 | 94 | if (TestCpuFlag(kCpuHasAVX2)) { |
7237 | 94 | I410ToARGBRow = I410ToARGBRow_Any_AVX2; |
7238 | 94 | if (IS_ALIGNED(width, 16)) { |
7239 | 3 | I410ToARGBRow = I410ToARGBRow_AVX2; |
7240 | 3 | } |
7241 | 94 | } |
7242 | 94 | #endif |
7243 | | |
7244 | 94 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_SSSE3) |
7245 | 94 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7246 | 94 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_SSSE3; |
7247 | 94 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7248 | 94 | } |
7249 | 94 | #endif |
7250 | | |
7251 | 94 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_AVX2) |
7252 | 94 | if (TestCpuFlag(kCpuHasAVX2)) { |
7253 | 94 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_AVX2; |
7254 | 94 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7255 | 94 | } |
7256 | 94 | #endif |
7257 | | |
7258 | | #if defined(HAS_SCALEROWUP2_BILINEAR_12_NEON) |
7259 | | if (TestCpuFlag(kCpuHasNEON)) { |
7260 | | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_NEON; |
7261 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7262 | | } |
7263 | | #endif |
7264 | | |
7265 | | // alloc 4 lines temp |
7266 | 94 | const int row_size = (width + 31) & ~31; |
7267 | 94 | align_buffer_64(row, row_size * 4 * sizeof(uint16_t)); |
7268 | 94 | uint16_t* temp_u_1 = (uint16_t*)(row); |
7269 | 94 | uint16_t* temp_u_2 = (uint16_t*)(row) + row_size; |
7270 | 94 | uint16_t* temp_v_1 = (uint16_t*)(row) + row_size * 2; |
7271 | 94 | uint16_t* temp_v_2 = (uint16_t*)(row) + row_size * 3; |
7272 | 94 | if (!row) |
7273 | 0 | return 1; |
7274 | | |
7275 | 94 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7276 | 94 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7277 | 94 | I410ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
7278 | 94 | dst_argb += dst_stride_argb; |
7279 | 94 | src_y += src_stride_y; |
7280 | | |
7281 | 1.31k | for (y = 0; y < height - 2; y += 2) { |
7282 | 1.21k | Scale2RowUp_Bilinear_12(src_u, src_stride_u, temp_u_1, row_size, width); |
7283 | 1.21k | Scale2RowUp_Bilinear_12(src_v, src_stride_v, temp_v_1, row_size, width); |
7284 | 1.21k | I410ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
7285 | 1.21k | dst_argb += dst_stride_argb; |
7286 | 1.21k | src_y += src_stride_y; |
7287 | 1.21k | I410ToARGBRow(src_y, temp_u_2, temp_v_2, dst_argb, yuvconstants, width); |
7288 | 1.21k | dst_argb += dst_stride_argb; |
7289 | 1.21k | src_y += src_stride_y; |
7290 | 1.21k | src_u += src_stride_u; |
7291 | 1.21k | src_v += src_stride_v; |
7292 | 1.21k | } |
7293 | | |
7294 | 94 | if (!(height & 1)) { |
7295 | 29 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7296 | 29 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7297 | 29 | I410ToARGBRow(src_y, temp_u_1, temp_v_1, dst_argb, yuvconstants, width); |
7298 | 29 | } |
7299 | | |
7300 | 94 | free_aligned_buffer_64(row); |
7301 | 94 | return 0; |
7302 | 94 | } |
7303 | | |
7304 | | static int I210ToARGBMatrixLinear(const uint16_t* src_y, |
7305 | | int src_stride_y, |
7306 | | const uint16_t* src_u, |
7307 | | int src_stride_u, |
7308 | | const uint16_t* src_v, |
7309 | | int src_stride_v, |
7310 | | uint8_t* dst_argb, |
7311 | | int dst_stride_argb, |
7312 | | const struct YuvConstants* yuvconstants, |
7313 | | int width, |
7314 | 138 | int height) { |
7315 | 138 | int y; |
7316 | 138 | void (*I410ToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
7317 | 138 | const uint16_t* v_buf, uint8_t* rgb_buf, |
7318 | 138 | const struct YuvConstants* yuvconstants, int width) = |
7319 | 138 | I410ToARGBRow_C; |
7320 | 138 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7321 | 138 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7322 | 138 | assert(yuvconstants); |
7323 | 138 | if (!src_y || !src_u || !src_v || !dst_argb || width <= 0 || height == 0) { |
7324 | 0 | return -1; |
7325 | 0 | } |
7326 | | // Negative height means invert the image. |
7327 | 138 | if (height < 0) { |
7328 | 0 | height = -height; |
7329 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7330 | 0 | dst_stride_argb = -dst_stride_argb; |
7331 | 0 | } |
7332 | 138 | #if defined(HAS_I410TOARGBROW_SSSE3) |
7333 | 138 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7334 | 138 | I410ToARGBRow = I410ToARGBRow_Any_SSSE3; |
7335 | 138 | if (IS_ALIGNED(width, 8)) { |
7336 | 34 | I410ToARGBRow = I410ToARGBRow_SSSE3; |
7337 | 34 | } |
7338 | 138 | } |
7339 | 138 | #endif |
7340 | | #if defined(HAS_I410TOARGBROW_NEON) |
7341 | | if (TestCpuFlag(kCpuHasNEON)) { |
7342 | | I410ToARGBRow = I410ToARGBRow_Any_NEON; |
7343 | | if (IS_ALIGNED(width, 8)) { |
7344 | | I410ToARGBRow = I410ToARGBRow_NEON; |
7345 | | } |
7346 | | } |
7347 | | #endif |
7348 | | #if defined(HAS_I410TOARGBROW_SVE2) |
7349 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7350 | | I410ToARGBRow = I410ToARGBRow_SVE2; |
7351 | | } |
7352 | | #endif |
7353 | | #if defined(HAS_I410TOARGBROW_SME) |
7354 | | if (TestCpuFlag(kCpuHasSME)) { |
7355 | | I410ToARGBRow = I410ToARGBRow_SME; |
7356 | | } |
7357 | | #endif |
7358 | 138 | #if defined(HAS_I410TOARGBROW_AVX2) |
7359 | 138 | if (TestCpuFlag(kCpuHasAVX2)) { |
7360 | 138 | I410ToARGBRow = I410ToARGBRow_Any_AVX2; |
7361 | 138 | if (IS_ALIGNED(width, 16)) { |
7362 | 17 | I410ToARGBRow = I410ToARGBRow_AVX2; |
7363 | 17 | } |
7364 | 138 | } |
7365 | 138 | #endif |
7366 | | |
7367 | 138 | #if defined(HAS_SCALEROWUP2_LINEAR_12_SSSE3) |
7368 | 138 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7369 | 138 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7370 | 138 | } |
7371 | 138 | #endif |
7372 | 138 | #if defined(HAS_SCALEROWUP2_LINEAR_12_AVX2) |
7373 | 138 | if (TestCpuFlag(kCpuHasAVX2)) { |
7374 | 138 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7375 | 138 | } |
7376 | 138 | #endif |
7377 | | #if defined(HAS_SCALEROWUP2_LINEAR_12_NEON) |
7378 | | if (TestCpuFlag(kCpuHasNEON)) { |
7379 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7380 | | } |
7381 | | #endif |
7382 | | |
7383 | | // alloc 2 lines temp |
7384 | 138 | const int row_size = (width + 31) & ~31; |
7385 | 138 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
7386 | 138 | uint16_t* temp_u = (uint16_t*)(row); |
7387 | 138 | uint16_t* temp_v = (uint16_t*)(row) + row_size; |
7388 | 138 | if (!row) |
7389 | 0 | return 1; |
7390 | | |
7391 | 4.66k | for (y = 0; y < height; ++y) { |
7392 | 4.52k | ScaleRowUp2_Linear_12(src_u, temp_u, width); |
7393 | 4.52k | ScaleRowUp2_Linear_12(src_v, temp_v, width); |
7394 | 4.52k | I410ToARGBRow(src_y, temp_u, temp_v, dst_argb, yuvconstants, width); |
7395 | 4.52k | dst_argb += dst_stride_argb; |
7396 | 4.52k | src_y += src_stride_y; |
7397 | 4.52k | src_u += src_stride_u; |
7398 | 4.52k | src_v += src_stride_v; |
7399 | 4.52k | } |
7400 | | |
7401 | 138 | free_aligned_buffer_64(row); |
7402 | 138 | return 0; |
7403 | 138 | } |
7404 | | |
7405 | | static int I420AlphaToARGBMatrixBilinear( |
7406 | | const uint8_t* src_y, |
7407 | | int src_stride_y, |
7408 | | const uint8_t* src_u, |
7409 | | int src_stride_u, |
7410 | | const uint8_t* src_v, |
7411 | | int src_stride_v, |
7412 | | const uint8_t* src_a, |
7413 | | int src_stride_a, |
7414 | | uint8_t* dst_argb, |
7415 | | int dst_stride_argb, |
7416 | | const struct YuvConstants* yuvconstants, |
7417 | | int width, |
7418 | | int height, |
7419 | 56 | int attenuate) { |
7420 | 56 | int y; |
7421 | 56 | void (*I444AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
7422 | 56 | const uint8_t* v_buf, const uint8_t* a_buf, |
7423 | 56 | uint8_t* dst_argb, |
7424 | 56 | const struct YuvConstants* yuvconstants, |
7425 | 56 | int width) = I444AlphaToARGBRow_C; |
7426 | 56 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
7427 | 56 | int width) = ARGBAttenuateRow_C; |
7428 | 56 | void (*Scale2RowUp_Bilinear)(const uint8_t* src_ptr, ptrdiff_t src_stride, |
7429 | 56 | uint8_t* dst_ptr, ptrdiff_t dst_stride, |
7430 | 56 | int dst_width) = ScaleRowUp2_Bilinear_Any_C; |
7431 | 56 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
7432 | 56 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
7433 | 56 | assert(yuvconstants); |
7434 | 56 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
7435 | 56 | height == 0) { |
7436 | 0 | return -1; |
7437 | 0 | } |
7438 | | // Negative height means invert the image. |
7439 | 56 | if (height < 0) { |
7440 | 0 | height = -height; |
7441 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7442 | 0 | dst_stride_argb = -dst_stride_argb; |
7443 | 0 | } |
7444 | 56 | #if defined(HAS_I444ALPHATOARGBROW_SSSE3) |
7445 | 56 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7446 | 56 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_SSSE3; |
7447 | 56 | if (IS_ALIGNED(width, 8)) { |
7448 | 5 | I444AlphaToARGBRow = I444AlphaToARGBRow_SSSE3; |
7449 | 5 | } |
7450 | 56 | } |
7451 | 56 | #endif |
7452 | 56 | #if defined(HAS_I444ALPHATOARGBROW_AVX2) |
7453 | 56 | if (TestCpuFlag(kCpuHasAVX2)) { |
7454 | 56 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_AVX2; |
7455 | 56 | if (IS_ALIGNED(width, 16)) { |
7456 | 3 | I444AlphaToARGBRow = I444AlphaToARGBRow_AVX2; |
7457 | 3 | } |
7458 | 56 | } |
7459 | 56 | #endif |
7460 | | #if defined(HAS_I444ALPHATOARGBROW_NEON) |
7461 | | if (TestCpuFlag(kCpuHasNEON)) { |
7462 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_NEON; |
7463 | | if (IS_ALIGNED(width, 8)) { |
7464 | | I444AlphaToARGBRow = I444AlphaToARGBRow_NEON; |
7465 | | } |
7466 | | } |
7467 | | #endif |
7468 | | #if defined(HAS_I444ALPHATOARGBROW_SVE2) |
7469 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7470 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SVE2; |
7471 | | } |
7472 | | #endif |
7473 | | #if defined(HAS_I444ALPHATOARGBROW_SME) |
7474 | | if (TestCpuFlag(kCpuHasSME)) { |
7475 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SME; |
7476 | | } |
7477 | | #endif |
7478 | | #if defined(HAS_I444ALPHATOARGBROW_LASX) |
7479 | | if (TestCpuFlag(kCpuHasLASX)) { |
7480 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_LASX; |
7481 | | if (IS_ALIGNED(width, 16)) { |
7482 | | I444AlphaToARGBRow = I444AlphaToARGBRow_LASX; |
7483 | | } |
7484 | | } |
7485 | | #endif |
7486 | | #if defined(HAS_I444ALPHATOARGBROW_RVV) |
7487 | | if (TestCpuFlag(kCpuHasRVV)) { |
7488 | | I444AlphaToARGBRow = I444AlphaToARGBRow_RVV; |
7489 | | } |
7490 | | #endif |
7491 | 56 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
7492 | 56 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7493 | 56 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
7494 | 56 | if (IS_ALIGNED(width, 4)) { |
7495 | 6 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
7496 | 6 | } |
7497 | 56 | } |
7498 | 56 | #endif |
7499 | 56 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
7500 | 56 | if (TestCpuFlag(kCpuHasAVX2)) { |
7501 | 56 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
7502 | 56 | if (IS_ALIGNED(width, 8)) { |
7503 | 5 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
7504 | 5 | } |
7505 | 56 | } |
7506 | 56 | #endif |
7507 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
7508 | | if (TestCpuFlag(kCpuHasNEON)) { |
7509 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
7510 | | if (IS_ALIGNED(width, 8)) { |
7511 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
7512 | | } |
7513 | | } |
7514 | | #endif |
7515 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
7516 | | if (TestCpuFlag(kCpuHasRVV)) { |
7517 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
7518 | | } |
7519 | | #endif |
7520 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
7521 | | if (TestCpuFlag(kCpuHasLSX)) { |
7522 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
7523 | | if (IS_ALIGNED(width, 8)) { |
7524 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
7525 | | } |
7526 | | } |
7527 | | #endif |
7528 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
7529 | | if (TestCpuFlag(kCpuHasLASX)) { |
7530 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
7531 | | if (IS_ALIGNED(width, 16)) { |
7532 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
7533 | | } |
7534 | | } |
7535 | | #endif |
7536 | | |
7537 | 56 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSE2) |
7538 | 56 | if (TestCpuFlag(kCpuHasSSE2)) { |
7539 | 56 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSE2; |
7540 | 56 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
7541 | 56 | } |
7542 | 56 | #endif |
7543 | | |
7544 | 56 | #if defined(HAS_SCALEROWUP2_BILINEAR_SSSE3) |
7545 | 56 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7546 | 56 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_SSSE3; |
7547 | 56 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
7548 | 56 | } |
7549 | 56 | #endif |
7550 | | |
7551 | 56 | #if defined(HAS_SCALEROWUP2_BILINEAR_AVX2) |
7552 | 56 | if (TestCpuFlag(kCpuHasAVX2)) { |
7553 | 56 | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_AVX2; |
7554 | 56 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
7555 | 56 | } |
7556 | 56 | #endif |
7557 | | |
7558 | | #if defined(HAS_SCALEROWUP2_BILINEAR_NEON) |
7559 | | if (TestCpuFlag(kCpuHasNEON)) { |
7560 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_Any_NEON; |
7561 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
7562 | | } |
7563 | | #endif |
7564 | | #if defined(HAS_SCALEROWUP2_BILINEAR_RVV) |
7565 | | if (TestCpuFlag(kCpuHasRVV)) { |
7566 | | Scale2RowUp_Bilinear = ScaleRowUp2_Bilinear_RVV; |
7567 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
7568 | | } |
7569 | | #endif |
7570 | | |
7571 | | // alloc 4 lines temp |
7572 | 56 | const int row_size = (width + 31) & ~31; |
7573 | 56 | align_buffer_64(row, row_size * 4); |
7574 | 56 | uint8_t* temp_u_1 = row; |
7575 | 56 | uint8_t* temp_u_2 = row + row_size; |
7576 | 56 | uint8_t* temp_v_1 = row + row_size * 2; |
7577 | 56 | uint8_t* temp_v_2 = row + row_size * 3; |
7578 | 56 | if (!row) |
7579 | 0 | return 1; |
7580 | | |
7581 | 56 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
7582 | 56 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
7583 | 56 | I444AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7584 | 56 | width); |
7585 | 56 | if (attenuate) { |
7586 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7587 | 0 | } |
7588 | 56 | dst_argb += dst_stride_argb; |
7589 | 56 | src_y += src_stride_y; |
7590 | 56 | src_a += src_stride_a; |
7591 | | |
7592 | 1.36k | for (y = 0; y < height - 2; y += 2) { |
7593 | 1.30k | Scale2RowUp_Bilinear(src_u, src_stride_u, temp_u_1, row_size, width); |
7594 | 1.30k | Scale2RowUp_Bilinear(src_v, src_stride_v, temp_v_1, row_size, width); |
7595 | 1.30k | I444AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7596 | 1.30k | width); |
7597 | 1.30k | if (attenuate) { |
7598 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7599 | 0 | } |
7600 | 1.30k | dst_argb += dst_stride_argb; |
7601 | 1.30k | src_y += src_stride_y; |
7602 | 1.30k | src_a += src_stride_a; |
7603 | 1.30k | I444AlphaToARGBRow(src_y, temp_u_2, temp_v_2, src_a, dst_argb, yuvconstants, |
7604 | 1.30k | width); |
7605 | 1.30k | if (attenuate) { |
7606 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7607 | 0 | } |
7608 | 1.30k | dst_argb += dst_stride_argb; |
7609 | 1.30k | src_y += src_stride_y; |
7610 | 1.30k | src_u += src_stride_u; |
7611 | 1.30k | src_v += src_stride_v; |
7612 | 1.30k | src_a += src_stride_a; |
7613 | 1.30k | } |
7614 | | |
7615 | 56 | if (!(height & 1)) { |
7616 | 22 | ScaleRowUp2_Linear(src_u, temp_u_1, width); |
7617 | 22 | ScaleRowUp2_Linear(src_v, temp_v_1, width); |
7618 | 22 | I444AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7619 | 22 | width); |
7620 | 22 | if (attenuate) { |
7621 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7622 | 0 | } |
7623 | 22 | } |
7624 | | |
7625 | 56 | free_aligned_buffer_64(row); |
7626 | 56 | return 0; |
7627 | 56 | } |
7628 | | |
7629 | | static int I422AlphaToARGBMatrixLinear(const uint8_t* src_y, |
7630 | | int src_stride_y, |
7631 | | const uint8_t* src_u, |
7632 | | int src_stride_u, |
7633 | | const uint8_t* src_v, |
7634 | | int src_stride_v, |
7635 | | const uint8_t* src_a, |
7636 | | int src_stride_a, |
7637 | | uint8_t* dst_argb, |
7638 | | int dst_stride_argb, |
7639 | | const struct YuvConstants* yuvconstants, |
7640 | | int width, |
7641 | | int height, |
7642 | 130 | int attenuate) { |
7643 | 130 | int y; |
7644 | 130 | void (*I444AlphaToARGBRow)(const uint8_t* y_buf, const uint8_t* u_buf, |
7645 | 130 | const uint8_t* v_buf, const uint8_t* a_buf, |
7646 | 130 | uint8_t* dst_argb, |
7647 | 130 | const struct YuvConstants* yuvconstants, |
7648 | 130 | int width) = I444AlphaToARGBRow_C; |
7649 | 130 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
7650 | 130 | int width) = ARGBAttenuateRow_C; |
7651 | 130 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
7652 | 130 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
7653 | 130 | assert(yuvconstants); |
7654 | 130 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
7655 | 130 | height == 0) { |
7656 | 0 | return -1; |
7657 | 0 | } |
7658 | | // Negative height means invert the image. |
7659 | 130 | if (height < 0) { |
7660 | 0 | height = -height; |
7661 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7662 | 0 | dst_stride_argb = -dst_stride_argb; |
7663 | 0 | } |
7664 | 130 | #if defined(HAS_I444ALPHATOARGBROW_SSSE3) |
7665 | 130 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7666 | 130 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_SSSE3; |
7667 | 130 | if (IS_ALIGNED(width, 8)) { |
7668 | 34 | I444AlphaToARGBRow = I444AlphaToARGBRow_SSSE3; |
7669 | 34 | } |
7670 | 130 | } |
7671 | 130 | #endif |
7672 | 130 | #if defined(HAS_I444ALPHATOARGBROW_AVX2) |
7673 | 130 | if (TestCpuFlag(kCpuHasAVX2)) { |
7674 | 130 | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_AVX2; |
7675 | 130 | if (IS_ALIGNED(width, 16)) { |
7676 | 17 | I444AlphaToARGBRow = I444AlphaToARGBRow_AVX2; |
7677 | 17 | } |
7678 | 130 | } |
7679 | 130 | #endif |
7680 | | #if defined(HAS_I444ALPHATOARGBROW_NEON) |
7681 | | if (TestCpuFlag(kCpuHasNEON)) { |
7682 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_NEON; |
7683 | | if (IS_ALIGNED(width, 8)) { |
7684 | | I444AlphaToARGBRow = I444AlphaToARGBRow_NEON; |
7685 | | } |
7686 | | } |
7687 | | #endif |
7688 | | #if defined(HAS_I444ALPHATOARGBROW_SVE2) |
7689 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7690 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SVE2; |
7691 | | } |
7692 | | #endif |
7693 | | #if defined(HAS_I444ALPHATOARGBROW_SME) |
7694 | | if (TestCpuFlag(kCpuHasSME)) { |
7695 | | I444AlphaToARGBRow = I444AlphaToARGBRow_SME; |
7696 | | } |
7697 | | #endif |
7698 | | #if defined(HAS_I444ALPHATOARGBROW_LASX) |
7699 | | if (TestCpuFlag(kCpuHasLASX)) { |
7700 | | I444AlphaToARGBRow = I444AlphaToARGBRow_Any_LASX; |
7701 | | if (IS_ALIGNED(width, 16)) { |
7702 | | I444AlphaToARGBRow = I444AlphaToARGBRow_LASX; |
7703 | | } |
7704 | | } |
7705 | | #endif |
7706 | | #if defined(HAS_I444ALPHATOARGBROW_RVV) |
7707 | | if (TestCpuFlag(kCpuHasRVV)) { |
7708 | | I444AlphaToARGBRow = I444AlphaToARGBRow_RVV; |
7709 | | } |
7710 | | #endif |
7711 | 130 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
7712 | 130 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7713 | 130 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
7714 | 130 | if (IS_ALIGNED(width, 4)) { |
7715 | 51 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
7716 | 51 | } |
7717 | 130 | } |
7718 | 130 | #endif |
7719 | 130 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
7720 | 130 | if (TestCpuFlag(kCpuHasAVX2)) { |
7721 | 130 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
7722 | 130 | if (IS_ALIGNED(width, 8)) { |
7723 | 34 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
7724 | 34 | } |
7725 | 130 | } |
7726 | 130 | #endif |
7727 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
7728 | | if (TestCpuFlag(kCpuHasNEON)) { |
7729 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
7730 | | if (IS_ALIGNED(width, 8)) { |
7731 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
7732 | | } |
7733 | | } |
7734 | | #endif |
7735 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
7736 | | if (TestCpuFlag(kCpuHasRVV)) { |
7737 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
7738 | | } |
7739 | | #endif |
7740 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
7741 | | if (TestCpuFlag(kCpuHasLSX)) { |
7742 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
7743 | | if (IS_ALIGNED(width, 8)) { |
7744 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
7745 | | } |
7746 | | } |
7747 | | #endif |
7748 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
7749 | | if (TestCpuFlag(kCpuHasLASX)) { |
7750 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
7751 | | if (IS_ALIGNED(width, 16)) { |
7752 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
7753 | | } |
7754 | | } |
7755 | | #endif |
7756 | | |
7757 | 130 | #if defined(HAS_SCALEROWUP2_LINEAR_SSE2) |
7758 | 130 | if (TestCpuFlag(kCpuHasSSE2)) { |
7759 | 130 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
7760 | 130 | } |
7761 | 130 | #endif |
7762 | 130 | #if defined(HAS_SCALEROWUP2_LINEAR_SSSE3) |
7763 | 130 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7764 | 130 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
7765 | 130 | } |
7766 | 130 | #endif |
7767 | 130 | #if defined(HAS_SCALEROWUP2_LINEAR_AVX2) |
7768 | 130 | if (TestCpuFlag(kCpuHasAVX2)) { |
7769 | 130 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
7770 | 130 | } |
7771 | 130 | #endif |
7772 | | #if defined(HAS_SCALEROWUP2_LINEAR_NEON) |
7773 | | if (TestCpuFlag(kCpuHasNEON)) { |
7774 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
7775 | | } |
7776 | | #endif |
7777 | | #if defined(HAS_SCALEROWUP2_LINEAR_RVV) |
7778 | | if (TestCpuFlag(kCpuHasRVV)) { |
7779 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
7780 | | } |
7781 | | #endif |
7782 | | |
7783 | | // alloc 2 lines temp |
7784 | 130 | const int row_size = (width + 31) & ~31; |
7785 | 130 | align_buffer_64(row, row_size * 2); |
7786 | 130 | uint8_t* temp_u = row; |
7787 | 130 | uint8_t* temp_v = row + row_size; |
7788 | 130 | if (!row) |
7789 | 0 | return 1; |
7790 | | |
7791 | 2.74k | for (y = 0; y < height; ++y) { |
7792 | 2.61k | ScaleRowUp2_Linear(src_u, temp_u, width); |
7793 | 2.61k | ScaleRowUp2_Linear(src_v, temp_v, width); |
7794 | 2.61k | I444AlphaToARGBRow(src_y, temp_u, temp_v, src_a, dst_argb, yuvconstants, |
7795 | 2.61k | width); |
7796 | 2.61k | if (attenuate) { |
7797 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7798 | 0 | } |
7799 | 2.61k | dst_argb += dst_stride_argb; |
7800 | 2.61k | src_a += src_stride_a; |
7801 | 2.61k | src_y += src_stride_y; |
7802 | 2.61k | src_u += src_stride_u; |
7803 | 2.61k | src_v += src_stride_v; |
7804 | 2.61k | } |
7805 | | |
7806 | 130 | free_aligned_buffer_64(row); |
7807 | 130 | return 0; |
7808 | 130 | } |
7809 | | |
7810 | | static int I010AlphaToARGBMatrixBilinear( |
7811 | | const uint16_t* src_y, |
7812 | | int src_stride_y, |
7813 | | const uint16_t* src_u, |
7814 | | int src_stride_u, |
7815 | | const uint16_t* src_v, |
7816 | | int src_stride_v, |
7817 | | const uint16_t* src_a, |
7818 | | int src_stride_a, |
7819 | | uint8_t* dst_argb, |
7820 | | int dst_stride_argb, |
7821 | | const struct YuvConstants* yuvconstants, |
7822 | | int width, |
7823 | | int height, |
7824 | 42 | int attenuate) { |
7825 | 42 | int y; |
7826 | 42 | void (*I410AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
7827 | 42 | const uint16_t* v_buf, const uint16_t* a_buf, |
7828 | 42 | uint8_t* dst_argb, |
7829 | 42 | const struct YuvConstants* yuvconstants, |
7830 | 42 | int width) = I410AlphaToARGBRow_C; |
7831 | 42 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
7832 | 42 | int width) = ARGBAttenuateRow_C; |
7833 | 42 | void (*Scale2RowUp_Bilinear_12)( |
7834 | 42 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
7835 | 42 | ptrdiff_t dst_stride, int dst_width) = ScaleRowUp2_Bilinear_16_Any_C; |
7836 | 42 | void (*ScaleRowUp2_Linear_12)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
7837 | 42 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
7838 | 42 | assert(yuvconstants); |
7839 | 42 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
7840 | 42 | height == 0) { |
7841 | 0 | return -1; |
7842 | 0 | } |
7843 | | // Negative height means invert the image. |
7844 | 42 | if (height < 0) { |
7845 | 0 | height = -height; |
7846 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
7847 | 0 | dst_stride_argb = -dst_stride_argb; |
7848 | 0 | } |
7849 | | #if defined(HAS_I410ALPHATOARGBROW_NEON) |
7850 | | if (TestCpuFlag(kCpuHasNEON)) { |
7851 | | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_NEON; |
7852 | | if (IS_ALIGNED(width, 8)) { |
7853 | | I410AlphaToARGBRow = I410AlphaToARGBRow_NEON; |
7854 | | } |
7855 | | } |
7856 | | #endif |
7857 | | #if defined(HAS_I410ALPHATOARGBROW_SVE2) |
7858 | | if (TestCpuFlag(kCpuHasSVE2)) { |
7859 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SVE2; |
7860 | | } |
7861 | | #endif |
7862 | | #if defined(HAS_I410ALPHATOARGBROW_SME) |
7863 | | if (TestCpuFlag(kCpuHasSME)) { |
7864 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SME; |
7865 | | } |
7866 | | #endif |
7867 | 42 | #if defined(HAS_I410ALPHATOARGBROW_SSSE3) |
7868 | 42 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7869 | 42 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_SSSE3; |
7870 | 42 | if (IS_ALIGNED(width, 8)) { |
7871 | 4 | I410AlphaToARGBRow = I410AlphaToARGBRow_SSSE3; |
7872 | 4 | } |
7873 | 42 | } |
7874 | 42 | #endif |
7875 | 42 | #if defined(HAS_I410ALPHATOARGBROW_AVX2) |
7876 | 42 | if (TestCpuFlag(kCpuHasAVX2)) { |
7877 | 42 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_AVX2; |
7878 | 42 | if (IS_ALIGNED(width, 16)) { |
7879 | 3 | I410AlphaToARGBRow = I410AlphaToARGBRow_AVX2; |
7880 | 3 | } |
7881 | 42 | } |
7882 | 42 | #endif |
7883 | 42 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
7884 | 42 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7885 | 42 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
7886 | 42 | if (IS_ALIGNED(width, 4)) { |
7887 | 7 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
7888 | 7 | } |
7889 | 42 | } |
7890 | 42 | #endif |
7891 | 42 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
7892 | 42 | if (TestCpuFlag(kCpuHasAVX2)) { |
7893 | 42 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
7894 | 42 | if (IS_ALIGNED(width, 8)) { |
7895 | 4 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
7896 | 4 | } |
7897 | 42 | } |
7898 | 42 | #endif |
7899 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
7900 | | if (TestCpuFlag(kCpuHasNEON)) { |
7901 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
7902 | | if (IS_ALIGNED(width, 8)) { |
7903 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
7904 | | } |
7905 | | } |
7906 | | #endif |
7907 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
7908 | | if (TestCpuFlag(kCpuHasRVV)) { |
7909 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
7910 | | } |
7911 | | #endif |
7912 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
7913 | | if (TestCpuFlag(kCpuHasLSX)) { |
7914 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
7915 | | if (IS_ALIGNED(width, 8)) { |
7916 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
7917 | | } |
7918 | | } |
7919 | | #endif |
7920 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
7921 | | if (TestCpuFlag(kCpuHasLASX)) { |
7922 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
7923 | | if (IS_ALIGNED(width, 16)) { |
7924 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
7925 | | } |
7926 | | } |
7927 | | #endif |
7928 | | |
7929 | 42 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_SSSE3) |
7930 | 42 | if (TestCpuFlag(kCpuHasSSSE3)) { |
7931 | 42 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_SSSE3; |
7932 | 42 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_SSSE3; |
7933 | 42 | } |
7934 | 42 | #endif |
7935 | | |
7936 | 42 | #if defined(HAS_SCALEROWUP2_BILINEAR_12_AVX2) |
7937 | 42 | if (TestCpuFlag(kCpuHasAVX2)) { |
7938 | 42 | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_AVX2; |
7939 | 42 | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_AVX2; |
7940 | 42 | } |
7941 | 42 | #endif |
7942 | | |
7943 | | #if defined(HAS_SCALEROWUP2_BILINEAR_12_NEON) |
7944 | | if (TestCpuFlag(kCpuHasNEON)) { |
7945 | | Scale2RowUp_Bilinear_12 = ScaleRowUp2_Bilinear_12_Any_NEON; |
7946 | | ScaleRowUp2_Linear_12 = ScaleRowUp2_Linear_12_Any_NEON; |
7947 | | } |
7948 | | #endif |
7949 | | |
7950 | | // alloc 4 lines temp |
7951 | 42 | const int row_size = (width + 31) & ~31; |
7952 | 42 | align_buffer_64(row, row_size * 4 * sizeof(uint16_t)); |
7953 | 42 | uint16_t* temp_u_1 = (uint16_t*)(row); |
7954 | 42 | uint16_t* temp_u_2 = (uint16_t*)(row) + row_size; |
7955 | 42 | uint16_t* temp_v_1 = (uint16_t*)(row) + row_size * 2; |
7956 | 42 | uint16_t* temp_v_2 = (uint16_t*)(row) + row_size * 3; |
7957 | 42 | if (!row) |
7958 | 0 | return 1; |
7959 | | |
7960 | 42 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7961 | 42 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7962 | 42 | I410AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7963 | 42 | width); |
7964 | 42 | if (attenuate) { |
7965 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7966 | 0 | } |
7967 | 42 | dst_argb += dst_stride_argb; |
7968 | 42 | src_y += src_stride_y; |
7969 | 42 | src_a += src_stride_a; |
7970 | | |
7971 | 811 | for (y = 0; y < height - 2; y += 2) { |
7972 | 769 | Scale2RowUp_Bilinear_12(src_u, src_stride_u, temp_u_1, row_size, width); |
7973 | 769 | Scale2RowUp_Bilinear_12(src_v, src_stride_v, temp_v_1, row_size, width); |
7974 | 769 | I410AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7975 | 769 | width); |
7976 | 769 | if (attenuate) { |
7977 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7978 | 0 | } |
7979 | 769 | dst_argb += dst_stride_argb; |
7980 | 769 | src_y += src_stride_y; |
7981 | 769 | src_a += src_stride_a; |
7982 | 769 | I410AlphaToARGBRow(src_y, temp_u_2, temp_v_2, src_a, dst_argb, yuvconstants, |
7983 | 769 | width); |
7984 | 769 | if (attenuate) { |
7985 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
7986 | 0 | } |
7987 | 769 | dst_argb += dst_stride_argb; |
7988 | 769 | src_y += src_stride_y; |
7989 | 769 | src_a += src_stride_a; |
7990 | 769 | src_u += src_stride_u; |
7991 | 769 | src_v += src_stride_v; |
7992 | 769 | } |
7993 | | |
7994 | 42 | if (!(height & 1)) { |
7995 | 15 | ScaleRowUp2_Linear_12(src_u, temp_u_1, width); |
7996 | 15 | ScaleRowUp2_Linear_12(src_v, temp_v_1, width); |
7997 | 15 | I410AlphaToARGBRow(src_y, temp_u_1, temp_v_1, src_a, dst_argb, yuvconstants, |
7998 | 15 | width); |
7999 | 15 | if (attenuate) { |
8000 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8001 | 0 | } |
8002 | 15 | } |
8003 | | |
8004 | 42 | free_aligned_buffer_64(row); |
8005 | 42 | return 0; |
8006 | 42 | } |
8007 | | |
8008 | | static int I210AlphaToARGBMatrixLinear(const uint16_t* src_y, |
8009 | | int src_stride_y, |
8010 | | const uint16_t* src_u, |
8011 | | int src_stride_u, |
8012 | | const uint16_t* src_v, |
8013 | | int src_stride_v, |
8014 | | const uint16_t* src_a, |
8015 | | int src_stride_a, |
8016 | | uint8_t* dst_argb, |
8017 | | int dst_stride_argb, |
8018 | | const struct YuvConstants* yuvconstants, |
8019 | | int width, |
8020 | | int height, |
8021 | 143 | int attenuate) { |
8022 | 143 | int y; |
8023 | 143 | void (*I410AlphaToARGBRow)(const uint16_t* y_buf, const uint16_t* u_buf, |
8024 | 143 | const uint16_t* v_buf, const uint16_t* a_buf, |
8025 | 143 | uint8_t* dst_argb, |
8026 | 143 | const struct YuvConstants* yuvconstants, |
8027 | 143 | int width) = I410AlphaToARGBRow_C; |
8028 | 143 | void (*ARGBAttenuateRow)(const uint8_t* src_argb, uint8_t* dst_argb, |
8029 | 143 | int width) = ARGBAttenuateRow_C; |
8030 | 143 | void (*ScaleRowUp2_Linear)(const uint16_t* src_ptr, uint16_t* dst_ptr, |
8031 | 143 | int dst_width) = ScaleRowUp2_Linear_16_Any_C; |
8032 | 143 | assert(yuvconstants); |
8033 | 143 | if (!src_y || !src_u || !src_v || !src_a || !dst_argb || width <= 0 || |
8034 | 143 | height == 0) { |
8035 | 0 | return -1; |
8036 | 0 | } |
8037 | | // Negative height means invert the image. |
8038 | 143 | if (height < 0) { |
8039 | 0 | height = -height; |
8040 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
8041 | 0 | dst_stride_argb = -dst_stride_argb; |
8042 | 0 | } |
8043 | | #if defined(HAS_I410ALPHATOARGBROW_NEON) |
8044 | | if (TestCpuFlag(kCpuHasNEON)) { |
8045 | | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_NEON; |
8046 | | if (IS_ALIGNED(width, 8)) { |
8047 | | I410AlphaToARGBRow = I410AlphaToARGBRow_NEON; |
8048 | | } |
8049 | | } |
8050 | | #endif |
8051 | | #if defined(HAS_I410ALPHATOARGBROW_SVE2) |
8052 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8053 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SVE2; |
8054 | | } |
8055 | | #endif |
8056 | | #if defined(HAS_I410ALPHATOARGBROW_SME) |
8057 | | if (TestCpuFlag(kCpuHasSME)) { |
8058 | | I410AlphaToARGBRow = I410AlphaToARGBRow_SME; |
8059 | | } |
8060 | | #endif |
8061 | 143 | #if defined(HAS_I410ALPHATOARGBROW_SSSE3) |
8062 | 143 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8063 | 143 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_SSSE3; |
8064 | 143 | if (IS_ALIGNED(width, 8)) { |
8065 | 37 | I410AlphaToARGBRow = I410AlphaToARGBRow_SSSE3; |
8066 | 37 | } |
8067 | 143 | } |
8068 | 143 | #endif |
8069 | 143 | #if defined(HAS_I410ALPHATOARGBROW_AVX2) |
8070 | 143 | if (TestCpuFlag(kCpuHasAVX2)) { |
8071 | 143 | I410AlphaToARGBRow = I410AlphaToARGBRow_Any_AVX2; |
8072 | 143 | if (IS_ALIGNED(width, 16)) { |
8073 | 18 | I410AlphaToARGBRow = I410AlphaToARGBRow_AVX2; |
8074 | 18 | } |
8075 | 143 | } |
8076 | 143 | #endif |
8077 | 143 | #if defined(HAS_ARGBATTENUATEROW_SSSE3) |
8078 | 143 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8079 | 143 | ARGBAttenuateRow = ARGBAttenuateRow_Any_SSSE3; |
8080 | 143 | if (IS_ALIGNED(width, 4)) { |
8081 | 54 | ARGBAttenuateRow = ARGBAttenuateRow_SSSE3; |
8082 | 54 | } |
8083 | 143 | } |
8084 | 143 | #endif |
8085 | 143 | #if defined(HAS_ARGBATTENUATEROW_AVX2) |
8086 | 143 | if (TestCpuFlag(kCpuHasAVX2)) { |
8087 | 143 | ARGBAttenuateRow = ARGBAttenuateRow_Any_AVX2; |
8088 | 143 | if (IS_ALIGNED(width, 8)) { |
8089 | 37 | ARGBAttenuateRow = ARGBAttenuateRow_AVX2; |
8090 | 37 | } |
8091 | 143 | } |
8092 | 143 | #endif |
8093 | | #if defined(HAS_ARGBATTENUATEROW_NEON) |
8094 | | if (TestCpuFlag(kCpuHasNEON)) { |
8095 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_NEON; |
8096 | | if (IS_ALIGNED(width, 8)) { |
8097 | | ARGBAttenuateRow = ARGBAttenuateRow_NEON; |
8098 | | } |
8099 | | } |
8100 | | #endif |
8101 | | #if defined(HAS_ARGBATTENUATEROW_RVV) |
8102 | | if (TestCpuFlag(kCpuHasRVV)) { |
8103 | | ARGBAttenuateRow = ARGBAttenuateRow_RVV; |
8104 | | } |
8105 | | #endif |
8106 | | #if defined(HAS_ARGBATTENUATEROW_LSX) |
8107 | | if (TestCpuFlag(kCpuHasLSX)) { |
8108 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LSX; |
8109 | | if (IS_ALIGNED(width, 8)) { |
8110 | | ARGBAttenuateRow = ARGBAttenuateRow_LSX; |
8111 | | } |
8112 | | } |
8113 | | #endif |
8114 | | #if defined(HAS_ARGBATTENUATEROW_LASX) |
8115 | | if (TestCpuFlag(kCpuHasLASX)) { |
8116 | | ARGBAttenuateRow = ARGBAttenuateRow_Any_LASX; |
8117 | | if (IS_ALIGNED(width, 16)) { |
8118 | | ARGBAttenuateRow = ARGBAttenuateRow_LASX; |
8119 | | } |
8120 | | } |
8121 | | #endif |
8122 | | |
8123 | 143 | #if defined(HAS_SCALEROWUP2_LINEAR_12_SSSE3) |
8124 | 143 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8125 | 143 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_12_Any_SSSE3; |
8126 | 143 | } |
8127 | 143 | #endif |
8128 | 143 | #if defined(HAS_SCALEROWUP2_LINEAR_12_AVX2) |
8129 | 143 | if (TestCpuFlag(kCpuHasAVX2)) { |
8130 | 143 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_12_Any_AVX2; |
8131 | 143 | } |
8132 | 143 | #endif |
8133 | | #if defined(HAS_SCALEROWUP2_LINEAR_12_NEON) |
8134 | | if (TestCpuFlag(kCpuHasNEON)) { |
8135 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_12_Any_NEON; |
8136 | | } |
8137 | | #endif |
8138 | | |
8139 | | // alloc 2 lines temp |
8140 | 143 | const int row_size = (width + 31) & ~31; |
8141 | 143 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
8142 | 143 | uint16_t* temp_u = (uint16_t*)(row); |
8143 | 143 | uint16_t* temp_v = (uint16_t*)(row) + row_size; |
8144 | 143 | if (!row) |
8145 | 0 | return 1; |
8146 | | |
8147 | 2.97k | for (y = 0; y < height; ++y) { |
8148 | 2.83k | ScaleRowUp2_Linear(src_u, temp_u, width); |
8149 | 2.83k | ScaleRowUp2_Linear(src_v, temp_v, width); |
8150 | 2.83k | I410AlphaToARGBRow(src_y, temp_u, temp_v, src_a, dst_argb, yuvconstants, |
8151 | 2.83k | width); |
8152 | 2.83k | if (attenuate) { |
8153 | 0 | ARGBAttenuateRow(dst_argb, dst_argb, width); |
8154 | 0 | } |
8155 | 2.83k | dst_argb += dst_stride_argb; |
8156 | 2.83k | src_a += src_stride_a; |
8157 | 2.83k | src_y += src_stride_y; |
8158 | 2.83k | src_u += src_stride_u; |
8159 | 2.83k | src_v += src_stride_v; |
8160 | 2.83k | } |
8161 | 143 | free_aligned_buffer_64(row); |
8162 | 143 | return 0; |
8163 | 143 | } |
8164 | | |
8165 | | static int P010ToARGBMatrixBilinear(const uint16_t* src_y, |
8166 | | int src_stride_y, |
8167 | | const uint16_t* src_uv, |
8168 | | int src_stride_uv, |
8169 | | uint8_t* dst_argb, |
8170 | | int dst_stride_argb, |
8171 | | const struct YuvConstants* yuvconstants, |
8172 | | int width, |
8173 | 0 | int height) { |
8174 | 0 | int y; |
8175 | 0 | void (*P410ToARGBRow)( |
8176 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8177 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToARGBRow_C; |
8178 | 0 | void (*Scale2RowUp_Bilinear_16)( |
8179 | 0 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
8180 | 0 | ptrdiff_t dst_stride, int dst_width) = ScaleUVRowUp2_Bilinear_16_Any_C; |
8181 | 0 | assert(yuvconstants); |
8182 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
8183 | 0 | return -1; |
8184 | 0 | } |
8185 | | // Negative height means invert the image. |
8186 | 0 | if (height < 0) { |
8187 | 0 | height = -height; |
8188 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
8189 | 0 | dst_stride_argb = -dst_stride_argb; |
8190 | 0 | } |
8191 | 0 | #if defined(HAS_P410TOARGBROW_SSSE3) |
8192 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8193 | 0 | P410ToARGBRow = P410ToARGBRow_Any_SSSE3; |
8194 | 0 | if (IS_ALIGNED(width, 8)) { |
8195 | 0 | P410ToARGBRow = P410ToARGBRow_SSSE3; |
8196 | 0 | } |
8197 | 0 | } |
8198 | 0 | #endif |
8199 | 0 | #if defined(HAS_P410TOARGBROW_AVX2) |
8200 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8201 | 0 | P410ToARGBRow = P410ToARGBRow_Any_AVX2; |
8202 | 0 | if (IS_ALIGNED(width, 16)) { |
8203 | 0 | P410ToARGBRow = P410ToARGBRow_AVX2; |
8204 | 0 | } |
8205 | 0 | } |
8206 | 0 | #endif |
8207 | | #if defined(HAS_P410TOARGBROW_NEON) |
8208 | | if (TestCpuFlag(kCpuHasNEON)) { |
8209 | | P410ToARGBRow = P410ToARGBRow_Any_NEON; |
8210 | | if (IS_ALIGNED(width, 8)) { |
8211 | | P410ToARGBRow = P410ToARGBRow_NEON; |
8212 | | } |
8213 | | } |
8214 | | #endif |
8215 | | #if defined(HAS_P410TOARGBROW_SVE2) |
8216 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8217 | | P410ToARGBRow = P410ToARGBRow_SVE2; |
8218 | | } |
8219 | | #endif |
8220 | | #if defined(HAS_P410TOARGBROW_SME) |
8221 | | if (TestCpuFlag(kCpuHasSME)) { |
8222 | | P410ToARGBRow = P410ToARGBRow_SME; |
8223 | | } |
8224 | | #endif |
8225 | |
|
8226 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_SSE41 |
8227 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8228 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_SSE41; |
8229 | 0 | } |
8230 | 0 | #endif |
8231 | |
|
8232 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_AVX2 |
8233 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8234 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_AVX2; |
8235 | 0 | } |
8236 | 0 | #endif |
8237 | |
|
8238 | | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_NEON |
8239 | | if (TestCpuFlag(kCpuHasNEON)) { |
8240 | | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_NEON; |
8241 | | } |
8242 | | #endif |
8243 | | |
8244 | | // alloc 2 lines temp |
8245 | 0 | const int row_size = (2 * width + 31) & ~31; |
8246 | 0 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
8247 | 0 | uint16_t* temp_uv_1 = (uint16_t*)(row); |
8248 | 0 | uint16_t* temp_uv_2 = (uint16_t*)(row) + row_size; |
8249 | 0 | if (!row) |
8250 | 0 | return 1; |
8251 | | |
8252 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8253 | 0 | P410ToARGBRow(src_y, temp_uv_1, dst_argb, yuvconstants, width); |
8254 | 0 | dst_argb += dst_stride_argb; |
8255 | 0 | src_y += src_stride_y; |
8256 | |
|
8257 | 0 | for (y = 0; y < height - 2; y += 2) { |
8258 | 0 | Scale2RowUp_Bilinear_16(src_uv, src_stride_uv, temp_uv_1, row_size, width); |
8259 | 0 | P410ToARGBRow(src_y, temp_uv_1, dst_argb, yuvconstants, width); |
8260 | 0 | dst_argb += dst_stride_argb; |
8261 | 0 | src_y += src_stride_y; |
8262 | 0 | P410ToARGBRow(src_y, temp_uv_2, dst_argb, yuvconstants, width); |
8263 | 0 | dst_argb += dst_stride_argb; |
8264 | 0 | src_y += src_stride_y; |
8265 | 0 | src_uv += src_stride_uv; |
8266 | 0 | } |
8267 | |
|
8268 | 0 | if (!(height & 1)) { |
8269 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8270 | 0 | P410ToARGBRow(src_y, temp_uv_1, dst_argb, yuvconstants, width); |
8271 | 0 | } |
8272 | |
|
8273 | 0 | free_aligned_buffer_64(row); |
8274 | 0 | return 0; |
8275 | 0 | } |
8276 | | |
8277 | | static int P210ToARGBMatrixLinear(const uint16_t* src_y, |
8278 | | int src_stride_y, |
8279 | | const uint16_t* src_uv, |
8280 | | int src_stride_uv, |
8281 | | uint8_t* dst_argb, |
8282 | | int dst_stride_argb, |
8283 | | const struct YuvConstants* yuvconstants, |
8284 | | int width, |
8285 | 0 | int height) { |
8286 | 0 | int y; |
8287 | 0 | void (*P410ToARGBRow)( |
8288 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8289 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToARGBRow_C; |
8290 | 0 | void (*ScaleRowUp2_Linear)(const uint16_t* src_uv, uint16_t* dst_uv, |
8291 | 0 | int dst_width) = ScaleUVRowUp2_Linear_16_Any_C; |
8292 | 0 | assert(yuvconstants); |
8293 | 0 | if (!src_y || !src_uv || !dst_argb || width <= 0 || height == 0) { |
8294 | 0 | return -1; |
8295 | 0 | } |
8296 | | // Negative height means invert the image. |
8297 | 0 | if (height < 0) { |
8298 | 0 | height = -height; |
8299 | 0 | dst_argb = dst_argb + (height - 1) * dst_stride_argb; |
8300 | 0 | dst_stride_argb = -dst_stride_argb; |
8301 | 0 | } |
8302 | 0 | #if defined(HAS_P410TOARGBROW_SSSE3) |
8303 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8304 | 0 | P410ToARGBRow = P410ToARGBRow_Any_SSSE3; |
8305 | 0 | if (IS_ALIGNED(width, 8)) { |
8306 | 0 | P410ToARGBRow = P410ToARGBRow_SSSE3; |
8307 | 0 | } |
8308 | 0 | } |
8309 | 0 | #endif |
8310 | 0 | #if defined(HAS_P410TOARGBROW_AVX2) |
8311 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8312 | 0 | P410ToARGBRow = P410ToARGBRow_Any_AVX2; |
8313 | 0 | if (IS_ALIGNED(width, 16)) { |
8314 | 0 | P410ToARGBRow = P410ToARGBRow_AVX2; |
8315 | 0 | } |
8316 | 0 | } |
8317 | 0 | #endif |
8318 | | #if defined(HAS_P410TOARGBROW_NEON) |
8319 | | if (TestCpuFlag(kCpuHasNEON)) { |
8320 | | P410ToARGBRow = P410ToARGBRow_Any_NEON; |
8321 | | if (IS_ALIGNED(width, 8)) { |
8322 | | P410ToARGBRow = P410ToARGBRow_NEON; |
8323 | | } |
8324 | | } |
8325 | | #endif |
8326 | | #if defined(HAS_P410TOARGBROW_SVE2) |
8327 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8328 | | P410ToARGBRow = P410ToARGBRow_SVE2; |
8329 | | } |
8330 | | #endif |
8331 | | #if defined(HAS_P410TOARGBROW_SME) |
8332 | | if (TestCpuFlag(kCpuHasSME)) { |
8333 | | P410ToARGBRow = P410ToARGBRow_SME; |
8334 | | } |
8335 | | #endif |
8336 | |
|
8337 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_SSE41 |
8338 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8339 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_SSE41; |
8340 | 0 | } |
8341 | 0 | #endif |
8342 | |
|
8343 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_AVX2 |
8344 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8345 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_AVX2; |
8346 | 0 | } |
8347 | 0 | #endif |
8348 | |
|
8349 | | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_NEON |
8350 | | if (TestCpuFlag(kCpuHasNEON)) { |
8351 | | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_NEON; |
8352 | | } |
8353 | | #endif |
8354 | |
|
8355 | 0 | const int row_size = (2 * width + 31) & ~31; |
8356 | 0 | align_buffer_64(row, row_size * sizeof(uint16_t)); |
8357 | 0 | uint16_t* temp_uv = (uint16_t*)(row); |
8358 | 0 | if (!row) |
8359 | 0 | return 1; |
8360 | | |
8361 | 0 | for (y = 0; y < height; ++y) { |
8362 | 0 | ScaleRowUp2_Linear(src_uv, temp_uv, width); |
8363 | 0 | P410ToARGBRow(src_y, temp_uv, dst_argb, yuvconstants, width); |
8364 | 0 | dst_argb += dst_stride_argb; |
8365 | 0 | src_y += src_stride_y; |
8366 | 0 | src_uv += src_stride_uv; |
8367 | 0 | } |
8368 | |
|
8369 | 0 | free_aligned_buffer_64(row); |
8370 | 0 | return 0; |
8371 | 0 | } |
8372 | | |
8373 | | static int P010ToAR30MatrixBilinear(const uint16_t* src_y, |
8374 | | int src_stride_y, |
8375 | | const uint16_t* src_uv, |
8376 | | int src_stride_uv, |
8377 | | uint8_t* dst_ar30, |
8378 | | int dst_stride_ar30, |
8379 | | const struct YuvConstants* yuvconstants, |
8380 | | int width, |
8381 | 0 | int height) { |
8382 | 0 | int y; |
8383 | 0 | void (*P410ToAR30Row)( |
8384 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8385 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToAR30Row_C; |
8386 | 0 | void (*Scale2RowUp_Bilinear_16)( |
8387 | 0 | const uint16_t* src_ptr, ptrdiff_t src_stride, uint16_t* dst_ptr, |
8388 | 0 | ptrdiff_t dst_stride, int dst_width) = ScaleUVRowUp2_Bilinear_16_Any_C; |
8389 | 0 | assert(yuvconstants); |
8390 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
8391 | 0 | return -1; |
8392 | 0 | } |
8393 | | // Negative height means invert the image. |
8394 | 0 | if (height < 0) { |
8395 | 0 | height = -height; |
8396 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
8397 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
8398 | 0 | } |
8399 | 0 | #if defined(HAS_P410TOAR30ROW_SSSE3) |
8400 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8401 | 0 | P410ToAR30Row = P410ToAR30Row_Any_SSSE3; |
8402 | 0 | if (IS_ALIGNED(width, 8)) { |
8403 | 0 | P410ToAR30Row = P410ToAR30Row_SSSE3; |
8404 | 0 | } |
8405 | 0 | } |
8406 | 0 | #endif |
8407 | 0 | #if defined(HAS_P410TOAR30ROW_AVX2) |
8408 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8409 | 0 | P410ToAR30Row = P410ToAR30Row_Any_AVX2; |
8410 | 0 | if (IS_ALIGNED(width, 16)) { |
8411 | 0 | P410ToAR30Row = P410ToAR30Row_AVX2; |
8412 | 0 | } |
8413 | 0 | } |
8414 | 0 | #endif |
8415 | | #if defined(HAS_P410TOAR30ROW_NEON) |
8416 | | if (TestCpuFlag(kCpuHasNEON)) { |
8417 | | P410ToAR30Row = P410ToAR30Row_Any_NEON; |
8418 | | if (IS_ALIGNED(width, 8)) { |
8419 | | P410ToAR30Row = P410ToAR30Row_NEON; |
8420 | | } |
8421 | | } |
8422 | | #endif |
8423 | | #if defined(HAS_P410TOAR30ROW_SVE2) |
8424 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8425 | | P410ToAR30Row = P410ToAR30Row_SVE2; |
8426 | | } |
8427 | | #endif |
8428 | | #if defined(HAS_P410TOAR30ROW_SME) |
8429 | | if (TestCpuFlag(kCpuHasSME)) { |
8430 | | P410ToAR30Row = P410ToAR30Row_SME; |
8431 | | } |
8432 | | #endif |
8433 | |
|
8434 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_SSE41 |
8435 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8436 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_SSE41; |
8437 | 0 | } |
8438 | 0 | #endif |
8439 | |
|
8440 | 0 | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_AVX2 |
8441 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8442 | 0 | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_AVX2; |
8443 | 0 | } |
8444 | 0 | #endif |
8445 | |
|
8446 | | #ifdef HAS_SCALEUVROWUP2_BILINEAR_16_NEON |
8447 | | if (TestCpuFlag(kCpuHasNEON)) { |
8448 | | Scale2RowUp_Bilinear_16 = ScaleUVRowUp2_Bilinear_16_Any_NEON; |
8449 | | } |
8450 | | #endif |
8451 | | |
8452 | | // alloc 2 lines temp |
8453 | 0 | const int row_size = (2 * width + 31) & ~31; |
8454 | 0 | align_buffer_64(row, row_size * 2 * sizeof(uint16_t)); |
8455 | 0 | uint16_t* temp_uv_1 = (uint16_t*)(row); |
8456 | 0 | uint16_t* temp_uv_2 = (uint16_t*)(row) + row_size; |
8457 | 0 | if (!row) |
8458 | 0 | return 1; |
8459 | | |
8460 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8461 | 0 | P410ToAR30Row(src_y, temp_uv_1, dst_ar30, yuvconstants, width); |
8462 | 0 | dst_ar30 += dst_stride_ar30; |
8463 | 0 | src_y += src_stride_y; |
8464 | |
|
8465 | 0 | for (y = 0; y < height - 2; y += 2) { |
8466 | 0 | Scale2RowUp_Bilinear_16(src_uv, src_stride_uv, temp_uv_1, row_size, width); |
8467 | 0 | P410ToAR30Row(src_y, temp_uv_1, dst_ar30, yuvconstants, width); |
8468 | 0 | dst_ar30 += dst_stride_ar30; |
8469 | 0 | src_y += src_stride_y; |
8470 | 0 | P410ToAR30Row(src_y, temp_uv_2, dst_ar30, yuvconstants, width); |
8471 | 0 | dst_ar30 += dst_stride_ar30; |
8472 | 0 | src_y += src_stride_y; |
8473 | 0 | src_uv += src_stride_uv; |
8474 | 0 | } |
8475 | |
|
8476 | 0 | if (!(height & 1)) { |
8477 | 0 | Scale2RowUp_Bilinear_16(src_uv, 0, temp_uv_1, row_size, width); |
8478 | 0 | P410ToAR30Row(src_y, temp_uv_1, dst_ar30, yuvconstants, width); |
8479 | 0 | } |
8480 | |
|
8481 | 0 | free_aligned_buffer_64(row); |
8482 | 0 | return 0; |
8483 | 0 | } |
8484 | | |
8485 | | static int P210ToAR30MatrixLinear(const uint16_t* src_y, |
8486 | | int src_stride_y, |
8487 | | const uint16_t* src_uv, |
8488 | | int src_stride_uv, |
8489 | | uint8_t* dst_ar30, |
8490 | | int dst_stride_ar30, |
8491 | | const struct YuvConstants* yuvconstants, |
8492 | | int width, |
8493 | 0 | int height) { |
8494 | 0 | int y; |
8495 | 0 | void (*P410ToAR30Row)( |
8496 | 0 | const uint16_t* y_buf, const uint16_t* uv_buf, uint8_t* rgb_buf, |
8497 | 0 | const struct YuvConstants* yuvconstants, int width) = P410ToAR30Row_C; |
8498 | 0 | void (*ScaleRowUp2_Linear)(const uint16_t* src_uv, uint16_t* dst_uv, |
8499 | 0 | int dst_width) = ScaleUVRowUp2_Linear_16_Any_C; |
8500 | 0 | assert(yuvconstants); |
8501 | 0 | if (!src_y || !src_uv || !dst_ar30 || width <= 0 || height == 0) { |
8502 | 0 | return -1; |
8503 | 0 | } |
8504 | | // Negative height means invert the image. |
8505 | 0 | if (height < 0) { |
8506 | 0 | height = -height; |
8507 | 0 | dst_ar30 = dst_ar30 + (height - 1) * dst_stride_ar30; |
8508 | 0 | dst_stride_ar30 = -dst_stride_ar30; |
8509 | 0 | } |
8510 | 0 | #if defined(HAS_P410TOAR30ROW_SSSE3) |
8511 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8512 | 0 | P410ToAR30Row = P410ToAR30Row_Any_SSSE3; |
8513 | 0 | if (IS_ALIGNED(width, 8)) { |
8514 | 0 | P410ToAR30Row = P410ToAR30Row_SSSE3; |
8515 | 0 | } |
8516 | 0 | } |
8517 | 0 | #endif |
8518 | 0 | #if defined(HAS_P410TOAR30ROW_AVX2) |
8519 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8520 | 0 | P410ToAR30Row = P410ToAR30Row_Any_AVX2; |
8521 | 0 | if (IS_ALIGNED(width, 16)) { |
8522 | 0 | P410ToAR30Row = P410ToAR30Row_AVX2; |
8523 | 0 | } |
8524 | 0 | } |
8525 | 0 | #endif |
8526 | | #if defined(HAS_P410TOAR30ROW_NEON) |
8527 | | if (TestCpuFlag(kCpuHasNEON)) { |
8528 | | P410ToAR30Row = P410ToAR30Row_Any_NEON; |
8529 | | if (IS_ALIGNED(width, 8)) { |
8530 | | P410ToAR30Row = P410ToAR30Row_NEON; |
8531 | | } |
8532 | | } |
8533 | | #endif |
8534 | | #if defined(HAS_P410TOAR30ROW_SVE2) |
8535 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8536 | | P410ToAR30Row = P410ToAR30Row_SVE2; |
8537 | | } |
8538 | | #endif |
8539 | | #if defined(HAS_P410TOAR30ROW_SME) |
8540 | | if (TestCpuFlag(kCpuHasSME)) { |
8541 | | P410ToAR30Row = P410ToAR30Row_SME; |
8542 | | } |
8543 | | #endif |
8544 | |
|
8545 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_SSE41 |
8546 | 0 | if (TestCpuFlag(kCpuHasSSE41)) { |
8547 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_SSE41; |
8548 | 0 | } |
8549 | 0 | #endif |
8550 | |
|
8551 | 0 | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_AVX2 |
8552 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
8553 | 0 | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_AVX2; |
8554 | 0 | } |
8555 | 0 | #endif |
8556 | |
|
8557 | | #ifdef HAS_SCALEUVROWUP2_LINEAR_16_NEON |
8558 | | if (TestCpuFlag(kCpuHasNEON)) { |
8559 | | ScaleRowUp2_Linear = ScaleUVRowUp2_Linear_16_Any_NEON; |
8560 | | } |
8561 | | #endif |
8562 | |
|
8563 | 0 | const int row_size = (2 * width + 31) & ~31; |
8564 | 0 | align_buffer_64(row, row_size * sizeof(uint16_t)); |
8565 | 0 | uint16_t* temp_uv = (uint16_t*)(row); |
8566 | 0 | if (!row) |
8567 | 0 | return 1; |
8568 | | |
8569 | 0 | for (y = 0; y < height; ++y) { |
8570 | 0 | ScaleRowUp2_Linear(src_uv, temp_uv, width); |
8571 | 0 | P410ToAR30Row(src_y, temp_uv, dst_ar30, yuvconstants, width); |
8572 | 0 | dst_ar30 += dst_stride_ar30; |
8573 | 0 | src_y += src_stride_y; |
8574 | 0 | src_uv += src_stride_uv; |
8575 | 0 | } |
8576 | |
|
8577 | 0 | free_aligned_buffer_64(row); |
8578 | 0 | return 0; |
8579 | 0 | } |
8580 | | |
8581 | | static int I422ToRGB24MatrixLinear(const uint8_t* src_y, |
8582 | | int src_stride_y, |
8583 | | const uint8_t* src_u, |
8584 | | int src_stride_u, |
8585 | | const uint8_t* src_v, |
8586 | | int src_stride_v, |
8587 | | uint8_t* dst_rgb24, |
8588 | | int dst_stride_rgb24, |
8589 | | const struct YuvConstants* yuvconstants, |
8590 | | int width, |
8591 | 158 | int height) { |
8592 | 158 | int y; |
8593 | 158 | void (*I444ToRGB24Row)(const uint8_t* y_buf, const uint8_t* u_buf, |
8594 | 158 | const uint8_t* v_buf, uint8_t* rgb_buf, |
8595 | 158 | const struct YuvConstants* yuvconstants, int width) = |
8596 | 158 | I444ToRGB24Row_C; |
8597 | 158 | void (*ScaleRowUp2_Linear)(const uint8_t* src_ptr, uint8_t* dst_ptr, |
8598 | 158 | int dst_width) = ScaleRowUp2_Linear_Any_C; |
8599 | 158 | assert(yuvconstants); |
8600 | 158 | if (!src_y || !src_u || !src_v || !dst_rgb24 || width <= 0 || height == 0) { |
8601 | 0 | return -1; |
8602 | 0 | } |
8603 | | // Negative height means invert the image. |
8604 | 158 | if (height < 0) { |
8605 | 0 | height = -height; |
8606 | 0 | dst_rgb24 = dst_rgb24 + (height - 1) * dst_stride_rgb24; |
8607 | 0 | dst_stride_rgb24 = -dst_stride_rgb24; |
8608 | 0 | } |
8609 | 158 | #if defined(HAS_I444TORGB24ROW_SSSE3) |
8610 | 158 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8611 | 158 | I444ToRGB24Row = I444ToRGB24Row_Any_SSSE3; |
8612 | 158 | if (IS_ALIGNED(width, 16)) { |
8613 | 34 | I444ToRGB24Row = I444ToRGB24Row_SSSE3; |
8614 | 34 | } |
8615 | 158 | } |
8616 | 158 | #endif |
8617 | 158 | #if defined(HAS_I444TORGB24ROW_AVX2) |
8618 | 158 | if (TestCpuFlag(kCpuHasAVX2)) { |
8619 | 158 | I444ToRGB24Row = I444ToRGB24Row_Any_AVX2; |
8620 | 158 | if (IS_ALIGNED(width, 32)) { |
8621 | 17 | I444ToRGB24Row = I444ToRGB24Row_AVX2; |
8622 | 17 | } |
8623 | 158 | } |
8624 | 158 | #endif |
8625 | | #if defined(HAS_I444TORGB24ROW_NEON) |
8626 | | if (TestCpuFlag(kCpuHasNEON)) { |
8627 | | I444ToRGB24Row = I444ToRGB24Row_Any_NEON; |
8628 | | if (IS_ALIGNED(width, 8)) { |
8629 | | I444ToRGB24Row = I444ToRGB24Row_NEON; |
8630 | | } |
8631 | | } |
8632 | | #endif |
8633 | | #if defined(HAS_I444TORGB24ROW_SVE2) |
8634 | | if (TestCpuFlag(kCpuHasSVE2)) { |
8635 | | I444ToRGB24Row = I444ToRGB24Row_SVE2; |
8636 | | } |
8637 | | #endif |
8638 | | #if defined(HAS_I444TORGB24ROW_SME) |
8639 | | if (TestCpuFlag(kCpuHasSME)) { |
8640 | | I444ToRGB24Row = I444ToRGB24Row_SME; |
8641 | | } |
8642 | | #endif |
8643 | | #if defined(HAS_I444TORGB24ROW_RVV) |
8644 | | if (TestCpuFlag(kCpuHasRVV)) { |
8645 | | I444ToRGB24Row = I444ToRGB24Row_RVV; |
8646 | | } |
8647 | | #endif |
8648 | 158 | #if defined(HAS_SCALEROWUP2_LINEAR_SSE2) |
8649 | 158 | if (TestCpuFlag(kCpuHasSSE2)) { |
8650 | 158 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSE2; |
8651 | 158 | } |
8652 | 158 | #endif |
8653 | 158 | #if defined(HAS_SCALEROWUP2_LINEAR_SSSE3) |
8654 | 158 | if (TestCpuFlag(kCpuHasSSSE3)) { |
8655 | 158 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_SSSE3; |
8656 | 158 | } |
8657 | 158 | #endif |
8658 | 158 | #if defined(HAS_SCALEROWUP2_LINEAR_AVX2) |
8659 | 158 | if (TestCpuFlag(kCpuHasAVX2)) { |
8660 | 158 | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_AVX2; |
8661 | 158 | } |
8662 | 158 | #endif |
8663 | | #if defined(HAS_SCALEROWUP2_LINEAR_NEON) |
8664 | | if (TestCpuFlag(kCpuHasNEON)) { |
8665 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_Any_NEON; |
8666 | | } |
8667 | | #endif |
8668 | | #if defined(HAS_SCALEROWUP2_LINEAR_RVV) |
8669 | | if (TestCpuFlag(kCpuHasRVV)) { |
8670 | | ScaleRowUp2_Linear = ScaleRowUp2_Linear_RVV; |
8671 | | } |
8672 | | #endif |
8673 | | |
8674 | | // alloc 2 lines temp |
8675 | 158 | const int row_size = (width + 31) & ~31; |
8676 | 158 | align_buffer_64(row, row_size * 2); |
8677 | 158 | uint8_t* temp_u = row; |
8678 | 158 | uint8_t* temp_v = row + row_size; |
8679 | 158 | if (!row) |
8680 | 0 | return 1; |
8681 | | |
8682 | 4.07k | for (y = 0; y < height; ++y) { |
8683 | 3.91k | ScaleRowUp2_Linear(src_u, temp_u, width); |
8684 | 3.91k | ScaleRowUp2_Linear(src_v, temp_v, width); |
8685 | 3.91k | I444ToRGB24Row(src_y, temp_u, temp_v, dst_rgb24, yuvconstants, width); |
8686 | 3.91k | dst_rgb24 += dst_stride_rgb24; |
8687 | 3.91k | src_y += src_stride_y; |
8688 | 3.91k | src_u += src_stride_u; |
8689 | 3.91k | src_v += src_stride_v; |
8690 | 3.91k | } |
8691 | | |
8692 | 158 | free_aligned_buffer_64(row); |
8693 | 158 | return 0; |
8694 | 158 | } |
8695 | | |
8696 | | LIBYUV_API |
8697 | | int I422ToRGB24MatrixFilter(const uint8_t* src_y, |
8698 | | int src_stride_y, |
8699 | | const uint8_t* src_u, |
8700 | | int src_stride_u, |
8701 | | const uint8_t* src_v, |
8702 | | int src_stride_v, |
8703 | | uint8_t* dst_rgb24, |
8704 | | int dst_stride_rgb24, |
8705 | | const struct YuvConstants* yuvconstants, |
8706 | | int width, |
8707 | | int height, |
8708 | 241 | enum FilterMode filter) { |
8709 | 241 | switch (filter) { |
8710 | 83 | case kFilterNone: |
8711 | 83 | return I422ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8712 | 83 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
8713 | 83 | yuvconstants, width, height); |
8714 | 158 | case kFilterBilinear: |
8715 | 158 | case kFilterBox: |
8716 | 158 | case kFilterLinear: |
8717 | 158 | return I422ToRGB24MatrixLinear( |
8718 | 158 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8719 | 158 | dst_rgb24, dst_stride_rgb24, yuvconstants, width, height); |
8720 | 241 | } |
8721 | | |
8722 | 0 | return -1; |
8723 | 241 | } |
8724 | | |
8725 | | LIBYUV_API |
8726 | | int I420ToARGBMatrixFilter(const uint8_t* src_y, |
8727 | | int src_stride_y, |
8728 | | const uint8_t* src_u, |
8729 | | int src_stride_u, |
8730 | | const uint8_t* src_v, |
8731 | | int src_stride_v, |
8732 | | uint8_t* dst_argb, |
8733 | | int dst_stride_argb, |
8734 | | const struct YuvConstants* yuvconstants, |
8735 | | int width, |
8736 | | int height, |
8737 | 320 | enum FilterMode filter) { |
8738 | 320 | switch (filter) { |
8739 | 141 | case kFilterNone: |
8740 | 141 | return I420ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8741 | 141 | src_stride_v, dst_argb, dst_stride_argb, |
8742 | 141 | yuvconstants, width, height); |
8743 | 179 | case kFilterBilinear: |
8744 | 179 | case kFilterBox: |
8745 | 179 | return I420ToARGBMatrixBilinear( |
8746 | 179 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8747 | 179 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
8748 | 0 | case kFilterLinear: |
8749 | | // Actually we can do this, but probably there's no usage. |
8750 | 0 | return -1; |
8751 | 320 | } |
8752 | | |
8753 | 0 | return -1; |
8754 | 320 | } |
8755 | | |
8756 | | LIBYUV_API |
8757 | | int I422ToARGBMatrixFilter(const uint8_t* src_y, |
8758 | | int src_stride_y, |
8759 | | const uint8_t* src_u, |
8760 | | int src_stride_u, |
8761 | | const uint8_t* src_v, |
8762 | | int src_stride_v, |
8763 | | uint8_t* dst_argb, |
8764 | | int dst_stride_argb, |
8765 | | const struct YuvConstants* yuvconstants, |
8766 | | int width, |
8767 | | int height, |
8768 | 289 | enum FilterMode filter) { |
8769 | 289 | switch (filter) { |
8770 | 95 | case kFilterNone: |
8771 | 95 | return I422ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8772 | 95 | src_stride_v, dst_argb, dst_stride_argb, |
8773 | 95 | yuvconstants, width, height); |
8774 | 194 | case kFilterBilinear: |
8775 | 194 | case kFilterBox: |
8776 | 194 | case kFilterLinear: |
8777 | 194 | return I422ToARGBMatrixLinear( |
8778 | 194 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8779 | 194 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
8780 | 289 | } |
8781 | | |
8782 | 0 | return -1; |
8783 | 289 | } |
8784 | | |
8785 | | LIBYUV_API |
8786 | | int I420ToRGB24MatrixFilter(const uint8_t* src_y, |
8787 | | int src_stride_y, |
8788 | | const uint8_t* src_u, |
8789 | | int src_stride_u, |
8790 | | const uint8_t* src_v, |
8791 | | int src_stride_v, |
8792 | | uint8_t* dst_rgb24, |
8793 | | int dst_stride_rgb24, |
8794 | | const struct YuvConstants* yuvconstants, |
8795 | | int width, |
8796 | | int height, |
8797 | 296 | enum FilterMode filter) { |
8798 | 296 | switch (filter) { |
8799 | 217 | case kFilterNone: |
8800 | 217 | return I420ToRGB24Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8801 | 217 | src_stride_v, dst_rgb24, dst_stride_rgb24, |
8802 | 217 | yuvconstants, width, height); |
8803 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
8804 | 79 | case kFilterBilinear: |
8805 | 79 | case kFilterBox: |
8806 | 79 | return I420ToRGB24MatrixBilinear( |
8807 | 79 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8808 | 79 | dst_rgb24, dst_stride_rgb24, yuvconstants, width, height); |
8809 | 296 | } |
8810 | | |
8811 | 0 | return -1; |
8812 | 296 | } |
8813 | | |
8814 | | LIBYUV_API |
8815 | | int I010ToAR30MatrixFilter(const uint16_t* src_y, |
8816 | | int src_stride_y, |
8817 | | const uint16_t* src_u, |
8818 | | int src_stride_u, |
8819 | | const uint16_t* src_v, |
8820 | | int src_stride_v, |
8821 | | uint8_t* dst_ar30, |
8822 | | int dst_stride_ar30, |
8823 | | const struct YuvConstants* yuvconstants, |
8824 | | int width, |
8825 | | int height, |
8826 | 0 | enum FilterMode filter) { |
8827 | 0 | switch (filter) { |
8828 | 0 | case kFilterNone: |
8829 | 0 | return I010ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8830 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
8831 | 0 | yuvconstants, width, height); |
8832 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
8833 | 0 | case kFilterBilinear: |
8834 | 0 | case kFilterBox: |
8835 | 0 | return I010ToAR30MatrixBilinear( |
8836 | 0 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8837 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, height); |
8838 | 0 | } |
8839 | | |
8840 | 0 | return -1; |
8841 | 0 | } |
8842 | | |
8843 | | LIBYUV_API |
8844 | | int I210ToAR30MatrixFilter(const uint16_t* src_y, |
8845 | | int src_stride_y, |
8846 | | const uint16_t* src_u, |
8847 | | int src_stride_u, |
8848 | | const uint16_t* src_v, |
8849 | | int src_stride_v, |
8850 | | uint8_t* dst_ar30, |
8851 | | int dst_stride_ar30, |
8852 | | const struct YuvConstants* yuvconstants, |
8853 | | int width, |
8854 | | int height, |
8855 | 0 | enum FilterMode filter) { |
8856 | 0 | switch (filter) { |
8857 | 0 | case kFilterNone: |
8858 | 0 | return I210ToAR30Matrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8859 | 0 | src_stride_v, dst_ar30, dst_stride_ar30, |
8860 | 0 | yuvconstants, width, height); |
8861 | 0 | case kFilterBilinear: |
8862 | 0 | case kFilterBox: |
8863 | 0 | case kFilterLinear: |
8864 | 0 | return I210ToAR30MatrixLinear( |
8865 | 0 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8866 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, height); |
8867 | 0 | } |
8868 | | |
8869 | 0 | return -1; |
8870 | 0 | } |
8871 | | |
8872 | | LIBYUV_API |
8873 | | int I010ToARGBMatrixFilter(const uint16_t* src_y, |
8874 | | int src_stride_y, |
8875 | | const uint16_t* src_u, |
8876 | | int src_stride_u, |
8877 | | const uint16_t* src_v, |
8878 | | int src_stride_v, |
8879 | | uint8_t* dst_argb, |
8880 | | int dst_stride_argb, |
8881 | | const struct YuvConstants* yuvconstants, |
8882 | | int width, |
8883 | | int height, |
8884 | 191 | enum FilterMode filter) { |
8885 | 191 | switch (filter) { |
8886 | 97 | case kFilterNone: |
8887 | 97 | return I010ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8888 | 97 | src_stride_v, dst_argb, dst_stride_argb, |
8889 | 97 | yuvconstants, width, height); |
8890 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
8891 | 94 | case kFilterBilinear: |
8892 | 94 | case kFilterBox: |
8893 | 94 | return I010ToARGBMatrixBilinear( |
8894 | 94 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8895 | 94 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
8896 | 191 | } |
8897 | | |
8898 | 0 | return -1; |
8899 | 191 | } |
8900 | | |
8901 | | LIBYUV_API |
8902 | | int I210ToARGBMatrixFilter(const uint16_t* src_y, |
8903 | | int src_stride_y, |
8904 | | const uint16_t* src_u, |
8905 | | int src_stride_u, |
8906 | | const uint16_t* src_v, |
8907 | | int src_stride_v, |
8908 | | uint8_t* dst_argb, |
8909 | | int dst_stride_argb, |
8910 | | const struct YuvConstants* yuvconstants, |
8911 | | int width, |
8912 | | int height, |
8913 | 225 | enum FilterMode filter) { |
8914 | 225 | switch (filter) { |
8915 | 87 | case kFilterNone: |
8916 | 87 | return I210ToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, src_v, |
8917 | 87 | src_stride_v, dst_argb, dst_stride_argb, |
8918 | 87 | yuvconstants, width, height); |
8919 | 138 | case kFilterBilinear: |
8920 | 138 | case kFilterBox: |
8921 | 138 | case kFilterLinear: |
8922 | 138 | return I210ToARGBMatrixLinear( |
8923 | 138 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, |
8924 | 138 | dst_argb, dst_stride_argb, yuvconstants, width, height); |
8925 | 225 | } |
8926 | | |
8927 | 0 | return -1; |
8928 | 225 | } |
8929 | | |
8930 | | LIBYUV_API |
8931 | | int I420AlphaToARGBMatrixFilter(const uint8_t* src_y, |
8932 | | int src_stride_y, |
8933 | | const uint8_t* src_u, |
8934 | | int src_stride_u, |
8935 | | const uint8_t* src_v, |
8936 | | int src_stride_v, |
8937 | | const uint8_t* src_a, |
8938 | | int src_stride_a, |
8939 | | uint8_t* dst_argb, |
8940 | | int dst_stride_argb, |
8941 | | const struct YuvConstants* yuvconstants, |
8942 | | int width, |
8943 | | int height, |
8944 | | int attenuate, |
8945 | 204 | enum FilterMode filter) { |
8946 | 204 | switch (filter) { |
8947 | 148 | case kFilterNone: |
8948 | 148 | return I420AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
8949 | 148 | src_v, src_stride_v, src_a, src_stride_a, |
8950 | 148 | dst_argb, dst_stride_argb, yuvconstants, |
8951 | 148 | width, height, attenuate); |
8952 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
8953 | 56 | case kFilterBilinear: |
8954 | 56 | case kFilterBox: |
8955 | 56 | return I420AlphaToARGBMatrixBilinear( |
8956 | 56 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
8957 | 56 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
8958 | 56 | attenuate); |
8959 | 204 | } |
8960 | | |
8961 | 0 | return -1; |
8962 | 204 | } |
8963 | | |
8964 | | LIBYUV_API |
8965 | | int I422AlphaToARGBMatrixFilter(const uint8_t* src_y, |
8966 | | int src_stride_y, |
8967 | | const uint8_t* src_u, |
8968 | | int src_stride_u, |
8969 | | const uint8_t* src_v, |
8970 | | int src_stride_v, |
8971 | | const uint8_t* src_a, |
8972 | | int src_stride_a, |
8973 | | uint8_t* dst_argb, |
8974 | | int dst_stride_argb, |
8975 | | const struct YuvConstants* yuvconstants, |
8976 | | int width, |
8977 | | int height, |
8978 | | int attenuate, |
8979 | 233 | enum FilterMode filter) { |
8980 | 233 | switch (filter) { |
8981 | 103 | case kFilterNone: |
8982 | 103 | return I422AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
8983 | 103 | src_v, src_stride_v, src_a, src_stride_a, |
8984 | 103 | dst_argb, dst_stride_argb, yuvconstants, |
8985 | 103 | width, height, attenuate); |
8986 | 130 | case kFilterBilinear: |
8987 | 130 | case kFilterBox: |
8988 | 130 | case kFilterLinear: |
8989 | 130 | return I422AlphaToARGBMatrixLinear( |
8990 | 130 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
8991 | 130 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
8992 | 130 | attenuate); |
8993 | 233 | } |
8994 | | |
8995 | 0 | return -1; |
8996 | 233 | } |
8997 | | |
8998 | | LIBYUV_API |
8999 | | int I010AlphaToARGBMatrixFilter(const uint16_t* src_y, |
9000 | | int src_stride_y, |
9001 | | const uint16_t* src_u, |
9002 | | int src_stride_u, |
9003 | | const uint16_t* src_v, |
9004 | | int src_stride_v, |
9005 | | const uint16_t* src_a, |
9006 | | int src_stride_a, |
9007 | | uint8_t* dst_argb, |
9008 | | int dst_stride_argb, |
9009 | | const struct YuvConstants* yuvconstants, |
9010 | | int width, |
9011 | | int height, |
9012 | | int attenuate, |
9013 | 175 | enum FilterMode filter) { |
9014 | 175 | switch (filter) { |
9015 | 133 | case kFilterNone: |
9016 | 133 | return I010AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
9017 | 133 | src_v, src_stride_v, src_a, src_stride_a, |
9018 | 133 | dst_argb, dst_stride_argb, yuvconstants, |
9019 | 133 | width, height, attenuate); |
9020 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9021 | 42 | case kFilterBilinear: |
9022 | 42 | case kFilterBox: |
9023 | 42 | return I010AlphaToARGBMatrixBilinear( |
9024 | 42 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
9025 | 42 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
9026 | 42 | attenuate); |
9027 | 175 | } |
9028 | | |
9029 | 0 | return -1; |
9030 | 175 | } |
9031 | | |
9032 | | LIBYUV_API |
9033 | | int I210AlphaToARGBMatrixFilter(const uint16_t* src_y, |
9034 | | int src_stride_y, |
9035 | | const uint16_t* src_u, |
9036 | | int src_stride_u, |
9037 | | const uint16_t* src_v, |
9038 | | int src_stride_v, |
9039 | | const uint16_t* src_a, |
9040 | | int src_stride_a, |
9041 | | uint8_t* dst_argb, |
9042 | | int dst_stride_argb, |
9043 | | const struct YuvConstants* yuvconstants, |
9044 | | int width, |
9045 | | int height, |
9046 | | int attenuate, |
9047 | 217 | enum FilterMode filter) { |
9048 | 217 | switch (filter) { |
9049 | 74 | case kFilterNone: |
9050 | 74 | return I210AlphaToARGBMatrix(src_y, src_stride_y, src_u, src_stride_u, |
9051 | 74 | src_v, src_stride_v, src_a, src_stride_a, |
9052 | 74 | dst_argb, dst_stride_argb, yuvconstants, |
9053 | 74 | width, height, attenuate); |
9054 | 143 | case kFilterBilinear: |
9055 | 143 | case kFilterBox: |
9056 | 143 | case kFilterLinear: |
9057 | 143 | return I210AlphaToARGBMatrixLinear( |
9058 | 143 | src_y, src_stride_y, src_u, src_stride_u, src_v, src_stride_v, src_a, |
9059 | 143 | src_stride_a, dst_argb, dst_stride_argb, yuvconstants, width, height, |
9060 | 143 | attenuate); |
9061 | 217 | } |
9062 | | |
9063 | 0 | return -1; |
9064 | 217 | } |
9065 | | |
9066 | | // TODO(fb): Verify this function works correctly. P010 is like NV12 but 10 bit |
9067 | | // UV is biplanar. |
9068 | | LIBYUV_API |
9069 | | int P010ToARGBMatrixFilter(const uint16_t* src_y, |
9070 | | int src_stride_y, |
9071 | | const uint16_t* src_uv, |
9072 | | int src_stride_uv, |
9073 | | uint8_t* dst_argb, |
9074 | | int dst_stride_argb, |
9075 | | const struct YuvConstants* yuvconstants, |
9076 | | int width, |
9077 | | int height, |
9078 | 0 | enum FilterMode filter) { |
9079 | 0 | switch (filter) { |
9080 | 0 | case kFilterNone: |
9081 | 0 | return P010ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9082 | 0 | dst_argb, dst_stride_argb, yuvconstants, width, |
9083 | 0 | height); |
9084 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9085 | 0 | case kFilterBilinear: |
9086 | 0 | case kFilterBox: |
9087 | 0 | return P010ToARGBMatrixBilinear(src_y, src_stride_y, src_uv, |
9088 | 0 | src_stride_uv, dst_argb, dst_stride_argb, |
9089 | 0 | yuvconstants, width, height); |
9090 | 0 | } |
9091 | | |
9092 | 0 | return -1; |
9093 | 0 | } |
9094 | | |
9095 | | LIBYUV_API |
9096 | | int P210ToARGBMatrixFilter(const uint16_t* src_y, |
9097 | | int src_stride_y, |
9098 | | const uint16_t* src_uv, |
9099 | | int src_stride_uv, |
9100 | | uint8_t* dst_argb, |
9101 | | int dst_stride_argb, |
9102 | | const struct YuvConstants* yuvconstants, |
9103 | | int width, |
9104 | | int height, |
9105 | 0 | enum FilterMode filter) { |
9106 | 0 | switch (filter) { |
9107 | 0 | case kFilterNone: |
9108 | 0 | return P210ToARGBMatrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9109 | 0 | dst_argb, dst_stride_argb, yuvconstants, width, |
9110 | 0 | height); |
9111 | 0 | case kFilterBilinear: |
9112 | 0 | case kFilterBox: |
9113 | 0 | case kFilterLinear: |
9114 | 0 | return P210ToARGBMatrixLinear(src_y, src_stride_y, src_uv, src_stride_uv, |
9115 | 0 | dst_argb, dst_stride_argb, yuvconstants, |
9116 | 0 | width, height); |
9117 | 0 | } |
9118 | | |
9119 | 0 | return -1; |
9120 | 0 | } |
9121 | | |
9122 | | LIBYUV_API |
9123 | | int P010ToAR30MatrixFilter(const uint16_t* src_y, |
9124 | | int src_stride_y, |
9125 | | const uint16_t* src_uv, |
9126 | | int src_stride_uv, |
9127 | | uint8_t* dst_ar30, |
9128 | | int dst_stride_ar30, |
9129 | | const struct YuvConstants* yuvconstants, |
9130 | | int width, |
9131 | | int height, |
9132 | 0 | enum FilterMode filter) { |
9133 | 0 | switch (filter) { |
9134 | 0 | case kFilterNone: |
9135 | 0 | return P010ToAR30Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9136 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, |
9137 | 0 | height); |
9138 | 0 | case kFilterLinear: // TODO(fb): Implement Linear using Bilinear stride 0 |
9139 | 0 | case kFilterBilinear: |
9140 | 0 | case kFilterBox: |
9141 | 0 | return P010ToAR30MatrixBilinear(src_y, src_stride_y, src_uv, |
9142 | 0 | src_stride_uv, dst_ar30, dst_stride_ar30, |
9143 | 0 | yuvconstants, width, height); |
9144 | 0 | } |
9145 | | |
9146 | 0 | return -1; |
9147 | 0 | } |
9148 | | |
9149 | | LIBYUV_API |
9150 | | int P210ToAR30MatrixFilter(const uint16_t* src_y, |
9151 | | int src_stride_y, |
9152 | | const uint16_t* src_uv, |
9153 | | int src_stride_uv, |
9154 | | uint8_t* dst_ar30, |
9155 | | int dst_stride_ar30, |
9156 | | const struct YuvConstants* yuvconstants, |
9157 | | int width, |
9158 | | int height, |
9159 | 0 | enum FilterMode filter) { |
9160 | 0 | switch (filter) { |
9161 | 0 | case kFilterNone: |
9162 | 0 | return P210ToAR30Matrix(src_y, src_stride_y, src_uv, src_stride_uv, |
9163 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, width, |
9164 | 0 | height); |
9165 | 0 | case kFilterBilinear: |
9166 | 0 | case kFilterBox: |
9167 | 0 | case kFilterLinear: |
9168 | 0 | return P210ToAR30MatrixLinear(src_y, src_stride_y, src_uv, src_stride_uv, |
9169 | 0 | dst_ar30, dst_stride_ar30, yuvconstants, |
9170 | 0 | width, height); |
9171 | 0 | } |
9172 | | |
9173 | 0 | return -1; |
9174 | 0 | } |
9175 | | |
9176 | | #ifdef __cplusplus |
9177 | | } // extern "C" |
9178 | | } // namespace libyuv |
9179 | | #endif |