/src/libavif/ext/libyuv/source/convert_from_argb.cc
Line | Count | Source |
1 | | /* |
2 | | * Copyright 2012 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_from_argb.h" |
12 | | |
13 | | #include <limits.h> |
14 | | #include <string.h> // For memcpy. |
15 | | |
16 | | #include "libyuv/basic_types.h" |
17 | | #include "libyuv/cpu_id.h" |
18 | | #include "libyuv/planar_functions.h" |
19 | | #include "libyuv/row.h" |
20 | | |
21 | | #ifdef __cplusplus |
22 | | namespace libyuv { |
23 | | extern "C" { |
24 | | #endif |
25 | | |
26 | | // ARGB little endian (bgra in memory) to I444 |
27 | | LIBYUV_API |
28 | | int ARGBToI444(const uint8_t* src_argb, |
29 | | int src_stride_argb, |
30 | | uint8_t* dst_y, |
31 | | int dst_stride_y, |
32 | | uint8_t* dst_u, |
33 | | int dst_stride_u, |
34 | | uint8_t* dst_v, |
35 | | int dst_stride_v, |
36 | | int width, |
37 | 0 | int height) { |
38 | 0 | return ARGBToI444Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u, |
39 | 0 | dst_stride_u, dst_v, dst_stride_v, |
40 | 0 | &kArgbI601Constants, width, height); |
41 | 0 | } |
42 | | |
43 | | LIBYUV_API |
44 | | int ARGBToI444Matrix(const uint8_t* src_argb, |
45 | | int src_stride_argb, |
46 | | uint8_t* dst_y, |
47 | | int dst_stride_y, |
48 | | uint8_t* dst_u, |
49 | | int dst_stride_u, |
50 | | uint8_t* dst_v, |
51 | | int dst_stride_v, |
52 | | const struct ArgbConstants* argbconstants, |
53 | | int width, |
54 | 0 | int height) { |
55 | 0 | int y; |
56 | 0 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
57 | 0 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
58 | 0 | void (*ARGBToUV444MatrixRow)( |
59 | 0 | const uint8_t* src_argb, uint8_t* dst_u, uint8_t* dst_v, int width, |
60 | 0 | const struct ArgbConstants* c) = ARGBToUV444MatrixRow_C; |
61 | |
|
62 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_SSSE3) |
63 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
64 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3; |
65 | 0 | if (IS_ALIGNED(width, 16)) { |
66 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3; |
67 | 0 | } |
68 | 0 | } |
69 | 0 | #endif |
70 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
71 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
72 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
73 | 0 | if (IS_ALIGNED(width, 32)) { |
74 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
75 | 0 | } |
76 | 0 | } |
77 | 0 | #endif |
78 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
79 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
80 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
81 | 0 | if (IS_ALIGNED(width, 64)) { |
82 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
83 | 0 | } |
84 | 0 | } |
85 | 0 | #endif |
86 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
87 | | if (TestCpuFlag(kCpuHasNEON)) { |
88 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
89 | | if (IS_ALIGNED(width, 16)) { |
90 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
91 | | } |
92 | | } |
93 | | #endif |
94 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
95 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
96 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
97 | | if (IS_ALIGNED(width, 16)) { |
98 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
99 | | } |
100 | | } |
101 | | #endif |
102 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
103 | | if (TestCpuFlag(kCpuHasSVE2)) { |
104 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
105 | | } |
106 | | #endif |
107 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
108 | | if (TestCpuFlag(kCpuHasSME)) { |
109 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
110 | | } |
111 | | #endif |
112 | | #if defined(HAS_ARGBTOYMATRIXROW_LSX) |
113 | | if (TestCpuFlag(kCpuHasLSX)) { |
114 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX; |
115 | | if (IS_ALIGNED(width, 16)) { |
116 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LSX; |
117 | | } |
118 | | } |
119 | | #endif |
120 | | #if defined(HAS_ARGBTOYMATRIXROW_LASX) |
121 | | if (TestCpuFlag(kCpuHasLASX)) { |
122 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX; |
123 | | if (IS_ALIGNED(width, 32)) { |
124 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LASX; |
125 | | } |
126 | | } |
127 | | #endif |
128 | | #if defined(HAS_ARGBTOYMATRIXROW_RVV) |
129 | | if (TestCpuFlag(kCpuHasRVV)) { |
130 | | ARGBToYMatrixRow = ARGBToYMatrixRow_RVV; |
131 | | } |
132 | | #endif |
133 | |
|
134 | 0 | #if defined(HAS_ARGBTOUV444MATRIXROW_SSSE3) |
135 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
136 | 0 | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_Any_SSSE3; |
137 | 0 | if (IS_ALIGNED(width, 16)) { |
138 | 0 | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_SSSE3; |
139 | 0 | } |
140 | 0 | } |
141 | 0 | #endif |
142 | 0 | #if defined(HAS_ARGBTOUV444MATRIXROW_AVX2) |
143 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
144 | 0 | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_Any_AVX2; |
145 | 0 | if (IS_ALIGNED(width, 32)) { |
146 | 0 | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_AVX2; |
147 | 0 | } |
148 | 0 | } |
149 | 0 | #endif |
150 | 0 | #if defined(HAS_ARGBTOUV444MATRIXROW_AVX512BW) |
151 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
152 | 0 | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_Any_AVX512BW; |
153 | 0 | if (IS_ALIGNED(width, 64)) { |
154 | 0 | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_AVX512BW; |
155 | 0 | } |
156 | 0 | } |
157 | 0 | #endif |
158 | | #if defined(HAS_ARGBTOUV444MATRIXROW_NEON) |
159 | | if (TestCpuFlag(kCpuHasNEON)) { |
160 | | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_Any_NEON; |
161 | | if (IS_ALIGNED(width, 8)) { |
162 | | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_NEON; |
163 | | } |
164 | | } |
165 | | #endif |
166 | | #if defined(HAS_ARGBTOUV444MATRIXROW_NEON_I8MM) |
167 | | if (TestCpuFlag(kCpuHasNeonI8MM)) { |
168 | | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_Any_NEON_I8MM; |
169 | | if (IS_ALIGNED(width, 8)) { |
170 | | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_NEON_I8MM; |
171 | | } |
172 | | } |
173 | | #endif |
174 | | #if defined(HAS_ARGBTOUV444MATRIXROW_RVV) |
175 | | if (TestCpuFlag(kCpuHasRVV)) { |
176 | | ARGBToUV444MatrixRow = ARGBToUV444MatrixRow_RVV; |
177 | | } |
178 | | #endif |
179 | 0 | if (!src_argb || !dst_y || !dst_u || !dst_v || !argbconstants || width <= 0 || |
180 | 0 | height == 0 || height == INT_MIN) { |
181 | 0 | return -1; |
182 | 0 | } |
183 | | // Negative height means invert the image. |
184 | 0 | if (height < 0) { |
185 | 0 | height = -height; |
186 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
187 | 0 | src_stride_argb = -src_stride_argb; |
188 | 0 | } |
189 | |
|
190 | 0 | for (y = 0; y < height; ++y) { |
191 | 0 | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
192 | 0 | ARGBToUV444MatrixRow(src_argb, dst_u, dst_v, width, argbconstants); |
193 | 0 | src_argb += src_stride_argb; |
194 | 0 | dst_y += dst_stride_y; |
195 | 0 | dst_u += dst_stride_u; |
196 | 0 | dst_v += dst_stride_v; |
197 | 0 | } |
198 | 0 | return 0; |
199 | 0 | } |
200 | | |
201 | | // ARGB little endian (bgra in memory) to I422 |
202 | | LIBYUV_API |
203 | | int ARGBToI422(const uint8_t* src_argb, |
204 | | int src_stride_argb, |
205 | | uint8_t* dst_y, |
206 | | int dst_stride_y, |
207 | | uint8_t* dst_u, |
208 | | int dst_stride_u, |
209 | | uint8_t* dst_v, |
210 | | int dst_stride_v, |
211 | | int width, |
212 | 0 | int height) { |
213 | 0 | return ARGBToI422Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u, |
214 | 0 | dst_stride_u, dst_v, dst_stride_v, |
215 | 0 | &kArgbI601Constants, width, height); |
216 | 0 | } |
217 | | |
218 | | LIBYUV_API |
219 | | int ARGBToI422Matrix(const uint8_t* src_argb, |
220 | | int src_stride_argb, |
221 | | uint8_t* dst_y, |
222 | | int dst_stride_y, |
223 | | uint8_t* dst_u, |
224 | | int dst_stride_u, |
225 | | uint8_t* dst_v, |
226 | | int dst_stride_v, |
227 | | const struct ArgbConstants* argbconstants, |
228 | | int width, |
229 | 184k | int height) { |
230 | 184k | int y; |
231 | 184k | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
232 | 184k | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
233 | 184k | void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb, |
234 | 184k | uint8_t* dst_u, uint8_t* dst_v, int width, |
235 | 184k | const struct ArgbConstants* c) = |
236 | 184k | ARGBToUVMatrixRow_C; |
237 | | |
238 | 184k | #if defined(HAS_ARGBTOYMATRIXROW_SSSE3) |
239 | 184k | if (TestCpuFlag(kCpuHasSSSE3)) { |
240 | 184k | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3; |
241 | 184k | if (IS_ALIGNED(width, 16)) { |
242 | 883 | ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3; |
243 | 883 | } |
244 | 184k | } |
245 | 184k | #endif |
246 | 184k | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
247 | 184k | if (TestCpuFlag(kCpuHasAVX2)) { |
248 | 184k | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
249 | 184k | if (IS_ALIGNED(width, 32)) { |
250 | 423 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
251 | 423 | } |
252 | 184k | } |
253 | 184k | #endif |
254 | 184k | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
255 | 184k | if (TestCpuFlag(kCpuHasAVX512BW)) { |
256 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
257 | 0 | if (IS_ALIGNED(width, 64)) { |
258 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
259 | 0 | } |
260 | 0 | } |
261 | 184k | #endif |
262 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
263 | | if (TestCpuFlag(kCpuHasNEON)) { |
264 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
265 | | if (IS_ALIGNED(width, 16)) { |
266 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
267 | | } |
268 | | } |
269 | | #endif |
270 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
271 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
272 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
273 | | if (IS_ALIGNED(width, 16)) { |
274 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
275 | | } |
276 | | } |
277 | | #endif |
278 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
279 | | if (TestCpuFlag(kCpuHasSVE2)) { |
280 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
281 | | } |
282 | | #endif |
283 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
284 | | if (TestCpuFlag(kCpuHasSME)) { |
285 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
286 | | } |
287 | | #endif |
288 | | #if defined(HAS_ARGBTOYMATRIXROW_LSX) |
289 | | if (TestCpuFlag(kCpuHasLSX)) { |
290 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX; |
291 | | if (IS_ALIGNED(width, 16)) { |
292 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LSX; |
293 | | } |
294 | | } |
295 | | #endif |
296 | | #if defined(HAS_ARGBTOYMATRIXROW_LASX) |
297 | | if (TestCpuFlag(kCpuHasLASX)) { |
298 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX; |
299 | | if (IS_ALIGNED(width, 32)) { |
300 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LASX; |
301 | | } |
302 | | } |
303 | | #endif |
304 | | #if defined(HAS_ARGBTOYMATRIXROW_RVV) |
305 | | if (TestCpuFlag(kCpuHasRVV)) { |
306 | | ARGBToYMatrixRow = ARGBToYMatrixRow_RVV; |
307 | | } |
308 | | #endif |
309 | | |
310 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON) |
311 | | if (TestCpuFlag(kCpuHasNEON)) { |
312 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON; |
313 | | if (IS_ALIGNED(width, 16)) { |
314 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON; |
315 | | } |
316 | | } |
317 | | #endif |
318 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM) |
319 | | if (TestCpuFlag(kCpuHasNEON) && TestCpuFlag(kCpuHasNeonI8MM)) { |
320 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM; |
321 | | if (IS_ALIGNED(width, 16)) { |
322 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM; |
323 | | } |
324 | | } |
325 | | #endif |
326 | | #if defined(HAS_ARGBTOUVMATRIXROW_SVE2) |
327 | | if (TestCpuFlag(kCpuHasSVE2)) { |
328 | | if (IS_ALIGNED(width, 2)) { |
329 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2; |
330 | | } |
331 | | } |
332 | | #endif |
333 | | #if defined(HAS_ARGBTOUVMATRIXROW_SME) |
334 | | if (TestCpuFlag(kCpuHasSME)) { |
335 | | if (IS_ALIGNED(width, 2)) { |
336 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME; |
337 | | } |
338 | | } |
339 | | #endif |
340 | 184k | #if defined(HAS_ARGBTOUVMATRIXROW_SSSE3) |
341 | 184k | if (TestCpuFlag(kCpuHasSSSE3)) { |
342 | 184k | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3; |
343 | 184k | if (IS_ALIGNED(width, 8)) { |
344 | 2.12k | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3; |
345 | 2.12k | } |
346 | 184k | } |
347 | 184k | #endif |
348 | 184k | #if defined(HAS_ARGBTOUVMATRIXROW_AVX2) |
349 | 184k | if (TestCpuFlag(kCpuHasAVX2)) { |
350 | 184k | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2; |
351 | 184k | if (IS_ALIGNED(width, 32)) { |
352 | 423 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2; |
353 | 423 | } |
354 | 184k | } |
355 | 184k | #endif |
356 | 184k | #if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW) |
357 | 184k | if (TestCpuFlag(kCpuHasAVX512BW)) { |
358 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX512BW; |
359 | 0 | if (IS_ALIGNED(width, 64)) { |
360 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW; |
361 | 0 | } |
362 | 0 | } |
363 | 184k | #endif |
364 | | #if defined(HAS_ARGBTOUVMATRIXROW_RVV) |
365 | | if (TestCpuFlag(kCpuHasRVV)) { |
366 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV; |
367 | | } |
368 | | #endif |
369 | 184k | if (!src_argb || !dst_y || !dst_u || !dst_v || !argbconstants || width <= 0 || |
370 | 184k | height == 0 || height == INT_MIN) { |
371 | 0 | return -1; |
372 | 0 | } |
373 | | // Negative height means invert the image. |
374 | 184k | if (height < 0) { |
375 | 0 | height = -height; |
376 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
377 | 0 | src_stride_argb = -src_stride_argb; |
378 | 0 | } |
379 | | |
380 | 1.49M | for (y = 0; y < height; ++y) { |
381 | 1.31M | ARGBToUVMatrixRow(src_argb, 0, dst_u, dst_v, width, argbconstants); |
382 | 1.31M | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
383 | 1.31M | src_argb += src_stride_argb; |
384 | 1.31M | dst_y += dst_stride_y; |
385 | 1.31M | dst_u += dst_stride_u; |
386 | 1.31M | dst_v += dst_stride_v; |
387 | 1.31M | } |
388 | 184k | return 0; |
389 | 184k | } |
390 | | |
391 | | LIBYUV_API |
392 | | int ARGBToNV12(const uint8_t* src_argb, |
393 | | int src_stride_argb, |
394 | | uint8_t* dst_y, |
395 | | int dst_stride_y, |
396 | | uint8_t* dst_uv, |
397 | | int dst_stride_uv, |
398 | | int width, |
399 | 0 | int height) { |
400 | 0 | return ARGBToNV12Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, |
401 | 0 | dst_uv, dst_stride_uv, &kArgbI601Constants, width, |
402 | 0 | height); |
403 | 0 | } |
404 | | |
405 | | LIBYUV_API |
406 | | int ARGBToNV12Matrix(const uint8_t* src_argb, |
407 | | int src_stride_argb, |
408 | | uint8_t* dst_y, |
409 | | int dst_stride_y, |
410 | | uint8_t* dst_uv, |
411 | | int dst_stride_uv, |
412 | | const struct ArgbConstants* argbconstants, |
413 | | int width, |
414 | 0 | int height) { |
415 | 0 | int y; |
416 | 0 | int halfwidth = (width + 1) >> 1; |
417 | 0 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
418 | 0 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
419 | 0 | void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb, |
420 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
421 | 0 | const struct ArgbConstants* c) = |
422 | 0 | ARGBToUVMatrixRow_C; |
423 | |
|
424 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_SSSE3) |
425 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
426 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3; |
427 | 0 | if (IS_ALIGNED(width, 16)) { |
428 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3; |
429 | 0 | } |
430 | 0 | } |
431 | 0 | #endif |
432 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
433 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
434 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
435 | 0 | if (IS_ALIGNED(width, 32)) { |
436 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
437 | 0 | } |
438 | 0 | } |
439 | 0 | #endif |
440 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
441 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
442 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
443 | 0 | if (IS_ALIGNED(width, 64)) { |
444 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
445 | 0 | } |
446 | 0 | } |
447 | 0 | #endif |
448 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
449 | | if (TestCpuFlag(kCpuHasNEON)) { |
450 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
451 | | if (IS_ALIGNED(width, 16)) { |
452 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
453 | | } |
454 | | } |
455 | | #endif |
456 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
457 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
458 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
459 | | if (IS_ALIGNED(width, 16)) { |
460 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
461 | | } |
462 | | } |
463 | | #endif |
464 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
465 | | if (TestCpuFlag(kCpuHasSVE2)) { |
466 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
467 | | } |
468 | | #endif |
469 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
470 | | if (TestCpuFlag(kCpuHasSME)) { |
471 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
472 | | } |
473 | | #endif |
474 | | #if defined(HAS_ARGBTOYMATRIXROW_LSX) |
475 | | if (TestCpuFlag(kCpuHasLSX)) { |
476 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX; |
477 | | if (IS_ALIGNED(width, 16)) { |
478 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LSX; |
479 | | } |
480 | | } |
481 | | #endif |
482 | | #if defined(HAS_ARGBTOYMATRIXROW_LASX) |
483 | | if (TestCpuFlag(kCpuHasLASX)) { |
484 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX; |
485 | | if (IS_ALIGNED(width, 32)) { |
486 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LASX; |
487 | | } |
488 | | } |
489 | | #endif |
490 | | #if defined(HAS_ARGBTOYMATRIXROW_RVV) |
491 | | if (TestCpuFlag(kCpuHasRVV)) { |
492 | | ARGBToYMatrixRow = ARGBToYMatrixRow_RVV; |
493 | | } |
494 | | #endif |
495 | |
|
496 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON) |
497 | | if (TestCpuFlag(kCpuHasNEON)) { |
498 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON; |
499 | | if (IS_ALIGNED(width, 16)) { |
500 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON; |
501 | | } |
502 | | } |
503 | | #endif |
504 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM) |
505 | | if (TestCpuFlag(kCpuHasNEON) && TestCpuFlag(kCpuHasNeonI8MM)) { |
506 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM; |
507 | | if (IS_ALIGNED(width, 16)) { |
508 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM; |
509 | | } |
510 | | } |
511 | | #endif |
512 | | #if defined(HAS_ARGBTOUVMATRIXROW_SVE2) |
513 | | if (TestCpuFlag(kCpuHasSVE2)) { |
514 | | if (IS_ALIGNED(width, 2)) { |
515 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2; |
516 | | } |
517 | | } |
518 | | #endif |
519 | | #if defined(HAS_ARGBTOUVMATRIXROW_SME) |
520 | | if (TestCpuFlag(kCpuHasSME)) { |
521 | | if (IS_ALIGNED(width, 2)) { |
522 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME; |
523 | | } |
524 | | } |
525 | | #endif |
526 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_SSSE3) |
527 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
528 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3; |
529 | 0 | if (IS_ALIGNED(width, 8)) { |
530 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3; |
531 | 0 | } |
532 | 0 | } |
533 | 0 | #endif |
534 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX2) |
535 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
536 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2; |
537 | 0 | if (IS_ALIGNED(width, 32)) { |
538 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2; |
539 | 0 | } |
540 | 0 | } |
541 | 0 | #endif |
542 | | #if defined(HAS_ARGBTOUVMATRIXROW_RVV) |
543 | | if (TestCpuFlag(kCpuHasRVV)) { |
544 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV; |
545 | | } |
546 | | #endif |
547 | 0 | void (*MergeUVRow)(const uint8_t* src_u, const uint8_t* src_v, |
548 | 0 | uint8_t* dst_uv, int width) = MergeUVRow_C; |
549 | 0 | if (!src_argb || !dst_y || !dst_uv || !argbconstants || width <= 0 || |
550 | 0 | height == 0 || height == INT_MIN) { |
551 | 0 | return -1; |
552 | 0 | } |
553 | | // Negative height means invert the image. |
554 | 0 | if (height < 0) { |
555 | 0 | height = -height; |
556 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
557 | 0 | src_stride_argb = -src_stride_argb; |
558 | 0 | } |
559 | 0 | #if defined(HAS_MERGEUVROW_SSE2) |
560 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
561 | 0 | MergeUVRow = MergeUVRow_Any_SSE2; |
562 | 0 | if (IS_ALIGNED(halfwidth, 16)) { |
563 | 0 | MergeUVRow = MergeUVRow_SSE2; |
564 | 0 | } |
565 | 0 | } |
566 | 0 | #endif |
567 | 0 | #if defined(HAS_MERGEUVROW_AVX2) |
568 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
569 | 0 | MergeUVRow = MergeUVRow_Any_AVX2; |
570 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
571 | 0 | MergeUVRow = MergeUVRow_AVX2; |
572 | 0 | } |
573 | 0 | } |
574 | 0 | #endif |
575 | 0 | #if defined(HAS_MERGEUVROW_AVX512BW) |
576 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
577 | 0 | MergeUVRow = MergeUVRow_Any_AVX512BW; |
578 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
579 | 0 | MergeUVRow = MergeUVRow_AVX512BW; |
580 | 0 | } |
581 | 0 | } |
582 | 0 | #endif |
583 | | #if defined(HAS_MERGEUVROW_NEON) |
584 | | if (TestCpuFlag(kCpuHasNEON)) { |
585 | | MergeUVRow = MergeUVRow_Any_NEON; |
586 | | if (IS_ALIGNED(halfwidth, 16)) { |
587 | | MergeUVRow = MergeUVRow_NEON; |
588 | | } |
589 | | } |
590 | | #endif |
591 | | #if defined(HAS_MERGEUVROW_SVE2) |
592 | | if (TestCpuFlag(kCpuHasSVE2)) { |
593 | | MergeUVRow = MergeUVRow_SVE2; |
594 | | } |
595 | | #endif |
596 | | #if defined(HAS_MERGEUVROW_SME) |
597 | | if (TestCpuFlag(kCpuHasSME)) { |
598 | | MergeUVRow = MergeUVRow_SME; |
599 | | } |
600 | | #endif |
601 | | #if defined(HAS_MERGEUVROW_LSX) |
602 | | if (TestCpuFlag(kCpuHasLSX)) { |
603 | | MergeUVRow = MergeUVRow_Any_LSX; |
604 | | if (IS_ALIGNED(halfwidth, 16)) { |
605 | | MergeUVRow = MergeUVRow_LSX; |
606 | | } |
607 | | } |
608 | | #endif |
609 | | #if defined(HAS_MERGEUVROW_RVV) |
610 | | if (TestCpuFlag(kCpuHasRVV)) { |
611 | | MergeUVRow = MergeUVRow_RVV; |
612 | | } |
613 | | #endif |
614 | | |
615 | | // Allocate a rows of uv. |
616 | 0 | align_buffer_64(row_u, ((halfwidth + 31) & ~31) * 2); |
617 | 0 | uint8_t* row_v = row_u + ((halfwidth + 31) & ~31); |
618 | 0 | if (!row_u) |
619 | 0 | return 1; |
620 | | |
621 | 0 | for (y = 0; y < height - 1; y += 2) { |
622 | 0 | ARGBToUVMatrixRow(src_argb, src_stride_argb, row_u, row_v, width, |
623 | 0 | argbconstants); |
624 | 0 | MergeUVRow(row_u, row_v, dst_uv, halfwidth); |
625 | 0 | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
626 | 0 | ARGBToYMatrixRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width, |
627 | 0 | argbconstants); |
628 | 0 | src_argb += src_stride_argb * 2; |
629 | 0 | dst_y += dst_stride_y * 2; |
630 | 0 | dst_uv += dst_stride_uv; |
631 | 0 | } |
632 | 0 | if (height & 1) { |
633 | 0 | ARGBToUVMatrixRow(src_argb, 0, row_u, row_v, width, argbconstants); |
634 | 0 | MergeUVRow(row_u, row_v, dst_uv, halfwidth); |
635 | 0 | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
636 | 0 | } |
637 | 0 | free_aligned_buffer_64(row_u); |
638 | 0 | return 0; |
639 | 0 | } |
640 | | |
641 | | LIBYUV_API |
642 | | int ARGBToNV21Matrix(const uint8_t* src_argb, |
643 | | int src_stride_argb, |
644 | | uint8_t* dst_y, |
645 | | int dst_stride_y, |
646 | | uint8_t* dst_vu, |
647 | | int dst_stride_vu, |
648 | | const struct ArgbConstants* argbconstants, |
649 | | int width, |
650 | 0 | int height) { |
651 | 0 | int y; |
652 | 0 | int halfwidth = (width + 1) >> 1; |
653 | 0 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
654 | 0 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
655 | 0 | void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb, |
656 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
657 | 0 | const struct ArgbConstants* c) = |
658 | 0 | ARGBToUVMatrixRow_C; |
659 | |
|
660 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_SSSE3) |
661 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
662 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_SSSE3; |
663 | 0 | if (IS_ALIGNED(width, 16)) { |
664 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_SSSE3; |
665 | 0 | } |
666 | 0 | } |
667 | 0 | #endif |
668 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
669 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
670 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
671 | 0 | if (IS_ALIGNED(width, 32)) { |
672 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
673 | 0 | } |
674 | 0 | } |
675 | 0 | #endif |
676 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
677 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
678 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
679 | 0 | if (IS_ALIGNED(width, 64)) { |
680 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
681 | 0 | } |
682 | 0 | } |
683 | 0 | #endif |
684 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
685 | | if (TestCpuFlag(kCpuHasNEON)) { |
686 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
687 | | if (IS_ALIGNED(width, 16)) { |
688 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
689 | | } |
690 | | } |
691 | | #endif |
692 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
693 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
694 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
695 | | if (IS_ALIGNED(width, 16)) { |
696 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
697 | | } |
698 | | } |
699 | | #endif |
700 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
701 | | if (TestCpuFlag(kCpuHasSVE2)) { |
702 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
703 | | } |
704 | | #endif |
705 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
706 | | if (TestCpuFlag(kCpuHasSME)) { |
707 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
708 | | } |
709 | | #endif |
710 | | #if defined(HAS_ARGBTOYMATRIXROW_LSX) |
711 | | if (TestCpuFlag(kCpuHasLSX)) { |
712 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LSX; |
713 | | if (IS_ALIGNED(width, 16)) { |
714 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LSX; |
715 | | } |
716 | | } |
717 | | #endif |
718 | | #if defined(HAS_ARGBTOYMATRIXROW_LASX) |
719 | | if (TestCpuFlag(kCpuHasLASX)) { |
720 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_LASX; |
721 | | if (IS_ALIGNED(width, 32)) { |
722 | | ARGBToYMatrixRow = ARGBToYMatrixRow_LASX; |
723 | | } |
724 | | } |
725 | | #endif |
726 | | #if defined(HAS_ARGBTOYMATRIXROW_RVV) |
727 | | if (TestCpuFlag(kCpuHasRVV)) { |
728 | | ARGBToYMatrixRow = ARGBToYMatrixRow_RVV; |
729 | | } |
730 | | #endif |
731 | |
|
732 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON) |
733 | | if (TestCpuFlag(kCpuHasNEON)) { |
734 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON; |
735 | | if (IS_ALIGNED(width, 16)) { |
736 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON; |
737 | | } |
738 | | } |
739 | | #endif |
740 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM) |
741 | | if (TestCpuFlag(kCpuHasNEON) && TestCpuFlag(kCpuHasNeonI8MM)) { |
742 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM; |
743 | | if (IS_ALIGNED(width, 16)) { |
744 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM; |
745 | | } |
746 | | } |
747 | | #endif |
748 | | #if defined(HAS_ARGBTOUVMATRIXROW_SVE2) |
749 | | if (TestCpuFlag(kCpuHasSVE2)) { |
750 | | if (IS_ALIGNED(width, 2)) { |
751 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2; |
752 | | } |
753 | | } |
754 | | #endif |
755 | | #if defined(HAS_ARGBTOUVMATRIXROW_SME) |
756 | | if (TestCpuFlag(kCpuHasSME)) { |
757 | | if (IS_ALIGNED(width, 2)) { |
758 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME; |
759 | | } |
760 | | } |
761 | | #endif |
762 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_SSSE3) |
763 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
764 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_SSSE3; |
765 | 0 | if (IS_ALIGNED(width, 8)) { |
766 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SSSE3; |
767 | 0 | } |
768 | 0 | } |
769 | 0 | #endif |
770 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX2) |
771 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
772 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2; |
773 | 0 | if (IS_ALIGNED(width, 32)) { |
774 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2; |
775 | 0 | } |
776 | 0 | } |
777 | 0 | #endif |
778 | | #if defined(HAS_ARGBTOUVMATRIXROW_RVV) |
779 | | if (TestCpuFlag(kCpuHasRVV)) { |
780 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_RVV; |
781 | | } |
782 | | #endif |
783 | 0 | void (*MergeUVRow)(const uint8_t* src_u, const uint8_t* src_v, |
784 | 0 | uint8_t* dst_vu, int width) = MergeUVRow_C; |
785 | 0 | if (!src_argb || !dst_y || !dst_vu || !argbconstants || width <= 0 || |
786 | 0 | height == 0 || height == INT_MIN) { |
787 | 0 | return -1; |
788 | 0 | } |
789 | | // Negative height means invert the image. |
790 | 0 | if (height < 0) { |
791 | 0 | height = -height; |
792 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
793 | 0 | src_stride_argb = -src_stride_argb; |
794 | 0 | } |
795 | 0 | #if defined(HAS_MERGEUVROW_SSE2) |
796 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
797 | 0 | MergeUVRow = MergeUVRow_Any_SSE2; |
798 | 0 | if (IS_ALIGNED(halfwidth, 16)) { |
799 | 0 | MergeUVRow = MergeUVRow_SSE2; |
800 | 0 | } |
801 | 0 | } |
802 | 0 | #endif |
803 | 0 | #if defined(HAS_MERGEUVROW_AVX2) |
804 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
805 | 0 | MergeUVRow = MergeUVRow_Any_AVX2; |
806 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
807 | 0 | MergeUVRow = MergeUVRow_AVX2; |
808 | 0 | } |
809 | 0 | } |
810 | 0 | #endif |
811 | 0 | #if defined(HAS_MERGEUVROW_AVX512BW) |
812 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
813 | 0 | MergeUVRow = MergeUVRow_Any_AVX512BW; |
814 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
815 | 0 | MergeUVRow = MergeUVRow_AVX512BW; |
816 | 0 | } |
817 | 0 | } |
818 | 0 | #endif |
819 | | #if defined(HAS_MERGEUVROW_NEON) |
820 | | if (TestCpuFlag(kCpuHasNEON)) { |
821 | | MergeUVRow = MergeUVRow_Any_NEON; |
822 | | if (IS_ALIGNED(halfwidth, 16)) { |
823 | | MergeUVRow = MergeUVRow_NEON; |
824 | | } |
825 | | } |
826 | | #endif |
827 | | #if defined(HAS_MERGEUVROW_SVE2) |
828 | | if (TestCpuFlag(kCpuHasSVE2)) { |
829 | | MergeUVRow = MergeUVRow_SVE2; |
830 | | } |
831 | | #endif |
832 | | #if defined(HAS_MERGEUVROW_SME) |
833 | | if (TestCpuFlag(kCpuHasSME)) { |
834 | | MergeUVRow = MergeUVRow_SME; |
835 | | } |
836 | | #endif |
837 | | #if defined(HAS_MERGEUVROW_LSX) |
838 | | if (TestCpuFlag(kCpuHasLSX)) { |
839 | | MergeUVRow = MergeUVRow_Any_LSX; |
840 | | if (IS_ALIGNED(halfwidth, 16)) { |
841 | | MergeUVRow = MergeUVRow_LSX; |
842 | | } |
843 | | } |
844 | | #endif |
845 | | #if defined(HAS_MERGEUVROW_RVV) |
846 | | if (TestCpuFlag(kCpuHasRVV)) { |
847 | | MergeUVRow = MergeUVRow_RVV; |
848 | | } |
849 | | #endif |
850 | | |
851 | | // Allocate a rows of uv. |
852 | 0 | align_buffer_64(row_u, ((halfwidth + 31) & ~31) * 2); |
853 | 0 | uint8_t* row_v = row_u + ((halfwidth + 31) & ~31); |
854 | 0 | if (!row_u) |
855 | 0 | return 1; |
856 | | |
857 | 0 | for (y = 0; y < height - 1; y += 2) { |
858 | 0 | ARGBToUVMatrixRow(src_argb, src_stride_argb, row_u, row_v, width, |
859 | 0 | argbconstants); |
860 | 0 | MergeUVRow(row_v, row_u, dst_vu, halfwidth); |
861 | 0 | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
862 | 0 | ARGBToYMatrixRow(src_argb + src_stride_argb, dst_y + dst_stride_y, width, |
863 | 0 | argbconstants); |
864 | 0 | src_argb += src_stride_argb * 2; |
865 | 0 | dst_y += dst_stride_y * 2; |
866 | 0 | dst_vu += dst_stride_vu; |
867 | 0 | } |
868 | 0 | if (height & 1) { |
869 | 0 | ARGBToUVMatrixRow(src_argb, 0, row_u, row_v, width, argbconstants); |
870 | 0 | MergeUVRow(row_v, row_u, dst_vu, halfwidth); |
871 | 0 | ARGBToYMatrixRow(src_argb, dst_y, width, argbconstants); |
872 | 0 | } |
873 | 0 | free_aligned_buffer_64(row_u); |
874 | 0 | return 0; |
875 | 0 | } |
876 | | LIBYUV_API |
877 | | int ARGBToI400Matrix(const uint8_t* src_argb, |
878 | | int src_stride_argb, |
879 | | uint8_t* dst_y, |
880 | | int dst_stride_y, |
881 | | const struct ArgbConstants* constants, |
882 | | int width, |
883 | 5 | int height) { |
884 | 5 | int y; |
885 | 5 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
886 | 5 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
887 | 5 | if (!src_argb || !dst_y || !constants || width <= 0 || height == 0 || |
888 | 5 | height == INT_MIN) { |
889 | 0 | return -1; |
890 | 0 | } |
891 | 5 | if (height < 0) { |
892 | 0 | height = -height; |
893 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
894 | 0 | src_stride_argb = -src_stride_argb; |
895 | 0 | } |
896 | 5 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
897 | 5 | if (TestCpuFlag(kCpuHasAVX2)) { |
898 | 5 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
899 | 5 | if (IS_ALIGNED(width, 32)) { |
900 | 1 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
901 | 1 | } |
902 | 5 | } |
903 | 5 | #endif |
904 | 5 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
905 | 5 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
906 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
907 | 0 | if (IS_ALIGNED(width, 64)) { |
908 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
909 | 0 | } |
910 | 0 | } |
911 | 5 | #endif |
912 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
913 | | if (TestCpuFlag(kCpuHasNEON)) { |
914 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
915 | | if (IS_ALIGNED(width, 16)) { |
916 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
917 | | } |
918 | | } |
919 | | #endif |
920 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
921 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
922 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
923 | | if (IS_ALIGNED(width, 16)) { |
924 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
925 | | } |
926 | | } |
927 | | #endif |
928 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
929 | | if (TestCpuFlag(kCpuHasSVE2)) { |
930 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
931 | | } |
932 | | #endif |
933 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
934 | | if (TestCpuFlag(kCpuHasSME)) { |
935 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
936 | | } |
937 | | #endif |
938 | | |
939 | 354 | for (y = 0; y < height; ++y) { |
940 | 349 | ARGBToYMatrixRow(src_argb, dst_y, width, constants); |
941 | 349 | src_argb += src_stride_argb; |
942 | 349 | dst_y += dst_stride_y; |
943 | 349 | } |
944 | 5 | return 0; |
945 | 5 | } |
946 | | LIBYUV_API |
947 | | int ARGBToYUY2Matrix(const uint8_t* src_argb, |
948 | | int src_stride_argb, |
949 | | uint8_t* dst_yuy2, |
950 | | int dst_stride_yuy2, |
951 | | const struct ArgbConstants* constants, |
952 | | int width, |
953 | 0 | int height) { |
954 | 0 | int y; |
955 | 0 | void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb, |
956 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
957 | 0 | const struct ArgbConstants* c) = |
958 | 0 | ARGBToUVMatrixRow_C; |
959 | 0 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
960 | 0 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
961 | 0 | void (*I422ToYUY2Row)(const uint8_t* src_y, const uint8_t* src_u, |
962 | 0 | const uint8_t* src_v, uint8_t* dst_yuy2, int width) = |
963 | 0 | I422ToYUY2Row_C; |
964 | |
|
965 | 0 | if (!src_argb || !dst_yuy2 || !constants || width <= 0 || height == 0 || |
966 | 0 | height == INT_MIN) { |
967 | 0 | return -1; |
968 | 0 | } |
969 | 0 | if (height < 0) { |
970 | 0 | height = -height; |
971 | 0 | dst_yuy2 = dst_yuy2 + (ptrdiff_t)(height - 1) * dst_stride_yuy2; |
972 | 0 | dst_stride_yuy2 = -dst_stride_yuy2; |
973 | 0 | } |
974 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
975 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
976 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
977 | 0 | if (IS_ALIGNED(width, 32)) { |
978 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
979 | 0 | } |
980 | 0 | } |
981 | 0 | #endif |
982 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
983 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
984 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
985 | 0 | if (IS_ALIGNED(width, 64)) { |
986 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
987 | 0 | } |
988 | 0 | } |
989 | 0 | #endif |
990 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
991 | | if (TestCpuFlag(kCpuHasNEON)) { |
992 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
993 | | if (IS_ALIGNED(width, 16)) { |
994 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
995 | | } |
996 | | } |
997 | | #endif |
998 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
999 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
1000 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
1001 | | if (IS_ALIGNED(width, 16)) { |
1002 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
1003 | | } |
1004 | | } |
1005 | | #endif |
1006 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
1007 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1008 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
1009 | | } |
1010 | | #endif |
1011 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
1012 | | if (TestCpuFlag(kCpuHasSME)) { |
1013 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
1014 | | } |
1015 | | #endif |
1016 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX2) |
1017 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1018 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2; |
1019 | 0 | if (IS_ALIGNED(width, 32)) { |
1020 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2; |
1021 | 0 | } |
1022 | 0 | } |
1023 | 0 | #endif |
1024 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW) |
1025 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
1026 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX512BW; |
1027 | 0 | if (IS_ALIGNED(width, 64)) { |
1028 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW; |
1029 | 0 | } |
1030 | 0 | } |
1031 | 0 | #endif |
1032 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON) |
1033 | | if (TestCpuFlag(kCpuHasNEON)) { |
1034 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON; |
1035 | | if (IS_ALIGNED(width, 16)) { |
1036 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON; |
1037 | | } |
1038 | | } |
1039 | | #endif |
1040 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM) |
1041 | | if (TestCpuFlag(kCpuHasNeonI8MM)) { |
1042 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM; |
1043 | | if (IS_ALIGNED(width, 16)) { |
1044 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM; |
1045 | | } |
1046 | | } |
1047 | | #endif |
1048 | | #if defined(HAS_ARGBTOUVMATRIXROW_SVE2) |
1049 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1050 | | if (IS_ALIGNED(width, 2)) { |
1051 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2; |
1052 | | } |
1053 | | } |
1054 | | #endif |
1055 | | #if defined(HAS_ARGBTOUVMATRIXROW_SME) |
1056 | | if (TestCpuFlag(kCpuHasSME)) { |
1057 | | if (IS_ALIGNED(width, 2)) { |
1058 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME; |
1059 | | } |
1060 | | } |
1061 | | #endif |
1062 | 0 | #if defined(HAS_I422TOYUY2ROW_AVX2) |
1063 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1064 | 0 | I422ToYUY2Row = I422ToYUY2Row_Any_AVX2; |
1065 | 0 | if (IS_ALIGNED(width, 32)) { |
1066 | 0 | I422ToYUY2Row = I422ToYUY2Row_AVX2; |
1067 | 0 | } |
1068 | 0 | } |
1069 | 0 | #endif |
1070 | | #if defined(HAS_I422TOYUY2ROW_NEON) |
1071 | | if (TestCpuFlag(kCpuHasNEON)) { |
1072 | | I422ToYUY2Row = I422ToYUY2Row_Any_NEON; |
1073 | | if (IS_ALIGNED(width, 16)) { |
1074 | | I422ToYUY2Row = I422ToYUY2Row_NEON; |
1075 | | } |
1076 | | } |
1077 | | #endif |
1078 | |
|
1079 | 0 | { |
1080 | 0 | align_buffer_64(row_y, ((width + 63) & ~63) * 2); |
1081 | 0 | uint8_t* row_u = row_y + ((width + 63) & ~63); |
1082 | 0 | uint8_t* row_v = row_u + ((width + 63) & ~63) / 2; |
1083 | 0 | if (!row_y) |
1084 | 0 | return 1; |
1085 | | |
1086 | 0 | for (y = 0; y < height; ++y) { |
1087 | 0 | ARGBToUVMatrixRow(src_argb, 0, row_u, row_v, width, constants); |
1088 | 0 | ARGBToYMatrixRow(src_argb, row_y, width, constants); |
1089 | 0 | I422ToYUY2Row(row_y, row_u, row_v, dst_yuy2, width); |
1090 | 0 | src_argb += src_stride_argb; |
1091 | 0 | dst_yuy2 += dst_stride_yuy2; |
1092 | 0 | } |
1093 | |
|
1094 | 0 | free_aligned_buffer_64(row_y); |
1095 | 0 | } |
1096 | 0 | return 0; |
1097 | 0 | } |
1098 | | |
1099 | | LIBYUV_API |
1100 | | int ARGBToUYVYMatrix(const uint8_t* src_argb, |
1101 | | int src_stride_argb, |
1102 | | uint8_t* dst_uyvy, |
1103 | | int dst_stride_uyvy, |
1104 | | const struct ArgbConstants* constants, |
1105 | | int width, |
1106 | 0 | int height) { |
1107 | 0 | int y; |
1108 | 0 | void (*ARGBToUVMatrixRow)(const uint8_t* src_argb, int src_stride_argb, |
1109 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
1110 | 0 | const struct ArgbConstants* c) = |
1111 | 0 | ARGBToUVMatrixRow_C; |
1112 | 0 | void (*ARGBToYMatrixRow)(const uint8_t* src_argb, uint8_t* dst_y, int width, |
1113 | 0 | const struct ArgbConstants* c) = ARGBToYMatrixRow_C; |
1114 | 0 | void (*I422ToUYVYRow)(const uint8_t* src_y, const uint8_t* src_u, |
1115 | 0 | const uint8_t* src_v, uint8_t* dst_uyvy, int width) = |
1116 | 0 | I422ToUYVYRow_C; |
1117 | |
|
1118 | 0 | if (!src_argb || !dst_uyvy || !constants || width <= 0 || height == 0 || |
1119 | 0 | height == INT_MIN) { |
1120 | 0 | return -1; |
1121 | 0 | } |
1122 | 0 | if (height < 0) { |
1123 | 0 | height = -height; |
1124 | 0 | dst_uyvy = dst_uyvy + (ptrdiff_t)(height - 1) * dst_stride_uyvy; |
1125 | 0 | dst_stride_uyvy = -dst_stride_uyvy; |
1126 | 0 | } |
1127 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX2) |
1128 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1129 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX2; |
1130 | 0 | if (IS_ALIGNED(width, 32)) { |
1131 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX2; |
1132 | 0 | } |
1133 | 0 | } |
1134 | 0 | #endif |
1135 | 0 | #if defined(HAS_ARGBTOYMATRIXROW_AVX512BW) |
1136 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
1137 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_AVX512BW; |
1138 | 0 | if (IS_ALIGNED(width, 64)) { |
1139 | 0 | ARGBToYMatrixRow = ARGBToYMatrixRow_AVX512BW; |
1140 | 0 | } |
1141 | 0 | } |
1142 | 0 | #endif |
1143 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON) |
1144 | | if (TestCpuFlag(kCpuHasNEON)) { |
1145 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON; |
1146 | | if (IS_ALIGNED(width, 16)) { |
1147 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON; |
1148 | | } |
1149 | | } |
1150 | | #endif |
1151 | | #if defined(HAS_ARGBTOYMATRIXROW_NEON_DOTPROD) |
1152 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
1153 | | ARGBToYMatrixRow = ARGBToYMatrixRow_Any_NEON_DotProd; |
1154 | | if (IS_ALIGNED(width, 16)) { |
1155 | | ARGBToYMatrixRow = ARGBToYMatrixRow_NEON_DotProd; |
1156 | | } |
1157 | | } |
1158 | | #endif |
1159 | | #if defined(HAS_ARGBTOYMATRIXROW_SVE2) |
1160 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1161 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SVE2; |
1162 | | } |
1163 | | #endif |
1164 | | #if defined(HAS_ARGBTOYMATRIXROW_SME) |
1165 | | if (TestCpuFlag(kCpuHasSME)) { |
1166 | | ARGBToYMatrixRow = ARGBToYMatrixRow_SME; |
1167 | | } |
1168 | | #endif |
1169 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX2) |
1170 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1171 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX2; |
1172 | 0 | if (IS_ALIGNED(width, 32)) { |
1173 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX2; |
1174 | 0 | } |
1175 | 0 | } |
1176 | 0 | #endif |
1177 | 0 | #if defined(HAS_ARGBTOUVMATRIXROW_AVX512BW) |
1178 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
1179 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_AVX512BW; |
1180 | 0 | if (IS_ALIGNED(width, 64)) { |
1181 | 0 | ARGBToUVMatrixRow = ARGBToUVMatrixRow_AVX512BW; |
1182 | 0 | } |
1183 | 0 | } |
1184 | 0 | #endif |
1185 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON) |
1186 | | if (TestCpuFlag(kCpuHasNEON)) { |
1187 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON; |
1188 | | if (IS_ALIGNED(width, 16)) { |
1189 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON; |
1190 | | } |
1191 | | } |
1192 | | #endif |
1193 | | #if defined(HAS_ARGBTOUVMATRIXROW_NEON_I8MM) |
1194 | | if (TestCpuFlag(kCpuHasNeonI8MM)) { |
1195 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_Any_NEON_I8MM; |
1196 | | if (IS_ALIGNED(width, 16)) { |
1197 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_NEON_I8MM; |
1198 | | } |
1199 | | } |
1200 | | #endif |
1201 | | #if defined(HAS_ARGBTOUVMATRIXROW_SVE2) |
1202 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1203 | | if (IS_ALIGNED(width, 2)) { |
1204 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SVE2; |
1205 | | } |
1206 | | } |
1207 | | #endif |
1208 | | #if defined(HAS_ARGBTOUVMATRIXROW_SME) |
1209 | | if (TestCpuFlag(kCpuHasSME)) { |
1210 | | if (IS_ALIGNED(width, 2)) { |
1211 | | ARGBToUVMatrixRow = ARGBToUVMatrixRow_SME; |
1212 | | } |
1213 | | } |
1214 | | #endif |
1215 | 0 | #if defined(HAS_I422TOUYVYROW_AVX2) |
1216 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1217 | 0 | I422ToUYVYRow = I422ToUYVYRow_Any_AVX2; |
1218 | 0 | if (IS_ALIGNED(width, 32)) { |
1219 | 0 | I422ToUYVYRow = I422ToUYVYRow_AVX2; |
1220 | 0 | } |
1221 | 0 | } |
1222 | 0 | #endif |
1223 | | #if defined(HAS_I422TOUYVYROW_NEON) |
1224 | | if (TestCpuFlag(kCpuHasNEON)) { |
1225 | | I422ToUYVYRow = I422ToUYVYRow_Any_NEON; |
1226 | | if (IS_ALIGNED(width, 16)) { |
1227 | | I422ToUYVYRow = I422ToUYVYRow_NEON; |
1228 | | } |
1229 | | } |
1230 | | #endif |
1231 | |
|
1232 | 0 | { |
1233 | 0 | align_buffer_64(row_y, ((width + 63) & ~63) * 2); |
1234 | 0 | uint8_t* row_u = row_y + ((width + 63) & ~63); |
1235 | 0 | uint8_t* row_v = row_u + ((width + 63) & ~63) / 2; |
1236 | 0 | if (!row_y) |
1237 | 0 | return 1; |
1238 | | |
1239 | 0 | for (y = 0; y < height; ++y) { |
1240 | 0 | ARGBToUVMatrixRow(src_argb, 0, row_u, row_v, width, constants); |
1241 | 0 | ARGBToYMatrixRow(src_argb, row_y, width, constants); |
1242 | 0 | I422ToUYVYRow(row_y, row_u, row_v, dst_uyvy, width); |
1243 | 0 | src_argb += src_stride_argb; |
1244 | 0 | dst_uyvy += dst_stride_uyvy; |
1245 | 0 | } |
1246 | |
|
1247 | 0 | free_aligned_buffer_64(row_y); |
1248 | 0 | } |
1249 | 0 | return 0; |
1250 | 0 | } |
1251 | | |
1252 | | // Same as NV12 but U and V swapped. |
1253 | | LIBYUV_API |
1254 | | int ARGBToNV21(const uint8_t* src_argb, |
1255 | | int src_stride_argb, |
1256 | | uint8_t* dst_y, |
1257 | | int dst_stride_y, |
1258 | | uint8_t* dst_vu, |
1259 | | int dst_stride_vu, |
1260 | | int width, |
1261 | 0 | int height) { |
1262 | 0 | return ARGBToNV21Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, |
1263 | 0 | dst_vu, dst_stride_vu, &kArgbI601Constants, width, |
1264 | 0 | height); |
1265 | 0 | } |
1266 | | |
1267 | | LIBYUV_API |
1268 | | int ABGRToNV12(const uint8_t* src_abgr, |
1269 | | int src_stride_abgr, |
1270 | | uint8_t* dst_y, |
1271 | | int dst_stride_y, |
1272 | | uint8_t* dst_uv, |
1273 | | int dst_stride_uv, |
1274 | | int width, |
1275 | 0 | int height) { |
1276 | 0 | return ARGBToNV12Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, |
1277 | 0 | dst_uv, dst_stride_uv, &kAbgrI601Constants, width, |
1278 | 0 | height); |
1279 | 0 | } |
1280 | | |
1281 | | // Same as NV12 but U and V swapped. |
1282 | | LIBYUV_API |
1283 | | int ABGRToNV21(const uint8_t* src_abgr, |
1284 | | int src_stride_abgr, |
1285 | | uint8_t* dst_y, |
1286 | | int dst_stride_y, |
1287 | | uint8_t* dst_vu, |
1288 | | int dst_stride_vu, |
1289 | | int width, |
1290 | 0 | int height) { |
1291 | 0 | return ARGBToNV21Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, |
1292 | 0 | dst_vu, dst_stride_vu, &kAbgrI601Constants, width, |
1293 | 0 | height); |
1294 | 0 | } |
1295 | | |
1296 | | // Convert ARGB to YUY2. |
1297 | | LIBYUV_API |
1298 | | int ARGBToYUY2(const uint8_t* src_argb, |
1299 | | int src_stride_argb, |
1300 | | uint8_t* dst_yuy2, |
1301 | | int dst_stride_yuy2, |
1302 | | int width, |
1303 | 0 | int height) { |
1304 | 0 | return ARGBToYUY2Matrix(src_argb, src_stride_argb, dst_yuy2, dst_stride_yuy2, |
1305 | 0 | &kArgbI601Constants, width, height); |
1306 | 0 | } |
1307 | | |
1308 | | // Convert ARGB to UYVY. |
1309 | | LIBYUV_API |
1310 | | int ARGBToUYVY(const uint8_t* src_argb, |
1311 | | int src_stride_argb, |
1312 | | uint8_t* dst_uyvy, |
1313 | | int dst_stride_uyvy, |
1314 | | int width, |
1315 | 0 | int height) { |
1316 | 0 | return ARGBToUYVYMatrix(src_argb, src_stride_argb, dst_uyvy, dst_stride_uyvy, |
1317 | 0 | &kArgbI601Constants, width, height); |
1318 | 0 | } |
1319 | | |
1320 | | // Convert ARGB to I400. |
1321 | | LIBYUV_API |
1322 | | int ARGBToI400(const uint8_t* src_argb, |
1323 | | int src_stride_argb, |
1324 | | uint8_t* dst_y, |
1325 | | int dst_stride_y, |
1326 | | int width, |
1327 | 0 | int height) { |
1328 | 0 | return ARGBToI400Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, |
1329 | 0 | &kArgbI601Constants, width, height); |
1330 | 0 | } |
1331 | | |
1332 | | #ifndef __riscv |
1333 | | // Shuffle table for converting ARGB to RGBA. |
1334 | | static const uvec8 kShuffleMaskARGBToRGBA = { |
1335 | | 3u, 0u, 1u, 2u, 7u, 4u, 5u, 6u, 11u, 8u, 9u, 10u, 15u, 12u, 13u, 14u}; |
1336 | | |
1337 | | // Convert ARGB to RGBA. |
1338 | | LIBYUV_API |
1339 | | int ARGBToRGBA(const uint8_t* src_argb, |
1340 | | int src_stride_argb, |
1341 | | uint8_t* dst_rgba, |
1342 | | int dst_stride_rgba, |
1343 | | int width, |
1344 | 0 | int height) { |
1345 | 0 | return ARGBShuffle(src_argb, src_stride_argb, dst_rgba, dst_stride_rgba, |
1346 | 0 | (const uint8_t*)(&kShuffleMaskARGBToRGBA), width, height); |
1347 | 0 | } |
1348 | | #else |
1349 | | // Convert ARGB to RGBA. |
1350 | | LIBYUV_API |
1351 | | int ARGBToRGBA(const uint8_t* src_argb, |
1352 | | int src_stride_argb, |
1353 | | uint8_t* dst_rgba, |
1354 | | int dst_stride_rgba, |
1355 | | int width, |
1356 | | int height) { |
1357 | | int y; |
1358 | | void (*ARGBToRGBARow)(const uint8_t* src_argb, uint8_t* dst_rgba, int width) = |
1359 | | ARGBToRGBARow_C; |
1360 | | if (!src_argb || !dst_rgba || width <= 0 || height == 0 || |
1361 | | height == INT_MIN) { |
1362 | | return -1; |
1363 | | } |
1364 | | if (height < 0) { |
1365 | | height = -height; |
1366 | | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1367 | | src_stride_argb = -src_stride_argb; |
1368 | | } |
1369 | | // Coalesce rows. |
1370 | | if (src_stride_argb == width * 4 && dst_stride_rgba == width * 4 && |
1371 | | (ptrdiff_t)width * height <= INT_MAX) { |
1372 | | width *= height; |
1373 | | height = 1; |
1374 | | src_stride_argb = dst_stride_rgba = 0; |
1375 | | } |
1376 | | |
1377 | | #if defined(HAS_ARGBTORGBAROW_RVV) |
1378 | | if (TestCpuFlag(kCpuHasRVV)) { |
1379 | | ARGBToRGBARow = ARGBToRGBARow_RVV; |
1380 | | } |
1381 | | #endif |
1382 | | |
1383 | | for (y = 0; y < height; ++y) { |
1384 | | ARGBToRGBARow(src_argb, dst_rgba, width); |
1385 | | src_argb += src_stride_argb; |
1386 | | dst_rgba += dst_stride_rgba; |
1387 | | } |
1388 | | return 0; |
1389 | | } |
1390 | | #endif |
1391 | | |
1392 | | // Convert ARGB To RGB24. |
1393 | | LIBYUV_API |
1394 | | int ARGBToRGB24(const uint8_t* src_argb, |
1395 | | int src_stride_argb, |
1396 | | uint8_t* dst_rgb24, |
1397 | | int dst_stride_rgb24, |
1398 | | int width, |
1399 | 0 | int height) { |
1400 | 0 | int y; |
1401 | 0 | void (*ARGBToRGB24Row)(const uint8_t* src_argb, uint8_t* dst_rgb, int width) = |
1402 | 0 | ARGBToRGB24Row_C; |
1403 | 0 | if (!src_argb || !dst_rgb24 || width <= 0 || height == 0 || |
1404 | 0 | height == INT_MIN) { |
1405 | 0 | return -1; |
1406 | 0 | } |
1407 | 0 | if (height < 0) { |
1408 | 0 | height = -height; |
1409 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1410 | 0 | src_stride_argb = -src_stride_argb; |
1411 | 0 | } |
1412 | | // Coalesce rows. |
1413 | 0 | if (src_stride_argb == width * 4 && dst_stride_rgb24 == width * 3 && |
1414 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1415 | 0 | width *= height; |
1416 | 0 | height = 1; |
1417 | 0 | src_stride_argb = dst_stride_rgb24 = 0; |
1418 | 0 | } |
1419 | 0 | #if defined(HAS_ARGBTORGB24ROW_SSSE3) |
1420 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1421 | 0 | ARGBToRGB24Row = ARGBToRGB24Row_Any_SSSE3; |
1422 | 0 | if (IS_ALIGNED(width, 16)) { |
1423 | 0 | ARGBToRGB24Row = ARGBToRGB24Row_SSSE3; |
1424 | 0 | } |
1425 | 0 | } |
1426 | 0 | #endif |
1427 | 0 | #if defined(HAS_ARGBTORGB24ROW_AVX2) |
1428 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1429 | 0 | ARGBToRGB24Row = ARGBToRGB24Row_Any_AVX2; |
1430 | 0 | if (IS_ALIGNED(width, 32)) { |
1431 | 0 | ARGBToRGB24Row = ARGBToRGB24Row_AVX2; |
1432 | 0 | } |
1433 | 0 | } |
1434 | 0 | #endif |
1435 | 0 | #if defined(HAS_ARGBTORGB24ROW_AVX512VBMI) |
1436 | 0 | if (TestCpuFlag(kCpuHasAVX512VBMI)) { |
1437 | 0 | ARGBToRGB24Row = ARGBToRGB24Row_Any_AVX512VBMI; |
1438 | 0 | if (IS_ALIGNED(width, 32)) { |
1439 | 0 | ARGBToRGB24Row = ARGBToRGB24Row_AVX512VBMI; |
1440 | 0 | } |
1441 | 0 | } |
1442 | 0 | #endif |
1443 | | #if defined(HAS_ARGBTORGB24ROW_NEON) |
1444 | | if (TestCpuFlag(kCpuHasNEON)) { |
1445 | | ARGBToRGB24Row = ARGBToRGB24Row_Any_NEON; |
1446 | | if (IS_ALIGNED(width, 16)) { |
1447 | | ARGBToRGB24Row = ARGBToRGB24Row_NEON; |
1448 | | } |
1449 | | } |
1450 | | #endif |
1451 | | #if defined(HAS_ARGBTORGB24ROW_SVE2) |
1452 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1453 | | ARGBToRGB24Row = ARGBToRGB24Row_SVE2; |
1454 | | } |
1455 | | #endif |
1456 | | #if defined(HAS_ARGBTORGB24ROW_LSX) |
1457 | | if (TestCpuFlag(kCpuHasLSX)) { |
1458 | | ARGBToRGB24Row = ARGBToRGB24Row_Any_LSX; |
1459 | | if (IS_ALIGNED(width, 16)) { |
1460 | | ARGBToRGB24Row = ARGBToRGB24Row_LSX; |
1461 | | } |
1462 | | } |
1463 | | #endif |
1464 | | #if defined(HAS_ARGBTORGB24ROW_LASX) |
1465 | | if (TestCpuFlag(kCpuHasLASX)) { |
1466 | | ARGBToRGB24Row = ARGBToRGB24Row_Any_LASX; |
1467 | | if (IS_ALIGNED(width, 32)) { |
1468 | | ARGBToRGB24Row = ARGBToRGB24Row_LASX; |
1469 | | } |
1470 | | } |
1471 | | #endif |
1472 | | #if defined(HAS_ARGBTORGB24ROW_RVV) |
1473 | | if (TestCpuFlag(kCpuHasRVV)) { |
1474 | | ARGBToRGB24Row = ARGBToRGB24Row_RVV; |
1475 | | } |
1476 | | #endif |
1477 | |
|
1478 | 0 | for (y = 0; y < height; ++y) { |
1479 | 0 | ARGBToRGB24Row(src_argb, dst_rgb24, width); |
1480 | 0 | src_argb += src_stride_argb; |
1481 | 0 | dst_rgb24 += dst_stride_rgb24; |
1482 | 0 | } |
1483 | 0 | return 0; |
1484 | 0 | } |
1485 | | |
1486 | | // Convert ARGB To RAW. |
1487 | | LIBYUV_API |
1488 | | int ARGBToRAW(const uint8_t* src_argb, |
1489 | | int src_stride_argb, |
1490 | | uint8_t* dst_raw, |
1491 | | int dst_stride_raw, |
1492 | | int width, |
1493 | 0 | int height) { |
1494 | 0 | int y; |
1495 | 0 | void (*ARGBToRAWRow)(const uint8_t* src_argb, uint8_t* dst_rgb, int width) = |
1496 | 0 | ARGBToRAWRow_C; |
1497 | 0 | if (!src_argb || !dst_raw || width <= 0 || height == 0 || height == INT_MIN) { |
1498 | 0 | return -1; |
1499 | 0 | } |
1500 | 0 | if (height < 0) { |
1501 | 0 | height = -height; |
1502 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1503 | 0 | src_stride_argb = -src_stride_argb; |
1504 | 0 | } |
1505 | | // Coalesce rows. |
1506 | 0 | if (src_stride_argb == width * 4 && dst_stride_raw == width * 3 && |
1507 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1508 | 0 | width *= height; |
1509 | 0 | height = 1; |
1510 | 0 | src_stride_argb = dst_stride_raw = 0; |
1511 | 0 | } |
1512 | 0 | #if defined(HAS_ARGBTORAWROW_SSSE3) |
1513 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1514 | 0 | ARGBToRAWRow = ARGBToRAWRow_Any_SSSE3; |
1515 | 0 | if (IS_ALIGNED(width, 16)) { |
1516 | 0 | ARGBToRAWRow = ARGBToRAWRow_SSSE3; |
1517 | 0 | } |
1518 | 0 | } |
1519 | 0 | #endif |
1520 | 0 | #if defined(HAS_ARGBTORAWROW_AVX2) |
1521 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1522 | 0 | ARGBToRAWRow = ARGBToRAWRow_Any_AVX2; |
1523 | 0 | if (IS_ALIGNED(width, 32)) { |
1524 | 0 | ARGBToRAWRow = ARGBToRAWRow_AVX2; |
1525 | 0 | } |
1526 | 0 | } |
1527 | 0 | #endif |
1528 | | #if defined(HAS_ARGBTORAWROW_NEON) |
1529 | | if (TestCpuFlag(kCpuHasNEON)) { |
1530 | | ARGBToRAWRow = ARGBToRAWRow_Any_NEON; |
1531 | | if (IS_ALIGNED(width, 8)) { |
1532 | | ARGBToRAWRow = ARGBToRAWRow_NEON; |
1533 | | } |
1534 | | } |
1535 | | #endif |
1536 | | #if defined(HAS_ARGBTORAWROW_SVE2) |
1537 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1538 | | ARGBToRAWRow = ARGBToRAWRow_SVE2; |
1539 | | } |
1540 | | #endif |
1541 | | #if defined(HAS_ARGBTORAWROW_LSX) |
1542 | | if (TestCpuFlag(kCpuHasLSX)) { |
1543 | | ARGBToRAWRow = ARGBToRAWRow_Any_LSX; |
1544 | | if (IS_ALIGNED(width, 16)) { |
1545 | | ARGBToRAWRow = ARGBToRAWRow_LSX; |
1546 | | } |
1547 | | } |
1548 | | #endif |
1549 | | #if defined(HAS_ARGBTORAWROW_LASX) |
1550 | | if (TestCpuFlag(kCpuHasLASX)) { |
1551 | | ARGBToRAWRow = ARGBToRAWRow_Any_LASX; |
1552 | | if (IS_ALIGNED(width, 32)) { |
1553 | | ARGBToRAWRow = ARGBToRAWRow_LASX; |
1554 | | } |
1555 | | } |
1556 | | #endif |
1557 | | #if defined(HAS_ARGBTORAWROW_RVV) |
1558 | | if (TestCpuFlag(kCpuHasRVV)) { |
1559 | | ARGBToRAWRow = ARGBToRAWRow_RVV; |
1560 | | } |
1561 | | #endif |
1562 | |
|
1563 | 0 | for (y = 0; y < height; ++y) { |
1564 | 0 | ARGBToRAWRow(src_argb, dst_raw, width); |
1565 | 0 | src_argb += src_stride_argb; |
1566 | 0 | dst_raw += dst_stride_raw; |
1567 | 0 | } |
1568 | 0 | return 0; |
1569 | 0 | } |
1570 | | |
1571 | | // Ordered 8x8 dither for 888 to 565. Values from 0 to 7. |
1572 | | static const uint8_t kDither565_4x4[16] = { |
1573 | | 0, 4, 1, 5, 6, 2, 7, 3, 1, 5, 0, 4, 7, 3, 6, 2, |
1574 | | }; |
1575 | | |
1576 | | // Convert ARGB To RGB565 with 4x4 dither matrix (16 bytes). |
1577 | | LIBYUV_API |
1578 | | int ARGBToRGB565Dither(const uint8_t* src_argb, |
1579 | | int src_stride_argb, |
1580 | | uint8_t* dst_rgb565, |
1581 | | int dst_stride_rgb565, |
1582 | | const uint8_t* dither4x4, |
1583 | | int width, |
1584 | 0 | int height) { |
1585 | 0 | int y; |
1586 | 0 | uint32_t dither4[4]; |
1587 | 0 | void (*ARGBToRGB565DitherRow)(const uint8_t* src_argb, uint8_t* dst_rgb, |
1588 | 0 | uint32_t dither4, int width) = |
1589 | 0 | ARGBToRGB565DitherRow_C; |
1590 | 0 | if (!src_argb || !dst_rgb565 || width <= 0 || height == 0 || |
1591 | 0 | height == INT_MIN) { |
1592 | 0 | return -1; |
1593 | 0 | } |
1594 | 0 | if (height < 0) { |
1595 | 0 | height = -height; |
1596 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1597 | 0 | src_stride_argb = -src_stride_argb; |
1598 | 0 | } |
1599 | 0 | if (!dither4x4) { |
1600 | 0 | dither4x4 = kDither565_4x4; |
1601 | 0 | } |
1602 | 0 | memcpy(dither4, dither4x4, 16); |
1603 | |
|
1604 | 0 | #if defined(HAS_ARGBTORGB565DITHERROW_AVX2) |
1605 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1606 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_AVX2; |
1607 | 0 | if (IS_ALIGNED(width, 8)) { |
1608 | 0 | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_AVX2; |
1609 | 0 | } |
1610 | 0 | } |
1611 | 0 | #endif |
1612 | | #if defined(HAS_ARGBTORGB565DITHERROW_NEON) |
1613 | | if (TestCpuFlag(kCpuHasNEON)) { |
1614 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_NEON; |
1615 | | if (IS_ALIGNED(width, 8)) { |
1616 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_NEON; |
1617 | | } |
1618 | | } |
1619 | | #endif |
1620 | | #if defined(HAS_ARGBTORGB565DITHERROW_SVE2) |
1621 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1622 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_SVE2; |
1623 | | } |
1624 | | #endif |
1625 | | #if defined(HAS_ARGBTORGB565DITHERROW_LSX) |
1626 | | if (TestCpuFlag(kCpuHasLSX)) { |
1627 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LSX; |
1628 | | if (IS_ALIGNED(width, 8)) { |
1629 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LSX; |
1630 | | } |
1631 | | } |
1632 | | #endif |
1633 | | #if defined(HAS_ARGBTORGB565DITHERROW_LASX) |
1634 | | if (TestCpuFlag(kCpuHasLASX)) { |
1635 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_Any_LASX; |
1636 | | if (IS_ALIGNED(width, 16)) { |
1637 | | ARGBToRGB565DitherRow = ARGBToRGB565DitherRow_LASX; |
1638 | | } |
1639 | | } |
1640 | | #endif |
1641 | |
|
1642 | 0 | for (y = 0; y < height; ++y) { |
1643 | 0 | ARGBToRGB565DitherRow(src_argb, dst_rgb565, dither4[y & 3], width); |
1644 | 0 | src_argb += src_stride_argb; |
1645 | 0 | dst_rgb565 += dst_stride_rgb565; |
1646 | 0 | } |
1647 | 0 | return 0; |
1648 | 0 | } |
1649 | | |
1650 | | // Convert ARGB To RGB565. |
1651 | | // TODO(fbarchard): Consider using dither function low level with zeros. |
1652 | | LIBYUV_API |
1653 | | int ARGBToRGB565(const uint8_t* src_argb, |
1654 | | int src_stride_argb, |
1655 | | uint8_t* dst_rgb565, |
1656 | | int dst_stride_rgb565, |
1657 | | int width, |
1658 | 0 | int height) { |
1659 | 0 | int y; |
1660 | 0 | void (*ARGBToRGB565Row)(const uint8_t* src_argb, uint8_t* dst_rgb, |
1661 | 0 | int width) = ARGBToRGB565Row_C; |
1662 | 0 | if (!src_argb || !dst_rgb565 || width <= 0 || height == 0 || |
1663 | 0 | height == INT_MIN) { |
1664 | 0 | return -1; |
1665 | 0 | } |
1666 | 0 | if (height < 0) { |
1667 | 0 | height = -height; |
1668 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1669 | 0 | src_stride_argb = -src_stride_argb; |
1670 | 0 | } |
1671 | | // Coalesce rows. |
1672 | 0 | if (src_stride_argb == width * 4 && dst_stride_rgb565 == width * 2 && |
1673 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1674 | 0 | width *= height; |
1675 | 0 | height = 1; |
1676 | 0 | src_stride_argb = dst_stride_rgb565 = 0; |
1677 | 0 | } |
1678 | |
|
1679 | | #if defined(HAS_ARGBTORGB565ROW_AVX2) |
1680 | | if (TestCpuFlag(kCpuHasAVX2)) { |
1681 | | ARGBToRGB565Row = ARGBToRGB565Row_Any_AVX2; |
1682 | | if (IS_ALIGNED(width, 8)) { |
1683 | | ARGBToRGB565Row = ARGBToRGB565Row_AVX2; |
1684 | | } |
1685 | | } |
1686 | | #endif |
1687 | | #if defined(HAS_ARGBTORGB565ROW_NEON) |
1688 | | if (TestCpuFlag(kCpuHasNEON)) { |
1689 | | ARGBToRGB565Row = ARGBToRGB565Row_Any_NEON; |
1690 | | if (IS_ALIGNED(width, 8)) { |
1691 | | ARGBToRGB565Row = ARGBToRGB565Row_NEON; |
1692 | | } |
1693 | | } |
1694 | | #endif |
1695 | | #if defined(HAS_ARGBTORGB565ROW_SVE2) |
1696 | | if (TestCpuFlag(kCpuHasSVE2)) { |
1697 | | ARGBToRGB565Row = ARGBToRGB565Row_SVE2; |
1698 | | } |
1699 | | #endif |
1700 | | #if defined(HAS_ARGBTORGB565ROW_LSX) |
1701 | | if (TestCpuFlag(kCpuHasLSX)) { |
1702 | | ARGBToRGB565Row = ARGBToRGB565Row_Any_LSX; |
1703 | | if (IS_ALIGNED(width, 8)) { |
1704 | | ARGBToRGB565Row = ARGBToRGB565Row_LSX; |
1705 | | } |
1706 | | } |
1707 | | #endif |
1708 | |
|
1709 | | #if defined(HAS_ARGBTORGB565ROW_LASX) |
1710 | | if (TestCpuFlag(kCpuHasLASX)) { |
1711 | | ARGBToRGB565Row = ARGBToRGB565Row_Any_LASX; |
1712 | | if (IS_ALIGNED(width, 16)) { |
1713 | | ARGBToRGB565Row = ARGBToRGB565Row_LASX; |
1714 | | } |
1715 | | } |
1716 | | #endif |
1717 | |
|
1718 | 0 | for (y = 0; y < height; ++y) { |
1719 | 0 | ARGBToRGB565Row(src_argb, dst_rgb565, width); |
1720 | 0 | src_argb += src_stride_argb; |
1721 | 0 | dst_rgb565 += dst_stride_rgb565; |
1722 | 0 | } |
1723 | 0 | return 0; |
1724 | 0 | } |
1725 | | |
1726 | | // Convert ARGB To ARGB1555. |
1727 | | LIBYUV_API |
1728 | | int ARGBToARGB1555(const uint8_t* src_argb, |
1729 | | int src_stride_argb, |
1730 | | uint8_t* dst_argb1555, |
1731 | | int dst_stride_argb1555, |
1732 | | int width, |
1733 | 0 | int height) { |
1734 | 0 | int y; |
1735 | 0 | void (*ARGBToARGB1555Row)(const uint8_t* src_argb, uint8_t* dst_rgb, |
1736 | 0 | int width) = ARGBToARGB1555Row_C; |
1737 | 0 | if (!src_argb || !dst_argb1555 || width <= 0 || height == 0 || |
1738 | 0 | height == INT_MIN) { |
1739 | 0 | return -1; |
1740 | 0 | } |
1741 | 0 | if (height < 0) { |
1742 | 0 | height = -height; |
1743 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1744 | 0 | src_stride_argb = -src_stride_argb; |
1745 | 0 | } |
1746 | | // Coalesce rows. |
1747 | 0 | if (src_stride_argb == width * 4 && dst_stride_argb1555 == width * 2 && |
1748 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1749 | 0 | width *= height; |
1750 | 0 | height = 1; |
1751 | 0 | src_stride_argb = dst_stride_argb1555 = 0; |
1752 | 0 | } |
1753 | |
|
1754 | | #if defined(HAS_ARGBTOARGB1555ROW_AVX2) |
1755 | | if (TestCpuFlag(kCpuHasAVX2)) { |
1756 | | ARGBToARGB1555Row = ARGBToARGB1555Row_Any_AVX2; |
1757 | | if (IS_ALIGNED(width, 8)) { |
1758 | | ARGBToARGB1555Row = ARGBToARGB1555Row_AVX2; |
1759 | | } |
1760 | | } |
1761 | | #endif |
1762 | | #if defined(HAS_ARGBTOARGB1555ROW_NEON) |
1763 | | if (TestCpuFlag(kCpuHasNEON)) { |
1764 | | ARGBToARGB1555Row = ARGBToARGB1555Row_Any_NEON; |
1765 | | if (IS_ALIGNED(width, 8)) { |
1766 | | ARGBToARGB1555Row = ARGBToARGB1555Row_NEON; |
1767 | | } |
1768 | | } |
1769 | | #endif |
1770 | | #if defined(HAS_ARGBTOARGB1555ROW_LSX) |
1771 | | if (TestCpuFlag(kCpuHasLSX)) { |
1772 | | ARGBToARGB1555Row = ARGBToARGB1555Row_Any_LSX; |
1773 | | if (IS_ALIGNED(width, 8)) { |
1774 | | ARGBToARGB1555Row = ARGBToARGB1555Row_LSX; |
1775 | | } |
1776 | | } |
1777 | | #endif |
1778 | | #if defined(HAS_ARGBTOARGB1555ROW_LASX) |
1779 | | if (TestCpuFlag(kCpuHasLASX)) { |
1780 | | ARGBToARGB1555Row = ARGBToARGB1555Row_Any_LASX; |
1781 | | if (IS_ALIGNED(width, 16)) { |
1782 | | ARGBToARGB1555Row = ARGBToARGB1555Row_LASX; |
1783 | | } |
1784 | | } |
1785 | | #endif |
1786 | |
|
1787 | 0 | for (y = 0; y < height; ++y) { |
1788 | 0 | ARGBToARGB1555Row(src_argb, dst_argb1555, width); |
1789 | 0 | src_argb += src_stride_argb; |
1790 | 0 | dst_argb1555 += dst_stride_argb1555; |
1791 | 0 | } |
1792 | 0 | return 0; |
1793 | 0 | } |
1794 | | |
1795 | | // Convert ARGB To ARGB4444. |
1796 | | LIBYUV_API |
1797 | | int ARGBToARGB4444(const uint8_t* src_argb, |
1798 | | int src_stride_argb, |
1799 | | uint8_t* dst_argb4444, |
1800 | | int dst_stride_argb4444, |
1801 | | int width, |
1802 | 0 | int height) { |
1803 | 0 | int y; |
1804 | 0 | void (*ARGBToARGB4444Row)(const uint8_t* src_argb, uint8_t* dst_rgb, |
1805 | 0 | int width) = ARGBToARGB4444Row_C; |
1806 | 0 | if (!src_argb || !dst_argb4444 || width <= 0 || height == 0 || |
1807 | 0 | height == INT_MIN) { |
1808 | 0 | return -1; |
1809 | 0 | } |
1810 | 0 | if (height < 0) { |
1811 | 0 | height = -height; |
1812 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1813 | 0 | src_stride_argb = -src_stride_argb; |
1814 | 0 | } |
1815 | | // Coalesce rows. |
1816 | 0 | if (src_stride_argb == width * 4 && dst_stride_argb4444 == width * 2 && |
1817 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1818 | 0 | width *= height; |
1819 | 0 | height = 1; |
1820 | 0 | src_stride_argb = dst_stride_argb4444 = 0; |
1821 | 0 | } |
1822 | |
|
1823 | | #if defined(HAS_ARGBTOARGB4444ROW_AVX2) |
1824 | | if (TestCpuFlag(kCpuHasAVX2)) { |
1825 | | ARGBToARGB4444Row = ARGBToARGB4444Row_Any_AVX2; |
1826 | | if (IS_ALIGNED(width, 8)) { |
1827 | | ARGBToARGB4444Row = ARGBToARGB4444Row_AVX2; |
1828 | | } |
1829 | | } |
1830 | | #endif |
1831 | | #if defined(HAS_ARGBTOARGB4444ROW_NEON) |
1832 | | if (TestCpuFlag(kCpuHasNEON)) { |
1833 | | ARGBToARGB4444Row = ARGBToARGB4444Row_Any_NEON; |
1834 | | if (IS_ALIGNED(width, 8)) { |
1835 | | ARGBToARGB4444Row = ARGBToARGB4444Row_NEON; |
1836 | | } |
1837 | | } |
1838 | | #endif |
1839 | | #if defined(HAS_ARGBTOARGB4444ROW_LSX) |
1840 | | if (TestCpuFlag(kCpuHasLSX)) { |
1841 | | ARGBToARGB4444Row = ARGBToARGB4444Row_Any_LSX; |
1842 | | if (IS_ALIGNED(width, 8)) { |
1843 | | ARGBToARGB4444Row = ARGBToARGB4444Row_LSX; |
1844 | | } |
1845 | | } |
1846 | | #endif |
1847 | | #if defined(HAS_ARGBTOARGB4444ROW_LASX) |
1848 | | if (TestCpuFlag(kCpuHasLASX)) { |
1849 | | ARGBToARGB4444Row = ARGBToARGB4444Row_Any_LASX; |
1850 | | if (IS_ALIGNED(width, 16)) { |
1851 | | ARGBToARGB4444Row = ARGBToARGB4444Row_LASX; |
1852 | | } |
1853 | | } |
1854 | | #endif |
1855 | |
|
1856 | 0 | for (y = 0; y < height; ++y) { |
1857 | 0 | ARGBToARGB4444Row(src_argb, dst_argb4444, width); |
1858 | 0 | src_argb += src_stride_argb; |
1859 | 0 | dst_argb4444 += dst_stride_argb4444; |
1860 | 0 | } |
1861 | 0 | return 0; |
1862 | 0 | } |
1863 | | |
1864 | | // Convert ABGR To AR30. |
1865 | | LIBYUV_API |
1866 | | int ABGRToAR30(const uint8_t* src_abgr, |
1867 | | int src_stride_abgr, |
1868 | | uint8_t* dst_ar30, |
1869 | | int dst_stride_ar30, |
1870 | | int width, |
1871 | 0 | int height) { |
1872 | 0 | int y; |
1873 | 0 | void (*ABGRToAR30Row)(const uint8_t* src_abgr, uint8_t* dst_rgb, int width) = |
1874 | 0 | ABGRToAR30Row_C; |
1875 | 0 | if (!src_abgr || !dst_ar30 || width <= 0 || height == 0 || |
1876 | 0 | height == INT_MIN) { |
1877 | 0 | return -1; |
1878 | 0 | } |
1879 | 0 | if (height < 0) { |
1880 | 0 | height = -height; |
1881 | 0 | src_abgr = src_abgr + (ptrdiff_t)(height - 1) * src_stride_abgr; |
1882 | 0 | src_stride_abgr = -src_stride_abgr; |
1883 | 0 | } |
1884 | | // Coalesce rows. |
1885 | 0 | if (src_stride_abgr == width * 4 && dst_stride_ar30 == width * 4 && |
1886 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1887 | 0 | width *= height; |
1888 | 0 | height = 1; |
1889 | 0 | src_stride_abgr = dst_stride_ar30 = 0; |
1890 | 0 | } |
1891 | | #if defined(HAS_ABGRTOAR30ROW_NEON) |
1892 | | if (TestCpuFlag(kCpuHasNEON)) { |
1893 | | ABGRToAR30Row = ABGRToAR30Row_Any_NEON; |
1894 | | if (IS_ALIGNED(width, 8)) { |
1895 | | ABGRToAR30Row = ABGRToAR30Row_NEON; |
1896 | | } |
1897 | | } |
1898 | | #endif |
1899 | 0 | #if defined(HAS_ABGRTOAR30ROW_SSSE3) |
1900 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1901 | 0 | ABGRToAR30Row = ABGRToAR30Row_Any_SSSE3; |
1902 | 0 | if (IS_ALIGNED(width, 4)) { |
1903 | 0 | ABGRToAR30Row = ABGRToAR30Row_SSSE3; |
1904 | 0 | } |
1905 | 0 | } |
1906 | 0 | #endif |
1907 | 0 | #if defined(HAS_ABGRTOAR30ROW_AVX2) |
1908 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1909 | 0 | ABGRToAR30Row = ABGRToAR30Row_Any_AVX2; |
1910 | 0 | if (IS_ALIGNED(width, 8)) { |
1911 | 0 | ABGRToAR30Row = ABGRToAR30Row_AVX2; |
1912 | 0 | } |
1913 | 0 | } |
1914 | 0 | #endif |
1915 | 0 | for (y = 0; y < height; ++y) { |
1916 | 0 | ABGRToAR30Row(src_abgr, dst_ar30, width); |
1917 | 0 | src_abgr += src_stride_abgr; |
1918 | 0 | dst_ar30 += dst_stride_ar30; |
1919 | 0 | } |
1920 | 0 | return 0; |
1921 | 0 | } |
1922 | | |
1923 | | // Convert ARGB To AR30. |
1924 | | LIBYUV_API |
1925 | | int ARGBToAR30(const uint8_t* src_argb, |
1926 | | int src_stride_argb, |
1927 | | uint8_t* dst_ar30, |
1928 | | int dst_stride_ar30, |
1929 | | int width, |
1930 | 0 | int height) { |
1931 | 0 | int y; |
1932 | 0 | void (*ARGBToAR30Row)(const uint8_t* src_argb, uint8_t* dst_rgb, int width) = |
1933 | 0 | ARGBToAR30Row_C; |
1934 | 0 | if (!src_argb || !dst_ar30 || width <= 0 || height == 0 || |
1935 | 0 | height == INT_MIN) { |
1936 | 0 | return -1; |
1937 | 0 | } |
1938 | 0 | if (height < 0) { |
1939 | 0 | height = -height; |
1940 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
1941 | 0 | src_stride_argb = -src_stride_argb; |
1942 | 0 | } |
1943 | | // Coalesce rows. |
1944 | 0 | if (src_stride_argb == width * 4 && dst_stride_ar30 == width * 4 && |
1945 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
1946 | 0 | width *= height; |
1947 | 0 | height = 1; |
1948 | 0 | src_stride_argb = dst_stride_ar30 = 0; |
1949 | 0 | } |
1950 | | #if defined(HAS_ARGBTOAR30ROW_NEON) |
1951 | | if (TestCpuFlag(kCpuHasNEON)) { |
1952 | | ARGBToAR30Row = ARGBToAR30Row_Any_NEON; |
1953 | | if (IS_ALIGNED(width, 8)) { |
1954 | | ARGBToAR30Row = ARGBToAR30Row_NEON; |
1955 | | } |
1956 | | } |
1957 | | #endif |
1958 | 0 | #if defined(HAS_ARGBTOAR30ROW_SSSE3) |
1959 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
1960 | 0 | ARGBToAR30Row = ARGBToAR30Row_Any_SSSE3; |
1961 | 0 | if (IS_ALIGNED(width, 4)) { |
1962 | 0 | ARGBToAR30Row = ARGBToAR30Row_SSSE3; |
1963 | 0 | } |
1964 | 0 | } |
1965 | 0 | #endif |
1966 | 0 | #if defined(HAS_ARGBTOAR30ROW_AVX2) |
1967 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
1968 | 0 | ARGBToAR30Row = ARGBToAR30Row_Any_AVX2; |
1969 | 0 | if (IS_ALIGNED(width, 8)) { |
1970 | 0 | ARGBToAR30Row = ARGBToAR30Row_AVX2; |
1971 | 0 | } |
1972 | 0 | } |
1973 | 0 | #endif |
1974 | 0 | for (y = 0; y < height; ++y) { |
1975 | 0 | ARGBToAR30Row(src_argb, dst_ar30, width); |
1976 | 0 | src_argb += src_stride_argb; |
1977 | 0 | dst_ar30 += dst_stride_ar30; |
1978 | 0 | } |
1979 | 0 | return 0; |
1980 | 0 | } |
1981 | | |
1982 | | // ARGB little endian (bgra in memory) to J444 |
1983 | | LIBYUV_API |
1984 | | int ARGBToJ444(const uint8_t* src_argb, |
1985 | | int src_stride_argb, |
1986 | | uint8_t* dst_y, |
1987 | | int dst_stride_y, |
1988 | | uint8_t* dst_u, |
1989 | | int dst_stride_u, |
1990 | | uint8_t* dst_v, |
1991 | | int dst_stride_v, |
1992 | | int width, |
1993 | 0 | int height) { |
1994 | 0 | return ARGBToI444Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u, |
1995 | 0 | dst_stride_u, dst_v, dst_stride_v, |
1996 | 0 | &kArgbJPEGConstants, width, height); |
1997 | 0 | } |
1998 | | |
1999 | | // Convert ARGB to J420. (JPeg full range I420). |
2000 | | LIBYUV_API |
2001 | | int ARGBToJ420(const uint8_t* src_argb, |
2002 | | int src_stride_argb, |
2003 | | uint8_t* dst_y, |
2004 | | int dst_stride_y, |
2005 | | uint8_t* dst_u, |
2006 | | int dst_stride_u, |
2007 | | uint8_t* dst_v, |
2008 | | int dst_stride_v, |
2009 | | int width, |
2010 | 0 | int height) { |
2011 | 0 | return ARGBToI420Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u, |
2012 | 0 | dst_stride_u, dst_v, dst_stride_v, |
2013 | 0 | &kArgbJPEGConstants, width, height); |
2014 | 0 | } |
2015 | | |
2016 | | // Convert ARGB to J422. (JPeg full range I422). |
2017 | | LIBYUV_API |
2018 | | int ARGBToJ422(const uint8_t* src_argb, |
2019 | | int src_stride_argb, |
2020 | | uint8_t* dst_y, |
2021 | | int dst_stride_y, |
2022 | | uint8_t* dst_u, |
2023 | | int dst_stride_u, |
2024 | | uint8_t* dst_v, |
2025 | | int dst_stride_v, |
2026 | | int width, |
2027 | 184k | int height) { |
2028 | 184k | return ARGBToI422Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, dst_u, |
2029 | 184k | dst_stride_u, dst_v, dst_stride_v, |
2030 | 184k | &kArgbJPEGConstants, width, height); |
2031 | 184k | } |
2032 | | |
2033 | | // Convert ARGB to J400. |
2034 | | LIBYUV_API |
2035 | | int ARGBToJ400(const uint8_t* src_argb, |
2036 | | int src_stride_argb, |
2037 | | uint8_t* dst_y, |
2038 | | int dst_stride_y, |
2039 | | int width, |
2040 | 0 | int height) { |
2041 | 0 | return ARGBToI400Matrix(src_argb, src_stride_argb, dst_y, dst_stride_y, |
2042 | 0 | &kArgbJPEGConstants, width, height); |
2043 | 0 | } |
2044 | | |
2045 | | // Convert RGBA to J400. |
2046 | | LIBYUV_API |
2047 | | int RGBAToJ400(const uint8_t* src_rgba, |
2048 | | int src_stride_rgba, |
2049 | | uint8_t* dst_yj, |
2050 | | int dst_stride_yj, |
2051 | | int width, |
2052 | 0 | int height) { |
2053 | 0 | return ARGBToI400Matrix(src_rgba, src_stride_rgba, dst_yj, dst_stride_yj, |
2054 | 0 | &kRgbaJPEGConstants, width, height); |
2055 | 0 | } |
2056 | | |
2057 | | |
2058 | | // Convert ABGR to J420. (JPeg full range I420). |
2059 | | LIBYUV_API |
2060 | | int ABGRToJ420(const uint8_t* src_abgr, |
2061 | | int src_stride_abgr, |
2062 | | uint8_t* dst_y, |
2063 | | int dst_stride_y, |
2064 | | uint8_t* dst_u, |
2065 | | int dst_stride_u, |
2066 | | uint8_t* dst_v, |
2067 | | int dst_stride_v, |
2068 | | int width, |
2069 | 7 | int height) { |
2070 | 7 | return ARGBToI420Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, dst_u, |
2071 | 7 | dst_stride_u, dst_v, dst_stride_v, |
2072 | 7 | &kAbgrJPEGConstants, width, height); |
2073 | 7 | } |
2074 | | |
2075 | | // Convert ABGR to J422. (JPeg full range I422). |
2076 | | LIBYUV_API |
2077 | | int ABGRToJ422(const uint8_t* src_abgr, |
2078 | | int src_stride_abgr, |
2079 | | uint8_t* dst_y, |
2080 | | int dst_stride_y, |
2081 | | uint8_t* dst_u, |
2082 | | int dst_stride_u, |
2083 | | uint8_t* dst_v, |
2084 | | int dst_stride_v, |
2085 | | int width, |
2086 | 2 | int height) { |
2087 | 2 | return ARGBToI422Matrix(src_abgr, src_stride_abgr, dst_y, dst_stride_y, dst_u, |
2088 | 2 | dst_stride_u, dst_v, dst_stride_v, |
2089 | 2 | &kAbgrJPEGConstants, width, height); |
2090 | 2 | } |
2091 | | |
2092 | | // Convert ABGR to J400. |
2093 | | LIBYUV_API |
2094 | | int ABGRToJ400(const uint8_t* src_abgr, |
2095 | | int src_stride_abgr, |
2096 | | uint8_t* dst_yj, |
2097 | | int dst_stride_yj, |
2098 | | int width, |
2099 | 5 | int height) { |
2100 | 5 | return ARGBToI400Matrix(src_abgr, src_stride_abgr, dst_yj, dst_stride_yj, |
2101 | 5 | &kAbgrJPEGConstants, width, height); |
2102 | 5 | } |
2103 | | |
2104 | | // Convert ARGB to AR64. |
2105 | | LIBYUV_API |
2106 | | int ARGBToAR64(const uint8_t* src_argb, |
2107 | | int src_stride_argb, |
2108 | | uint16_t* dst_ar64, |
2109 | | int dst_stride_ar64, |
2110 | | int width, |
2111 | 0 | int height) { |
2112 | 0 | int y; |
2113 | 0 | void (*ARGBToAR64Row)(const uint8_t* src_argb, uint16_t* dst_ar64, |
2114 | 0 | int width) = ARGBToAR64Row_C; |
2115 | 0 | if (!src_argb || !dst_ar64 || width <= 0 || height == 0 || |
2116 | 0 | height == INT_MIN) { |
2117 | 0 | return -1; |
2118 | 0 | } |
2119 | | // Negative height means invert the image. |
2120 | 0 | if (height < 0) { |
2121 | 0 | height = -height; |
2122 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
2123 | 0 | src_stride_argb = -src_stride_argb; |
2124 | 0 | } |
2125 | | // Coalesce rows. |
2126 | 0 | if (src_stride_argb == width * 4 && dst_stride_ar64 == width * 4 && |
2127 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
2128 | 0 | width *= height; |
2129 | 0 | height = 1; |
2130 | 0 | src_stride_argb = dst_stride_ar64 = 0; |
2131 | 0 | } |
2132 | 0 | #if defined(HAS_ARGBTOAR64ROW_SSSE3) |
2133 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2134 | 0 | ARGBToAR64Row = ARGBToAR64Row_Any_SSSE3; |
2135 | 0 | if (IS_ALIGNED(width, 4)) { |
2136 | 0 | ARGBToAR64Row = ARGBToAR64Row_SSSE3; |
2137 | 0 | } |
2138 | 0 | } |
2139 | 0 | #endif |
2140 | 0 | #if defined(HAS_ARGBTOAR64ROW_AVX2) |
2141 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2142 | 0 | ARGBToAR64Row = ARGBToAR64Row_Any_AVX2; |
2143 | 0 | if (IS_ALIGNED(width, 8)) { |
2144 | 0 | ARGBToAR64Row = ARGBToAR64Row_AVX2; |
2145 | 0 | } |
2146 | 0 | } |
2147 | 0 | #endif |
2148 | | #if defined(HAS_ARGBTOAR64ROW_NEON) |
2149 | | if (TestCpuFlag(kCpuHasNEON)) { |
2150 | | ARGBToAR64Row = ARGBToAR64Row_Any_NEON; |
2151 | | if (IS_ALIGNED(width, 8)) { |
2152 | | ARGBToAR64Row = ARGBToAR64Row_NEON; |
2153 | | } |
2154 | | } |
2155 | | #endif |
2156 | | #if defined(HAS_ARGBTOAR64ROW_RVV) |
2157 | | if (TestCpuFlag(kCpuHasRVV)) { |
2158 | | ARGBToAR64Row = ARGBToAR64Row_RVV; |
2159 | | } |
2160 | | #endif |
2161 | |
|
2162 | 0 | for (y = 0; y < height; ++y) { |
2163 | 0 | ARGBToAR64Row(src_argb, dst_ar64, width); |
2164 | 0 | src_argb += src_stride_argb; |
2165 | 0 | dst_ar64 += dst_stride_ar64; |
2166 | 0 | } |
2167 | 0 | return 0; |
2168 | 0 | } |
2169 | | |
2170 | | // Convert ARGB to AB64. |
2171 | | LIBYUV_API |
2172 | | int ARGBToAB64(const uint8_t* src_argb, |
2173 | | int src_stride_argb, |
2174 | | uint16_t* dst_ab64, |
2175 | | int dst_stride_ab64, |
2176 | | int width, |
2177 | 0 | int height) { |
2178 | 0 | int y; |
2179 | 0 | void (*ARGBToAB64Row)(const uint8_t* src_argb, uint16_t* dst_ar64, |
2180 | 0 | int width) = ARGBToAB64Row_C; |
2181 | 0 | if (!src_argb || !dst_ab64 || width <= 0 || height == 0 || |
2182 | 0 | height == INT_MIN) { |
2183 | 0 | return -1; |
2184 | 0 | } |
2185 | | // Negative height means invert the image. |
2186 | 0 | if (height < 0) { |
2187 | 0 | height = -height; |
2188 | 0 | src_argb = src_argb + (ptrdiff_t)(height - 1) * src_stride_argb; |
2189 | 0 | src_stride_argb = -src_stride_argb; |
2190 | 0 | } |
2191 | | // Coalesce rows. |
2192 | 0 | if (src_stride_argb == width * 4 && dst_stride_ab64 == width * 4 && |
2193 | 0 | (ptrdiff_t)width * height <= INT_MAX) { |
2194 | 0 | width *= height; |
2195 | 0 | height = 1; |
2196 | 0 | src_stride_argb = dst_stride_ab64 = 0; |
2197 | 0 | } |
2198 | 0 | #if defined(HAS_ARGBTOAB64ROW_SSSE3) |
2199 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2200 | 0 | ARGBToAB64Row = ARGBToAB64Row_Any_SSSE3; |
2201 | 0 | if (IS_ALIGNED(width, 4)) { |
2202 | 0 | ARGBToAB64Row = ARGBToAB64Row_SSSE3; |
2203 | 0 | } |
2204 | 0 | } |
2205 | 0 | #endif |
2206 | 0 | #if defined(HAS_ARGBTOAB64ROW_AVX2) |
2207 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2208 | 0 | ARGBToAB64Row = ARGBToAB64Row_Any_AVX2; |
2209 | 0 | if (IS_ALIGNED(width, 8)) { |
2210 | 0 | ARGBToAB64Row = ARGBToAB64Row_AVX2; |
2211 | 0 | } |
2212 | 0 | } |
2213 | 0 | #endif |
2214 | | #if defined(HAS_ARGBTOAB64ROW_NEON) |
2215 | | if (TestCpuFlag(kCpuHasNEON)) { |
2216 | | ARGBToAB64Row = ARGBToAB64Row_Any_NEON; |
2217 | | if (IS_ALIGNED(width, 8)) { |
2218 | | ARGBToAB64Row = ARGBToAB64Row_NEON; |
2219 | | } |
2220 | | } |
2221 | | #endif |
2222 | | #if defined(HAS_ARGBTOAB64ROW_RVV) |
2223 | | if (TestCpuFlag(kCpuHasRVV)) { |
2224 | | ARGBToAB64Row = ARGBToAB64Row_RVV; |
2225 | | } |
2226 | | #endif |
2227 | |
|
2228 | 0 | for (y = 0; y < height; ++y) { |
2229 | 0 | ARGBToAB64Row(src_argb, dst_ab64, width); |
2230 | 0 | src_argb += src_stride_argb; |
2231 | 0 | dst_ab64 += dst_stride_ab64; |
2232 | 0 | } |
2233 | 0 | return 0; |
2234 | 0 | } |
2235 | | |
2236 | | // Convert RAW to NV21 with Matrix. |
2237 | | static int RAWToNVMatrix(const uint8_t* src_raw, |
2238 | | int src_stride_raw, |
2239 | | uint8_t* dst_y, |
2240 | | int dst_stride_y, |
2241 | | uint8_t* dst_uv, |
2242 | | int dst_stride_uv, |
2243 | | const struct ArgbConstants* argbconstants, |
2244 | | int width, |
2245 | | int height, |
2246 | 0 | int is_nv21) { |
2247 | 0 | int y; |
2248 | 0 | int halfwidth = (width + 1) >> 1; |
2249 | 0 | void (*RGBToUVMatrixRow)(const uint8_t* src_rgb, int src_stride_rgb, |
2250 | 0 | uint8_t* dst_u, uint8_t* dst_v, int width, |
2251 | 0 | const struct ArgbConstants* c) = |
2252 | 0 | RGBToUVMatrixRow_C; |
2253 | 0 | void (*RGBToYMatrixRow)(const uint8_t* src_rgb, uint8_t* dst_y, int width, |
2254 | 0 | const struct ArgbConstants* c) = RGBToYMatrixRow_C; |
2255 | 0 | void (*MergeUVRow)(const uint8_t* src_uj, const uint8_t* src_vj, |
2256 | 0 | uint8_t* dst_vu, int width) = MergeUVRow_C; |
2257 | |
|
2258 | 0 | if (!src_raw || !dst_y || !dst_uv || !argbconstants || width <= 0 || |
2259 | 0 | height == 0 || height == INT_MIN) { |
2260 | 0 | return -1; |
2261 | 0 | } |
2262 | | // Negative height means invert the image. |
2263 | 0 | if (height < 0) { |
2264 | 0 | height = -height; |
2265 | 0 | src_raw = src_raw + (ptrdiff_t)(height - 1) * src_stride_raw; |
2266 | 0 | src_stride_raw = -src_stride_raw; |
2267 | 0 | } |
2268 | |
|
2269 | 0 | #if defined(HAS_RGBTOYMATRIXROW_SSSE3) |
2270 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2271 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_SSSE3; |
2272 | 0 | if (IS_ALIGNED(width, 16)) { |
2273 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_SSSE3; |
2274 | 0 | } |
2275 | 0 | } |
2276 | 0 | #endif |
2277 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX2) |
2278 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2279 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX2; |
2280 | 0 | if (IS_ALIGNED(width, 32)) { |
2281 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX2; |
2282 | 0 | } |
2283 | 0 | } |
2284 | 0 | #endif |
2285 | 0 | #if defined(HAS_RGBTOYMATRIXROW_AVX512BW) |
2286 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2287 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_Any_AVX512BW; |
2288 | 0 | if (IS_ALIGNED(width, 64)) { |
2289 | 0 | RGBToYMatrixRow = RGBToYMatrixRow_AVX512BW; |
2290 | 0 | } |
2291 | 0 | } |
2292 | 0 | #endif |
2293 | | #if defined(HAS_RGBTOYMATRIXROW_NEON) |
2294 | | if (TestCpuFlag(kCpuHasNEON)) { |
2295 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON; |
2296 | | if (IS_ALIGNED(width, 16)) { |
2297 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON; |
2298 | | } |
2299 | | } |
2300 | | #endif |
2301 | | #if defined(HAS_RGBTOYMATRIXROW_NEON_DOTPROD) |
2302 | | if (TestCpuFlag(kCpuHasNeonDotProd)) { |
2303 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_NEON_DotProd; |
2304 | | if (IS_ALIGNED(width, 16)) { |
2305 | | RGBToYMatrixRow = RGBToYMatrixRow_NEON_DotProd; |
2306 | | } |
2307 | | } |
2308 | | #endif |
2309 | | #if defined(HAS_RGBTOYMATRIXROW_LASX) |
2310 | | if (TestCpuFlag(kCpuHasLASX)) { |
2311 | | RGBToYMatrixRow = RGBToYMatrixRow_Any_LASX; |
2312 | | if (IS_ALIGNED(width, 32)) { |
2313 | | RGBToYMatrixRow = RGBToYMatrixRow_LASX; |
2314 | | } |
2315 | | } |
2316 | | #endif |
2317 | | #if defined(HAS_RGBTOYMATRIXROW_RVV) |
2318 | | if (TestCpuFlag(kCpuHasRVV)) { |
2319 | | RGBToYMatrixRow = RGBToYMatrixRow_RVV; |
2320 | | } |
2321 | | #endif |
2322 | |
|
2323 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_SSSE3) |
2324 | 0 | if (TestCpuFlag(kCpuHasSSSE3)) { |
2325 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_SSSE3; |
2326 | 0 | if (IS_ALIGNED(width, 16)) { |
2327 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_SSSE3; |
2328 | 0 | } |
2329 | 0 | } |
2330 | 0 | #endif |
2331 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX2) |
2332 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2333 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX2; |
2334 | 0 | if (IS_ALIGNED(width, 32)) { |
2335 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX2; |
2336 | 0 | } |
2337 | 0 | } |
2338 | 0 | #endif |
2339 | 0 | #if defined(HAS_RGBTOUVMATRIXROW_AVX512BW) |
2340 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2341 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_AVX512BW; |
2342 | 0 | if (IS_ALIGNED(width, 64)) { |
2343 | 0 | RGBToUVMatrixRow = RGBToUVMatrixRow_AVX512BW; |
2344 | 0 | } |
2345 | 0 | } |
2346 | 0 | #endif |
2347 | | #if defined(HAS_RGBTOUVMATRIXROW_NEON) |
2348 | | if (TestCpuFlag(kCpuHasNEON)) { |
2349 | | RGBToUVMatrixRow = RGBToUVMatrixRow_Any_NEON; |
2350 | | if (IS_ALIGNED(width, 16)) { |
2351 | | RGBToUVMatrixRow = RGBToUVMatrixRow_NEON; |
2352 | | } |
2353 | | } |
2354 | | #endif |
2355 | |
|
2356 | 0 | #if defined(HAS_MERGEUVROW_SSE2) |
2357 | 0 | if (TestCpuFlag(kCpuHasSSE2)) { |
2358 | 0 | MergeUVRow = MergeUVRow_Any_SSE2; |
2359 | 0 | if (IS_ALIGNED(halfwidth, 16)) { |
2360 | 0 | MergeUVRow = MergeUVRow_SSE2; |
2361 | 0 | } |
2362 | 0 | } |
2363 | 0 | #endif |
2364 | 0 | #if defined(HAS_MERGEUVROW_AVX2) |
2365 | 0 | if (TestCpuFlag(kCpuHasAVX2)) { |
2366 | 0 | MergeUVRow = MergeUVRow_Any_AVX2; |
2367 | 0 | if (IS_ALIGNED(halfwidth, 32)) { |
2368 | 0 | MergeUVRow = MergeUVRow_AVX2; |
2369 | 0 | } |
2370 | 0 | } |
2371 | 0 | #endif |
2372 | 0 | #if defined(HAS_MERGEUVROW_AVX512BW) |
2373 | 0 | if (TestCpuFlag(kCpuHasAVX512BW)) { |
2374 | 0 | MergeUVRow = MergeUVRow_Any_AVX512BW; |
2375 | 0 | if (IS_ALIGNED(halfwidth, 64)) { |
2376 | 0 | MergeUVRow = MergeUVRow_AVX512BW; |
2377 | 0 | } |
2378 | 0 | } |
2379 | 0 | #endif |
2380 | | #if defined(HAS_MERGEUVROW_NEON) |
2381 | | if (TestCpuFlag(kCpuHasNEON)) { |
2382 | | MergeUVRow = MergeUVRow_Any_NEON; |
2383 | | if (IS_ALIGNED(halfwidth, 16)) { |
2384 | | MergeUVRow = MergeUVRow_NEON; |
2385 | | } |
2386 | | } |
2387 | | #endif |
2388 | | #if defined(HAS_MERGEUVROW_SVE2) |
2389 | | if (TestCpuFlag(kCpuHasSVE2)) { |
2390 | | MergeUVRow = MergeUVRow_SVE2; |
2391 | | } |
2392 | | #endif |
2393 | | #if defined(HAS_MERGEUVROW_SME) |
2394 | | if (TestCpuFlag(kCpuHasSME)) { |
2395 | | MergeUVRow = MergeUVRow_SME; |
2396 | | } |
2397 | | #endif |
2398 | | #if defined(HAS_MERGEUVROW_LSX) |
2399 | | if (TestCpuFlag(kCpuHasLSX)) { |
2400 | | MergeUVRow = MergeUVRow_Any_LSX; |
2401 | | if (IS_ALIGNED(halfwidth, 16)) { |
2402 | | MergeUVRow = MergeUVRow_LSX; |
2403 | | } |
2404 | | } |
2405 | | #endif |
2406 | | #if defined(HAS_MERGEUVROW_RVV) |
2407 | | if (TestCpuFlag(kCpuHasRVV)) { |
2408 | | MergeUVRow = MergeUVRow_RVV; |
2409 | | } |
2410 | | #endif |
2411 | |
|
2412 | 0 | { |
2413 | | // Allocate 1 row of U and 1 row of V. |
2414 | 0 | align_buffer_64(row_u, halfwidth); |
2415 | 0 | align_buffer_64(row_v, halfwidth); |
2416 | |
|
2417 | 0 | if (!row_u || !row_v) { |
2418 | 0 | free_aligned_buffer_64(row_u); |
2419 | 0 | free_aligned_buffer_64(row_v); |
2420 | 0 | return 1; |
2421 | 0 | } |
2422 | | |
2423 | 0 | const uint8_t* first_chroma = is_nv21 ? row_v : row_u; |
2424 | 0 | const uint8_t* second_chroma = is_nv21 ? row_u : row_v; |
2425 | |
|
2426 | 0 | for (y = 0; y < height - 1; y += 2) { |
2427 | 0 | RGBToUVMatrixRow(src_raw, src_stride_raw, row_u, row_v, width, |
2428 | 0 | argbconstants); |
2429 | 0 | MergeUVRow(first_chroma, second_chroma, dst_uv, halfwidth); |
2430 | 0 | RGBToYMatrixRow(src_raw, dst_y, width, argbconstants); |
2431 | 0 | RGBToYMatrixRow(src_raw + src_stride_raw, dst_y + dst_stride_y, width, |
2432 | 0 | argbconstants); |
2433 | 0 | src_raw += src_stride_raw * 2; |
2434 | 0 | dst_y += dst_stride_y * 2; |
2435 | 0 | dst_uv += dst_stride_uv; |
2436 | 0 | } |
2437 | 0 | if (height & 1) { |
2438 | 0 | RGBToUVMatrixRow(src_raw, 0, row_u, row_v, width, argbconstants); |
2439 | 0 | MergeUVRow(first_chroma, second_chroma, dst_uv, halfwidth); |
2440 | 0 | RGBToYMatrixRow(src_raw, dst_y, width, argbconstants); |
2441 | 0 | } |
2442 | 0 | free_aligned_buffer_64(row_v); |
2443 | 0 | free_aligned_buffer_64(row_u); |
2444 | 0 | } |
2445 | 0 | return 0; |
2446 | 0 | } |
2447 | | |
2448 | | LIBYUV_API |
2449 | | int RAWToNV21Matrix(const uint8_t* src_raw, |
2450 | | int src_stride_raw, |
2451 | | uint8_t* dst_y, |
2452 | | int dst_stride_y, |
2453 | | uint8_t* dst_vu, |
2454 | | int dst_stride_vu, |
2455 | | const struct ArgbConstants* argbconstants, |
2456 | | int width, |
2457 | 0 | int height) { |
2458 | 0 | return RAWToNVMatrix(src_raw, src_stride_raw, dst_y, dst_stride_y, dst_vu, |
2459 | 0 | dst_stride_vu, argbconstants, width, height, 1); |
2460 | 0 | } |
2461 | | |
2462 | | LIBYUV_API |
2463 | | int RAWToJNV21(const uint8_t* src_raw, |
2464 | | int src_stride_raw, |
2465 | | uint8_t* dst_y, |
2466 | | int dst_stride_y, |
2467 | | uint8_t* dst_vu, |
2468 | | int dst_stride_vu, |
2469 | | int width, |
2470 | 0 | int height) { |
2471 | 0 | return RAWToNV21Matrix(src_raw, src_stride_raw, dst_y, dst_stride_y, dst_vu, |
2472 | 0 | dst_stride_vu, &kAbgrJPEGConstants, width, height); |
2473 | 0 | } |
2474 | | |
2475 | | LIBYUV_API |
2476 | | int RAWToNV21(const uint8_t* src_raw, |
2477 | | int src_stride_raw, |
2478 | | uint8_t* dst_y, |
2479 | | int dst_stride_y, |
2480 | | uint8_t* dst_vu, |
2481 | | int dst_stride_vu, |
2482 | | int width, |
2483 | 0 | int height) { |
2484 | 0 | return RAWToNV21Matrix(src_raw, src_stride_raw, dst_y, dst_stride_y, dst_vu, |
2485 | 0 | dst_stride_vu, &kAbgrI601Constants, width, height); |
2486 | 0 | } |
2487 | | |
2488 | | LIBYUV_API |
2489 | | int RGB24ToNV12(const uint8_t* src_rgb24, |
2490 | | int src_stride_rgb24, |
2491 | | uint8_t* dst_y, |
2492 | | int dst_stride_y, |
2493 | | uint8_t* dst_uv, |
2494 | | int dst_stride_uv, |
2495 | | int width, |
2496 | 0 | int height) { |
2497 | 0 | return RAWToNVMatrix(src_rgb24, src_stride_rgb24, dst_y, dst_stride_y, |
2498 | 0 | dst_uv, dst_stride_uv, &kArgbI601Constants, width, |
2499 | 0 | height, 0); |
2500 | 0 | } |
2501 | | |
2502 | | #ifdef __cplusplus |
2503 | | } // extern "C" |
2504 | | } // namespace libyuv |
2505 | | #endif |