2 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of Libav.
6 * Libav is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * Libav is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 #include "libswscale/swscale.h"
24 #include "libswscale/swscale_internal.h"
25 #include "libavutil/attributes.h"
26 #include "libavutil/intreadwrite.h"
27 #include "libavutil/x86/asm.h"
28 #include "libavutil/x86/cpu.h"
29 #include "libavutil/cpu.h"
30 #include "libavutil/pixdesc.h"
36 DECLARE_ASM_CONST(8, uint64_t, bF8)= 0xF8F8F8F8F8F8F8F8LL;
37 DECLARE_ASM_CONST(8, uint64_t, bFC)= 0xFCFCFCFCFCFCFCFCLL;
38 DECLARE_ASM_CONST(8, uint64_t, w10)= 0x0010001000100010LL;
39 DECLARE_ASM_CONST(8, uint64_t, w02)= 0x0002000200020002LL;
41 const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
43 0x0200020002000200LL,};
45 const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
47 0x0004000400040004LL,};
49 DECLARE_ASM_CONST(8, uint64_t, b16Mask)= 0x001F001F001F001FLL;
50 DECLARE_ASM_CONST(8, uint64_t, g16Mask)= 0x07E007E007E007E0LL;
51 DECLARE_ASM_CONST(8, uint64_t, r16Mask)= 0xF800F800F800F800LL;
52 DECLARE_ASM_CONST(8, uint64_t, b15Mask)= 0x001F001F001F001FLL;
53 DECLARE_ASM_CONST(8, uint64_t, g15Mask)= 0x03E003E003E003E0LL;
54 DECLARE_ASM_CONST(8, uint64_t, r15Mask)= 0x7C007C007C007C00LL;
56 DECLARE_ALIGNED(8, const uint64_t, ff_M24A) = 0x00FF0000FF0000FFLL;
57 DECLARE_ALIGNED(8, const uint64_t, ff_M24B) = 0xFF0000FF0000FF00LL;
58 DECLARE_ALIGNED(8, const uint64_t, ff_M24C) = 0x0000FF0000FF0000LL;
61 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000000210041000DULL;
62 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000FFEEFFDC0038ULL;
63 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00000038FFD2FFF8ULL;
65 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000020E540830C8BULL;
66 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000ED0FDAC23831ULL;
67 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00003831D0E6F6EAULL;
68 #endif /* FAST_BGR2YV12 */
69 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset) = 0x1010101010101010ULL;
70 DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
71 DECLARE_ALIGNED(8, const uint64_t, ff_w1111) = 0x0001000100010001ULL;
76 #define COMPILE_TEMPLATE_MMXEXT 0
77 #define RENAME(a) a ## _mmx
78 #include "swscale_template.c"
82 #if HAVE_MMXEXT_INLINE
84 #undef COMPILE_TEMPLATE_MMXEXT
85 #define COMPILE_TEMPLATE_MMXEXT 1
86 #define RENAME(a) a ## _mmxext
87 #include "swscale_template.c"
90 void updateMMXDitherTables(SwsContext *c, int dstY, int lumBufIndex, int chrBufIndex,
91 int lastInLumBuf, int lastInChrBuf)
93 const int dstH= c->dstH;
94 const int flags= c->flags;
95 int16_t **lumPixBuf= c->lumPixBuf;
96 int16_t **chrUPixBuf= c->chrUPixBuf;
97 int16_t **alpPixBuf= c->alpPixBuf;
98 const int vLumBufSize= c->vLumBufSize;
99 const int vChrBufSize= c->vChrBufSize;
100 int32_t *vLumFilterPos= c->vLumFilterPos;
101 int32_t *vChrFilterPos= c->vChrFilterPos;
102 int16_t *vLumFilter= c->vLumFilter;
103 int16_t *vChrFilter= c->vChrFilter;
104 int32_t *lumMmxFilter= c->lumMmxFilter;
105 int32_t *chrMmxFilter= c->chrMmxFilter;
106 int32_t av_unused *alpMmxFilter= c->alpMmxFilter;
107 const int vLumFilterSize= c->vLumFilterSize;
108 const int vChrFilterSize= c->vChrFilterSize;
109 const int chrDstY= dstY>>c->chrDstVSubSample;
110 const int firstLumSrcY= vLumFilterPos[dstY]; //First line needed as input
111 const int firstChrSrcY= vChrFilterPos[chrDstY]; //First line needed as input
113 c->blueDither= ff_dither8[dstY&1];
114 if (c->dstFormat == AV_PIX_FMT_RGB555 || c->dstFormat == AV_PIX_FMT_BGR555)
115 c->greenDither= ff_dither8[dstY&1];
117 c->greenDither= ff_dither4[dstY&1];
118 c->redDither= ff_dither8[(dstY+1)&1];
119 if (dstY < dstH - 2) {
120 const int16_t **lumSrcPtr= (const int16_t **) lumPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize;
121 const int16_t **chrUSrcPtr= (const int16_t **) chrUPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
122 const int16_t **alpSrcPtr= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? (const int16_t **) alpPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize : NULL;
125 if (firstLumSrcY < 0 || firstLumSrcY + vLumFilterSize > c->srcH) {
126 const int16_t **tmpY = (const int16_t **) lumPixBuf + 2 * vLumBufSize;
127 int neg = -firstLumSrcY, i, end = FFMIN(c->srcH - firstLumSrcY, vLumFilterSize);
128 for (i = 0; i < neg; i++)
129 tmpY[i] = lumSrcPtr[neg];
130 for ( ; i < end; i++)
131 tmpY[i] = lumSrcPtr[i];
132 for ( ; i < vLumFilterSize; i++)
137 const int16_t **tmpA = (const int16_t **) alpPixBuf + 2 * vLumBufSize;
138 for (i = 0; i < neg; i++)
139 tmpA[i] = alpSrcPtr[neg];
140 for ( ; i < end; i++)
141 tmpA[i] = alpSrcPtr[i];
142 for ( ; i < vLumFilterSize; i++)
143 tmpA[i] = tmpA[i - 1];
147 if (firstChrSrcY < 0 || firstChrSrcY + vChrFilterSize > c->chrSrcH) {
148 const int16_t **tmpU = (const int16_t **) chrUPixBuf + 2 * vChrBufSize;
149 int neg = -firstChrSrcY, i, end = FFMIN(c->chrSrcH - firstChrSrcY, vChrFilterSize);
150 for (i = 0; i < neg; i++) {
151 tmpU[i] = chrUSrcPtr[neg];
153 for ( ; i < end; i++) {
154 tmpU[i] = chrUSrcPtr[i];
156 for ( ; i < vChrFilterSize; i++) {
157 tmpU[i] = tmpU[i - 1];
162 if (flags & SWS_ACCURATE_RND) {
163 int s= APCK_SIZE / 8;
164 for (i=0; i<vLumFilterSize; i+=2) {
165 *(const void**)&lumMmxFilter[s*i ]= lumSrcPtr[i ];
166 *(const void**)&lumMmxFilter[s*i+APCK_PTR2/4 ]= lumSrcPtr[i+(vLumFilterSize>1)];
167 lumMmxFilter[s*i+APCK_COEF/4 ]=
168 lumMmxFilter[s*i+APCK_COEF/4+1]= vLumFilter[dstY*vLumFilterSize + i ]
169 + (vLumFilterSize>1 ? vLumFilter[dstY*vLumFilterSize + i + 1]<<16 : 0);
170 if (CONFIG_SWSCALE_ALPHA && alpPixBuf) {
171 *(const void**)&alpMmxFilter[s*i ]= alpSrcPtr[i ];
172 *(const void**)&alpMmxFilter[s*i+APCK_PTR2/4 ]= alpSrcPtr[i+(vLumFilterSize>1)];
173 alpMmxFilter[s*i+APCK_COEF/4 ]=
174 alpMmxFilter[s*i+APCK_COEF/4+1]= lumMmxFilter[s*i+APCK_COEF/4 ];
177 for (i=0; i<vChrFilterSize; i+=2) {
178 *(const void**)&chrMmxFilter[s*i ]= chrUSrcPtr[i ];
179 *(const void**)&chrMmxFilter[s*i+APCK_PTR2/4 ]= chrUSrcPtr[i+(vChrFilterSize>1)];
180 chrMmxFilter[s*i+APCK_COEF/4 ]=
181 chrMmxFilter[s*i+APCK_COEF/4+1]= vChrFilter[chrDstY*vChrFilterSize + i ]
182 + (vChrFilterSize>1 ? vChrFilter[chrDstY*vChrFilterSize + i + 1]<<16 : 0);
185 for (i=0; i<vLumFilterSize; i++) {
186 *(const void**)&lumMmxFilter[4*i+0]= lumSrcPtr[i];
189 ((uint16_t)vLumFilter[dstY*vLumFilterSize + i])*0x10001;
190 if (CONFIG_SWSCALE_ALPHA && alpPixBuf) {
191 *(const void**)&alpMmxFilter[4*i+0]= alpSrcPtr[i];
193 alpMmxFilter[4*i+3]= lumMmxFilter[4*i+2];
196 for (i=0; i<vChrFilterSize; i++) {
197 *(const void**)&chrMmxFilter[4*i+0]= chrUSrcPtr[i];
200 ((uint16_t)vChrFilter[chrDstY*vChrFilterSize + i])*0x10001;
206 #endif /* HAVE_INLINE_ASM */
208 #define SCALE_FUNC(filter_n, from_bpc, to_bpc, opt) \
209 void ff_hscale ## from_bpc ## to ## to_bpc ## _ ## filter_n ## _ ## opt( \
210 SwsContext *c, int16_t *data, \
211 int dstW, const uint8_t *src, \
212 const int16_t *filter, \
213 const int32_t *filterPos, int filterSize)
215 #define SCALE_FUNCS(filter_n, opt) \
216 SCALE_FUNC(filter_n, 8, 15, opt); \
217 SCALE_FUNC(filter_n, 9, 15, opt); \
218 SCALE_FUNC(filter_n, 10, 15, opt); \
219 SCALE_FUNC(filter_n, 16, 15, opt); \
220 SCALE_FUNC(filter_n, 8, 19, opt); \
221 SCALE_FUNC(filter_n, 9, 19, opt); \
222 SCALE_FUNC(filter_n, 10, 19, opt); \
223 SCALE_FUNC(filter_n, 16, 19, opt)
225 #define SCALE_FUNCS_MMX(opt) \
226 SCALE_FUNCS(4, opt); \
227 SCALE_FUNCS(8, opt); \
230 #define SCALE_FUNCS_SSE(opt) \
231 SCALE_FUNCS(4, opt); \
232 SCALE_FUNCS(8, opt); \
233 SCALE_FUNCS(X4, opt); \
237 SCALE_FUNCS_MMX(mmx);
239 SCALE_FUNCS_SSE(sse2);
240 SCALE_FUNCS_SSE(ssse3);
241 SCALE_FUNCS_SSE(sse4);
243 #define VSCALEX_FUNC(size, opt) \
244 void ff_yuv2planeX_ ## size ## _ ## opt(const int16_t *filter, int filterSize, \
245 const int16_t **src, uint8_t *dest, int dstW, \
246 const uint8_t *dither, int offset)
247 #define VSCALEX_FUNCS(opt) \
248 VSCALEX_FUNC(8, opt); \
249 VSCALEX_FUNC(9, opt); \
250 VSCALEX_FUNC(10, opt)
253 VSCALEX_FUNCS(mmxext);
257 VSCALEX_FUNC(16, sse4);
260 #define VSCALE_FUNC(size, opt) \
261 void ff_yuv2plane1_ ## size ## _ ## opt(const int16_t *src, uint8_t *dst, int dstW, \
262 const uint8_t *dither, int offset)
263 #define VSCALE_FUNCS(opt1, opt2) \
264 VSCALE_FUNC(8, opt1); \
265 VSCALE_FUNC(9, opt2); \
266 VSCALE_FUNC(10, opt2); \
267 VSCALE_FUNC(16, opt1)
270 VSCALE_FUNCS(mmx, mmxext);
272 VSCALE_FUNCS(sse2, sse2);
273 VSCALE_FUNC(16, sse4);
274 VSCALE_FUNCS(avx, avx);
276 #define INPUT_Y_FUNC(fmt, opt) \
277 void ff_ ## fmt ## ToY_ ## opt(uint8_t *dst, const uint8_t *src, \
278 int w, uint32_t *unused)
279 #define INPUT_UV_FUNC(fmt, opt) \
280 void ff_ ## fmt ## ToUV_ ## opt(uint8_t *dstU, uint8_t *dstV, \
281 const uint8_t *src, const uint8_t *unused1, \
282 int w, uint32_t *unused2)
283 #define INPUT_FUNC(fmt, opt) \
284 INPUT_Y_FUNC(fmt, opt); \
285 INPUT_UV_FUNC(fmt, opt)
286 #define INPUT_FUNCS(opt) \
287 INPUT_FUNC(uyvy, opt); \
288 INPUT_FUNC(yuyv, opt); \
289 INPUT_UV_FUNC(nv12, opt); \
290 INPUT_UV_FUNC(nv21, opt); \
291 INPUT_FUNC(rgba, opt); \
292 INPUT_FUNC(bgra, opt); \
293 INPUT_FUNC(argb, opt); \
294 INPUT_FUNC(abgr, opt); \
295 INPUT_FUNC(rgb24, opt); \
296 INPUT_FUNC(bgr24, opt)
305 av_cold void ff_sws_init_swscale_x86(SwsContext *c)
307 int cpu_flags = av_get_cpu_flags();
310 if (INLINE_MMX(cpu_flags))
311 sws_init_swscale_mmx(c);
313 #if HAVE_MMXEXT_INLINE
314 if (INLINE_MMXEXT(cpu_flags))
315 sws_init_swscale_mmxext(c);
318 #define ASSIGN_SCALE_FUNC2(hscalefn, filtersize, opt1, opt2) do { \
319 if (c->srcBpc == 8) { \
320 hscalefn = c->dstBpc <= 10 ? ff_hscale8to15_ ## filtersize ## _ ## opt2 : \
321 ff_hscale8to19_ ## filtersize ## _ ## opt1; \
322 } else if (c->srcBpc == 9) { \
323 hscalefn = c->dstBpc <= 10 ? ff_hscale9to15_ ## filtersize ## _ ## opt2 : \
324 ff_hscale9to19_ ## filtersize ## _ ## opt1; \
325 } else if (c->srcBpc == 10) { \
326 hscalefn = c->dstBpc <= 10 ? ff_hscale10to15_ ## filtersize ## _ ## opt2 : \
327 ff_hscale10to19_ ## filtersize ## _ ## opt1; \
328 } else /* c->srcBpc == 16 */ { \
329 hscalefn = c->dstBpc <= 10 ? ff_hscale16to15_ ## filtersize ## _ ## opt2 : \
330 ff_hscale16to19_ ## filtersize ## _ ## opt1; \
333 #define ASSIGN_MMX_SCALE_FUNC(hscalefn, filtersize, opt1, opt2) \
334 switch (filtersize) { \
335 case 4: ASSIGN_SCALE_FUNC2(hscalefn, 4, opt1, opt2); break; \
336 case 8: ASSIGN_SCALE_FUNC2(hscalefn, 8, opt1, opt2); break; \
337 default: ASSIGN_SCALE_FUNC2(hscalefn, X, opt1, opt2); break; \
339 #define ASSIGN_VSCALEX_FUNC(vscalefn, opt, do_16_case, condition_8bit) \
341 case 16: do_16_case; break; \
342 case 10: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2planeX_10_ ## opt; break; \
343 case 9: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2planeX_9_ ## opt; break; \
344 default: if (condition_8bit) vscalefn = ff_yuv2planeX_8_ ## opt; break; \
346 #define ASSIGN_VSCALE_FUNC(vscalefn, opt1, opt2, opt2chk) \
348 case 16: if (!isBE(c->dstFormat)) vscalefn = ff_yuv2plane1_16_ ## opt1; break; \
349 case 10: if (!isBE(c->dstFormat) && opt2chk) vscalefn = ff_yuv2plane1_10_ ## opt2; break; \
350 case 9: if (!isBE(c->dstFormat) && opt2chk) vscalefn = ff_yuv2plane1_9_ ## opt2; break; \
351 default: vscalefn = ff_yuv2plane1_8_ ## opt1; break; \
353 #define case_rgb(x, X, opt) \
354 case AV_PIX_FMT_ ## X: \
355 c->lumToYV12 = ff_ ## x ## ToY_ ## opt; \
356 if (!c->chrSrcHSubSample) \
357 c->chrToYV12 = ff_ ## x ## ToUV_ ## opt; \
360 if (EXTERNAL_MMX(cpu_flags)) {
361 ASSIGN_MMX_SCALE_FUNC(c->hyScale, c->hLumFilterSize, mmx, mmx);
362 ASSIGN_MMX_SCALE_FUNC(c->hcScale, c->hChrFilterSize, mmx, mmx);
363 ASSIGN_VSCALE_FUNC(c->yuv2plane1, mmx, mmxext, cpu_flags & AV_CPU_FLAG_MMXEXT);
365 switch (c->srcFormat) {
367 c->lumToYV12 = ff_yuyvToY_mmx;
369 c->alpToYV12 = ff_uyvyToY_mmx;
371 case AV_PIX_FMT_YUYV422:
372 c->lumToYV12 = ff_yuyvToY_mmx;
373 c->chrToYV12 = ff_yuyvToUV_mmx;
375 case AV_PIX_FMT_UYVY422:
376 c->lumToYV12 = ff_uyvyToY_mmx;
377 c->chrToYV12 = ff_uyvyToUV_mmx;
379 case AV_PIX_FMT_NV12:
380 c->chrToYV12 = ff_nv12ToUV_mmx;
382 case AV_PIX_FMT_NV21:
383 c->chrToYV12 = ff_nv21ToUV_mmx;
385 case_rgb(rgb24, RGB24, mmx);
386 case_rgb(bgr24, BGR24, mmx);
387 case_rgb(bgra, BGRA, mmx);
388 case_rgb(rgba, RGBA, mmx);
389 case_rgb(abgr, ABGR, mmx);
390 case_rgb(argb, ARGB, mmx);
395 if (EXTERNAL_MMXEXT(cpu_flags)) {
396 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, mmxext, , 1);
398 #endif /* ARCH_X86_32 */
399 #define ASSIGN_SSE_SCALE_FUNC(hscalefn, filtersize, opt1, opt2) \
400 switch (filtersize) { \
401 case 4: ASSIGN_SCALE_FUNC2(hscalefn, 4, opt1, opt2); break; \
402 case 8: ASSIGN_SCALE_FUNC2(hscalefn, 8, opt1, opt2); break; \
403 default: if (filtersize & 4) ASSIGN_SCALE_FUNC2(hscalefn, X4, opt1, opt2); \
404 else ASSIGN_SCALE_FUNC2(hscalefn, X8, opt1, opt2); \
407 if (EXTERNAL_SSE2(cpu_flags)) {
408 ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, sse2, sse2);
409 ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, sse2, sse2);
410 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse2, ,
411 HAVE_ALIGNED_STACK || ARCH_X86_64);
412 ASSIGN_VSCALE_FUNC(c->yuv2plane1, sse2, sse2, 1);
414 switch (c->srcFormat) {
416 c->lumToYV12 = ff_yuyvToY_sse2;
418 c->alpToYV12 = ff_uyvyToY_sse2;
420 case AV_PIX_FMT_YUYV422:
421 c->lumToYV12 = ff_yuyvToY_sse2;
422 c->chrToYV12 = ff_yuyvToUV_sse2;
424 case AV_PIX_FMT_UYVY422:
425 c->lumToYV12 = ff_uyvyToY_sse2;
426 c->chrToYV12 = ff_uyvyToUV_sse2;
428 case AV_PIX_FMT_NV12:
429 c->chrToYV12 = ff_nv12ToUV_sse2;
431 case AV_PIX_FMT_NV21:
432 c->chrToYV12 = ff_nv21ToUV_sse2;
434 case_rgb(rgb24, RGB24, sse2);
435 case_rgb(bgr24, BGR24, sse2);
436 case_rgb(bgra, BGRA, sse2);
437 case_rgb(rgba, RGBA, sse2);
438 case_rgb(abgr, ABGR, sse2);
439 case_rgb(argb, ARGB, sse2);
444 if (EXTERNAL_SSSE3(cpu_flags)) {
445 ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, ssse3, ssse3);
446 ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, ssse3, ssse3);
447 switch (c->srcFormat) {
448 case_rgb(rgb24, RGB24, ssse3);
449 case_rgb(bgr24, BGR24, ssse3);
454 if (EXTERNAL_SSE4(cpu_flags)) {
455 /* Xto15 don't need special sse4 functions */
456 ASSIGN_SSE_SCALE_FUNC(c->hyScale, c->hLumFilterSize, sse4, ssse3);
457 ASSIGN_SSE_SCALE_FUNC(c->hcScale, c->hChrFilterSize, sse4, ssse3);
458 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, sse4,
459 if (!isBE(c->dstFormat)) c->yuv2planeX = ff_yuv2planeX_16_sse4,
460 HAVE_ALIGNED_STACK || ARCH_X86_64);
461 if (c->dstBpc == 16 && !isBE(c->dstFormat))
462 c->yuv2plane1 = ff_yuv2plane1_16_sse4;
465 if (EXTERNAL_AVX(cpu_flags)) {
466 ASSIGN_VSCALEX_FUNC(c->yuv2planeX, avx, ,
467 HAVE_ALIGNED_STACK || ARCH_X86_64);
468 ASSIGN_VSCALE_FUNC(c->yuv2plane1, avx, avx, 1);
470 switch (c->srcFormat) {
471 case AV_PIX_FMT_YUYV422:
472 c->chrToYV12 = ff_yuyvToUV_avx;
474 case AV_PIX_FMT_UYVY422:
475 c->chrToYV12 = ff_uyvyToUV_avx;
477 case AV_PIX_FMT_NV12:
478 c->chrToYV12 = ff_nv12ToUV_avx;
480 case AV_PIX_FMT_NV21:
481 c->chrToYV12 = ff_nv21ToUV_avx;
483 case_rgb(rgb24, RGB24, avx);
484 case_rgb(bgr24, BGR24, avx);
485 case_rgb(bgra, BGRA, avx);
486 case_rgb(rgba, RGBA, avx);
487 case_rgb(abgr, ABGR, avx);
488 case_rgb(argb, ARGB, avx);