1 /*
2 * software RGB to RGB converter
3 * pluralize by software PAL8 to RGB converter
4 * software YUV to YUV converter
5 * software YUV to RGB converter
6 * Written by Nick Kurshev.
7 * palette & YUV & runtime CPU stuff by Michael (michaelni@gmx.at)
8 * lot of big-endian byte order fixes by Alex Beregszaszi
9 *
10 * This file is part of FFmpeg.
11 *
12 * FFmpeg is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU Lesser General Public
14 * License as published by the Free Software Foundation; either
15 * version 2.1 of the License, or (at your option) any later version.
16 *
17 * FFmpeg is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 * Lesser General Public License for more details.
21 *
22 * You should have received a copy of the GNU Lesser General Public
23 * License along with FFmpeg; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 */
26
27 #include <stddef.h>
28
29 #include "libavutil/attributes.h"
30
rgb24tobgr32_c(const uint8_t * src,uint8_t * dst,int src_size)31 static inline void rgb24tobgr32_c(const uint8_t *src, uint8_t *dst,
32 int src_size)
33 {
34 uint8_t *dest = dst;
35 const uint8_t *s = src;
36 const uint8_t *end = s + src_size;
37
38 while (s < end) {
39 #if HAVE_BIGENDIAN
40 /* RGB24 (= R, G, B) -> RGB32 (= A, B, G, R) */
41 *dest++ = 255;
42 *dest++ = s[2];
43 *dest++ = s[1];
44 *dest++ = s[0];
45 s += 3;
46 #else
47 *dest++ = *s++;
48 *dest++ = *s++;
49 *dest++ = *s++;
50 *dest++ = 255;
51 #endif
52 }
53 }
54
rgb32tobgr24_c(const uint8_t * src,uint8_t * dst,int src_size)55 static inline void rgb32tobgr24_c(const uint8_t *src, uint8_t *dst,
56 int src_size)
57 {
58 uint8_t *dest = dst;
59 const uint8_t *s = src;
60 const uint8_t *end = s + src_size;
61
62 while (s < end) {
63 #if HAVE_BIGENDIAN
64 /* RGB32 (= A, B, G, R) -> RGB24 (= R, G, B) */
65 s++;
66 dest[2] = *s++;
67 dest[1] = *s++;
68 dest[0] = *s++;
69 dest += 3;
70 #else
71 *dest++ = *s++;
72 *dest++ = *s++;
73 *dest++ = *s++;
74 s++;
75 #endif
76 }
77 }
78
79 /*
80 * original by Strepto/Astral
81 * ported to gcc & bugfixed: A'rpi
82 * MMXEXT, 3DNOW optimization by Nick Kurshev
83 * 32-bit C version, and and&add trick by Michael Niedermayer
84 */
rgb15to16_c(const uint8_t * src,uint8_t * dst,int src_size)85 static inline void rgb15to16_c(const uint8_t *src, uint8_t *dst, int src_size)
86 {
87 register uint8_t *d = dst;
88 register const uint8_t *s = src;
89 register const uint8_t *end = s + src_size;
90 const uint8_t *mm_end = end - 3;
91
92 while (s < mm_end) {
93 register unsigned x = *((const uint32_t *)s);
94 *((uint32_t *)d) = (x & 0x7FFF7FFF) + (x & 0x7FE07FE0);
95 d += 4;
96 s += 4;
97 }
98 if (s < end) {
99 register unsigned short x = *((const uint16_t *)s);
100 *((uint16_t *)d) = (x & 0x7FFF) + (x & 0x7FE0);
101 }
102 }
103
rgb16to15_c(const uint8_t * src,uint8_t * dst,int src_size)104 static inline void rgb16to15_c(const uint8_t *src, uint8_t *dst, int src_size)
105 {
106 register uint8_t *d = dst;
107 register const uint8_t *s = src;
108 register const uint8_t *end = s + src_size;
109 const uint8_t *mm_end = end - 3;
110
111 while (s < mm_end) {
112 register uint32_t x = *((const uint32_t *)s);
113 *((uint32_t *)d) = ((x >> 1) & 0x7FE07FE0) | (x & 0x001F001F);
114 s += 4;
115 d += 4;
116 }
117 if (s < end) {
118 register uint16_t x = *((const uint16_t *)s);
119 *((uint16_t *)d) = ((x >> 1) & 0x7FE0) | (x & 0x001F);
120 }
121 }
122
rgb32to16_c(const uint8_t * src,uint8_t * dst,int src_size)123 static inline void rgb32to16_c(const uint8_t *src, uint8_t *dst, int src_size)
124 {
125 uint16_t *d = (uint16_t *)dst;
126 const uint8_t *s = src;
127 const uint8_t *end = s + src_size;
128
129 while (s < end) {
130 register int rgb = *(const uint32_t *)s;
131 s += 4;
132 *d++ = ((rgb & 0xFF) >> 3) +
133 ((rgb & 0xFC00) >> 5) +
134 ((rgb & 0xF80000) >> 8);
135 }
136 }
137
rgb32tobgr16_c(const uint8_t * src,uint8_t * dst,int src_size)138 static inline void rgb32tobgr16_c(const uint8_t *src, uint8_t *dst,
139 int src_size)
140 {
141 uint16_t *d = (uint16_t *)dst;
142 const uint8_t *s = src;
143 const uint8_t *end = s + src_size;
144
145 while (s < end) {
146 register int rgb = *(const uint32_t *)s;
147 s += 4;
148 *d++ = ((rgb & 0xF8) << 8) +
149 ((rgb & 0xFC00) >> 5) +
150 ((rgb & 0xF80000) >> 19);
151 }
152 }
153
rgb32to15_c(const uint8_t * src,uint8_t * dst,int src_size)154 static inline void rgb32to15_c(const uint8_t *src, uint8_t *dst, int src_size)
155 {
156 uint16_t *d = (uint16_t *)dst;
157 const uint8_t *s = src;
158 const uint8_t *end = s + src_size;
159
160 while (s < end) {
161 register int rgb = *(const uint32_t *)s;
162 s += 4;
163 *d++ = ((rgb & 0xFF) >> 3) +
164 ((rgb & 0xF800) >> 6) +
165 ((rgb & 0xF80000) >> 9);
166 }
167 }
168
rgb32tobgr15_c(const uint8_t * src,uint8_t * dst,int src_size)169 static inline void rgb32tobgr15_c(const uint8_t *src, uint8_t *dst,
170 int src_size)
171 {
172 uint16_t *d = (uint16_t *)dst;
173 const uint8_t *s = src;
174 const uint8_t *end = s + src_size;
175
176 while (s < end) {
177 register int rgb = *(const uint32_t *)s;
178 s += 4;
179 *d++ = ((rgb & 0xF8) << 7) +
180 ((rgb & 0xF800) >> 6) +
181 ((rgb & 0xF80000) >> 19);
182 }
183 }
184
rgb24tobgr16_c(const uint8_t * src,uint8_t * dst,int src_size)185 static inline void rgb24tobgr16_c(const uint8_t *src, uint8_t *dst,
186 int src_size)
187 {
188 uint16_t *d = (uint16_t *)dst;
189 const uint8_t *s = src;
190 const uint8_t *end = s + src_size;
191
192 while (s < end) {
193 const int b = *s++;
194 const int g = *s++;
195 const int r = *s++;
196 *d++ = (b >> 3) | ((g & 0xFC) << 3) | ((r & 0xF8) << 8);
197 }
198 }
199
rgb24to16_c(const uint8_t * src,uint8_t * dst,int src_size)200 static inline void rgb24to16_c(const uint8_t *src, uint8_t *dst, int src_size)
201 {
202 uint16_t *d = (uint16_t *)dst;
203 const uint8_t *s = src;
204 const uint8_t *end = s + src_size;
205
206 while (s < end) {
207 const int r = *s++;
208 const int g = *s++;
209 const int b = *s++;
210 *d++ = (b >> 3) | ((g & 0xFC) << 3) | ((r & 0xF8) << 8);
211 }
212 }
213
rgb24tobgr15_c(const uint8_t * src,uint8_t * dst,int src_size)214 static inline void rgb24tobgr15_c(const uint8_t *src, uint8_t *dst,
215 int src_size)
216 {
217 uint16_t *d = (uint16_t *)dst;
218 const uint8_t *s = src;
219 const uint8_t *end = s + src_size;
220
221 while (s < end) {
222 const int b = *s++;
223 const int g = *s++;
224 const int r = *s++;
225 *d++ = (b >> 3) | ((g & 0xF8) << 2) | ((r & 0xF8) << 7);
226 }
227 }
228
rgb24to15_c(const uint8_t * src,uint8_t * dst,int src_size)229 static inline void rgb24to15_c(const uint8_t *src, uint8_t *dst, int src_size)
230 {
231 uint16_t *d = (uint16_t *)dst;
232 const uint8_t *s = src;
233 const uint8_t *end = s + src_size;
234
235 while (s < end) {
236 const int r = *s++;
237 const int g = *s++;
238 const int b = *s++;
239 *d++ = (b >> 3) | ((g & 0xF8) << 2) | ((r & 0xF8) << 7);
240 }
241 }
242
rgb15tobgr24_c(const uint8_t * src,uint8_t * dst,int src_size)243 static inline void rgb15tobgr24_c(const uint8_t *src, uint8_t *dst,
244 int src_size)
245 {
246 uint8_t *d = dst;
247 const uint16_t *s = (const uint16_t *)src;
248 const uint16_t *end = s + src_size / 2;
249
250 while (s < end) {
251 register uint16_t bgr = *s++;
252 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
253 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
254 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
255 }
256 }
257
rgb16tobgr24_c(const uint8_t * src,uint8_t * dst,int src_size)258 static inline void rgb16tobgr24_c(const uint8_t *src, uint8_t *dst,
259 int src_size)
260 {
261 uint8_t *d = (uint8_t *)dst;
262 const uint16_t *s = (const uint16_t *)src;
263 const uint16_t *end = s + src_size / 2;
264
265 while (s < end) {
266 register uint16_t bgr = *s++;
267 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
268 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
269 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
270 }
271 }
272
rgb15to32_c(const uint8_t * src,uint8_t * dst,int src_size)273 static inline void rgb15to32_c(const uint8_t *src, uint8_t *dst, int src_size)
274 {
275 uint8_t *d = dst;
276 const uint16_t *s = (const uint16_t *)src;
277 const uint16_t *end = s + src_size / 2;
278
279 while (s < end) {
280 register uint16_t bgr = *s++;
281 #if HAVE_BIGENDIAN
282 *d++ = 255;
283 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
284 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
285 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
286 #else
287 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
288 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
289 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
290 *d++ = 255;
291 #endif
292 }
293 }
294
rgb16to32_c(const uint8_t * src,uint8_t * dst,int src_size)295 static inline void rgb16to32_c(const uint8_t *src, uint8_t *dst, int src_size)
296 {
297 uint8_t *d = dst;
298 const uint16_t *s = (const uint16_t *)src;
299 const uint16_t *end = s + src_size / 2;
300
301 while (s < end) {
302 register uint16_t bgr = *s++;
303 #if HAVE_BIGENDIAN
304 *d++ = 255;
305 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
306 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
307 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
308 #else
309 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
310 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
311 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
312 *d++ = 255;
313 #endif
314 }
315 }
316
shuffle_bytes_2103_c(const uint8_t * src,uint8_t * dst,int src_size)317 static inline void shuffle_bytes_2103_c(const uint8_t *src, uint8_t *dst,
318 int src_size)
319 {
320 int idx = 15 - src_size;
321 const uint8_t *s = src - idx;
322 uint8_t *d = dst - idx;
323
324 for (; idx < 15; idx += 4) {
325 register unsigned v = *(const uint32_t *)&s[idx], g = v & 0xff00ff00;
326 v &= 0xff00ff;
327 *(uint32_t *)&d[idx] = (v >> 16) + g + (v << 16);
328 }
329 }
330
shuffle_bytes_0321_c(const uint8_t * src,uint8_t * dst,int src_size)331 static inline void shuffle_bytes_0321_c(const uint8_t *src, uint8_t *dst,
332 int src_size)
333 {
334 int idx = 15 - src_size;
335 const uint8_t *s = src - idx;
336 uint8_t *d = dst - idx;
337
338 for (; idx < 15; idx += 4) {
339 register unsigned v = *(const uint32_t *)&s[idx], g = v & 0x00ff00ff;
340 v &= 0xff00ff00;
341 *(uint32_t *)&d[idx] = (v >> 16) + g + (v << 16);
342 }
343 }
344
345 #define DEFINE_SHUFFLE_BYTES(name, a, b, c, d) \
346 static void shuffle_bytes_##name (const uint8_t *src, \
347 uint8_t *dst, int src_size) \
348 { \
349 int i; \
350 \
351 for (i = 0; i < src_size; i += 4) { \
352 dst[i + 0] = src[i + a]; \
353 dst[i + 1] = src[i + b]; \
354 dst[i + 2] = src[i + c]; \
355 dst[i + 3] = src[i + d]; \
356 } \
357 }
358
359 DEFINE_SHUFFLE_BYTES(1230_c, 1, 2, 3, 0)
360 DEFINE_SHUFFLE_BYTES(3012_c, 3, 0, 1, 2)
361 DEFINE_SHUFFLE_BYTES(3210_c, 3, 2, 1, 0)
362
rgb24tobgr24_c(const uint8_t * src,uint8_t * dst,int src_size)363 static inline void rgb24tobgr24_c(const uint8_t *src, uint8_t *dst, int src_size)
364 {
365 unsigned i;
366
367 for (i = 0; i < src_size; i += 3) {
368 register uint8_t x = src[i + 2];
369 dst[i + 1] = src[i + 1];
370 dst[i + 2] = src[i + 0];
371 dst[i + 0] = x;
372 }
373 }
374
yuvPlanartoyuy2_c(const uint8_t * ysrc,const uint8_t * usrc,const uint8_t * vsrc,uint8_t * dst,int width,int height,int lumStride,int chromStride,int dstStride,int vertLumPerChroma)375 static inline void yuvPlanartoyuy2_c(const uint8_t *ysrc, const uint8_t *usrc,
376 const uint8_t *vsrc, uint8_t *dst,
377 int width, int height,
378 int lumStride, int chromStride,
379 int dstStride, int vertLumPerChroma)
380 {
381 int y, i;
382 const int chromWidth = width >> 1;
383
384 for (y = 0; y < height; y++) {
385 #if HAVE_FAST_64BIT
386 uint64_t *ldst = (uint64_t *)dst;
387 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
388 for (i = 0; i < chromWidth; i += 2) {
389 uint64_t k = yc[0] + (uc[0] << 8) +
390 (yc[1] << 16) + ((unsigned) vc[0] << 24);
391 uint64_t l = yc[2] + (uc[1] << 8) +
392 (yc[3] << 16) + ((unsigned) vc[1] << 24);
393 *ldst++ = k + (l << 32);
394 yc += 4;
395 uc += 2;
396 vc += 2;
397 }
398
399 #else
400 int *idst = (int32_t *)dst;
401 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
402
403 for (i = 0; i < chromWidth; i++) {
404 #if HAVE_BIGENDIAN
405 *idst++ = (yc[0] << 24) + (uc[0] << 16) +
406 (yc[1] << 8) + (vc[0] << 0);
407 #else
408 *idst++ = yc[0] + (uc[0] << 8) +
409 (yc[1] << 16) + (vc[0] << 24);
410 #endif
411 yc += 2;
412 uc++;
413 vc++;
414 }
415 #endif
416 if ((y & (vertLumPerChroma - 1)) == vertLumPerChroma - 1) {
417 usrc += chromStride;
418 vsrc += chromStride;
419 }
420 ysrc += lumStride;
421 dst += dstStride;
422 }
423 }
424
425 /**
426 * Height should be a multiple of 2 and width should be a multiple of 16.
427 * (If this is a problem for anyone then tell me, and I will fix it.)
428 */
yv12toyuy2_c(const uint8_t * ysrc,const uint8_t * usrc,const uint8_t * vsrc,uint8_t * dst,int width,int height,int lumStride,int chromStride,int dstStride)429 static inline void yv12toyuy2_c(const uint8_t *ysrc, const uint8_t *usrc,
430 const uint8_t *vsrc, uint8_t *dst,
431 int width, int height, int lumStride,
432 int chromStride, int dstStride)
433 {
434 //FIXME interpolate chroma
435 yuvPlanartoyuy2_c(ysrc, usrc, vsrc, dst, width, height, lumStride,
436 chromStride, dstStride, 2);
437 }
438
yuvPlanartouyvy_c(const uint8_t * ysrc,const uint8_t * usrc,const uint8_t * vsrc,uint8_t * dst,int width,int height,int lumStride,int chromStride,int dstStride,int vertLumPerChroma)439 static inline void yuvPlanartouyvy_c(const uint8_t *ysrc, const uint8_t *usrc,
440 const uint8_t *vsrc, uint8_t *dst,
441 int width, int height,
442 int lumStride, int chromStride,
443 int dstStride, int vertLumPerChroma)
444 {
445 int y, i;
446 const int chromWidth = width >> 1;
447
448 for (y = 0; y < height; y++) {
449 #if HAVE_FAST_64BIT
450 uint64_t *ldst = (uint64_t *)dst;
451 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
452 for (i = 0; i < chromWidth; i += 2) {
453 uint64_t k = uc[0] + (yc[0] << 8) +
454 (vc[0] << 16) + ((unsigned) yc[1] << 24);
455 uint64_t l = uc[1] + (yc[2] << 8) +
456 (vc[1] << 16) + ((unsigned) yc[3] << 24);
457 *ldst++ = k + (l << 32);
458 yc += 4;
459 uc += 2;
460 vc += 2;
461 }
462
463 #else
464 int *idst = (int32_t *)dst;
465 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
466
467 for (i = 0; i < chromWidth; i++) {
468 #if HAVE_BIGENDIAN
469 *idst++ = (uc[0] << 24) + (yc[0] << 16) +
470 (vc[0] << 8) + (yc[1] << 0);
471 #else
472 *idst++ = uc[0] + (yc[0] << 8) +
473 (vc[0] << 16) + (yc[1] << 24);
474 #endif
475 yc += 2;
476 uc++;
477 vc++;
478 }
479 #endif
480 if ((y & (vertLumPerChroma - 1)) == vertLumPerChroma - 1) {
481 usrc += chromStride;
482 vsrc += chromStride;
483 }
484 ysrc += lumStride;
485 dst += dstStride;
486 }
487 }
488
489 /**
490 * Height should be a multiple of 2 and width should be a multiple of 16
491 * (If this is a problem for anyone then tell me, and I will fix it.)
492 */
yv12touyvy_c(const uint8_t * ysrc,const uint8_t * usrc,const uint8_t * vsrc,uint8_t * dst,int width,int height,int lumStride,int chromStride,int dstStride)493 static inline void yv12touyvy_c(const uint8_t *ysrc, const uint8_t *usrc,
494 const uint8_t *vsrc, uint8_t *dst,
495 int width, int height, int lumStride,
496 int chromStride, int dstStride)
497 {
498 //FIXME interpolate chroma
499 yuvPlanartouyvy_c(ysrc, usrc, vsrc, dst, width, height, lumStride,
500 chromStride, dstStride, 2);
501 }
502
503 /**
504 * Width should be a multiple of 16.
505 */
yuv422ptouyvy_c(const uint8_t * ysrc,const uint8_t * usrc,const uint8_t * vsrc,uint8_t * dst,int width,int height,int lumStride,int chromStride,int dstStride)506 static inline void yuv422ptouyvy_c(const uint8_t *ysrc, const uint8_t *usrc,
507 const uint8_t *vsrc, uint8_t *dst,
508 int width, int height, int lumStride,
509 int chromStride, int dstStride)
510 {
511 yuvPlanartouyvy_c(ysrc, usrc, vsrc, dst, width, height, lumStride,
512 chromStride, dstStride, 1);
513 }
514
515 /**
516 * Width should be a multiple of 16.
517 */
yuv422ptoyuy2_c(const uint8_t * ysrc,const uint8_t * usrc,const uint8_t * vsrc,uint8_t * dst,int width,int height,int lumStride,int chromStride,int dstStride)518 static inline void yuv422ptoyuy2_c(const uint8_t *ysrc, const uint8_t *usrc,
519 const uint8_t *vsrc, uint8_t *dst,
520 int width, int height, int lumStride,
521 int chromStride, int dstStride)
522 {
523 yuvPlanartoyuy2_c(ysrc, usrc, vsrc, dst, width, height, lumStride,
524 chromStride, dstStride, 1);
525 }
526
527 /**
528 * Height should be a multiple of 2 and width should be a multiple of 16.
529 * (If this is a problem for anyone then tell me, and I will fix it.)
530 */
yuy2toyv12_c(const uint8_t * src,uint8_t * ydst,uint8_t * udst,uint8_t * vdst,int width,int height,int lumStride,int chromStride,int srcStride)531 static inline void yuy2toyv12_c(const uint8_t *src, uint8_t *ydst,
532 uint8_t *udst, uint8_t *vdst,
533 int width, int height, int lumStride,
534 int chromStride, int srcStride)
535 {
536 int y;
537 const int chromWidth = width >> 1;
538
539 for (y = 0; y < height; y += 2) {
540 int i;
541 for (i = 0; i < chromWidth; i++) {
542 ydst[2 * i + 0] = src[4 * i + 0];
543 udst[i] = src[4 * i + 1];
544 ydst[2 * i + 1] = src[4 * i + 2];
545 vdst[i] = src[4 * i + 3];
546 }
547 ydst += lumStride;
548 src += srcStride;
549
550 for (i = 0; i < chromWidth; i++) {
551 ydst[2 * i + 0] = src[4 * i + 0];
552 ydst[2 * i + 1] = src[4 * i + 2];
553 }
554 udst += chromStride;
555 vdst += chromStride;
556 ydst += lumStride;
557 src += srcStride;
558 }
559 }
560
planar2x_c(const uint8_t * src,uint8_t * dst,int srcWidth,int srcHeight,int srcStride,int dstStride)561 static inline void planar2x_c(const uint8_t *src, uint8_t *dst, int srcWidth,
562 int srcHeight, int srcStride, int dstStride)
563 {
564 int x, y;
565
566 dst[0] = src[0];
567
568 // first line
569 for (x = 0; x < srcWidth - 1; x++) {
570 dst[2 * x + 1] = (3 * src[x] + src[x + 1]) >> 2;
571 dst[2 * x + 2] = (src[x] + 3 * src[x + 1]) >> 2;
572 }
573 dst[2 * srcWidth - 1] = src[srcWidth - 1];
574
575 dst += dstStride;
576
577 for (y = 1; y < srcHeight; y++) {
578 const int mmxSize = 1;
579
580 dst[0] = (src[0] * 3 + src[srcStride]) >> 2;
581 dst[dstStride] = (src[0] + 3 * src[srcStride]) >> 2;
582
583 for (x = mmxSize - 1; x < srcWidth - 1; x++) {
584 dst[2 * x + 1] = (src[x + 0] * 3 + src[x + srcStride + 1]) >> 2;
585 dst[2 * x + dstStride + 2] = (src[x + 0] + 3 * src[x + srcStride + 1]) >> 2;
586 dst[2 * x + dstStride + 1] = (src[x + 1] + 3 * src[x + srcStride]) >> 2;
587 dst[2 * x + 2] = (src[x + 1] * 3 + src[x + srcStride]) >> 2;
588 }
589 dst[srcWidth * 2 - 1] = (src[srcWidth - 1] * 3 + src[srcWidth - 1 + srcStride]) >> 2;
590 dst[srcWidth * 2 - 1 + dstStride] = (src[srcWidth - 1] + 3 * src[srcWidth - 1 + srcStride]) >> 2;
591
592 dst += dstStride * 2;
593 src += srcStride;
594 }
595
596 // last line
597 dst[0] = src[0];
598
599 for (x = 0; x < srcWidth - 1; x++) {
600 dst[2 * x + 1] = (src[x] * 3 + src[x + 1]) >> 2;
601 dst[2 * x + 2] = (src[x] + 3 * src[x + 1]) >> 2;
602 }
603 dst[2 * srcWidth - 1] = src[srcWidth - 1];
604 }
605
606 /**
607 * Height should be a multiple of 2 and width should be a multiple of 16.
608 * (If this is a problem for anyone then tell me, and I will fix it.)
609 * Chrominance data is only taken from every second line, others are ignored.
610 * FIXME: Write HQ version.
611 */
uyvytoyv12_c(const uint8_t * src,uint8_t * ydst,uint8_t * udst,uint8_t * vdst,int width,int height,int lumStride,int chromStride,int srcStride)612 static inline void uyvytoyv12_c(const uint8_t *src, uint8_t *ydst,
613 uint8_t *udst, uint8_t *vdst,
614 int width, int height, int lumStride,
615 int chromStride, int srcStride)
616 {
617 int y;
618 const int chromWidth = width >> 1;
619
620 for (y = 0; y < height; y += 2) {
621 int i;
622 for (i = 0; i < chromWidth; i++) {
623 udst[i] = src[4 * i + 0];
624 ydst[2 * i + 0] = src[4 * i + 1];
625 vdst[i] = src[4 * i + 2];
626 ydst[2 * i + 1] = src[4 * i + 3];
627 }
628 ydst += lumStride;
629 src += srcStride;
630
631 for (i = 0; i < chromWidth; i++) {
632 ydst[2 * i + 0] = src[4 * i + 1];
633 ydst[2 * i + 1] = src[4 * i + 3];
634 }
635 udst += chromStride;
636 vdst += chromStride;
637 ydst += lumStride;
638 src += srcStride;
639 }
640 }
641
642 /**
643 * Height should be a multiple of 2 and width should be a multiple of 2.
644 * (If this is a problem for anyone then tell me, and I will fix it.)
645 * Chrominance data is only taken from every second line,
646 * others are ignored in the C version.
647 * FIXME: Write HQ version.
648 */
ff_rgb24toyv12_c(const uint8_t * src,uint8_t * ydst,uint8_t * udst,uint8_t * vdst,int width,int height,int lumStride,int chromStride,int srcStride,int32_t * rgb2yuv)649 void ff_rgb24toyv12_c(const uint8_t *src, uint8_t *ydst, uint8_t *udst,
650 uint8_t *vdst, int width, int height, int lumStride,
651 int chromStride, int srcStride, int32_t *rgb2yuv)
652 {
653 int32_t ry = rgb2yuv[RY_IDX], gy = rgb2yuv[GY_IDX], by = rgb2yuv[BY_IDX];
654 int32_t ru = rgb2yuv[RU_IDX], gu = rgb2yuv[GU_IDX], bu = rgb2yuv[BU_IDX];
655 int32_t rv = rgb2yuv[RV_IDX], gv = rgb2yuv[GV_IDX], bv = rgb2yuv[BV_IDX];
656 int y;
657 const int chromWidth = width >> 1;
658
659 for (y = 0; y < height; y += 2) {
660 int i;
661 for (i = 0; i < chromWidth; i++) {
662 unsigned int b = src[6 * i + 0];
663 unsigned int g = src[6 * i + 1];
664 unsigned int r = src[6 * i + 2];
665
666 unsigned int Y = ((ry * r + gy * g + by * b) >> RGB2YUV_SHIFT) + 16;
667 unsigned int V = ((rv * r + gv * g + bv * b) >> RGB2YUV_SHIFT) + 128;
668 unsigned int U = ((ru * r + gu * g + bu * b) >> RGB2YUV_SHIFT) + 128;
669
670 udst[i] = U;
671 vdst[i] = V;
672 ydst[2 * i] = Y;
673
674 b = src[6 * i + 3];
675 g = src[6 * i + 4];
676 r = src[6 * i + 5];
677
678 Y = ((ry * r + gy * g + by * b) >> RGB2YUV_SHIFT) + 16;
679 ydst[2 * i + 1] = Y;
680 }
681 ydst += lumStride;
682 src += srcStride;
683
684 if (y+1 == height)
685 break;
686
687 for (i = 0; i < chromWidth; i++) {
688 unsigned int b = src[6 * i + 0];
689 unsigned int g = src[6 * i + 1];
690 unsigned int r = src[6 * i + 2];
691
692 unsigned int Y = ((ry * r + gy * g + by * b) >> RGB2YUV_SHIFT) + 16;
693
694 ydst[2 * i] = Y;
695
696 b = src[6 * i + 3];
697 g = src[6 * i + 4];
698 r = src[6 * i + 5];
699
700 Y = ((ry * r + gy * g + by * b) >> RGB2YUV_SHIFT) + 16;
701 ydst[2 * i + 1] = Y;
702 }
703 udst += chromStride;
704 vdst += chromStride;
705 ydst += lumStride;
706 src += srcStride;
707 }
708 }
709
interleaveBytes_c(const uint8_t * src1,const uint8_t * src2,uint8_t * dest,int width,int height,int src1Stride,int src2Stride,int dstStride)710 static void interleaveBytes_c(const uint8_t *src1, const uint8_t *src2,
711 uint8_t *dest, int width, int height,
712 int src1Stride, int src2Stride, int dstStride)
713 {
714 int h;
715
716 for (h = 0; h < height; h++) {
717 int w;
718 for (w = 0; w < width; w++) {
719 dest[2 * w + 0] = src1[w];
720 dest[2 * w + 1] = src2[w];
721 }
722 dest += dstStride;
723 src1 += src1Stride;
724 src2 += src2Stride;
725 }
726 }
727
deinterleaveBytes_c(const uint8_t * src,uint8_t * dst1,uint8_t * dst2,int width,int height,int srcStride,int dst1Stride,int dst2Stride)728 static void deinterleaveBytes_c(const uint8_t *src, uint8_t *dst1, uint8_t *dst2,
729 int width, int height, int srcStride,
730 int dst1Stride, int dst2Stride)
731 {
732 int h;
733
734 for (h = 0; h < height; h++) {
735 int w;
736 for (w = 0; w < width; w++) {
737 dst1[w] = src[2 * w + 0];
738 dst2[w] = src[2 * w + 1];
739 }
740 src += srcStride;
741 dst1 += dst1Stride;
742 dst2 += dst2Stride;
743 }
744 }
745
vu9_to_vu12_c(const uint8_t * src1,const uint8_t * src2,uint8_t * dst1,uint8_t * dst2,int width,int height,int srcStride1,int srcStride2,int dstStride1,int dstStride2)746 static inline void vu9_to_vu12_c(const uint8_t *src1, const uint8_t *src2,
747 uint8_t *dst1, uint8_t *dst2,
748 int width, int height,
749 int srcStride1, int srcStride2,
750 int dstStride1, int dstStride2)
751 {
752 int x, y;
753 int w = width / 2;
754 int h = height / 2;
755
756 for (y = 0; y < h; y++) {
757 const uint8_t *s1 = src1 + srcStride1 * (y >> 1);
758 uint8_t *d = dst1 + dstStride1 * y;
759 for (x = 0; x < w; x++)
760 d[2 * x] = d[2 * x + 1] = s1[x];
761 }
762 for (y = 0; y < h; y++) {
763 const uint8_t *s2 = src2 + srcStride2 * (y >> 1);
764 uint8_t *d = dst2 + dstStride2 * y;
765 for (x = 0; x < w; x++)
766 d[2 * x] = d[2 * x + 1] = s2[x];
767 }
768 }
769
yvu9_to_yuy2_c(const uint8_t * src1,const uint8_t * src2,const uint8_t * src3,uint8_t * dst,int width,int height,int srcStride1,int srcStride2,int srcStride3,int dstStride)770 static inline void yvu9_to_yuy2_c(const uint8_t *src1, const uint8_t *src2,
771 const uint8_t *src3, uint8_t *dst,
772 int width, int height,
773 int srcStride1, int srcStride2,
774 int srcStride3, int dstStride)
775 {
776 int x, y;
777 int w = width / 2;
778 int h = height;
779
780 for (y = 0; y < h; y++) {
781 const uint8_t *yp = src1 + srcStride1 * y;
782 const uint8_t *up = src2 + srcStride2 * (y >> 2);
783 const uint8_t *vp = src3 + srcStride3 * (y >> 2);
784 uint8_t *d = dst + dstStride * y;
785 for (x = 0; x < w; x++) {
786 const int x2 = x << 2;
787 d[8 * x + 0] = yp[x2];
788 d[8 * x + 1] = up[x];
789 d[8 * x + 2] = yp[x2 + 1];
790 d[8 * x + 3] = vp[x];
791 d[8 * x + 4] = yp[x2 + 2];
792 d[8 * x + 5] = up[x];
793 d[8 * x + 6] = yp[x2 + 3];
794 d[8 * x + 7] = vp[x];
795 }
796 }
797 }
798
extract_even_c(const uint8_t * src,uint8_t * dst,int count)799 static void extract_even_c(const uint8_t *src, uint8_t *dst, int count)
800 {
801 dst += count;
802 src += count * 2;
803 count = -count;
804 while (count < 0) {
805 dst[count] = src[2 * count];
806 count++;
807 }
808 }
809
extract_even2_c(const uint8_t * src,uint8_t * dst0,uint8_t * dst1,int count)810 static void extract_even2_c(const uint8_t *src, uint8_t *dst0, uint8_t *dst1,
811 int count)
812 {
813 dst0 += count;
814 dst1 += count;
815 src += count * 4;
816 count = -count;
817 while (count < 0) {
818 dst0[count] = src[4 * count + 0];
819 dst1[count] = src[4 * count + 2];
820 count++;
821 }
822 }
823
extract_even2avg_c(const uint8_t * src0,const uint8_t * src1,uint8_t * dst0,uint8_t * dst1,int count)824 static void extract_even2avg_c(const uint8_t *src0, const uint8_t *src1,
825 uint8_t *dst0, uint8_t *dst1, int count)
826 {
827 dst0 += count;
828 dst1 += count;
829 src0 += count * 4;
830 src1 += count * 4;
831 count = -count;
832 while (count < 0) {
833 dst0[count] = (src0[4 * count + 0] + src1[4 * count + 0]) >> 1;
834 dst1[count] = (src0[4 * count + 2] + src1[4 * count + 2]) >> 1;
835 count++;
836 }
837 }
838
extract_odd2_c(const uint8_t * src,uint8_t * dst0,uint8_t * dst1,int count)839 static void extract_odd2_c(const uint8_t *src, uint8_t *dst0, uint8_t *dst1,
840 int count)
841 {
842 dst0 += count;
843 dst1 += count;
844 src += count * 4;
845 count = -count;
846 src++;
847 while (count < 0) {
848 dst0[count] = src[4 * count + 0];
849 dst1[count] = src[4 * count + 2];
850 count++;
851 }
852 }
853
extract_odd2avg_c(const uint8_t * src0,const uint8_t * src1,uint8_t * dst0,uint8_t * dst1,int count)854 static void extract_odd2avg_c(const uint8_t *src0, const uint8_t *src1,
855 uint8_t *dst0, uint8_t *dst1, int count)
856 {
857 dst0 += count;
858 dst1 += count;
859 src0 += count * 4;
860 src1 += count * 4;
861 count = -count;
862 src0++;
863 src1++;
864 while (count < 0) {
865 dst0[count] = (src0[4 * count + 0] + src1[4 * count + 0]) >> 1;
866 dst1[count] = (src0[4 * count + 2] + src1[4 * count + 2]) >> 1;
867 count++;
868 }
869 }
870
yuyvtoyuv420_c(uint8_t * ydst,uint8_t * udst,uint8_t * vdst,const uint8_t * src,int width,int height,int lumStride,int chromStride,int srcStride)871 static void yuyvtoyuv420_c(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
872 const uint8_t *src, int width, int height,
873 int lumStride, int chromStride, int srcStride)
874 {
875 int y;
876 const int chromWidth = AV_CEIL_RSHIFT(width, 1);
877
878 for (y = 0; y < height; y++) {
879 extract_even_c(src, ydst, width);
880 if (y & 1) {
881 extract_odd2avg_c(src - srcStride, src, udst, vdst, chromWidth);
882 udst += chromStride;
883 vdst += chromStride;
884 }
885
886 src += srcStride;
887 ydst += lumStride;
888 }
889 }
890
yuyvtoyuv422_c(uint8_t * ydst,uint8_t * udst,uint8_t * vdst,const uint8_t * src,int width,int height,int lumStride,int chromStride,int srcStride)891 static void yuyvtoyuv422_c(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
892 const uint8_t *src, int width, int height,
893 int lumStride, int chromStride, int srcStride)
894 {
895 int y;
896 const int chromWidth = AV_CEIL_RSHIFT(width, 1);
897
898 for (y = 0; y < height; y++) {
899 extract_even_c(src, ydst, width);
900 extract_odd2_c(src, udst, vdst, chromWidth);
901
902 src += srcStride;
903 ydst += lumStride;
904 udst += chromStride;
905 vdst += chromStride;
906 }
907 }
908
uyvytoyuv420_c(uint8_t * ydst,uint8_t * udst,uint8_t * vdst,const uint8_t * src,int width,int height,int lumStride,int chromStride,int srcStride)909 static void uyvytoyuv420_c(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
910 const uint8_t *src, int width, int height,
911 int lumStride, int chromStride, int srcStride)
912 {
913 int y;
914 const int chromWidth = AV_CEIL_RSHIFT(width, 1);
915
916 for (y = 0; y < height; y++) {
917 extract_even_c(src + 1, ydst, width);
918 if (y & 1) {
919 extract_even2avg_c(src - srcStride, src, udst, vdst, chromWidth);
920 udst += chromStride;
921 vdst += chromStride;
922 }
923
924 src += srcStride;
925 ydst += lumStride;
926 }
927 }
928
uyvytoyuv422_c(uint8_t * ydst,uint8_t * udst,uint8_t * vdst,const uint8_t * src,int width,int height,int lumStride,int chromStride,int srcStride)929 static void uyvytoyuv422_c(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
930 const uint8_t *src, int width, int height,
931 int lumStride, int chromStride, int srcStride)
932 {
933 int y;
934 const int chromWidth = AV_CEIL_RSHIFT(width, 1);
935
936 for (y = 0; y < height; y++) {
937 extract_even_c(src + 1, ydst, width);
938 extract_even2_c(src, udst, vdst, chromWidth);
939
940 src += srcStride;
941 ydst += lumStride;
942 udst += chromStride;
943 vdst += chromStride;
944 }
945 }
946
rgb2rgb_init_c(void)947 static av_cold void rgb2rgb_init_c(void)
948 {
949 rgb15to16 = rgb15to16_c;
950 rgb15tobgr24 = rgb15tobgr24_c;
951 rgb15to32 = rgb15to32_c;
952 rgb16tobgr24 = rgb16tobgr24_c;
953 rgb16to32 = rgb16to32_c;
954 rgb16to15 = rgb16to15_c;
955 rgb24tobgr16 = rgb24tobgr16_c;
956 rgb24tobgr15 = rgb24tobgr15_c;
957 rgb24tobgr32 = rgb24tobgr32_c;
958 rgb32to16 = rgb32to16_c;
959 rgb32to15 = rgb32to15_c;
960 rgb32tobgr24 = rgb32tobgr24_c;
961 rgb24to15 = rgb24to15_c;
962 rgb24to16 = rgb24to16_c;
963 rgb24tobgr24 = rgb24tobgr24_c;
964 #if HAVE_BIGENDIAN
965 shuffle_bytes_0321 = shuffle_bytes_2103_c;
966 shuffle_bytes_2103 = shuffle_bytes_0321_c;
967 #else
968 shuffle_bytes_0321 = shuffle_bytes_0321_c;
969 shuffle_bytes_2103 = shuffle_bytes_2103_c;
970 #endif
971 shuffle_bytes_1230 = shuffle_bytes_1230_c;
972 shuffle_bytes_3012 = shuffle_bytes_3012_c;
973 shuffle_bytes_3210 = shuffle_bytes_3210_c;
974 rgb32tobgr16 = rgb32tobgr16_c;
975 rgb32tobgr15 = rgb32tobgr15_c;
976 yv12toyuy2 = yv12toyuy2_c;
977 yv12touyvy = yv12touyvy_c;
978 yuv422ptoyuy2 = yuv422ptoyuy2_c;
979 yuv422ptouyvy = yuv422ptouyvy_c;
980 yuy2toyv12 = yuy2toyv12_c;
981 planar2x = planar2x_c;
982 ff_rgb24toyv12 = ff_rgb24toyv12_c;
983 interleaveBytes = interleaveBytes_c;
984 deinterleaveBytes = deinterleaveBytes_c;
985 vu9_to_vu12 = vu9_to_vu12_c;
986 yvu9_to_yuy2 = yvu9_to_yuy2_c;
987
988 uyvytoyuv420 = uyvytoyuv420_c;
989 uyvytoyuv422 = uyvytoyuv422_c;
990 yuyvtoyuv420 = yuyvtoyuv420_c;
991 yuyvtoyuv422 = yuyvtoyuv422_c;
992 }
993