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
30
32 int src_size)
33 {
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
56 int src_size)
57 {
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 */
86 {
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
105 {
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
124 {
125 uint16_t *d = (uint16_t *)dst;
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
139 int src_size)
140 {
141 uint16_t *d = (uint16_t *)dst;
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
155 {
156 uint16_t *d = (uint16_t *)dst;
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
170 int src_size)
171 {
172 uint16_t *d = (uint16_t *)dst;
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
186 int src_size)
187 {
188 uint16_t *d = (uint16_t *)dst;
191
192 while (s < end) {
196 *d++ = (b >> 3) | ((g & 0xFC) << 3) | ((r & 0xF8) << 8);
197 }
198 }
199
201 {
202 uint16_t *d = (uint16_t *)dst;
205
206 while (s < end) {
210 *d++ = (b >> 3) | ((g & 0xFC) << 3) | ((r & 0xF8) << 8);
211 }
212 }
213
215 int src_size)
216 {
217 uint16_t *d = (uint16_t *)dst;
220
221 while (s < end) {
225 *d++ = (b >> 3) | ((g & 0xF8) << 2) | ((r & 0xF8) << 7);
226 }
227 }
228
230 {
231 uint16_t *d = (uint16_t *)dst;
234
235 while (s < end) {
239 *d++ = (b >> 3) | ((g & 0xF8) << 2) | ((r & 0xF8) << 7);
240 }
241 }
242
244 int src_size)
245 {
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
259 int src_size)
260 {
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
274 {
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
296 {
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
318 int src_size)
319 {
320 int idx = 15 - src_size;
323
324 for (; idx < 15; idx += 4) {
325 register int 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
332 {
333 unsigned i;
334
335 for (i = 0; i < src_size; i += 3) {
336 register uint8_t x = src[i + 2];
337 dst[i + 1] = src[i + 1];
338 dst[i + 2] = src[i + 0];
339 dst[i + 0] = x;
340 }
341 }
342
346 int lumStride, int chromStride,
347 int dstStride, int vertLumPerChroma)
348 {
350 const int chromWidth = width >> 1;
351
352 for (y = 0; y <
height; y++) {
353 #if HAVE_FAST_64BIT
354 uint64_t *ldst = (uint64_t *)dst;
355 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
356 for (i = 0; i < chromWidth; i += 2) {
357 uint64_t k = yc[0] + (uc[0] << 8) +
358 (yc[1] << 16) + (unsigned)(vc[0] << 24);
359 uint64_t l = yc[2] + (uc[1] << 8) +
360 (yc[3] << 16) + (unsigned)(vc[1] << 24);
361 *ldst++ = k + (l << 32);
362 yc += 4;
363 uc += 2;
364 vc += 2;
365 }
366
367 #else
369 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
370
371 for (i = 0; i < chromWidth; i++) {
372 #if HAVE_BIGENDIAN
373 *idst++ = (yc[0] << 24) + (uc[0] << 16) +
374 (yc[1] << 8) + (vc[0] << 0);
375 #else
376 *idst++ = yc[0] + (uc[0] << 8) +
377 (yc[1] << 16) + (vc[0] << 24);
378 #endif
379 yc += 2;
380 uc++;
381 vc++;
382 }
383 #endif
384 if ((y & (vertLumPerChroma - 1)) == vertLumPerChroma - 1) {
385 usrc += chromStride;
386 vsrc += chromStride;
387 }
388 ysrc += lumStride;
389 dst += dstStride;
390 }
391 }
392
393 /**
394 * Height should be a multiple of 2 and width should be a multiple of 16.
395 * (If this is a problem for anyone then tell me, and I will fix it.)
396 */
400 int chromStride, int dstStride)
401 {
402 //FIXME interpolate chroma
404 chromStride, dstStride, 2);
405 }
406
410 int lumStride, int chromStride,
411 int dstStride, int vertLumPerChroma)
412 {
414 const int chromWidth = width >> 1;
415
416 for (y = 0; y <
height; y++) {
417 #if HAVE_FAST_64BIT
418 uint64_t *ldst = (uint64_t *)dst;
419 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
420 for (i = 0; i < chromWidth; i += 2) {
421 uint64_t k = uc[0] + (yc[0] << 8) +
422 (vc[0] << 16) + (unsigned)(yc[1] << 24);
423 uint64_t l = uc[1] + (yc[2] << 8) +
424 (vc[1] << 16) + (unsigned)(yc[3] << 24);
425 *ldst++ = k + (l << 32);
426 yc += 4;
427 uc += 2;
428 vc += 2;
429 }
430
431 #else
433 const uint8_t *yc = ysrc, *uc = usrc, *vc = vsrc;
434
435 for (i = 0; i < chromWidth; i++) {
436 #if HAVE_BIGENDIAN
437 *idst++ = (uc[0] << 24) + (yc[0] << 16) +
438 (vc[0] << 8) + (yc[1] << 0);
439 #else
440 *idst++ = uc[0] + (yc[0] << 8) +
441 (vc[0] << 16) + (yc[1] << 24);
442 #endif
443 yc += 2;
444 uc++;
445 vc++;
446 }
447 #endif
448 if ((y & (vertLumPerChroma - 1)) == vertLumPerChroma - 1) {
449 usrc += chromStride;
450 vsrc += chromStride;
451 }
452 ysrc += lumStride;
453 dst += dstStride;
454 }
455 }
456
457 /**
458 * Height should be a multiple of 2 and width should be a multiple of 16
459 * (If this is a problem for anyone then tell me, and I will fix it.)
460 */
464 int chromStride, int dstStride)
465 {
466 //FIXME interpolate chroma
468 chromStride, dstStride, 2);
469 }
470
471 /**
472 * Width should be a multiple of 16.
473 */
477 int chromStride, int dstStride)
478 {
480 chromStride, dstStride, 1);
481 }
482
483 /**
484 * Width should be a multiple of 16.
485 */
489 int chromStride, int dstStride)
490 {
492 chromStride, dstStride, 1);
493 }
494
495 /**
496 * Height should be a multiple of 2 and width should be a multiple of 16.
497 * (If this is a problem for anyone then tell me, and I will fix it.)
498 */
502 int chromStride, int srcStride)
503 {
505 const int chromWidth = width >> 1;
506
507 for (y = 0; y <
height; y += 2) {
508 int i;
509 for (i = 0; i < chromWidth; i++) {
510 ydst[2 * i + 0] = src[4 * i + 0];
511 udst[i] = src[4 * i + 1];
512 ydst[2 * i + 1] = src[4 * i + 2];
513 vdst[i] = src[4 * i + 3];
514 }
515 ydst += lumStride;
516 src += srcStride;
517
518 for (i = 0; i < chromWidth; i++) {
519 ydst[2 * i + 0] = src[4 * i + 0];
520 ydst[2 * i + 1] = src[4 * i + 2];
521 }
522 udst += chromStride;
523 vdst += chromStride;
524 ydst += lumStride;
525 src += srcStride;
526 }
527 }
528
530 int srcHeight, int srcStride, int dstStride)
531 {
533
534 dst[0] = src[0];
535
536 // first line
537 for (x = 0; x < srcWidth - 1; x++) {
538 dst[2 * x + 1] = (3 * src[x] + src[x + 1]) >> 2;
539 dst[2 * x + 2] = (src[x] + 3 * src[x + 1]) >> 2;
540 }
541 dst[2 * srcWidth - 1] = src[srcWidth - 1];
542
543 dst += dstStride;
544
545 for (y = 1; y < srcHeight; y++) {
546 const int mmxSize = 1;
547
548 dst[0] = (src[0] * 3 + src[srcStride]) >> 2;
549 dst[dstStride] = (src[0] + 3 * src[srcStride]) >> 2;
550
551 for (x = mmxSize - 1; x < srcWidth - 1; x++) {
552 dst[2 * x + 1] = (src[x + 0] * 3 + src[x + srcStride + 1]) >> 2;
553 dst[2 * x + dstStride + 2] = (src[x + 0] + 3 * src[x + srcStride + 1]) >> 2;
554 dst[2 * x + dstStride + 1] = (src[x + 1] + 3 * src[x + srcStride]) >> 2;
555 dst[2 * x + 2] = (src[x + 1] * 3 + src[x + srcStride]) >> 2;
556 }
557 dst[srcWidth * 2 - 1] = (src[srcWidth - 1] * 3 + src[srcWidth - 1 + srcStride]) >> 2;
558 dst[srcWidth * 2 - 1 + dstStride] = (src[srcWidth - 1] + 3 * src[srcWidth - 1 + srcStride]) >> 2;
559
560 dst += dstStride * 2;
561 src += srcStride;
562 }
563
564 // last line
565 dst[0] = src[0];
566
567 for (x = 0; x < srcWidth - 1; x++) {
568 dst[2 * x + 1] = (src[x] * 3 + src[x + 1]) >> 2;
569 dst[2 * x + 2] = (src[x] + 3 * src[x + 1]) >> 2;
570 }
571 dst[2 * srcWidth - 1] = src[srcWidth - 1];
572 }
573
574 /**
575 * Height should be a multiple of 2 and width should be a multiple of 16.
576 * (If this is a problem for anyone then tell me, and I will fix it.)
577 * Chrominance data is only taken from every second line, others are ignored.
578 * FIXME: Write HQ version.
579 */
583 int chromStride, int srcStride)
584 {
586 const int chromWidth = width >> 1;
587
588 for (y = 0; y <
height; y += 2) {
589 int i;
590 for (i = 0; i < chromWidth; i++) {
591 udst[i] = src[4 * i + 0];
592 ydst[2 * i + 0] = src[4 * i + 1];
593 vdst[i] = src[4 * i + 2];
594 ydst[2 * i + 1] = src[4 * i + 3];
595 }
596 ydst += lumStride;
597 src += srcStride;
598
599 for (i = 0; i < chromWidth; i++) {
600 ydst[2 * i + 0] = src[4 * i + 1];
601 ydst[2 * i + 1] = src[4 * i + 3];
602 }
603 udst += chromStride;
604 vdst += chromStride;
605 ydst += lumStride;
606 src += srcStride;
607 }
608 }
609
610 /**
611 * Height should be a multiple of 2 and width should be a multiple of 2.
612 * (If this is a problem for anyone then tell me, and I will fix it.)
613 * Chrominance data is only taken from every second line,
614 * others are ignored in the C version.
615 * FIXME: Write HQ version.
616 */
619 int chromStride,
int srcStride,
int32_t *rgb2yuv)
620 {
625 const int chromWidth = width >> 1;
626
627 for (y = 0; y <
height; y += 2) {
628 int i;
629 for (i = 0; i < chromWidth; i++) {
630 unsigned int b = src[6 * i + 0];
631 unsigned int g = src[6 * i + 1];
632 unsigned int r = src[6 * i + 2];
633
637
641
642 b = src[6 * i + 3];
643 g = src[6 * i + 4];
644 r = src[6 * i + 5];
645
648 }
649 ydst += lumStride;
650 src += srcStride;
651
652 if (y+1 == height)
653 break;
654
655 for (i = 0; i < chromWidth; i++) {
656 unsigned int b = src[6 * i + 0];
657 unsigned int g = src[6 * i + 1];
658 unsigned int r = src[6 * i + 2];
659
661
663
664 b = src[6 * i + 3];
665 g = src[6 * i + 4];
666 r = src[6 * i + 5];
667
670 }
671 udst += chromStride;
672 vdst += chromStride;
673 ydst += lumStride;
674 src += srcStride;
675 }
676 }
677
680 int src1Stride, int src2Stride, int dstStride)
681 {
682 int h;
683
684 for (h = 0; h <
height; h++) {
685 int w;
686 for (w = 0; w <
width; w++) {
687 dest[2 * w + 0] = src1[w];
688 dest[2 * w + 1] = src2[w];
689 }
690 dest += dstStride;
691 src1 += src1Stride;
692 src2 += src2Stride;
693 }
694 }
695
699 int srcStride1, int srcStride2,
700 int dstStride1, int dstStride2)
701 {
703 int w = width / 2;
704 int h = height / 2;
705
706 for (y = 0; y < h; y++) {
707 const uint8_t *
s1 = src1 + srcStride1 * (y >> 1);
709 for (x = 0; x < w; x++)
710 d[2 * x] = d[2 * x + 1] = s1[x];
711 }
712 for (y = 0; y < h; y++) {
713 const uint8_t *
s2 = src2 + srcStride2 * (y >> 1);
715 for (x = 0; x < w; x++)
716 d[2 * x] = d[2 * x + 1] = s2[x];
717 }
718 }
719
723 int srcStride1, int srcStride2,
724 int srcStride3, int dstStride)
725 {
727 int w = width / 2;
729
730 for (y = 0; y < h; y++) {
731 const uint8_t *yp = src1 + srcStride1 *
y;
732 const uint8_t *up = src2 + srcStride2 * (y >> 2);
733 const uint8_t *vp = src3 + srcStride3 * (y >> 2);
735 for (x = 0; x < w; x++) {
736 const int x2 = x << 2;
737 d[8 * x + 0] = yp[x2];
738 d[8 * x + 1] = up[x];
739 d[8 * x + 2] = yp[x2 + 1];
740 d[8 * x + 3] = vp[x];
741 d[8 * x + 4] = yp[x2 + 2];
742 d[8 * x + 5] = up[x];
743 d[8 * x + 6] = yp[x2 + 3];
744 d[8 * x + 7] = vp[x];
745 }
746 }
747 }
748
750 {
752 src += count * 2;
754 while (count < 0) {
756 count++;
757 }
758 }
759
762 {
765 src += count * 4;
767 while (count < 0) {
768 dst0[
count] = src[4 * count + 0];
769 dst1[
count] = src[4 * count + 2];
770 count++;
771 }
772 }
773
776 {
779 src0 += count * 4;
780 src1 += count * 4;
782 while (count < 0) {
783 dst0[
count] = (src0[4 * count + 0] + src1[4 * count + 0]) >> 1;
784 dst1[
count] = (src0[4 * count + 2] + src1[4 * count + 2]) >> 1;
785 count++;
786 }
787 }
788
791 {
794 src += count * 4;
796 src++;
797 while (count < 0) {
798 dst0[
count] = src[4 * count + 0];
799 dst1[
count] = src[4 * count + 2];
800 count++;
801 }
802 }
803
806 {
809 src0 += count * 4;
810 src1 += count * 4;
812 src0++;
813 src1++;
814 while (count < 0) {
815 dst0[
count] = (src0[4 * count + 0] + src1[4 * count + 0]) >> 1;
816 dst1[
count] = (src0[4 * count + 2] + src1[4 * count + 2]) >> 1;
817 count++;
818 }
819 }
820
823 int lumStride, int chromStride, int srcStride)
824 {
827
828 for (y = 0; y <
height; y++) {
830 if (y & 1) {
832 udst += chromStride;
833 vdst += chromStride;
834 }
835
836 src += srcStride;
837 ydst += lumStride;
838 }
839 }
840
843 int lumStride, int chromStride, int srcStride)
844 {
847
848 for (y = 0; y <
height; y++) {
851
852 src += srcStride;
853 ydst += lumStride;
854 udst += chromStride;
855 vdst += chromStride;
856 }
857 }
858
861 int lumStride, int chromStride, int srcStride)
862 {
865
866 for (y = 0; y <
height; y++) {
868 if (y & 1) {
870 udst += chromStride;
871 vdst += chromStride;
872 }
873
874 src += srcStride;
875 ydst += lumStride;
876 }
877 }
878
881 int lumStride, int chromStride, int srcStride)
882 {
885
886 for (y = 0; y <
height; y++) {
889
890 src += srcStride;
891 ydst += lumStride;
892 udst += chromStride;
893 vdst += chromStride;
894 }
895 }
896
898 {
927
932 }