2 * software YUV to RGB converter
4 * Copyright (C) 2009 Konstantin Shishkov
6 * 1,4,8bpp support and context / deglobalize stuff
7 * by Michael Niedermayer (michaelni@gmx.at)
9 * This file is part of FFmpeg.
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
30 #include "libavutil/cpu.h"
31 #include "libavutil/bswap.h"
35 #include "swscale_internal.h"
36 #include "libavutil/pixdesc.h"
38 extern const uint8_t dither_2x2_4[3][8];
39 extern const uint8_t dither_2x2_8[3][8];
40 extern const uint8_t dither_4x4_16[5][8];
41 extern const uint8_t dither_8x8_32[9][8];
42 extern const uint8_t dither_8x8_73[9][8];
43 extern const uint8_t dither_8x8_220[9][8];
45 const int32_t ff_yuv2rgb_coeffs[8][4] = {
46 { 117504, 138453, 13954, 34903 }, /* no sequence_display_extension */
47 { 117504, 138453, 13954, 34903 }, /* ITU-R Rec. 709 (1990) */
48 { 104597, 132201, 25675, 53279 }, /* unspecified */
49 { 104597, 132201, 25675, 53279 }, /* reserved */
50 { 104448, 132798, 24759, 53109 }, /* FCC */
51 { 104597, 132201, 25675, 53279 }, /* ITU-R Rec. 624-4 System B, G */
52 { 104597, 132201, 25675, 53279 }, /* SMPTE 170M */
53 { 117579, 136230, 16907, 35559 } /* SMPTE 240M (1987) */
56 const int *sws_getCoefficients(int colorspace)
58 if (colorspace > 7 || colorspace < 0)
59 colorspace = SWS_CS_DEFAULT;
60 return ff_yuv2rgb_coeffs[colorspace];
63 #define LOADCHROMA(i) \
66 r = (void *)c->table_rV[V+YUVRGB_TABLE_HEADROOM]; \
67 g = (void *)(c->table_gU[U+YUVRGB_TABLE_HEADROOM] + c->table_gV[V+YUVRGB_TABLE_HEADROOM]); \
68 b = (void *)c->table_bU[U+YUVRGB_TABLE_HEADROOM];
70 #define PUTRGB(dst, src, i) \
72 dst[2 * i] = r[Y] + g[Y] + b[Y]; \
74 dst[2 * i + 1] = r[Y] + g[Y] + b[Y];
76 #define PUTRGB24(dst, src, i) \
78 dst[6 * i + 0] = r[Y]; \
79 dst[6 * i + 1] = g[Y]; \
80 dst[6 * i + 2] = b[Y]; \
82 dst[6 * i + 3] = r[Y]; \
83 dst[6 * i + 4] = g[Y]; \
84 dst[6 * i + 5] = b[Y];
86 #define PUTBGR24(dst, src, i) \
88 dst[6 * i + 0] = b[Y]; \
89 dst[6 * i + 1] = g[Y]; \
90 dst[6 * i + 2] = r[Y]; \
92 dst[6 * i + 3] = b[Y]; \
93 dst[6 * i + 4] = g[Y]; \
94 dst[6 * i + 5] = r[Y];
96 #define PUTRGBA(dst, ysrc, asrc, i, s) \
98 dst[2 * i] = r[Y] + g[Y] + b[Y] + (asrc[2 * i] << s); \
99 Y = ysrc[2 * i + 1]; \
100 dst[2 * i + 1] = r[Y] + g[Y] + b[Y] + (asrc[2 * i + 1] << s);
102 #define PUTRGB48(dst, src, i) \
104 dst[12 * i + 0] = dst[12 * i + 1] = r[Y]; \
105 dst[12 * i + 2] = dst[12 * i + 3] = g[Y]; \
106 dst[12 * i + 4] = dst[12 * i + 5] = b[Y]; \
107 Y = src[ 2 * i + 1]; \
108 dst[12 * i + 6] = dst[12 * i + 7] = r[Y]; \
109 dst[12 * i + 8] = dst[12 * i + 9] = g[Y]; \
110 dst[12 * i + 10] = dst[12 * i + 11] = b[Y];
112 #define PUTBGR48(dst, src, i) \
114 dst[12 * i + 0] = dst[12 * i + 1] = b[Y]; \
115 dst[12 * i + 2] = dst[12 * i + 3] = g[Y]; \
116 dst[12 * i + 4] = dst[12 * i + 5] = r[Y]; \
117 Y = src[2 * i + 1]; \
118 dst[12 * i + 6] = dst[12 * i + 7] = b[Y]; \
119 dst[12 * i + 8] = dst[12 * i + 9] = g[Y]; \
120 dst[12 * i + 10] = dst[12 * i + 11] = r[Y];
122 #define YUV2RGBFUNC(func_name, dst_type, alpha) \
123 static int func_name(SwsContext *c, const uint8_t *src[], \
124 int srcStride[], int srcSliceY, int srcSliceH, \
125 uint8_t *dst[], int dstStride[]) \
129 if (!alpha && c->srcFormat == AV_PIX_FMT_YUV422P) { \
133 for (y = 0; y < srcSliceH; y += 2) { \
135 (dst_type *)(dst[0] + (y + srcSliceY) * dstStride[0]); \
137 (dst_type *)(dst[0] + (y + srcSliceY + 1) * dstStride[0]); \
138 dst_type av_unused *r, *g, *b; \
139 const uint8_t *py_1 = src[0] + y * srcStride[0]; \
140 const uint8_t *py_2 = py_1 + srcStride[0]; \
141 const uint8_t *pu = src[1] + (y >> 1) * srcStride[1]; \
142 const uint8_t *pv = src[2] + (y >> 1) * srcStride[2]; \
143 const uint8_t av_unused *pa_1, *pa_2; \
144 unsigned int h_size = c->dstW >> 3; \
146 pa_1 = src[3] + y * srcStride[3]; \
147 pa_2 = pa_1 + srcStride[3]; \
150 int av_unused U, V, Y; \
152 #define ENDYUV2RGBLINE(dst_delta, ss) \
157 dst_1 += dst_delta >> ss; \
158 dst_2 += dst_delta >> ss; \
160 if (c->dstW & (4 >> ss)) { \
161 int av_unused Y, U, V; \
163 #define ENDYUV2RGBFUNC() \
169 #define CLOSEYUV2RGBFUNC(dst_delta) \
170 ENDYUV2RGBLINE(dst_delta, 0) \
173 YUV2RGBFUNC(yuv2rgb_c_48, uint8_t, 0)
175 PUTRGB48(dst_1, py_1, 0);
176 PUTRGB48(dst_2, py_2, 0);
179 PUTRGB48(dst_2, py_2, 1);
180 PUTRGB48(dst_1, py_1, 1);
183 PUTRGB48(dst_1, py_1, 2);
184 PUTRGB48(dst_2, py_2, 2);
187 PUTRGB48(dst_2, py_2, 3);
188 PUTRGB48(dst_1, py_1, 3);
189 ENDYUV2RGBLINE(48, 0)
191 PUTRGB48(dst_1, py_1, 0);
192 PUTRGB48(dst_2, py_2, 0);
195 PUTRGB48(dst_2, py_2, 1);
196 PUTRGB48(dst_1, py_1, 1);
197 ENDYUV2RGBLINE(48, 1)
199 PUTRGB48(dst_1, py_1, 0);
200 PUTRGB48(dst_2, py_2, 0);
203 YUV2RGBFUNC(yuv2rgb_c_bgr48, uint8_t, 0)
205 PUTBGR48(dst_1, py_1, 0);
206 PUTBGR48(dst_2, py_2, 0);
209 PUTBGR48(dst_2, py_2, 1);
210 PUTBGR48(dst_1, py_1, 1);
213 PUTBGR48(dst_1, py_1, 2);
214 PUTBGR48(dst_2, py_2, 2);
217 PUTBGR48(dst_2, py_2, 3);
218 PUTBGR48(dst_1, py_1, 3);
219 ENDYUV2RGBLINE(48, 0)
221 PUTBGR48(dst_1, py_1, 0);
222 PUTBGR48(dst_2, py_2, 0);
225 PUTBGR48(dst_2, py_2, 1);
226 PUTBGR48(dst_1, py_1, 1);
227 ENDYUV2RGBLINE(48, 1)
229 PUTBGR48(dst_1, py_1, 0);
230 PUTBGR48(dst_2, py_2, 0);
233 YUV2RGBFUNC(yuv2rgb_c_32, uint32_t, 0)
235 PUTRGB(dst_1, py_1, 0);
236 PUTRGB(dst_2, py_2, 0);
239 PUTRGB(dst_2, py_2, 1);
240 PUTRGB(dst_1, py_1, 1);
243 PUTRGB(dst_1, py_1, 2);
244 PUTRGB(dst_2, py_2, 2);
247 PUTRGB(dst_2, py_2, 3);
248 PUTRGB(dst_1, py_1, 3);
251 PUTRGB(dst_1, py_1, 0);
252 PUTRGB(dst_2, py_2, 0);
255 PUTRGB(dst_2, py_2, 1);
256 PUTRGB(dst_1, py_1, 1);
259 PUTRGB(dst_1, py_1, 0);
260 PUTRGB(dst_2, py_2, 0);
263 YUV2RGBFUNC(yuva2rgba_c, uint32_t, 1)
265 PUTRGBA(dst_1, py_1, pa_1, 0, 24);
266 PUTRGBA(dst_2, py_2, pa_2, 0, 24);
269 PUTRGBA(dst_2, py_2, pa_1, 1, 24);
270 PUTRGBA(dst_1, py_1, pa_2, 1, 24);
273 PUTRGBA(dst_1, py_1, pa_1, 2, 24);
274 PUTRGBA(dst_2, py_2, pa_2, 2, 24);
277 PUTRGBA(dst_2, py_2, pa_1, 3, 24);
278 PUTRGBA(dst_1, py_1, pa_2, 3, 24);
283 PUTRGBA(dst_1, py_1, pa_1, 0, 24);
284 PUTRGBA(dst_2, py_2, pa_2, 0, 24);
287 PUTRGBA(dst_2, py_2, pa_1, 1, 24);
288 PUTRGBA(dst_1, py_1, pa_2, 1, 24);
293 PUTRGBA(dst_1, py_1, pa_1, 0, 24);
294 PUTRGBA(dst_2, py_2, pa_2, 0, 24);
297 YUV2RGBFUNC(yuva2argb_c, uint32_t, 1)
299 PUTRGBA(dst_1, py_1, pa_1, 0, 0);
300 PUTRGBA(dst_2, py_2, pa_2, 0, 0);
303 PUTRGBA(dst_2, py_2, pa_2, 1, 0);
304 PUTRGBA(dst_1, py_1, pa_1, 1, 0);
307 PUTRGBA(dst_1, py_1, pa_1, 2, 0);
308 PUTRGBA(dst_2, py_2, pa_2, 2, 0);
311 PUTRGBA(dst_2, py_2, pa_2, 3, 0);
312 PUTRGBA(dst_1, py_1, pa_1, 3, 0);
317 PUTRGBA(dst_1, py_1, pa_1, 0, 0);
318 PUTRGBA(dst_2, py_2, pa_2, 0, 0);
321 PUTRGBA(dst_2, py_2, pa_2, 1, 0);
322 PUTRGBA(dst_1, py_1, pa_1, 1, 0);
327 PUTRGBA(dst_1, py_1, pa_1, 0, 0);
328 PUTRGBA(dst_2, py_2, pa_2, 0, 0);
331 YUV2RGBFUNC(yuv2rgb_c_24_rgb, uint8_t, 0)
333 PUTRGB24(dst_1, py_1, 0);
334 PUTRGB24(dst_2, py_2, 0);
337 PUTRGB24(dst_2, py_2, 1);
338 PUTRGB24(dst_1, py_1, 1);
341 PUTRGB24(dst_1, py_1, 2);
342 PUTRGB24(dst_2, py_2, 2);
345 PUTRGB24(dst_2, py_2, 3);
346 PUTRGB24(dst_1, py_1, 3);
347 ENDYUV2RGBLINE(24, 0)
349 PUTRGB24(dst_1, py_1, 0);
350 PUTRGB24(dst_2, py_2, 0);
353 PUTRGB24(dst_2, py_2, 1);
354 PUTRGB24(dst_1, py_1, 1);
355 ENDYUV2RGBLINE(24, 1)
357 PUTRGB24(dst_1, py_1, 0);
358 PUTRGB24(dst_2, py_2, 0);
361 // only trivial mods from yuv2rgb_c_24_rgb
362 YUV2RGBFUNC(yuv2rgb_c_24_bgr, uint8_t, 0)
364 PUTBGR24(dst_1, py_1, 0);
365 PUTBGR24(dst_2, py_2, 0);
368 PUTBGR24(dst_2, py_2, 1);
369 PUTBGR24(dst_1, py_1, 1);
372 PUTBGR24(dst_1, py_1, 2);
373 PUTBGR24(dst_2, py_2, 2);
376 PUTBGR24(dst_2, py_2, 3);
377 PUTBGR24(dst_1, py_1, 3);
378 ENDYUV2RGBLINE(24, 0)
380 PUTBGR24(dst_1, py_1, 0);
381 PUTBGR24(dst_2, py_2, 0);
384 PUTBGR24(dst_2, py_2, 1);
385 PUTBGR24(dst_1, py_1, 1);
386 ENDYUV2RGBLINE(24, 1)
388 PUTBGR24(dst_1, py_1, 0);
389 PUTBGR24(dst_2, py_2, 0);
392 YUV2RGBFUNC(yuv2rgb_c_16_ordered_dither, uint16_t, 0)
393 const uint8_t *d16 = dither_2x2_8[y & 1];
394 const uint8_t *e16 = dither_2x2_4[y & 1];
395 const uint8_t *f16 = dither_2x2_8[(y & 1)^1];
397 #define PUTRGB16(dst, src, i, o) \
399 dst[2 * i] = r[Y + d16[0 + o]] + \
400 g[Y + e16[0 + o]] + \
402 Y = src[2 * i + 1]; \
403 dst[2 * i + 1] = r[Y + d16[1 + o]] + \
404 g[Y + e16[1 + o]] + \
407 PUTRGB16(dst_1, py_1, 0, 0);
408 PUTRGB16(dst_2, py_2, 0, 0 + 8);
411 PUTRGB16(dst_2, py_2, 1, 2 + 8);
412 PUTRGB16(dst_1, py_1, 1, 2);
415 PUTRGB16(dst_1, py_1, 2, 4);
416 PUTRGB16(dst_2, py_2, 2, 4 + 8);
419 PUTRGB16(dst_2, py_2, 3, 6 + 8);
420 PUTRGB16(dst_1, py_1, 3, 6);
423 YUV2RGBFUNC(yuv2rgb_c_15_ordered_dither, uint16_t, 0)
424 const uint8_t *d16 = dither_2x2_8[y & 1];
425 const uint8_t *e16 = dither_2x2_8[(y & 1)^1];
427 #define PUTRGB15(dst, src, i, o) \
429 dst[2 * i] = r[Y + d16[0 + o]] + \
430 g[Y + d16[1 + o]] + \
432 Y = src[2 * i + 1]; \
433 dst[2 * i + 1] = r[Y + d16[1 + o]] + \
434 g[Y + d16[0 + o]] + \
437 PUTRGB15(dst_1, py_1, 0, 0);
438 PUTRGB15(dst_2, py_2, 0, 0 + 8);
441 PUTRGB15(dst_2, py_2, 1, 2 + 8);
442 PUTRGB15(dst_1, py_1, 1, 2);
445 PUTRGB15(dst_1, py_1, 2, 4);
446 PUTRGB15(dst_2, py_2, 2, 4 + 8);
449 PUTRGB15(dst_2, py_2, 3, 6 + 8);
450 PUTRGB15(dst_1, py_1, 3, 6);
453 // r, g, b, dst_1, dst_2
454 YUV2RGBFUNC(yuv2rgb_c_12_ordered_dither, uint16_t, 0)
455 const uint8_t *d16 = dither_4x4_16[y & 3];
457 #define PUTRGB12(dst, src, i, o) \
459 dst[2 * i] = r[Y + d16[0 + o]] + \
460 g[Y + d16[0 + o]] + \
462 Y = src[2 * i + 1]; \
463 dst[2 * i + 1] = r[Y + d16[1 + o]] + \
464 g[Y + d16[1 + o]] + \
468 PUTRGB12(dst_1, py_1, 0, 0);
469 PUTRGB12(dst_2, py_2, 0, 0 + 8);
472 PUTRGB12(dst_2, py_2, 1, 2 + 8);
473 PUTRGB12(dst_1, py_1, 1, 2);
476 PUTRGB12(dst_1, py_1, 2, 4);
477 PUTRGB12(dst_2, py_2, 2, 4 + 8);
480 PUTRGB12(dst_2, py_2, 3, 6 + 8);
481 PUTRGB12(dst_1, py_1, 3, 6);
484 // r, g, b, dst_1, dst_2
485 YUV2RGBFUNC(yuv2rgb_c_8_ordered_dither, uint8_t, 0)
486 const uint8_t *d32 = dither_8x8_32[y & 7];
487 const uint8_t *d64 = dither_8x8_73[y & 7];
489 #define PUTRGB8(dst, src, i, o) \
491 dst[2 * i] = r[Y + d32[0 + o]] + \
492 g[Y + d32[0 + o]] + \
494 Y = src[2 * i + 1]; \
495 dst[2 * i + 1] = r[Y + d32[1 + o]] + \
496 g[Y + d32[1 + o]] + \
500 PUTRGB8(dst_1, py_1, 0, 0);
501 PUTRGB8(dst_2, py_2, 0, 0 + 8);
504 PUTRGB8(dst_2, py_2, 1, 2 + 8);
505 PUTRGB8(dst_1, py_1, 1, 2);
508 PUTRGB8(dst_1, py_1, 2, 4);
509 PUTRGB8(dst_2, py_2, 2, 4 + 8);
512 PUTRGB8(dst_2, py_2, 3, 6 + 8);
513 PUTRGB8(dst_1, py_1, 3, 6);
516 YUV2RGBFUNC(yuv2rgb_c_4_ordered_dither, uint8_t, 0)
517 const uint8_t * d64 = dither_8x8_73[y & 7];
518 const uint8_t *d128 = dither_8x8_220[y & 7];
521 #define PUTRGB4D(dst, src, i, o) \
523 acc = r[Y + d128[0 + o]] + \
524 g[Y + d64[0 + o]] + \
525 b[Y + d128[0 + o]]; \
526 Y = src[2 * i + 1]; \
527 acc |= (r[Y + d128[1 + o]] + \
528 g[Y + d64[1 + o]] + \
529 b[Y + d128[1 + o]]) << 4; \
533 PUTRGB4D(dst_1, py_1, 0, 0);
534 PUTRGB4D(dst_2, py_2, 0, 0 + 8);
537 PUTRGB4D(dst_2, py_2, 1, 2 + 8);
538 PUTRGB4D(dst_1, py_1, 1, 2);
541 PUTRGB4D(dst_1, py_1, 2, 4);
542 PUTRGB4D(dst_2, py_2, 2, 4 + 8);
545 PUTRGB4D(dst_2, py_2, 3, 6 + 8);
546 PUTRGB4D(dst_1, py_1, 3, 6);
549 YUV2RGBFUNC(yuv2rgb_c_4b_ordered_dither, uint8_t, 0)
550 const uint8_t *d64 = dither_8x8_73[y & 7];
551 const uint8_t *d128 = dither_8x8_220[y & 7];
553 #define PUTRGB4DB(dst, src, i, o) \
555 dst[2 * i] = r[Y + d128[0 + o]] + \
556 g[Y + d64[0 + o]] + \
557 b[Y + d128[0 + o]]; \
558 Y = src[2 * i + 1]; \
559 dst[2 * i + 1] = r[Y + d128[1 + o]] + \
560 g[Y + d64[1 + o]] + \
564 PUTRGB4DB(dst_1, py_1, 0, 0);
565 PUTRGB4DB(dst_2, py_2, 0, 0 + 8);
568 PUTRGB4DB(dst_2, py_2, 1, 2 + 8);
569 PUTRGB4DB(dst_1, py_1, 1, 2);
572 PUTRGB4DB(dst_1, py_1, 2, 4);
573 PUTRGB4DB(dst_2, py_2, 2, 4 + 8);
576 PUTRGB4DB(dst_2, py_2, 3, 6 + 8);
577 PUTRGB4DB(dst_1, py_1, 3, 6);
580 YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0)
581 const uint8_t *d128 = dither_8x8_220[y & 7];
582 char out_1 = 0, out_2 = 0;
583 g = c->table_gU[128 + YUVRGB_TABLE_HEADROOM] + c->table_gV[128 + YUVRGB_TABLE_HEADROOM];
585 #define PUTRGB1(out, src, i, o) \
587 out += out + g[Y + d128[0 + o]]; \
588 Y = src[2 * i + 1]; \
589 out += out + g[Y + d128[1 + o]];
591 PUTRGB1(out_1, py_1, 0, 0);
592 PUTRGB1(out_2, py_2, 0, 0 + 8);
594 PUTRGB1(out_2, py_2, 1, 2 + 8);
595 PUTRGB1(out_1, py_1, 1, 2);
597 PUTRGB1(out_1, py_1, 2, 4);
598 PUTRGB1(out_2, py_2, 2, 4 + 8);
600 PUTRGB1(out_2, py_2, 3, 6 + 8);
601 PUTRGB1(out_1, py_1, 3, 6);
607 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c)
612 t = ff_yuv2rgb_init_mmx(c);
614 t = ff_yuv2rgb_init_vis(c);
615 else if (HAVE_ALTIVEC)
616 t = ff_yuv2rgb_init_altivec(c);
618 t = ff_yuv2rgb_get_func_ptr_bfin(c);
623 av_log(c, AV_LOG_WARNING,
624 "No accelerated colorspace conversion found from %s to %s.\n",
625 av_get_pix_fmt_name(c->srcFormat), av_get_pix_fmt_name(c->dstFormat));
627 switch (c->dstFormat) {
628 case AV_PIX_FMT_BGR48BE:
629 case AV_PIX_FMT_BGR48LE:
630 return yuv2rgb_c_bgr48;
631 case AV_PIX_FMT_RGB48BE:
632 case AV_PIX_FMT_RGB48LE:
634 case AV_PIX_FMT_ARGB:
635 case AV_PIX_FMT_ABGR:
636 if (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat))
638 case AV_PIX_FMT_RGBA:
639 case AV_PIX_FMT_BGRA:
640 return (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat)) ? yuva2rgba_c : yuv2rgb_c_32;
641 case AV_PIX_FMT_RGB24:
642 return yuv2rgb_c_24_rgb;
643 case AV_PIX_FMT_BGR24:
644 return yuv2rgb_c_24_bgr;
645 case AV_PIX_FMT_RGB565:
646 case AV_PIX_FMT_BGR565:
647 return yuv2rgb_c_16_ordered_dither;
648 case AV_PIX_FMT_RGB555:
649 case AV_PIX_FMT_BGR555:
650 return yuv2rgb_c_15_ordered_dither;
651 case AV_PIX_FMT_RGB444:
652 case AV_PIX_FMT_BGR444:
653 return yuv2rgb_c_12_ordered_dither;
654 case AV_PIX_FMT_RGB8:
655 case AV_PIX_FMT_BGR8:
656 return yuv2rgb_c_8_ordered_dither;
657 case AV_PIX_FMT_RGB4:
658 case AV_PIX_FMT_BGR4:
659 return yuv2rgb_c_4_ordered_dither;
660 case AV_PIX_FMT_RGB4_BYTE:
661 case AV_PIX_FMT_BGR4_BYTE:
662 return yuv2rgb_c_4b_ordered_dither;
663 case AV_PIX_FMT_MONOBLACK:
664 return yuv2rgb_c_1_ordered_dither;
669 static void fill_table(uint8_t* table[256 + 2*YUVRGB_TABLE_HEADROOM], const int elemsize,
670 const int64_t inc, void *y_tab)
673 uint8_t *y_table = y_tab;
675 y_table -= elemsize * (inc >> 9);
677 for (i = 0; i < 256 + 2*YUVRGB_TABLE_HEADROOM; i++) {
678 int64_t cb = av_clip(i-YUVRGB_TABLE_HEADROOM, 0, 255)*inc;
679 table[i] = y_table + elemsize * (cb >> 16);
683 static void fill_gv_table(int table[256 + 2*YUVRGB_TABLE_HEADROOM], const int elemsize, const int64_t inc)
686 int off = -(inc >> 9);
688 for (i = 0; i < 256 + 2*YUVRGB_TABLE_HEADROOM; i++) {
689 int64_t cb = av_clip(i-YUVRGB_TABLE_HEADROOM, 0, 255)*inc;
690 table[i] = elemsize * (off + (cb >> 16));
694 static uint16_t roundToInt16(int64_t f)
696 int r = (f + (1 << 15)) >> 16;
706 av_cold int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
707 int fullRange, int brightness,
708 int contrast, int saturation)
710 const int isRgb = c->dstFormat == AV_PIX_FMT_RGB32 ||
711 c->dstFormat == AV_PIX_FMT_RGB32_1 ||
712 c->dstFormat == AV_PIX_FMT_BGR24 ||
713 c->dstFormat == AV_PIX_FMT_RGB565BE ||
714 c->dstFormat == AV_PIX_FMT_RGB565LE ||
715 c->dstFormat == AV_PIX_FMT_RGB555BE ||
716 c->dstFormat == AV_PIX_FMT_RGB555LE ||
717 c->dstFormat == AV_PIX_FMT_RGB444BE ||
718 c->dstFormat == AV_PIX_FMT_RGB444LE ||
719 c->dstFormat == AV_PIX_FMT_RGB8 ||
720 c->dstFormat == AV_PIX_FMT_RGB4 ||
721 c->dstFormat == AV_PIX_FMT_RGB4_BYTE ||
722 c->dstFormat == AV_PIX_FMT_MONOBLACK;
723 const int isNotNe = c->dstFormat == AV_PIX_FMT_NE(RGB565LE, RGB565BE) ||
724 c->dstFormat == AV_PIX_FMT_NE(RGB555LE, RGB555BE) ||
725 c->dstFormat == AV_PIX_FMT_NE(RGB444LE, RGB444BE) ||
726 c->dstFormat == AV_PIX_FMT_NE(BGR565LE, BGR565BE) ||
727 c->dstFormat == AV_PIX_FMT_NE(BGR555LE, BGR555BE) ||
728 c->dstFormat == AV_PIX_FMT_NE(BGR444LE, BGR444BE);
729 const int bpp = c->dstFormatBpp;
733 int i, base, rbase, gbase, bbase, av_uninit(abase), needAlpha;
734 const int yoffs = fullRange ? 384 : 326;
736 int64_t crv = inv_table[0];
737 int64_t cbu = inv_table[1];
738 int64_t cgu = -inv_table[2];
739 int64_t cgv = -inv_table[3];
740 int64_t cy = 1 << 16;
745 cy = (cy * 255) / 219;
748 crv = (crv * 224) / 255;
749 cbu = (cbu * 224) / 255;
750 cgu = (cgu * 224) / 255;
751 cgv = (cgv * 224) / 255;
754 cy = (cy * contrast) >> 16;
755 crv = (crv * contrast * saturation) >> 32;
756 cbu = (cbu * contrast * saturation) >> 32;
757 cgu = (cgu * contrast * saturation) >> 32;
758 cgv = (cgv * contrast * saturation) >> 32;
759 oy -= 256 * brightness;
761 c->uOffset = 0x0400040004000400LL;
762 c->vOffset = 0x0400040004000400LL;
763 c->yCoeff = roundToInt16(cy * 8192) * 0x0001000100010001ULL;
764 c->vrCoeff = roundToInt16(crv * 8192) * 0x0001000100010001ULL;
765 c->ubCoeff = roundToInt16(cbu * 8192) * 0x0001000100010001ULL;
766 c->vgCoeff = roundToInt16(cgv * 8192) * 0x0001000100010001ULL;
767 c->ugCoeff = roundToInt16(cgu * 8192) * 0x0001000100010001ULL;
768 c->yOffset = roundToInt16(oy * 8) * 0x0001000100010001ULL;
770 c->yuv2rgb_y_coeff = (int16_t)roundToInt16(cy << 13);
771 c->yuv2rgb_y_offset = (int16_t)roundToInt16(oy << 9);
772 c->yuv2rgb_v2r_coeff = (int16_t)roundToInt16(crv << 13);
773 c->yuv2rgb_v2g_coeff = (int16_t)roundToInt16(cgv << 13);
774 c->yuv2rgb_u2g_coeff = (int16_t)roundToInt16(cgu << 13);
775 c->yuv2rgb_u2b_coeff = (int16_t)roundToInt16(cbu << 13);
777 //scale coefficients by cy
778 crv = ((crv << 16) + 0x8000) / cy;
779 cbu = ((cbu << 16) + 0x8000) / cy;
780 cgu = ((cgu << 16) + 0x8000) / cy;
781 cgv = ((cgv << 16) + 0x8000) / cy;
783 av_free(c->yuvTable);
787 c->yuvTable = av_malloc(1024);
788 y_table = c->yuvTable;
789 yb = -(384 << 16) - oy;
790 for (i = 0; i < 1024 - 110; i++) {
791 y_table[i + 110] = av_clip_uint8((yb + 0x8000) >> 16) >> 7;
794 fill_table(c->table_gU, 1, cgu, y_table + yoffs);
795 fill_gv_table(c->table_gV, 1, cgv);
799 rbase = isRgb ? 3 : 0;
801 bbase = isRgb ? 0 : 3;
802 c->yuvTable = av_malloc(1024 * 3);
803 y_table = c->yuvTable;
804 yb = -(384 << 16) - oy;
805 for (i = 0; i < 1024 - 110; i++) {
806 int yval = av_clip_uint8((yb + 0x8000) >> 16);
807 y_table[i + 110] = (yval >> 7) << rbase;
808 y_table[i + 37 + 1024] = ((yval + 43) / 85) << gbase;
809 y_table[i + 110 + 2048] = (yval >> 7) << bbase;
812 fill_table(c->table_rV, 1, crv, y_table + yoffs);
813 fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024);
814 fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048);
815 fill_gv_table(c->table_gV, 1, cgv);
818 rbase = isRgb ? 5 : 0;
819 gbase = isRgb ? 2 : 3;
820 bbase = isRgb ? 0 : 6;
821 c->yuvTable = av_malloc(1024 * 3);
822 y_table = c->yuvTable;
823 yb = -(384 << 16) - oy;
824 for (i = 0; i < 1024 - 38; i++) {
825 int yval = av_clip_uint8((yb + 0x8000) >> 16);
826 y_table[i + 16] = ((yval + 18) / 36) << rbase;
827 y_table[i + 16 + 1024] = ((yval + 18) / 36) << gbase;
828 y_table[i + 37 + 2048] = ((yval + 43) / 85) << bbase;
831 fill_table(c->table_rV, 1, crv, y_table + yoffs);
832 fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024);
833 fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048);
834 fill_gv_table(c->table_gV, 1, cgv);
837 rbase = isRgb ? 8 : 0;
839 bbase = isRgb ? 0 : 8;
840 c->yuvTable = av_malloc(1024 * 3 * 2);
841 y_table16 = c->yuvTable;
842 yb = -(384 << 16) - oy;
843 for (i = 0; i < 1024; i++) {
844 uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
845 y_table16[i] = (yval >> 4) << rbase;
846 y_table16[i + 1024] = (yval >> 4) << gbase;
847 y_table16[i + 2048] = (yval >> 4) << bbase;
851 for (i = 0; i < 1024 * 3; i++)
852 y_table16[i] = av_bswap16(y_table16[i]);
853 fill_table(c->table_rV, 2, crv, y_table16 + yoffs);
854 fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + 1024);
855 fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2048);
856 fill_gv_table(c->table_gV, 2, cgv);
860 rbase = isRgb ? bpp - 5 : 0;
862 bbase = isRgb ? 0 : (bpp - 5);
863 c->yuvTable = av_malloc(1024 * 3 * 2);
864 y_table16 = c->yuvTable;
865 yb = -(384 << 16) - oy;
866 for (i = 0; i < 1024; i++) {
867 uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
868 y_table16[i] = (yval >> 3) << rbase;
869 y_table16[i + 1024] = (yval >> (18 - bpp)) << gbase;
870 y_table16[i + 2048] = (yval >> 3) << bbase;
874 for (i = 0; i < 1024 * 3; i++)
875 y_table16[i] = av_bswap16(y_table16[i]);
876 fill_table(c->table_rV, 2, crv, y_table16 + yoffs);
877 fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + 1024);
878 fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2048);
879 fill_gv_table(c->table_gV, 2, cgv);
883 c->yuvTable = av_malloc(1024);
884 y_table = c->yuvTable;
885 yb = -(384 << 16) - oy;
886 for (i = 0; i < 1024; i++) {
887 y_table[i] = av_clip_uint8((yb + 0x8000) >> 16);
890 fill_table(c->table_rV, 1, crv, y_table + yoffs);
891 fill_table(c->table_gU, 1, cgu, y_table + yoffs);
892 fill_table(c->table_bU, 1, cbu, y_table + yoffs);
893 fill_gv_table(c->table_gV, 1, cgv);
896 base = (c->dstFormat == AV_PIX_FMT_RGB32_1 ||
897 c->dstFormat == AV_PIX_FMT_BGR32_1) ? 8 : 0;
898 rbase = base + (isRgb ? 16 : 0);
900 bbase = base + (isRgb ? 0 : 16);
901 needAlpha = CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat);
903 abase = (base + 24) & 31;
904 c->yuvTable = av_malloc(1024 * 3 * 4);
905 y_table32 = c->yuvTable;
906 yb = -(384 << 16) - oy;
907 for (i = 0; i < 1024; i++) {
908 unsigned yval = av_clip_uint8((yb + 0x8000) >> 16);
909 y_table32[i] = (yval << rbase) +
910 (needAlpha ? 0 : (255u << abase));
911 y_table32[i + 1024] = yval << gbase;
912 y_table32[i + 2048] = yval << bbase;
915 fill_table(c->table_rV, 4, crv, y_table32 + yoffs);
916 fill_table(c->table_gU, 4, cgu, y_table32 + yoffs + 1024);
917 fill_table(c->table_bU, 4, cbu, y_table32 + yoffs + 2048);
918 fill_gv_table(c->table_gV, 4, cgv);
922 if(!isPlanar(c->dstFormat) || bpp<=24)
923 av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);