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 /* Color space conversion coefficients for YCbCr -> RGB mapping.
40 * Entries are {crv, cbu, cgu, cgv}
42 * crv = (255 / 224) * 65536 * (1 - cr) / 0.5
43 * cbu = (255 / 224) * 65536 * (1 - cb) / 0.5
44 * cgu = (255 / 224) * 65536 * (cb / cg) * (1 - cb) / 0.5
45 * cgv = (255 / 224) * 65536 * (cr / cg) * (1 - cr) / 0.5
47 * where Y = cr * R + cg * G + cb * B and cr + cg + cb = 1.
49 const int32_t ff_yuv2rgb_coeffs[11][4] = {
50 { 117489, 138438, 13975, 34925 }, /* no sequence_display_extension */
51 { 117489, 138438, 13975, 34925 }, /* ITU-R Rec. 709 (1990) */
52 { 104597, 132201, 25675, 53279 }, /* unspecified */
53 { 104597, 132201, 25675, 53279 }, /* reserved */
54 { 104448, 132798, 24759, 53109 }, /* FCC */
55 { 104597, 132201, 25675, 53279 }, /* ITU-R Rec. 624-4 System B, G */
56 { 104597, 132201, 25675, 53279 }, /* SMPTE 170M */
57 { 117579, 136230, 16907, 35559 }, /* SMPTE 240M (1987) */
59 { 110013, 140363, 12277, 42626 }, /* Bt-2020-NCL */
60 { 110013, 140363, 12277, 42626 }, /* Bt-2020-CL */
63 const int *sws_getCoefficients(int colorspace)
65 if (colorspace > 10 || colorspace < 0 || colorspace == 8)
66 colorspace = SWS_CS_DEFAULT;
67 return ff_yuv2rgb_coeffs[colorspace];
70 #define LOADCHROMA(i) \
73 r = (void *)c->table_rV[V+YUVRGB_TABLE_HEADROOM]; \
74 g = (void *)(c->table_gU[U+YUVRGB_TABLE_HEADROOM] + c->table_gV[V+YUVRGB_TABLE_HEADROOM]); \
75 b = (void *)c->table_bU[U+YUVRGB_TABLE_HEADROOM];
77 #define PUTRGB(dst, src, i) \
79 dst[2 * i] = r[Y] + g[Y] + b[Y]; \
81 dst[2 * i + 1] = r[Y] + g[Y] + b[Y];
83 #define PUTRGB24(dst, src, i) \
85 dst[6 * i + 0] = r[Y]; \
86 dst[6 * i + 1] = g[Y]; \
87 dst[6 * i + 2] = b[Y]; \
89 dst[6 * i + 3] = r[Y]; \
90 dst[6 * i + 4] = g[Y]; \
91 dst[6 * i + 5] = b[Y];
93 #define PUTBGR24(dst, src, i) \
95 dst[6 * i + 0] = b[Y]; \
96 dst[6 * i + 1] = g[Y]; \
97 dst[6 * i + 2] = r[Y]; \
99 dst[6 * i + 3] = b[Y]; \
100 dst[6 * i + 4] = g[Y]; \
101 dst[6 * i + 5] = r[Y];
103 #define PUTRGBA(dst, ysrc, asrc, i, s) \
105 dst[2 * i] = r[Y] + g[Y] + b[Y] + (asrc[2 * i] << s); \
106 Y = ysrc[2 * i + 1]; \
107 dst[2 * i + 1] = r[Y] + g[Y] + b[Y] + (asrc[2 * i + 1] << s);
109 #define PUTRGB48(dst, src, i) \
111 dst[12 * i + 0] = dst[12 * i + 1] = r[Y]; \
112 dst[12 * i + 2] = dst[12 * i + 3] = g[Y]; \
113 dst[12 * i + 4] = dst[12 * i + 5] = b[Y]; \
114 Y = src[ 2 * i + 1]; \
115 dst[12 * i + 6] = dst[12 * i + 7] = r[Y]; \
116 dst[12 * i + 8] = dst[12 * i + 9] = g[Y]; \
117 dst[12 * i + 10] = dst[12 * i + 11] = b[Y];
119 #define PUTBGR48(dst, src, i) \
121 dst[12 * i + 0] = dst[12 * i + 1] = b[Y]; \
122 dst[12 * i + 2] = dst[12 * i + 3] = g[Y]; \
123 dst[12 * i + 4] = dst[12 * i + 5] = r[Y]; \
124 Y = src[2 * i + 1]; \
125 dst[12 * i + 6] = dst[12 * i + 7] = b[Y]; \
126 dst[12 * i + 8] = dst[12 * i + 9] = g[Y]; \
127 dst[12 * i + 10] = dst[12 * i + 11] = r[Y];
129 #define YUV2RGBFUNC(func_name, dst_type, alpha) \
130 static int func_name(SwsContext *c, const uint8_t *src[], \
131 int srcStride[], int srcSliceY, int srcSliceH, \
132 uint8_t *dst[], int dstStride[]) \
136 if (!alpha && c->srcFormat == AV_PIX_FMT_YUV422P) { \
140 for (y = 0; y < srcSliceH; y += 2) { \
141 int yd = y + srcSliceY; \
143 (dst_type *)(dst[0] + (yd) * dstStride[0]); \
145 (dst_type *)(dst[0] + (yd + 1) * dstStride[0]); \
146 dst_type av_unused *r, *g, *b; \
147 const uint8_t *py_1 = src[0] + y * srcStride[0]; \
148 const uint8_t *py_2 = py_1 + srcStride[0]; \
149 const uint8_t *pu = src[1] + (y >> 1) * srcStride[1]; \
150 const uint8_t *pv = src[2] + (y >> 1) * srcStride[2]; \
151 const uint8_t av_unused *pa_1, *pa_2; \
152 unsigned int h_size = c->dstW >> 3; \
154 pa_1 = src[3] + y * srcStride[3]; \
155 pa_2 = pa_1 + srcStride[3]; \
158 int av_unused U, V, Y; \
160 #define ENDYUV2RGBLINE(dst_delta, ss) \
165 dst_1 += dst_delta >> ss; \
166 dst_2 += dst_delta >> ss; \
168 if (c->dstW & (4 >> ss)) { \
169 int av_unused Y, U, V; \
171 #define ENDYUV2RGBFUNC() \
177 #define CLOSEYUV2RGBFUNC(dst_delta) \
178 ENDYUV2RGBLINE(dst_delta, 0) \
181 YUV2RGBFUNC(yuv2rgb_c_48, uint8_t, 0)
183 PUTRGB48(dst_1, py_1, 0);
184 PUTRGB48(dst_2, py_2, 0);
187 PUTRGB48(dst_2, py_2, 1);
188 PUTRGB48(dst_1, py_1, 1);
191 PUTRGB48(dst_1, py_1, 2);
192 PUTRGB48(dst_2, py_2, 2);
195 PUTRGB48(dst_2, py_2, 3);
196 PUTRGB48(dst_1, py_1, 3);
197 ENDYUV2RGBLINE(48, 0)
199 PUTRGB48(dst_1, py_1, 0);
200 PUTRGB48(dst_2, py_2, 0);
203 PUTRGB48(dst_2, py_2, 1);
204 PUTRGB48(dst_1, py_1, 1);
205 ENDYUV2RGBLINE(48, 1)
207 PUTRGB48(dst_1, py_1, 0);
208 PUTRGB48(dst_2, py_2, 0);
211 YUV2RGBFUNC(yuv2rgb_c_bgr48, uint8_t, 0)
213 PUTBGR48(dst_1, py_1, 0);
214 PUTBGR48(dst_2, py_2, 0);
217 PUTBGR48(dst_2, py_2, 1);
218 PUTBGR48(dst_1, py_1, 1);
221 PUTBGR48(dst_1, py_1, 2);
222 PUTBGR48(dst_2, py_2, 2);
225 PUTBGR48(dst_2, py_2, 3);
226 PUTBGR48(dst_1, py_1, 3);
227 ENDYUV2RGBLINE(48, 0)
229 PUTBGR48(dst_1, py_1, 0);
230 PUTBGR48(dst_2, py_2, 0);
233 PUTBGR48(dst_2, py_2, 1);
234 PUTBGR48(dst_1, py_1, 1);
235 ENDYUV2RGBLINE(48, 1)
237 PUTBGR48(dst_1, py_1, 0);
238 PUTBGR48(dst_2, py_2, 0);
241 YUV2RGBFUNC(yuv2rgb_c_32, uint32_t, 0)
243 PUTRGB(dst_1, py_1, 0);
244 PUTRGB(dst_2, py_2, 0);
247 PUTRGB(dst_2, py_2, 1);
248 PUTRGB(dst_1, py_1, 1);
251 PUTRGB(dst_1, py_1, 2);
252 PUTRGB(dst_2, py_2, 2);
255 PUTRGB(dst_2, py_2, 3);
256 PUTRGB(dst_1, py_1, 3);
259 PUTRGB(dst_1, py_1, 0);
260 PUTRGB(dst_2, py_2, 0);
263 PUTRGB(dst_2, py_2, 1);
264 PUTRGB(dst_1, py_1, 1);
267 PUTRGB(dst_1, py_1, 0);
268 PUTRGB(dst_2, py_2, 0);
272 YUV2RGBFUNC(yuva2argb_c, uint32_t, 1)
274 YUV2RGBFUNC(yuva2rgba_c, uint32_t, 1)
277 PUTRGBA(dst_1, py_1, pa_1, 0, 24);
278 PUTRGBA(dst_2, py_2, pa_2, 0, 24);
281 PUTRGBA(dst_2, py_2, pa_2, 1, 24);
282 PUTRGBA(dst_1, py_1, pa_1, 1, 24);
285 PUTRGBA(dst_1, py_1, pa_1, 2, 24);
286 PUTRGBA(dst_2, py_2, pa_2, 2, 24);
289 PUTRGBA(dst_2, py_2, pa_2, 3, 24);
290 PUTRGBA(dst_1, py_1, pa_1, 3, 24);
295 PUTRGBA(dst_1, py_1, pa_1, 0, 24);
296 PUTRGBA(dst_2, py_2, pa_2, 0, 24);
299 PUTRGBA(dst_2, py_2, pa_2, 1, 24);
300 PUTRGBA(dst_1, py_1, pa_1, 1, 24);
305 PUTRGBA(dst_1, py_1, pa_1, 0, 24);
306 PUTRGBA(dst_2, py_2, pa_2, 0, 24);
310 YUV2RGBFUNC(yuva2rgba_c, uint32_t, 1)
312 YUV2RGBFUNC(yuva2argb_c, uint32_t, 1)
315 PUTRGBA(dst_1, py_1, pa_1, 0, 0);
316 PUTRGBA(dst_2, py_2, pa_2, 0, 0);
319 PUTRGBA(dst_2, py_2, pa_2, 1, 0);
320 PUTRGBA(dst_1, py_1, pa_1, 1, 0);
323 PUTRGBA(dst_1, py_1, pa_1, 2, 0);
324 PUTRGBA(dst_2, py_2, pa_2, 2, 0);
327 PUTRGBA(dst_2, py_2, pa_2, 3, 0);
328 PUTRGBA(dst_1, py_1, pa_1, 3, 0);
333 PUTRGBA(dst_1, py_1, pa_1, 0, 0);
334 PUTRGBA(dst_2, py_2, pa_2, 0, 0);
337 PUTRGBA(dst_2, py_2, pa_2, 1, 0);
338 PUTRGBA(dst_1, py_1, pa_1, 1, 0);
343 PUTRGBA(dst_1, py_1, pa_1, 0, 0);
344 PUTRGBA(dst_2, py_2, pa_2, 0, 0);
347 YUV2RGBFUNC(yuv2rgb_c_24_rgb, uint8_t, 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);
357 PUTRGB24(dst_1, py_1, 2);
358 PUTRGB24(dst_2, py_2, 2);
361 PUTRGB24(dst_2, py_2, 3);
362 PUTRGB24(dst_1, py_1, 3);
363 ENDYUV2RGBLINE(24, 0)
365 PUTRGB24(dst_1, py_1, 0);
366 PUTRGB24(dst_2, py_2, 0);
369 PUTRGB24(dst_2, py_2, 1);
370 PUTRGB24(dst_1, py_1, 1);
371 ENDYUV2RGBLINE(24, 1)
373 PUTRGB24(dst_1, py_1, 0);
374 PUTRGB24(dst_2, py_2, 0);
377 // only trivial mods from yuv2rgb_c_24_rgb
378 YUV2RGBFUNC(yuv2rgb_c_24_bgr, uint8_t, 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);
388 PUTBGR24(dst_1, py_1, 2);
389 PUTBGR24(dst_2, py_2, 2);
392 PUTBGR24(dst_2, py_2, 3);
393 PUTBGR24(dst_1, py_1, 3);
394 ENDYUV2RGBLINE(24, 0)
396 PUTBGR24(dst_1, py_1, 0);
397 PUTBGR24(dst_2, py_2, 0);
400 PUTBGR24(dst_2, py_2, 1);
401 PUTBGR24(dst_1, py_1, 1);
402 ENDYUV2RGBLINE(24, 1)
404 PUTBGR24(dst_1, py_1, 0);
405 PUTBGR24(dst_2, py_2, 0);
408 YUV2RGBFUNC(yuv2rgb_c_16_ordered_dither, uint16_t, 0)
409 const uint8_t *d16 = ff_dither_2x2_8[y & 1];
410 const uint8_t *e16 = ff_dither_2x2_4[y & 1];
411 const uint8_t *f16 = ff_dither_2x2_8[(y & 1)^1];
413 #define PUTRGB16(dst, src, i, o) \
415 dst[2 * i] = r[Y + d16[0 + o]] + \
416 g[Y + e16[0 + o]] + \
418 Y = src[2 * i + 1]; \
419 dst[2 * i + 1] = r[Y + d16[1 + o]] + \
420 g[Y + e16[1 + o]] + \
423 PUTRGB16(dst_1, py_1, 0, 0);
424 PUTRGB16(dst_2, py_2, 0, 0 + 8);
427 PUTRGB16(dst_2, py_2, 1, 2 + 8);
428 PUTRGB16(dst_1, py_1, 1, 2);
431 PUTRGB16(dst_1, py_1, 2, 4);
432 PUTRGB16(dst_2, py_2, 2, 4 + 8);
435 PUTRGB16(dst_2, py_2, 3, 6 + 8);
436 PUTRGB16(dst_1, py_1, 3, 6);
439 YUV2RGBFUNC(yuv2rgb_c_15_ordered_dither, uint16_t, 0)
440 const uint8_t *d16 = ff_dither_2x2_8[y & 1];
441 const uint8_t *e16 = ff_dither_2x2_8[(y & 1)^1];
443 #define PUTRGB15(dst, src, i, o) \
445 dst[2 * i] = r[Y + d16[0 + o]] + \
446 g[Y + d16[1 + o]] + \
448 Y = src[2 * i + 1]; \
449 dst[2 * i + 1] = r[Y + d16[1 + o]] + \
450 g[Y + d16[0 + o]] + \
453 PUTRGB15(dst_1, py_1, 0, 0);
454 PUTRGB15(dst_2, py_2, 0, 0 + 8);
457 PUTRGB15(dst_2, py_2, 1, 2 + 8);
458 PUTRGB15(dst_1, py_1, 1, 2);
461 PUTRGB15(dst_1, py_1, 2, 4);
462 PUTRGB15(dst_2, py_2, 2, 4 + 8);
465 PUTRGB15(dst_2, py_2, 3, 6 + 8);
466 PUTRGB15(dst_1, py_1, 3, 6);
469 // r, g, b, dst_1, dst_2
470 YUV2RGBFUNC(yuv2rgb_c_12_ordered_dither, uint16_t, 0)
471 const uint8_t *d16 = ff_dither_4x4_16[y & 3];
473 #define PUTRGB12(dst, src, i, o) \
475 dst[2 * i] = r[Y + d16[0 + o]] + \
476 g[Y + d16[0 + o]] + \
478 Y = src[2 * i + 1]; \
479 dst[2 * i + 1] = r[Y + d16[1 + o]] + \
480 g[Y + d16[1 + o]] + \
484 PUTRGB12(dst_1, py_1, 0, 0);
485 PUTRGB12(dst_2, py_2, 0, 0 + 8);
488 PUTRGB12(dst_2, py_2, 1, 2 + 8);
489 PUTRGB12(dst_1, py_1, 1, 2);
492 PUTRGB12(dst_1, py_1, 2, 4);
493 PUTRGB12(dst_2, py_2, 2, 4 + 8);
496 PUTRGB12(dst_2, py_2, 3, 6 + 8);
497 PUTRGB12(dst_1, py_1, 3, 6);
500 // r, g, b, dst_1, dst_2
501 YUV2RGBFUNC(yuv2rgb_c_8_ordered_dither, uint8_t, 0)
502 const uint8_t *d32 = ff_dither_8x8_32[yd & 7];
503 const uint8_t *d64 = ff_dither_8x8_73[yd & 7];
505 #define PUTRGB8(dst, src, i, o) \
507 dst[2 * i] = r[Y + d32[0 + o]] + \
508 g[Y + d32[0 + o]] + \
510 Y = src[2 * i + 1]; \
511 dst[2 * i + 1] = r[Y + d32[1 + o]] + \
512 g[Y + d32[1 + o]] + \
516 PUTRGB8(dst_1, py_1, 0, 0);
517 PUTRGB8(dst_2, py_2, 0, 0 + 8);
520 PUTRGB8(dst_2, py_2, 1, 2 + 8);
521 PUTRGB8(dst_1, py_1, 1, 2);
524 PUTRGB8(dst_1, py_1, 2, 4);
525 PUTRGB8(dst_2, py_2, 2, 4 + 8);
528 PUTRGB8(dst_2, py_2, 3, 6 + 8);
529 PUTRGB8(dst_1, py_1, 3, 6);
532 const uint8_t *d32 = ff_dither_8x8_32[yd & 7];
533 const uint8_t *d64 = ff_dither_8x8_73[yd & 7];
535 PUTRGB8(dst_1, py_1, 0, 0);
536 PUTRGB8(dst_2, py_2, 0, 0 + 8);
539 PUTRGB8(dst_2, py_2, 1, 2 + 8);
540 PUTRGB8(dst_1, py_1, 1, 2);
543 const uint8_t *d32 = ff_dither_8x8_32[yd & 7];
544 const uint8_t *d64 = ff_dither_8x8_73[yd & 7];
546 PUTRGB8(dst_1, py_1, 0, 0);
547 PUTRGB8(dst_2, py_2, 0, 0 + 8);
552 YUV2RGBFUNC(yuv2rgb_c_4_ordered_dither, uint8_t, 0)
553 const uint8_t * d64 = ff_dither_8x8_73[yd & 7];
554 const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
557 #define PUTRGB4D(dst, src, i, o) \
559 acc = r[Y + d128[0 + o]] + \
560 g[Y + d64[0 + o]] + \
561 b[Y + d128[0 + o]]; \
562 Y = src[2 * i + 1]; \
563 acc |= (r[Y + d128[1 + o]] + \
564 g[Y + d64[1 + o]] + \
565 b[Y + d128[1 + o]]) << 4; \
569 PUTRGB4D(dst_1, py_1, 0, 0);
570 PUTRGB4D(dst_2, py_2, 0, 0 + 8);
573 PUTRGB4D(dst_2, py_2, 1, 2 + 8);
574 PUTRGB4D(dst_1, py_1, 1, 2);
577 PUTRGB4D(dst_1, py_1, 2, 4);
578 PUTRGB4D(dst_2, py_2, 2, 4 + 8);
581 PUTRGB4D(dst_2, py_2, 3, 6 + 8);
582 PUTRGB4D(dst_1, py_1, 3, 6);
585 const uint8_t * d64 = ff_dither_8x8_73[yd & 7];
586 const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
589 PUTRGB4D(dst_1, py_1, 0, 0);
590 PUTRGB4D(dst_2, py_2, 0, 0 + 8);
593 PUTRGB4D(dst_2, py_2, 1, 2 + 8);
594 PUTRGB4D(dst_1, py_1, 1, 2);
597 const uint8_t * d64 = ff_dither_8x8_73[yd & 7];
598 const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
601 PUTRGB4D(dst_1, py_1, 0, 0);
602 PUTRGB4D(dst_2, py_2, 0, 0 + 8);
605 YUV2RGBFUNC(yuv2rgb_c_4b_ordered_dither, uint8_t, 0)
606 const uint8_t *d64 = ff_dither_8x8_73[yd & 7];
607 const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
609 #define PUTRGB4DB(dst, src, i, o) \
611 dst[2 * i] = r[Y + d128[0 + o]] + \
612 g[Y + d64[0 + o]] + \
613 b[Y + d128[0 + o]]; \
614 Y = src[2 * i + 1]; \
615 dst[2 * i + 1] = r[Y + d128[1 + o]] + \
616 g[Y + d64[1 + o]] + \
620 PUTRGB4DB(dst_1, py_1, 0, 0);
621 PUTRGB4DB(dst_2, py_2, 0, 0 + 8);
624 PUTRGB4DB(dst_2, py_2, 1, 2 + 8);
625 PUTRGB4DB(dst_1, py_1, 1, 2);
628 PUTRGB4DB(dst_1, py_1, 2, 4);
629 PUTRGB4DB(dst_2, py_2, 2, 4 + 8);
632 PUTRGB4DB(dst_2, py_2, 3, 6 + 8);
633 PUTRGB4DB(dst_1, py_1, 3, 6);
635 const uint8_t *d64 = ff_dither_8x8_73[yd & 7];
636 const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
638 PUTRGB4DB(dst_1, py_1, 0, 0);
639 PUTRGB4DB(dst_2, py_2, 0, 0 + 8);
642 PUTRGB4DB(dst_2, py_2, 1, 2 + 8);
643 PUTRGB4DB(dst_1, py_1, 1, 2);
645 const uint8_t *d64 = ff_dither_8x8_73[yd & 7];
646 const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
648 PUTRGB4DB(dst_1, py_1, 0, 0);
649 PUTRGB4DB(dst_2, py_2, 0, 0 + 8);
652 YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0)
653 const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
654 char out_1 = 0, out_2 = 0;
655 g = c->table_gU[128 + YUVRGB_TABLE_HEADROOM] + c->table_gV[128 + YUVRGB_TABLE_HEADROOM];
657 #define PUTRGB1(out, src, i, o) \
659 out += out + g[Y + d128[0 + o]]; \
660 Y = src[2 * i + 1]; \
661 out += out + g[Y + d128[1 + o]];
663 PUTRGB1(out_1, py_1, 0, 0);
664 PUTRGB1(out_2, py_2, 0, 0 + 8);
666 PUTRGB1(out_2, py_2, 1, 2 + 8);
667 PUTRGB1(out_1, py_1, 1, 2);
669 PUTRGB1(out_1, py_1, 2, 4);
670 PUTRGB1(out_2, py_2, 2, 4 + 8);
672 PUTRGB1(out_2, py_2, 3, 6 + 8);
673 PUTRGB1(out_1, py_1, 3, 6);
679 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c)
684 t = ff_yuv2rgb_init_ppc(c);
686 t = ff_yuv2rgb_init_x86(c);
691 av_log(c, AV_LOG_WARNING,
692 "No accelerated colorspace conversion found from %s to %s.\n",
693 av_get_pix_fmt_name(c->srcFormat), av_get_pix_fmt_name(c->dstFormat));
695 switch (c->dstFormat) {
696 case AV_PIX_FMT_BGR48BE:
697 case AV_PIX_FMT_BGR48LE:
698 return yuv2rgb_c_bgr48;
699 case AV_PIX_FMT_RGB48BE:
700 case AV_PIX_FMT_RGB48LE:
702 case AV_PIX_FMT_ARGB:
703 case AV_PIX_FMT_ABGR:
704 if (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat))
706 case AV_PIX_FMT_RGBA:
707 case AV_PIX_FMT_BGRA:
708 return (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat)) ? yuva2rgba_c : yuv2rgb_c_32;
709 case AV_PIX_FMT_RGB24:
710 return yuv2rgb_c_24_rgb;
711 case AV_PIX_FMT_BGR24:
712 return yuv2rgb_c_24_bgr;
713 case AV_PIX_FMT_RGB565:
714 case AV_PIX_FMT_BGR565:
715 return yuv2rgb_c_16_ordered_dither;
716 case AV_PIX_FMT_RGB555:
717 case AV_PIX_FMT_BGR555:
718 return yuv2rgb_c_15_ordered_dither;
719 case AV_PIX_FMT_RGB444:
720 case AV_PIX_FMT_BGR444:
721 return yuv2rgb_c_12_ordered_dither;
722 case AV_PIX_FMT_RGB8:
723 case AV_PIX_FMT_BGR8:
724 return yuv2rgb_c_8_ordered_dither;
725 case AV_PIX_FMT_RGB4:
726 case AV_PIX_FMT_BGR4:
727 return yuv2rgb_c_4_ordered_dither;
728 case AV_PIX_FMT_RGB4_BYTE:
729 case AV_PIX_FMT_BGR4_BYTE:
730 return yuv2rgb_c_4b_ordered_dither;
731 case AV_PIX_FMT_MONOBLACK:
732 return yuv2rgb_c_1_ordered_dither;
737 static void fill_table(uint8_t* table[256 + 2*YUVRGB_TABLE_HEADROOM], const int elemsize,
738 const int64_t inc, void *y_tab)
741 uint8_t *y_table = y_tab;
743 y_table -= elemsize * (inc >> 9);
745 for (i = 0; i < 256 + 2*YUVRGB_TABLE_HEADROOM; i++) {
746 int64_t cb = av_clip_uint8(i-YUVRGB_TABLE_HEADROOM)*inc;
747 table[i] = y_table + elemsize * (cb >> 16);
751 static void fill_gv_table(int table[256 + 2*YUVRGB_TABLE_HEADROOM], const int elemsize, const int64_t inc)
754 int off = -(inc >> 9);
756 for (i = 0; i < 256 + 2*YUVRGB_TABLE_HEADROOM; i++) {
757 int64_t cb = av_clip_uint8(i-YUVRGB_TABLE_HEADROOM)*inc;
758 table[i] = elemsize * (off + (cb >> 16));
762 static uint16_t roundToInt16(int64_t f)
764 int r = (f + (1 << 15)) >> 16;
774 av_cold int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
775 int fullRange, int brightness,
776 int contrast, int saturation)
778 const int isRgb = c->dstFormat == AV_PIX_FMT_RGB32 ||
779 c->dstFormat == AV_PIX_FMT_RGB32_1 ||
780 c->dstFormat == AV_PIX_FMT_BGR24 ||
781 c->dstFormat == AV_PIX_FMT_RGB565BE ||
782 c->dstFormat == AV_PIX_FMT_RGB565LE ||
783 c->dstFormat == AV_PIX_FMT_RGB555BE ||
784 c->dstFormat == AV_PIX_FMT_RGB555LE ||
785 c->dstFormat == AV_PIX_FMT_RGB444BE ||
786 c->dstFormat == AV_PIX_FMT_RGB444LE ||
787 c->dstFormat == AV_PIX_FMT_RGB8 ||
788 c->dstFormat == AV_PIX_FMT_RGB4 ||
789 c->dstFormat == AV_PIX_FMT_RGB4_BYTE ||
790 c->dstFormat == AV_PIX_FMT_MONOBLACK;
791 const int isNotNe = c->dstFormat == AV_PIX_FMT_NE(RGB565LE, RGB565BE) ||
792 c->dstFormat == AV_PIX_FMT_NE(RGB555LE, RGB555BE) ||
793 c->dstFormat == AV_PIX_FMT_NE(RGB444LE, RGB444BE) ||
794 c->dstFormat == AV_PIX_FMT_NE(BGR565LE, BGR565BE) ||
795 c->dstFormat == AV_PIX_FMT_NE(BGR555LE, BGR555BE) ||
796 c->dstFormat == AV_PIX_FMT_NE(BGR444LE, BGR444BE) ||
797 c->dstFormat == AV_PIX_FMT_NE(X2RGB10LE, X2RGB10BE);
798 const int bpp = c->dstFormatBpp;
802 int i, base, rbase, gbase, bbase, av_uninit(abase), needAlpha;
803 const int yoffs = (fullRange ? 384 : 326) + YUVRGB_TABLE_LUMA_HEADROOM;
804 const int table_plane_size = 1024 + 2*YUVRGB_TABLE_LUMA_HEADROOM;
806 int64_t crv = inv_table[0];
807 int64_t cbu = inv_table[1];
808 int64_t cgu = -inv_table[2];
809 int64_t cgv = -inv_table[3];
810 int64_t cy = 1 << 16;
815 cy = (cy * 255) / 219;
818 crv = (crv * 224) / 255;
819 cbu = (cbu * 224) / 255;
820 cgu = (cgu * 224) / 255;
821 cgv = (cgv * 224) / 255;
824 cy = (cy * contrast) >> 16;
825 crv = (crv * contrast * saturation) >> 32;
826 cbu = (cbu * contrast * saturation) >> 32;
827 cgu = (cgu * contrast * saturation) >> 32;
828 cgv = (cgv * contrast * saturation) >> 32;
829 oy -= 256 * brightness;
831 c->uOffset = 0x0400040004000400LL;
832 c->vOffset = 0x0400040004000400LL;
833 c->yCoeff = roundToInt16(cy * (1 << 13)) * 0x0001000100010001ULL;
834 c->vrCoeff = roundToInt16(crv * (1 << 13)) * 0x0001000100010001ULL;
835 c->ubCoeff = roundToInt16(cbu * (1 << 13)) * 0x0001000100010001ULL;
836 c->vgCoeff = roundToInt16(cgv * (1 << 13)) * 0x0001000100010001ULL;
837 c->ugCoeff = roundToInt16(cgu * (1 << 13)) * 0x0001000100010001ULL;
838 c->yOffset = roundToInt16(oy * (1 << 3)) * 0x0001000100010001ULL;
840 c->yuv2rgb_y_coeff = (int16_t)roundToInt16(cy * (1 << 13));
841 c->yuv2rgb_y_offset = (int16_t)roundToInt16(oy * (1 << 9));
842 c->yuv2rgb_v2r_coeff = (int16_t)roundToInt16(crv * (1 << 13));
843 c->yuv2rgb_v2g_coeff = (int16_t)roundToInt16(cgv * (1 << 13));
844 c->yuv2rgb_u2g_coeff = (int16_t)roundToInt16(cgu * (1 << 13));
845 c->yuv2rgb_u2b_coeff = (int16_t)roundToInt16(cbu * (1 << 13));
847 //scale coefficients by cy
848 crv = ((crv * (1 << 16)) + 0x8000) / FFMAX(cy, 1);
849 cbu = ((cbu * (1 << 16)) + 0x8000) / FFMAX(cy, 1);
850 cgu = ((cgu * (1 << 16)) + 0x8000) / FFMAX(cy, 1);
851 cgv = ((cgv * (1 << 16)) + 0x8000) / FFMAX(cy, 1);
853 av_freep(&c->yuvTable);
855 #define ALLOC_YUV_TABLE(x) \
856 c->yuvTable = av_malloc(x); \
858 return AVERROR(ENOMEM);
861 ALLOC_YUV_TABLE(table_plane_size);
862 y_table = c->yuvTable;
863 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
864 for (i = 0; i < table_plane_size - 110; i++) {
865 y_table[i + 110] = av_clip_uint8((yb + 0x8000) >> 16) >> 7;
868 fill_table(c->table_gU, 1, cgu, y_table + yoffs);
869 fill_gv_table(c->table_gV, 1, cgv);
873 rbase = isRgb ? 3 : 0;
875 bbase = isRgb ? 0 : 3;
876 ALLOC_YUV_TABLE(table_plane_size * 3);
877 y_table = c->yuvTable;
878 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
879 for (i = 0; i < table_plane_size - 110; i++) {
880 int yval = av_clip_uint8((yb + 0x8000) >> 16);
881 y_table[i + 110] = (yval >> 7) << rbase;
882 y_table[i + 37 + table_plane_size] = ((yval + 43) / 85) << gbase;
883 y_table[i + 110 + 2*table_plane_size] = (yval >> 7) << bbase;
886 fill_table(c->table_rV, 1, crv, y_table + yoffs);
887 fill_table(c->table_gU, 1, cgu, y_table + yoffs + table_plane_size);
888 fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2*table_plane_size);
889 fill_gv_table(c->table_gV, 1, cgv);
892 rbase = isRgb ? 5 : 0;
893 gbase = isRgb ? 2 : 3;
894 bbase = isRgb ? 0 : 6;
895 ALLOC_YUV_TABLE(table_plane_size * 3);
896 y_table = c->yuvTable;
897 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
898 for (i = 0; i < table_plane_size - 38; i++) {
899 int yval = av_clip_uint8((yb + 0x8000) >> 16);
900 y_table[i + 16] = ((yval + 18) / 36) << rbase;
901 y_table[i + 16 + table_plane_size] = ((yval + 18) / 36) << gbase;
902 y_table[i + 37 + 2*table_plane_size] = ((yval + 43) / 85) << bbase;
905 fill_table(c->table_rV, 1, crv, y_table + yoffs);
906 fill_table(c->table_gU, 1, cgu, y_table + yoffs + table_plane_size);
907 fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2*table_plane_size);
908 fill_gv_table(c->table_gV, 1, cgv);
911 rbase = isRgb ? 8 : 0;
913 bbase = isRgb ? 0 : 8;
914 ALLOC_YUV_TABLE(table_plane_size * 3 * 2);
915 y_table16 = c->yuvTable;
916 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
917 for (i = 0; i < table_plane_size; i++) {
918 uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
919 y_table16[i] = (yval >> 4) << rbase;
920 y_table16[i + table_plane_size] = (yval >> 4) << gbase;
921 y_table16[i + 2*table_plane_size] = (yval >> 4) << bbase;
925 for (i = 0; i < table_plane_size * 3; i++)
926 y_table16[i] = av_bswap16(y_table16[i]);
927 fill_table(c->table_rV, 2, crv, y_table16 + yoffs);
928 fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + table_plane_size);
929 fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2*table_plane_size);
930 fill_gv_table(c->table_gV, 2, cgv);
934 rbase = isRgb ? bpp - 5 : 0;
936 bbase = isRgb ? 0 : (bpp - 5);
937 ALLOC_YUV_TABLE(table_plane_size * 3 * 2);
938 y_table16 = c->yuvTable;
939 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
940 for (i = 0; i < table_plane_size; i++) {
941 uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
942 y_table16[i] = (yval >> 3) << rbase;
943 y_table16[i + table_plane_size] = (yval >> (18 - bpp)) << gbase;
944 y_table16[i + 2*table_plane_size] = (yval >> 3) << bbase;
948 for (i = 0; i < table_plane_size * 3; i++)
949 y_table16[i] = av_bswap16(y_table16[i]);
950 fill_table(c->table_rV, 2, crv, y_table16 + yoffs);
951 fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + table_plane_size);
952 fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2*table_plane_size);
953 fill_gv_table(c->table_gV, 2, cgv);
957 ALLOC_YUV_TABLE(table_plane_size);
958 y_table = c->yuvTable;
959 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
960 for (i = 0; i < table_plane_size; i++) {
961 y_table[i] = av_clip_uint8((yb + 0x8000) >> 16);
964 fill_table(c->table_rV, 1, crv, y_table + yoffs);
965 fill_table(c->table_gU, 1, cgu, y_table + yoffs);
966 fill_table(c->table_bU, 1, cbu, y_table + yoffs);
967 fill_gv_table(c->table_gV, 1, cgv);
973 needAlpha = CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat);
976 ALLOC_YUV_TABLE(table_plane_size * 3 * 4);
977 y_table32 = c->yuvTable;
978 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
979 for (i = 0; i < table_plane_size; i++) {
980 unsigned yval = av_clip_uint8((yb + 0x8000) >> 16);
981 y_table32[i]= (yval << rbase) + (needAlpha ? 0 : (255u << abase));
982 y_table32[i + table_plane_size] = yval << gbase;
983 y_table32[i + 2 * table_plane_size] = yval << bbase;
987 for (i = 0; i < table_plane_size * 3; i++)
988 y_table32[i] = av_bswap32(y_table32[i]);
990 fill_table(c->table_rV, 4, crv, y_table32 + yoffs);
991 fill_table(c->table_gU, 4, cgu, y_table32 + yoffs + table_plane_size);
992 fill_table(c->table_bU, 4, cbu, y_table32 + yoffs + 2 * table_plane_size);
993 fill_gv_table(c->table_gV, 4, cgv);
997 base = (c->dstFormat == AV_PIX_FMT_RGB32_1 ||
998 c->dstFormat == AV_PIX_FMT_BGR32_1) ? 8 : 0;
999 rbase = base + (isRgb ? 16 : 0);
1001 bbase = base + (isRgb ? 0 : 16);
1002 needAlpha = CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat);
1004 abase = (base + 24) & 31;
1005 ALLOC_YUV_TABLE(table_plane_size * 3 * 4);
1006 y_table32 = c->yuvTable;
1007 yb = -(384 << 16) - YUVRGB_TABLE_LUMA_HEADROOM*cy - oy;
1008 for (i = 0; i < table_plane_size; i++) {
1009 unsigned yval = av_clip_uint8((yb + 0x8000) >> 16);
1010 y_table32[i] = (yval << rbase) +
1011 (needAlpha ? 0 : (255u << abase));
1012 y_table32[i + table_plane_size] = yval << gbase;
1013 y_table32[i + 2*table_plane_size] = yval << bbase;
1016 fill_table(c->table_rV, 4, crv, y_table32 + yoffs);
1017 fill_table(c->table_gU, 4, cgu, y_table32 + yoffs + table_plane_size);
1018 fill_table(c->table_bU, 4, cbu, y_table32 + yoffs + 2*table_plane_size);
1019 fill_gv_table(c->table_gV, 4, cgv);
1022 if(!isPlanar(c->dstFormat) || bpp <= 24)
1023 av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);
1024 return AVERROR(EINVAL);