2 * Copyright (C) 2001-2012 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
26 #include "libavutil/attributes.h"
27 #include "libavutil/avutil.h"
28 #include "libavutil/avassert.h"
29 #include "libavutil/bswap.h"
30 #include "libavutil/cpu.h"
31 #include "libavutil/intreadwrite.h"
32 #include "libavutil/mathematics.h"
33 #include "libavutil/pixdesc.h"
37 #include "swscale_internal.h"
39 DECLARE_ALIGNED(8, const uint8_t, ff_dither_2x2_4)[][8] = {
40 { 1, 3, 1, 3, 1, 3, 1, 3, },
41 { 2, 0, 2, 0, 2, 0, 2, 0, },
42 { 1, 3, 1, 3, 1, 3, 1, 3, },
45 DECLARE_ALIGNED(8, const uint8_t, ff_dither_2x2_8)[][8] = {
46 { 6, 2, 6, 2, 6, 2, 6, 2, },
47 { 0, 4, 0, 4, 0, 4, 0, 4, },
48 { 6, 2, 6, 2, 6, 2, 6, 2, },
51 DECLARE_ALIGNED(8, const uint8_t, ff_dither_4x4_16)[][8] = {
52 { 8, 4, 11, 7, 8, 4, 11, 7, },
53 { 2, 14, 1, 13, 2, 14, 1, 13, },
54 { 10, 6, 9, 5, 10, 6, 9, 5, },
55 { 0, 12, 3, 15, 0, 12, 3, 15, },
56 { 8, 4, 11, 7, 8, 4, 11, 7, },
59 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_32)[][8] = {
60 { 17, 9, 23, 15, 16, 8, 22, 14, },
61 { 5, 29, 3, 27, 4, 28, 2, 26, },
62 { 21, 13, 19, 11, 20, 12, 18, 10, },
63 { 0, 24, 6, 30, 1, 25, 7, 31, },
64 { 16, 8, 22, 14, 17, 9, 23, 15, },
65 { 4, 28, 2, 26, 5, 29, 3, 27, },
66 { 20, 12, 18, 10, 21, 13, 19, 11, },
67 { 1, 25, 7, 31, 0, 24, 6, 30, },
68 { 17, 9, 23, 15, 16, 8, 22, 14, },
71 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_73)[][8] = {
72 { 0, 55, 14, 68, 3, 58, 17, 72, },
73 { 37, 18, 50, 32, 40, 22, 54, 35, },
74 { 9, 64, 5, 59, 13, 67, 8, 63, },
75 { 46, 27, 41, 23, 49, 31, 44, 26, },
76 { 2, 57, 16, 71, 1, 56, 15, 70, },
77 { 39, 21, 52, 34, 38, 19, 51, 33, },
78 { 11, 66, 7, 62, 10, 65, 6, 60, },
79 { 48, 30, 43, 25, 47, 29, 42, 24, },
80 { 0, 55, 14, 68, 3, 58, 17, 72, },
84 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
85 {117, 62, 158, 103, 113, 58, 155, 100, },
86 { 34, 199, 21, 186, 31, 196, 17, 182, },
87 {144, 89, 131, 76, 141, 86, 127, 72, },
88 { 0, 165, 41, 206, 10, 175, 52, 217, },
89 {110, 55, 151, 96, 120, 65, 162, 107, },
90 { 28, 193, 14, 179, 38, 203, 24, 189, },
91 {138, 83, 124, 69, 148, 93, 134, 79, },
92 { 7, 172, 48, 213, 3, 168, 45, 210, },
93 {117, 62, 158, 103, 113, 58, 155, 100, },
96 // tries to correct a gamma of 1.5
97 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
98 { 0, 143, 18, 200, 2, 156, 25, 215, },
99 { 78, 28, 125, 64, 89, 36, 138, 74, },
100 { 10, 180, 3, 161, 16, 195, 8, 175, },
101 {109, 51, 93, 38, 121, 60, 105, 47, },
102 { 1, 152, 23, 210, 0, 147, 20, 205, },
103 { 85, 33, 134, 71, 81, 30, 130, 67, },
104 { 14, 190, 6, 171, 12, 185, 5, 166, },
105 {117, 57, 101, 44, 113, 54, 97, 41, },
106 { 0, 143, 18, 200, 2, 156, 25, 215, },
109 // tries to correct a gamma of 2.0
110 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
111 { 0, 124, 8, 193, 0, 140, 12, 213, },
112 { 55, 14, 104, 42, 66, 19, 119, 52, },
113 { 3, 168, 1, 145, 6, 187, 3, 162, },
114 { 86, 31, 70, 21, 99, 39, 82, 28, },
115 { 0, 134, 11, 206, 0, 129, 9, 200, },
116 { 62, 17, 114, 48, 58, 16, 109, 45, },
117 { 5, 181, 2, 157, 4, 175, 1, 151, },
118 { 95, 36, 78, 26, 90, 34, 74, 24, },
119 { 0, 124, 8, 193, 0, 140, 12, 213, },
122 // tries to correct a gamma of 2.5
123 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
124 { 0, 107, 3, 187, 0, 125, 6, 212, },
125 { 39, 7, 86, 28, 49, 11, 102, 36, },
126 { 1, 158, 0, 131, 3, 180, 1, 151, },
127 { 68, 19, 52, 12, 81, 25, 64, 17, },
128 { 0, 119, 5, 203, 0, 113, 4, 195, },
129 { 45, 9, 96, 33, 42, 8, 91, 30, },
130 { 2, 172, 1, 144, 2, 165, 0, 137, },
131 { 77, 23, 60, 15, 72, 21, 56, 14, },
132 { 0, 107, 3, 187, 0, 125, 6, 212, },
136 #define output_pixel(pos, val, bias, signedness) \
138 AV_WB16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
140 AV_WL16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
143 static av_always_inline void
144 yuv2plane1_16_c_template(const int32_t *src, uint16_t *dest, int dstW,
145 int big_endian, int output_bits)
149 av_assert0(output_bits == 16);
151 for (i = 0; i < dstW; i++) {
152 int val = src[i] + (1 << (shift - 1));
153 output_pixel(&dest[i], val, 0, uint);
157 static av_always_inline void
158 yuv2planeX_16_c_template(const int16_t *filter, int filterSize,
159 const int32_t **src, uint16_t *dest, int dstW,
160 int big_endian, int output_bits)
164 av_assert0(output_bits == 16);
166 for (i = 0; i < dstW; i++) {
167 int val = 1 << (shift - 1);
170 /* range of val is [0,0x7FFFFFFF], so 31 bits, but with lanczos/spline
171 * filters (or anything with negative coeffs, the range can be slightly
172 * wider in both directions. To account for this overflow, we subtract
173 * a constant so it always fits in the signed range (assuming a
174 * reasonable filterSize), and re-add that at the end. */
176 for (j = 0; j < filterSize; j++)
177 val += src[j][i] * (unsigned)filter[j];
179 output_pixel(&dest[i], val, 0x8000, int);
185 #define output_pixel(pos, val) \
187 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
189 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
192 static av_always_inline void
193 yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
194 int big_endian, int output_bits)
197 int shift = 15 - output_bits;
199 for (i = 0; i < dstW; i++) {
200 int val = src[i] + (1 << (shift - 1));
201 output_pixel(&dest[i], val);
205 static av_always_inline void
206 yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
207 const int16_t **src, uint16_t *dest, int dstW,
208 int big_endian, int output_bits)
211 int shift = 11 + 16 - output_bits;
213 for (i = 0; i < dstW; i++) {
214 int val = 1 << (shift - 1);
217 for (j = 0; j < filterSize; j++)
218 val += src[j][i] * filter[j];
220 output_pixel(&dest[i], val);
226 #define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
227 static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
228 uint8_t *dest, int dstW, \
229 const uint8_t *dither, int offset)\
231 yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
232 (uint16_t *) dest, dstW, is_be, bits); \
234 static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
235 const int16_t **src, uint8_t *dest, int dstW, \
236 const uint8_t *dither, int offset)\
238 yuv2planeX_## template_size ## _c_template(filter, \
239 filterSize, (const typeX_t **) src, \
240 (uint16_t *) dest, dstW, is_be, bits); \
242 yuv2NBPS( 9, BE, 1, 10, int16_t)
243 yuv2NBPS( 9, LE, 0, 10, int16_t)
244 yuv2NBPS(10, BE, 1, 10, int16_t)
245 yuv2NBPS(10, LE, 0, 10, int16_t)
246 yuv2NBPS(12, BE, 1, 10, int16_t)
247 yuv2NBPS(12, LE, 0, 10, int16_t)
248 yuv2NBPS(14, BE, 1, 10, int16_t)
249 yuv2NBPS(14, LE, 0, 10, int16_t)
250 yuv2NBPS(16, BE, 1, 16, int32_t)
251 yuv2NBPS(16, LE, 0, 16, int32_t)
253 static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
254 const int16_t **src, uint8_t *dest, int dstW,
255 const uint8_t *dither, int offset)
258 for (i=0; i<dstW; i++) {
259 int val = dither[(i + offset) & 7] << 12;
261 for (j=0; j<filterSize; j++)
262 val += src[j][i] * filter[j];
264 dest[i]= av_clip_uint8(val>>19);
268 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
269 const uint8_t *dither, int offset)
272 for (i=0; i<dstW; i++) {
273 int val = (src[i] + dither[(i + offset) & 7]) >> 7;
274 dest[i]= av_clip_uint8(val);
278 static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
279 const int16_t **chrUSrc, const int16_t **chrVSrc,
280 uint8_t *dest, int chrDstW)
282 enum AVPixelFormat dstFormat = c->dstFormat;
283 const uint8_t *chrDither = c->chrDither8;
286 if (dstFormat == AV_PIX_FMT_NV12)
287 for (i=0; i<chrDstW; i++) {
288 int u = chrDither[i & 7] << 12;
289 int v = chrDither[(i + 3) & 7] << 12;
291 for (j=0; j<chrFilterSize; j++) {
292 u += chrUSrc[j][i] * chrFilter[j];
293 v += chrVSrc[j][i] * chrFilter[j];
296 dest[2*i]= av_clip_uint8(u>>19);
297 dest[2*i+1]= av_clip_uint8(v>>19);
300 for (i=0; i<chrDstW; i++) {
301 int u = chrDither[i & 7] << 12;
302 int v = chrDither[(i + 3) & 7] << 12;
304 for (j=0; j<chrFilterSize; j++) {
305 u += chrUSrc[j][i] * chrFilter[j];
306 v += chrVSrc[j][i] * chrFilter[j];
309 dest[2*i]= av_clip_uint8(v>>19);
310 dest[2*i+1]= av_clip_uint8(u>>19);
314 #define accumulate_bit(acc, val) \
316 acc |= (val) >= (128 + 110)
317 #define output_pixel(pos, acc) \
318 if (target == AV_PIX_FMT_MONOBLACK) { \
324 static av_always_inline void
325 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
326 const int16_t **lumSrc, int lumFilterSize,
327 const int16_t *chrFilter, const int16_t **chrUSrc,
328 const int16_t **chrVSrc, int chrFilterSize,
329 const int16_t **alpSrc, uint8_t *dest, int dstW,
330 int y, enum AVPixelFormat target)
332 const uint8_t * const d128 = ff_dither_8x8_220[y&7];
337 for (i = 0; i < dstW; i += 2) {
342 for (j = 0; j < lumFilterSize; j++) {
343 Y1 += lumSrc[j][i] * lumFilter[j];
344 Y2 += lumSrc[j][i+1] * lumFilter[j];
348 if ((Y1 | Y2) & 0x100) {
349 Y1 = av_clip_uint8(Y1);
350 Y2 = av_clip_uint8(Y2);
352 if (c->dither == SWS_DITHER_ED) {
353 Y1 += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
354 c->dither_error[0][i] = err;
355 acc = 2*acc + (Y1 >= 128);
358 err = Y2 + ((7*Y1 + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4);
359 c->dither_error[0][i+1] = Y1;
360 acc = 2*acc + (err >= 128);
363 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
364 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
367 output_pixel(*dest++, acc);
370 c->dither_error[0][i] = err;
373 output_pixel(*dest, acc);
377 static av_always_inline void
378 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
379 const int16_t *ubuf[2], const int16_t *vbuf[2],
380 const int16_t *abuf[2], uint8_t *dest, int dstW,
381 int yalpha, int uvalpha, int y,
382 enum AVPixelFormat target)
384 const int16_t *buf0 = buf[0], *buf1 = buf[1];
385 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
386 int yalpha1 = 4096 - yalpha;
389 if (c->dither == SWS_DITHER_ED) {
392 for (i = 0; i < dstW; i +=2) {
395 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
396 Y += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
397 c->dither_error[0][i] = err;
398 acc = 2*acc + (Y >= 128);
401 err = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
402 err += (7*Y + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4;
403 c->dither_error[0][i+1] = Y;
404 acc = 2*acc + (err >= 128);
408 output_pixel(*dest++, acc);
410 c->dither_error[0][i] = err;
412 for (i = 0; i < dstW; i += 8) {
415 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
416 accumulate_bit(acc, Y + d128[0]);
417 Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
418 accumulate_bit(acc, Y + d128[1]);
419 Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
420 accumulate_bit(acc, Y + d128[2]);
421 Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
422 accumulate_bit(acc, Y + d128[3]);
423 Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
424 accumulate_bit(acc, Y + d128[4]);
425 Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
426 accumulate_bit(acc, Y + d128[5]);
427 Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
428 accumulate_bit(acc, Y + d128[6]);
429 Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
430 accumulate_bit(acc, Y + d128[7]);
432 output_pixel(*dest++, acc);
437 static av_always_inline void
438 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
439 const int16_t *ubuf[2], const int16_t *vbuf[2],
440 const int16_t *abuf0, uint8_t *dest, int dstW,
441 int uvalpha, int y, enum AVPixelFormat target)
443 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
446 if (c->dither == SWS_DITHER_ED) {
449 for (i = 0; i < dstW; i +=2) {
452 Y = ((buf0[i + 0] + 64) >> 7);
453 Y += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
454 c->dither_error[0][i] = err;
455 acc = 2*acc + (Y >= 128);
458 err = ((buf0[i + 1] + 64) >> 7);
459 err += (7*Y + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4;
460 c->dither_error[0][i+1] = Y;
461 acc = 2*acc + (err >= 128);
465 output_pixel(*dest++, acc);
467 c->dither_error[0][i] = err;
469 for (i = 0; i < dstW; i += 8) {
471 accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
472 accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
473 accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
474 accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
475 accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
476 accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
477 accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
478 accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
480 output_pixel(*dest++, acc);
486 #undef accumulate_bit
488 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
489 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
490 const int16_t **lumSrc, int lumFilterSize, \
491 const int16_t *chrFilter, const int16_t **chrUSrc, \
492 const int16_t **chrVSrc, int chrFilterSize, \
493 const int16_t **alpSrc, uint8_t *dest, int dstW, \
496 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
497 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
498 alpSrc, dest, dstW, y, fmt); \
501 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
502 const int16_t *ubuf[2], const int16_t *vbuf[2], \
503 const int16_t *abuf[2], uint8_t *dest, int dstW, \
504 int yalpha, int uvalpha, int y) \
506 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
507 dest, dstW, yalpha, uvalpha, y, fmt); \
510 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
511 const int16_t *ubuf[2], const int16_t *vbuf[2], \
512 const int16_t *abuf0, uint8_t *dest, int dstW, \
513 int uvalpha, int y) \
515 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
516 abuf0, dest, dstW, uvalpha, \
520 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
521 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
523 #define output_pixels(pos, Y1, U, Y2, V) \
524 if (target == AV_PIX_FMT_YUYV422) { \
525 dest[pos + 0] = Y1; \
527 dest[pos + 2] = Y2; \
529 } else if (target == AV_PIX_FMT_YVYU422) { \
530 dest[pos + 0] = Y1; \
532 dest[pos + 2] = Y2; \
534 } else { /* AV_PIX_FMT_UYVY422 */ \
536 dest[pos + 1] = Y1; \
538 dest[pos + 3] = Y2; \
541 static av_always_inline void
542 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
543 const int16_t **lumSrc, int lumFilterSize,
544 const int16_t *chrFilter, const int16_t **chrUSrc,
545 const int16_t **chrVSrc, int chrFilterSize,
546 const int16_t **alpSrc, uint8_t *dest, int dstW,
547 int y, enum AVPixelFormat target)
551 for (i = 0; i < ((dstW + 1) >> 1); i++) {
558 for (j = 0; j < lumFilterSize; j++) {
559 Y1 += lumSrc[j][i * 2] * lumFilter[j];
560 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
562 for (j = 0; j < chrFilterSize; j++) {
563 U += chrUSrc[j][i] * chrFilter[j];
564 V += chrVSrc[j][i] * chrFilter[j];
570 if ((Y1 | Y2 | U | V) & 0x100) {
571 Y1 = av_clip_uint8(Y1);
572 Y2 = av_clip_uint8(Y2);
573 U = av_clip_uint8(U);
574 V = av_clip_uint8(V);
576 output_pixels(4*i, Y1, U, Y2, V);
580 static av_always_inline void
581 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
582 const int16_t *ubuf[2], const int16_t *vbuf[2],
583 const int16_t *abuf[2], uint8_t *dest, int dstW,
584 int yalpha, int uvalpha, int y,
585 enum AVPixelFormat target)
587 const int16_t *buf0 = buf[0], *buf1 = buf[1],
588 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
589 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
590 int yalpha1 = 4096 - yalpha;
591 int uvalpha1 = 4096 - uvalpha;
594 for (i = 0; i < ((dstW + 1) >> 1); i++) {
595 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
596 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
597 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
598 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
600 if ((Y1 | Y2 | U | V) & 0x100) {
601 Y1 = av_clip_uint8(Y1);
602 Y2 = av_clip_uint8(Y2);
603 U = av_clip_uint8(U);
604 V = av_clip_uint8(V);
607 output_pixels(i * 4, Y1, U, Y2, V);
611 static av_always_inline void
612 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
613 const int16_t *ubuf[2], const int16_t *vbuf[2],
614 const int16_t *abuf0, uint8_t *dest, int dstW,
615 int uvalpha, int y, enum AVPixelFormat target)
617 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
620 if (uvalpha < 2048) {
621 for (i = 0; i < ((dstW + 1) >> 1); i++) {
622 int Y1 = (buf0[i * 2 ]+64) >> 7;
623 int Y2 = (buf0[i * 2 + 1]+64) >> 7;
624 int U = (ubuf0[i] +64) >> 7;
625 int V = (vbuf0[i] +64) >> 7;
627 if ((Y1 | Y2 | U | V) & 0x100) {
628 Y1 = av_clip_uint8(Y1);
629 Y2 = av_clip_uint8(Y2);
630 U = av_clip_uint8(U);
631 V = av_clip_uint8(V);
634 Y1 = av_clip_uint8(Y1);
635 Y2 = av_clip_uint8(Y2);
636 U = av_clip_uint8(U);
637 V = av_clip_uint8(V);
639 output_pixels(i * 4, Y1, U, Y2, V);
642 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
643 for (i = 0; i < ((dstW + 1) >> 1); i++) {
644 int Y1 = (buf0[i * 2 ] + 64) >> 7;
645 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
646 int U = (ubuf0[i] + ubuf1[i]+128) >> 8;
647 int V = (vbuf0[i] + vbuf1[i]+128) >> 8;
649 if ((Y1 | Y2 | U | V) & 0x100) {
650 Y1 = av_clip_uint8(Y1);
651 Y2 = av_clip_uint8(Y2);
652 U = av_clip_uint8(U);
653 V = av_clip_uint8(V);
656 Y1 = av_clip_uint8(Y1);
657 Y2 = av_clip_uint8(Y2);
658 U = av_clip_uint8(U);
659 V = av_clip_uint8(V);
661 output_pixels(i * 4, Y1, U, Y2, V);
668 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422)
669 YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422)
670 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422)
672 #define R_B ((target == AV_PIX_FMT_RGB48LE || target == AV_PIX_FMT_RGB48BE || target == AV_PIX_FMT_RGBA64LE || target == AV_PIX_FMT_RGBA64BE) ? R : B)
673 #define B_R ((target == AV_PIX_FMT_RGB48LE || target == AV_PIX_FMT_RGB48BE || target == AV_PIX_FMT_RGBA64LE || target == AV_PIX_FMT_RGBA64BE) ? B : R)
674 #define output_pixel(pos, val) \
675 if (isBE(target)) { \
681 static av_always_inline void
682 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
683 const int32_t **lumSrc, int lumFilterSize,
684 const int16_t *chrFilter, const int32_t **chrUSrc,
685 const int32_t **chrVSrc, int chrFilterSize,
686 const int32_t **alpSrc, uint16_t *dest, int dstW,
687 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
690 int A1 = 0xffff<<14, A2 = 0xffff<<14;
692 for (i = 0; i < ((dstW + 1) >> 1); i++) {
694 int Y1 = -0x40000000;
695 int Y2 = -0x40000000;
696 int U = -128 << 23; // 19
700 for (j = 0; j < lumFilterSize; j++) {
701 Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
702 Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
704 for (j = 0; j < chrFilterSize; j++) {;
705 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
706 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
712 for (j = 0; j < lumFilterSize; j++) {
713 A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
714 A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
722 // 8bit: 12+15=27; 16-bit: 12+19=31
730 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
731 Y1 -= c->yuv2rgb_y_offset;
732 Y2 -= c->yuv2rgb_y_offset;
733 Y1 *= c->yuv2rgb_y_coeff;
734 Y2 *= c->yuv2rgb_y_coeff;
737 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
739 R = V * c->yuv2rgb_v2r_coeff;
740 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
741 B = U * c->yuv2rgb_u2b_coeff;
743 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
744 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
745 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
746 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
748 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
749 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
750 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
751 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
752 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
755 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
756 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
757 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
763 static av_always_inline void
764 yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
765 const int32_t *ubuf[2], const int32_t *vbuf[2],
766 const int32_t *abuf[2], uint16_t *dest, int dstW,
767 int yalpha, int uvalpha, int y,
768 enum AVPixelFormat target, int hasAlpha, int eightbytes)
770 const int32_t *buf0 = buf[0], *buf1 = buf[1],
771 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
772 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
773 *abuf0 = hasAlpha ? abuf[0] : NULL,
774 *abuf1 = hasAlpha ? abuf[1] : NULL;
775 int yalpha1 = 4096 - yalpha;
776 int uvalpha1 = 4096 - uvalpha;
778 int A1 = 0xffff<<14, A2 = 0xffff<<14;
780 for (i = 0; i < ((dstW + 1) >> 1); i++) {
781 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
782 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
783 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha + (-128 << 23)) >> 14;
784 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha + (-128 << 23)) >> 14;
787 Y1 -= c->yuv2rgb_y_offset;
788 Y2 -= c->yuv2rgb_y_offset;
789 Y1 *= c->yuv2rgb_y_coeff;
790 Y2 *= c->yuv2rgb_y_coeff;
794 R = V * c->yuv2rgb_v2r_coeff;
795 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
796 B = U * c->yuv2rgb_u2b_coeff;
799 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
800 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
806 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
807 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
808 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
810 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
811 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
812 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
813 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
814 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
817 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
818 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
819 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
825 static av_always_inline void
826 yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
827 const int32_t *ubuf[2], const int32_t *vbuf[2],
828 const int32_t *abuf0, uint16_t *dest, int dstW,
829 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
831 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
833 int A1 = 0xffff<<14, A2= 0xffff<<14;
835 if (uvalpha < 2048) {
836 for (i = 0; i < ((dstW + 1) >> 1); i++) {
837 int Y1 = (buf0[i * 2] ) >> 2;
838 int Y2 = (buf0[i * 2 + 1]) >> 2;
839 int U = (ubuf0[i] + (-128 << 11)) >> 2;
840 int V = (vbuf0[i] + (-128 << 11)) >> 2;
843 Y1 -= c->yuv2rgb_y_offset;
844 Y2 -= c->yuv2rgb_y_offset;
845 Y1 *= c->yuv2rgb_y_coeff;
846 Y2 *= c->yuv2rgb_y_coeff;
851 A1 = abuf0[i * 2 ] << 11;
852 A2 = abuf0[i * 2 + 1] << 11;
858 R = V * c->yuv2rgb_v2r_coeff;
859 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
860 B = U * c->yuv2rgb_u2b_coeff;
862 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
863 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
864 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
866 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
867 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
868 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
869 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
870 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
873 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
874 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
875 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
880 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
881 int A1 = 0xffff<<14, A2 = 0xffff<<14;
882 for (i = 0; i < ((dstW + 1) >> 1); i++) {
883 int Y1 = (buf0[i * 2] ) >> 2;
884 int Y2 = (buf0[i * 2 + 1]) >> 2;
885 int U = (ubuf0[i] + ubuf1[i] + (-128 << 12)) >> 3;
886 int V = (vbuf0[i] + vbuf1[i] + (-128 << 12)) >> 3;
889 Y1 -= c->yuv2rgb_y_offset;
890 Y2 -= c->yuv2rgb_y_offset;
891 Y1 *= c->yuv2rgb_y_coeff;
892 Y2 *= c->yuv2rgb_y_coeff;
897 A1 = abuf0[i * 2 ] << 11;
898 A2 = abuf0[i * 2 + 1] << 11;
904 R = V * c->yuv2rgb_v2r_coeff;
905 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
906 B = U * c->yuv2rgb_u2b_coeff;
908 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
909 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
910 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
912 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
913 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
914 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
915 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
916 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
919 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
920 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
921 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
928 static av_always_inline void
929 yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
930 const int32_t **lumSrc, int lumFilterSize,
931 const int16_t *chrFilter, const int32_t **chrUSrc,
932 const int32_t **chrVSrc, int chrFilterSize,
933 const int32_t **alpSrc, uint16_t *dest, int dstW,
934 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
939 for (i = 0; i < dstW; i++) {
942 int U = -128 << 23; // 19
946 for (j = 0; j < lumFilterSize; j++) {
947 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
949 for (j = 0; j < chrFilterSize; j++) {;
950 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
951 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
956 for (j = 0; j < lumFilterSize; j++) {
957 A += alpSrc[j][i] * (unsigned)lumFilter[j];
963 // 8bit: 12+15=27; 16-bit: 12+19=31
969 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
970 Y -= c->yuv2rgb_y_offset;
971 Y *= c->yuv2rgb_y_coeff;
973 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
975 R = V * c->yuv2rgb_v2r_coeff;
976 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
977 B = U * c->yuv2rgb_u2b_coeff;
979 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
980 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
981 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
982 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
984 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
992 static av_always_inline void
993 yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
994 const int32_t *ubuf[2], const int32_t *vbuf[2],
995 const int32_t *abuf[2], uint16_t *dest, int dstW,
996 int yalpha, int uvalpha, int y,
997 enum AVPixelFormat target, int hasAlpha, int eightbytes)
999 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1000 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1001 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1002 *abuf0 = hasAlpha ? abuf[0] : NULL,
1003 *abuf1 = hasAlpha ? abuf[1] : NULL;
1004 int yalpha1 = 4096 - yalpha;
1005 int uvalpha1 = 4096 - uvalpha;
1009 for (i = 0; i < dstW; i++) {
1010 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 14;
1011 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha + (-128 << 23)) >> 14;
1012 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha + (-128 << 23)) >> 14;
1015 Y -= c->yuv2rgb_y_offset;
1016 Y *= c->yuv2rgb_y_coeff;
1019 R = V * c->yuv2rgb_v2r_coeff;
1020 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1021 B = U * c->yuv2rgb_u2b_coeff;
1024 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1029 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1030 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1031 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1033 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1041 static av_always_inline void
1042 yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1043 const int32_t *ubuf[2], const int32_t *vbuf[2],
1044 const int32_t *abuf0, uint16_t *dest, int dstW,
1045 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1047 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1051 if (uvalpha < 2048) {
1052 for (i = 0; i < dstW; i++) {
1053 int Y = (buf0[i]) >> 2;
1054 int U = (ubuf0[i] + (-128 << 11)) >> 2;
1055 int V = (vbuf0[i] + (-128 << 11)) >> 2;
1058 Y -= c->yuv2rgb_y_offset;
1059 Y *= c->yuv2rgb_y_coeff;
1068 R = V * c->yuv2rgb_v2r_coeff;
1069 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1070 B = U * c->yuv2rgb_u2b_coeff;
1072 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1073 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1074 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1076 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1083 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1085 for (i = 0; i < dstW; i++) {
1086 int Y = (buf0[i] ) >> 2;
1087 int U = (ubuf0[i] + ubuf1[i] + (-128 << 12)) >> 3;
1088 int V = (vbuf0[i] + vbuf1[i] + (-128 << 12)) >> 3;
1091 Y -= c->yuv2rgb_y_offset;
1092 Y *= c->yuv2rgb_y_coeff;
1101 R = V * c->yuv2rgb_v2r_coeff;
1102 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1103 B = U * c->yuv2rgb_u2b_coeff;
1105 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1106 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1107 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1109 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1122 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1123 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1124 const int16_t **_lumSrc, int lumFilterSize, \
1125 const int16_t *chrFilter, const int16_t **_chrUSrc, \
1126 const int16_t **_chrVSrc, int chrFilterSize, \
1127 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1130 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1131 **chrUSrc = (const int32_t **) _chrUSrc, \
1132 **chrVSrc = (const int32_t **) _chrVSrc, \
1133 **alpSrc = (const int32_t **) _alpSrc; \
1134 uint16_t *dest = (uint16_t *) _dest; \
1135 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1136 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1137 alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1140 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1141 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1142 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1143 int yalpha, int uvalpha, int y) \
1145 const int32_t **buf = (const int32_t **) _buf, \
1146 **ubuf = (const int32_t **) _ubuf, \
1147 **vbuf = (const int32_t **) _vbuf, \
1148 **abuf = (const int32_t **) _abuf; \
1149 uint16_t *dest = (uint16_t *) _dest; \
1150 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1151 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1154 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1155 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1156 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1157 int uvalpha, int y) \
1159 const int32_t *buf0 = (const int32_t *) _buf0, \
1160 **ubuf = (const int32_t **) _ubuf, \
1161 **vbuf = (const int32_t **) _vbuf, \
1162 *abuf0 = (const int32_t *) _abuf0; \
1163 uint16_t *dest = (uint16_t *) _dest; \
1164 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1165 dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1168 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1169 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1170 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1171 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1172 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1173 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1174 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1175 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1176 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1177 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1178 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1179 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1181 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1182 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1183 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1184 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1185 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1186 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1187 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1188 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1189 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1190 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1191 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1192 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1195 * Write out 2 RGB pixels in the target pixel format. This function takes a
1196 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1197 * things like endianness conversion and shifting. The caller takes care of
1198 * setting the correct offset in these tables from the chroma (U/V) values.
1199 * This function then uses the luminance (Y1/Y2) values to write out the
1200 * correct RGB values into the destination buffer.
1202 static av_always_inline void
1203 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1204 unsigned A1, unsigned A2,
1205 const void *_r, const void *_g, const void *_b, int y,
1206 enum AVPixelFormat target, int hasAlpha)
1208 if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1209 target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1210 uint32_t *dest = (uint32_t *) _dest;
1211 const uint32_t *r = (const uint32_t *) _r;
1212 const uint32_t *g = (const uint32_t *) _g;
1213 const uint32_t *b = (const uint32_t *) _b;
1216 int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1218 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1219 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1222 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1224 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1225 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1226 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1228 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1229 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1231 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1233 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1234 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1237 } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1238 uint8_t *dest = (uint8_t *) _dest;
1239 const uint8_t *r = (const uint8_t *) _r;
1240 const uint8_t *g = (const uint8_t *) _g;
1241 const uint8_t *b = (const uint8_t *) _b;
1243 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1244 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1246 dest[i * 6 + 0] = r_b[Y1];
1247 dest[i * 6 + 1] = g[Y1];
1248 dest[i * 6 + 2] = b_r[Y1];
1249 dest[i * 6 + 3] = r_b[Y2];
1250 dest[i * 6 + 4] = g[Y2];
1251 dest[i * 6 + 5] = b_r[Y2];
1254 } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1255 target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1256 target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1257 uint16_t *dest = (uint16_t *) _dest;
1258 const uint16_t *r = (const uint16_t *) _r;
1259 const uint16_t *g = (const uint16_t *) _g;
1260 const uint16_t *b = (const uint16_t *) _b;
1261 int dr1, dg1, db1, dr2, dg2, db2;
1263 if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1264 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1265 dg1 = ff_dither_2x2_4[ y & 1 ][0];
1266 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1267 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1268 dg2 = ff_dither_2x2_4[ y & 1 ][1];
1269 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1270 } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1271 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1272 dg1 = ff_dither_2x2_8[ y & 1 ][1];
1273 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1274 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1275 dg2 = ff_dither_2x2_8[ y & 1 ][0];
1276 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1278 dr1 = ff_dither_4x4_16[ y & 3 ][0];
1279 dg1 = ff_dither_4x4_16[ y & 3 ][1];
1280 db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1281 dr2 = ff_dither_4x4_16[ y & 3 ][1];
1282 dg2 = ff_dither_4x4_16[ y & 3 ][0];
1283 db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1286 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1287 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1288 } else /* 8/4-bit */ {
1289 uint8_t *dest = (uint8_t *) _dest;
1290 const uint8_t *r = (const uint8_t *) _r;
1291 const uint8_t *g = (const uint8_t *) _g;
1292 const uint8_t *b = (const uint8_t *) _b;
1293 int dr1, dg1, db1, dr2, dg2, db2;
1295 if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1296 const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1297 const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1298 dr1 = dg1 = d32[(i * 2 + 0) & 7];
1299 db1 = d64[(i * 2 + 0) & 7];
1300 dr2 = dg2 = d32[(i * 2 + 1) & 7];
1301 db2 = d64[(i * 2 + 1) & 7];
1303 const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1304 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1305 dr1 = db1 = d128[(i * 2 + 0) & 7];
1306 dg1 = d64[(i * 2 + 0) & 7];
1307 dr2 = db2 = d128[(i * 2 + 1) & 7];
1308 dg2 = d64[(i * 2 + 1) & 7];
1311 if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1312 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1313 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1315 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1316 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1321 static av_always_inline void
1322 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1323 const int16_t **lumSrc, int lumFilterSize,
1324 const int16_t *chrFilter, const int16_t **chrUSrc,
1325 const int16_t **chrVSrc, int chrFilterSize,
1326 const int16_t **alpSrc, uint8_t *dest, int dstW,
1327 int y, enum AVPixelFormat target, int hasAlpha)
1331 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1337 const void *r, *g, *b;
1339 for (j = 0; j < lumFilterSize; j++) {
1340 Y1 += lumSrc[j][i * 2] * lumFilter[j];
1341 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1343 for (j = 0; j < chrFilterSize; j++) {
1344 U += chrUSrc[j][i] * chrFilter[j];
1345 V += chrVSrc[j][i] * chrFilter[j];
1354 for (j = 0; j < lumFilterSize; j++) {
1355 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1356 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1360 if ((A1 | A2) & 0x100) {
1361 A1 = av_clip_uint8(A1);
1362 A2 = av_clip_uint8(A2);
1366 r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1367 g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1368 b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1370 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1371 r, g, b, y, target, hasAlpha);
1375 static av_always_inline void
1376 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1377 const int16_t *ubuf[2], const int16_t *vbuf[2],
1378 const int16_t *abuf[2], uint8_t *dest, int dstW,
1379 int yalpha, int uvalpha, int y,
1380 enum AVPixelFormat target, int hasAlpha)
1382 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1383 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1384 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1385 *abuf0 = hasAlpha ? abuf[0] : NULL,
1386 *abuf1 = hasAlpha ? abuf[1] : NULL;
1387 int yalpha1 = 4096 - yalpha;
1388 int uvalpha1 = 4096 - uvalpha;
1391 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1392 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1393 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1394 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1395 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1397 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1398 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1399 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1402 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1403 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1404 A1 = av_clip_uint8(A1);
1405 A2 = av_clip_uint8(A2);
1408 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1409 r, g, b, y, target, hasAlpha);
1413 static av_always_inline void
1414 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1415 const int16_t *ubuf[2], const int16_t *vbuf[2],
1416 const int16_t *abuf0, uint8_t *dest, int dstW,
1417 int uvalpha, int y, enum AVPixelFormat target,
1420 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1423 if (uvalpha < 2048) {
1424 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1425 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1426 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1427 int U = (ubuf0[i] + 64) >> 7;
1428 int V = (vbuf0[i] + 64) >> 7;
1430 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1431 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1432 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1435 A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1436 A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1437 A1 = av_clip_uint8(A1);
1438 A2 = av_clip_uint8(A2);
1441 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1442 r, g, b, y, target, hasAlpha);
1445 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1446 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1447 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1448 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1449 int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1450 int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1452 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1453 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1454 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1457 A1 = (abuf0[i * 2 ] + 64) >> 7;
1458 A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1459 A1 = av_clip_uint8(A1);
1460 A2 = av_clip_uint8(A2);
1463 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1464 r, g, b, y, target, hasAlpha);
1469 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1470 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1471 const int16_t **lumSrc, int lumFilterSize, \
1472 const int16_t *chrFilter, const int16_t **chrUSrc, \
1473 const int16_t **chrVSrc, int chrFilterSize, \
1474 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1477 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1478 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1479 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1482 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1483 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1484 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1485 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1486 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1487 int yalpha, int uvalpha, int y) \
1489 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1490 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1493 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1494 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1495 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1496 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1497 const int16_t *abuf0, uint8_t *dest, int dstW, \
1498 int uvalpha, int y) \
1500 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1501 dstW, uvalpha, y, fmt, hasAlpha); \
1505 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1506 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1508 #if CONFIG_SWSCALE_ALPHA
1509 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1510 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1512 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1513 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1515 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1516 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1517 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1518 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1519 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1520 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1521 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1522 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1524 static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1525 uint8_t *dest, int i, int Y, int A, int U, int V,
1526 int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1529 int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1531 Y -= c->yuv2rgb_y_offset;
1532 Y *= c->yuv2rgb_y_coeff;
1534 R = Y + V*c->yuv2rgb_v2r_coeff;
1535 G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1536 B = Y + U*c->yuv2rgb_u2b_coeff;
1537 if ((R | G | B) & 0xC0000000) {
1538 R = av_clip_uintp2(R, 30);
1539 G = av_clip_uintp2(G, 30);
1540 B = av_clip_uintp2(B, 30);
1544 case AV_PIX_FMT_ARGB:
1545 dest[0] = hasAlpha ? A : 255;
1550 case AV_PIX_FMT_RGB24:
1555 case AV_PIX_FMT_RGBA:
1559 dest[3] = hasAlpha ? A : 255;
1561 case AV_PIX_FMT_ABGR:
1562 dest[0] = hasAlpha ? A : 255;
1567 case AV_PIX_FMT_BGR24:
1572 case AV_PIX_FMT_BGRA:
1576 dest[3] = hasAlpha ? A : 255;
1578 case AV_PIX_FMT_BGR4_BYTE:
1579 case AV_PIX_FMT_RGB4_BYTE:
1580 case AV_PIX_FMT_BGR8:
1581 case AV_PIX_FMT_RGB8:
1585 switch (c->dither) {
1587 case SWS_DITHER_AUTO:
1592 R += (7*err[0] + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2])>>4;
1593 G += (7*err[1] + 1*c->dither_error[1][i] + 5*c->dither_error[1][i+1] + 3*c->dither_error[1][i+2])>>4;
1594 B += (7*err[2] + 1*c->dither_error[2][i] + 5*c->dither_error[2][i+1] + 3*c->dither_error[2][i+2])>>4;
1595 c->dither_error[0][i] = err[0];
1596 c->dither_error[1][i] = err[1];
1597 c->dither_error[2][i] = err[2];
1598 r = R >> (isrgb8 ? 5 : 7);
1599 g = G >> (isrgb8 ? 5 : 6);
1600 b = B >> (isrgb8 ? 6 : 7);
1601 r = av_clip(r, 0, isrgb8 ? 7 : 1);
1602 g = av_clip(g, 0, isrgb8 ? 7 : 3);
1603 b = av_clip(b, 0, isrgb8 ? 3 : 1);
1604 err[0] = R - r*(isrgb8 ? 36 : 255);
1605 err[1] = G - g*(isrgb8 ? 36 : 85);
1606 err[2] = B - b*(isrgb8 ? 85 : 255);
1608 case SWS_DITHER_A_DITHER:
1610 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1611 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1612 r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1613 g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1614 b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1615 r = av_clip_uintp2(r, 3);
1616 g = av_clip_uintp2(g, 3);
1617 b = av_clip_uintp2(b, 2);
1619 r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1620 g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1621 b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1622 r = av_clip_uintp2(r, 1);
1623 g = av_clip_uintp2(g, 2);
1624 b = av_clip_uintp2(b, 1);
1627 case SWS_DITHER_X_DITHER:
1629 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1630 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1631 r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1632 g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1633 b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1634 r = av_clip_uintp2(r, 3);
1635 g = av_clip_uintp2(g, 3);
1636 b = av_clip_uintp2(b, 2);
1638 r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1639 g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1640 b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1641 r = av_clip_uintp2(r, 1);
1642 g = av_clip_uintp2(g, 2);
1643 b = av_clip_uintp2(b, 1);
1649 if(target == AV_PIX_FMT_BGR4_BYTE) {
1650 dest[0] = r + 2*g + 8*b;
1651 } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1652 dest[0] = b + 2*g + 8*r;
1653 } else if(target == AV_PIX_FMT_BGR8) {
1654 dest[0] = r + 8*g + 64*b;
1655 } else if(target == AV_PIX_FMT_RGB8) {
1656 dest[0] = b + 4*g + 32*r;
1663 static av_always_inline void
1664 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1665 const int16_t **lumSrc, int lumFilterSize,
1666 const int16_t *chrFilter, const int16_t **chrUSrc,
1667 const int16_t **chrVSrc, int chrFilterSize,
1668 const int16_t **alpSrc, uint8_t *dest,
1669 int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1672 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1674 int A = 0; //init to silence warning
1676 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1677 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1680 for (i = 0; i < dstW; i++) {
1683 int U = (1<<9)-(128 << 19);
1684 int V = (1<<9)-(128 << 19);
1686 for (j = 0; j < lumFilterSize; j++) {
1687 Y += lumSrc[j][i] * lumFilter[j];
1689 for (j = 0; j < chrFilterSize; j++) {
1690 U += chrUSrc[j][i] * chrFilter[j];
1691 V += chrVSrc[j][i] * chrFilter[j];
1698 for (j = 0; j < lumFilterSize; j++) {
1699 A += alpSrc[j][i] * lumFilter[j];
1703 A = av_clip_uint8(A);
1705 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1708 c->dither_error[0][i] = err[0];
1709 c->dither_error[1][i] = err[1];
1710 c->dither_error[2][i] = err[2];
1713 static av_always_inline void
1714 yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
1715 const int16_t *ubuf[2], const int16_t *vbuf[2],
1716 const int16_t *abuf[2], uint8_t *dest, int dstW,
1717 int yalpha, int uvalpha, int y,
1718 enum AVPixelFormat target, int hasAlpha)
1720 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1721 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1722 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1723 *abuf0 = hasAlpha ? abuf[0] : NULL,
1724 *abuf1 = hasAlpha ? abuf[1] : NULL;
1725 int yalpha1 = 4096 - yalpha;
1726 int uvalpha1 = 4096 - uvalpha;
1728 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1730 int A = 0; // init to silcene warning
1732 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1733 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1736 for (i = 0; i < dstW; i++) {
1737 int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
1738 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
1739 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
1742 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
1744 A = av_clip_uint8(A);
1747 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1750 c->dither_error[0][i] = err[0];
1751 c->dither_error[1][i] = err[1];
1752 c->dither_error[2][i] = err[2];
1755 static av_always_inline void
1756 yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
1757 const int16_t *ubuf[2], const int16_t *vbuf[2],
1758 const int16_t *abuf0, uint8_t *dest, int dstW,
1759 int uvalpha, int y, enum AVPixelFormat target,
1762 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1764 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1767 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1768 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1771 if (uvalpha < 2048) {
1772 int A = 0; //init to silence warning
1773 for (i = 0; i < dstW; i++) {
1774 int Y = buf0[i] << 2;
1775 int U = (ubuf0[i] - (128<<7)) * 4;
1776 int V = (vbuf0[i] - (128<<7)) * 4;
1779 A = (abuf0[i] + 64) >> 7;
1781 A = av_clip_uint8(A);
1784 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1788 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1789 int A = 0; //init to silence warning
1790 for (i = 0; i < dstW; i++) {
1791 int Y = buf0[i] << 2;
1792 int U = (ubuf0[i] + ubuf1[i] - (128<<8)) << 1;
1793 int V = (vbuf0[i] + vbuf1[i] - (128<<8)) << 1;
1796 A = (abuf0[i] + 64) >> 7;
1798 A = av_clip_uint8(A);
1801 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1806 c->dither_error[0][i] = err[0];
1807 c->dither_error[1][i] = err[1];
1808 c->dither_error[2][i] = err[2];
1812 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1813 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1814 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1815 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1817 #if CONFIG_SWSCALE_ALPHA
1818 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
1819 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
1820 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
1821 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
1823 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
1824 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
1825 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
1826 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
1828 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
1829 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
1831 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
1832 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
1833 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
1834 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
1837 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
1838 const int16_t **lumSrc, int lumFilterSize,
1839 const int16_t *chrFilter, const int16_t **chrUSrc,
1840 const int16_t **chrVSrc, int chrFilterSize,
1841 const int16_t **alpSrc, uint8_t **dest,
1844 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
1846 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
1847 uint16_t **dest16 = (uint16_t**)dest;
1848 int SH = 22 + 7 - desc->comp[0].depth_minus1;
1849 int A = 0; // init to silence warning
1851 for (i = 0; i < dstW; i++) {
1854 int U = (1 << 9) - (128 << 19);
1855 int V = (1 << 9) - (128 << 19);
1858 for (j = 0; j < lumFilterSize; j++)
1859 Y += lumSrc[j][i] * lumFilter[j];
1861 for (j = 0; j < chrFilterSize; j++) {
1862 U += chrUSrc[j][i] * chrFilter[j];
1863 V += chrVSrc[j][i] * chrFilter[j];
1873 for (j = 0; j < lumFilterSize; j++)
1874 A += alpSrc[j][i] * lumFilter[j];
1879 A = av_clip_uint8(A);
1882 Y -= c->yuv2rgb_y_offset;
1883 Y *= c->yuv2rgb_y_coeff;
1885 R = Y + V * c->yuv2rgb_v2r_coeff;
1886 G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1887 B = Y + U * c->yuv2rgb_u2b_coeff;
1889 if ((R | G | B) & 0xC0000000) {
1890 R = av_clip_uintp2(R, 30);
1891 G = av_clip_uintp2(G, 30);
1892 B = av_clip_uintp2(B, 30);
1896 dest16[0][i] = G >> SH;
1897 dest16[1][i] = B >> SH;
1898 dest16[2][i] = R >> SH;
1902 dest[0][i] = G >> 22;
1903 dest[1][i] = B >> 22;
1904 dest[2][i] = R >> 22;
1909 if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
1910 for (i = 0; i < dstW; i++) {
1911 dest16[0][i] = av_bswap16(dest16[0][i]);
1912 dest16[1][i] = av_bswap16(dest16[1][i]);
1913 dest16[2][i] = av_bswap16(dest16[2][i]);
1915 dest16[3][i] = av_bswap16(dest16[3][i]);
1921 yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
1922 const int16_t *ubuf[2], const int16_t *vbuf[2],
1923 const int16_t *abuf0, uint8_t *dest, int dstW,
1926 int hasAlpha = !!abuf0;
1929 for (i = 0; i < dstW; i++) {
1930 int Y = (buf0[i] + 64) >> 7;
1933 Y = av_clip_uint8(Y);
1936 A = (abuf0[i] + 64) >> 7;
1938 A = av_clip_uint8(A);
1942 dest[i * 2 + 1] = hasAlpha ? A : 255;
1947 yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
1948 const int16_t *ubuf[2], const int16_t *vbuf[2],
1949 const int16_t *abuf[2], uint8_t *dest, int dstW,
1950 int yalpha, int uvalpha, int y)
1952 int hasAlpha = abuf[0] && abuf[1];
1953 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1954 *abuf0 = hasAlpha ? abuf[0] : NULL,
1955 *abuf1 = hasAlpha ? abuf[1] : NULL;
1956 int yalpha1 = 4096 - yalpha;
1959 for (i = 0; i < dstW; i++) {
1960 int Y = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1963 Y = av_clip_uint8(Y);
1966 A = (abuf0[i * 2] * yalpha1 + abuf1[i * 2] * yalpha) >> 19;
1967 A = av_clip_uint8(A);
1971 dest[i * 2 + 1] = hasAlpha ? A : 255;
1976 yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
1977 const int16_t **lumSrc, int lumFilterSize,
1978 const int16_t *chrFilter, const int16_t **chrUSrc,
1979 const int16_t **chrVSrc, int chrFilterSize,
1980 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1982 int hasAlpha = !!alpSrc;
1985 for (i = 0; i < dstW; i++) {
1987 int Y = 1 << 18, A = 1 << 18;
1989 for (j = 0; j < lumFilterSize; j++)
1990 Y += lumSrc[j][i] * lumFilter[j];
1994 Y = av_clip_uint8(Y);
1997 for (j = 0; j < lumFilterSize; j++)
1998 A += alpSrc[j][i] * lumFilter[j];
2003 A = av_clip_uint8(A);
2007 dest[2 * i + 1] = hasAlpha ? A : 255;
2011 av_cold void ff_sws_init_output_funcs(SwsContext *c,
2012 yuv2planar1_fn *yuv2plane1,
2013 yuv2planarX_fn *yuv2planeX,
2014 yuv2interleavedX_fn *yuv2nv12cX,
2015 yuv2packed1_fn *yuv2packed1,
2016 yuv2packed2_fn *yuv2packed2,
2017 yuv2packedX_fn *yuv2packedX,
2018 yuv2anyX_fn *yuv2anyX)
2020 enum AVPixelFormat dstFormat = c->dstFormat;
2021 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2023 if (is16BPS(dstFormat)) {
2024 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
2025 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
2026 } else if (is9_OR_10BPS(dstFormat)) {
2027 if (desc->comp[0].depth_minus1 == 8) {
2028 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
2029 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
2030 } else if (desc->comp[0].depth_minus1 == 9) {
2031 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
2032 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
2033 } else if (desc->comp[0].depth_minus1 == 11) {
2034 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
2035 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
2036 } else if (desc->comp[0].depth_minus1 == 13) {
2037 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
2038 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
2042 *yuv2plane1 = yuv2plane1_8_c;
2043 *yuv2planeX = yuv2planeX_8_c;
2044 if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21)
2045 *yuv2nv12cX = yuv2nv12cX_c;
2048 if(c->flags & SWS_FULL_CHR_H_INT) {
2049 switch (dstFormat) {
2050 case AV_PIX_FMT_RGBA:
2052 *yuv2packedX = yuv2rgba32_full_X_c;
2053 *yuv2packed2 = yuv2rgba32_full_2_c;
2054 *yuv2packed1 = yuv2rgba32_full_1_c;
2056 #if CONFIG_SWSCALE_ALPHA
2058 *yuv2packedX = yuv2rgba32_full_X_c;
2059 *yuv2packed2 = yuv2rgba32_full_2_c;
2060 *yuv2packed1 = yuv2rgba32_full_1_c;
2062 #endif /* CONFIG_SWSCALE_ALPHA */
2064 *yuv2packedX = yuv2rgbx32_full_X_c;
2065 *yuv2packed2 = yuv2rgbx32_full_2_c;
2066 *yuv2packed1 = yuv2rgbx32_full_1_c;
2068 #endif /* !CONFIG_SMALL */
2070 case AV_PIX_FMT_ARGB:
2072 *yuv2packedX = yuv2argb32_full_X_c;
2073 *yuv2packed2 = yuv2argb32_full_2_c;
2074 *yuv2packed1 = yuv2argb32_full_1_c;
2076 #if CONFIG_SWSCALE_ALPHA
2078 *yuv2packedX = yuv2argb32_full_X_c;
2079 *yuv2packed2 = yuv2argb32_full_2_c;
2080 *yuv2packed1 = yuv2argb32_full_1_c;
2082 #endif /* CONFIG_SWSCALE_ALPHA */
2084 *yuv2packedX = yuv2xrgb32_full_X_c;
2085 *yuv2packed2 = yuv2xrgb32_full_2_c;
2086 *yuv2packed1 = yuv2xrgb32_full_1_c;
2088 #endif /* !CONFIG_SMALL */
2090 case AV_PIX_FMT_BGRA:
2092 *yuv2packedX = yuv2bgra32_full_X_c;
2093 *yuv2packed2 = yuv2bgra32_full_2_c;
2094 *yuv2packed1 = yuv2bgra32_full_1_c;
2096 #if CONFIG_SWSCALE_ALPHA
2098 *yuv2packedX = yuv2bgra32_full_X_c;
2099 *yuv2packed2 = yuv2bgra32_full_2_c;
2100 *yuv2packed1 = yuv2bgra32_full_1_c;
2102 #endif /* CONFIG_SWSCALE_ALPHA */
2104 *yuv2packedX = yuv2bgrx32_full_X_c;
2105 *yuv2packed2 = yuv2bgrx32_full_2_c;
2106 *yuv2packed1 = yuv2bgrx32_full_1_c;
2108 #endif /* !CONFIG_SMALL */
2110 case AV_PIX_FMT_ABGR:
2112 *yuv2packedX = yuv2abgr32_full_X_c;
2113 *yuv2packed2 = yuv2abgr32_full_2_c;
2114 *yuv2packed1 = yuv2abgr32_full_1_c;
2116 #if CONFIG_SWSCALE_ALPHA
2118 *yuv2packedX = yuv2abgr32_full_X_c;
2119 *yuv2packed2 = yuv2abgr32_full_2_c;
2120 *yuv2packed1 = yuv2abgr32_full_1_c;
2122 #endif /* CONFIG_SWSCALE_ALPHA */
2124 *yuv2packedX = yuv2xbgr32_full_X_c;
2125 *yuv2packed2 = yuv2xbgr32_full_2_c;
2126 *yuv2packed1 = yuv2xbgr32_full_1_c;
2128 #endif /* !CONFIG_SMALL */
2130 case AV_PIX_FMT_RGBA64LE:
2131 #if CONFIG_SWSCALE_ALPHA
2133 *yuv2packedX = yuv2rgba64le_full_X_c;
2134 *yuv2packed2 = yuv2rgba64le_full_2_c;
2135 *yuv2packed1 = yuv2rgba64le_full_1_c;
2137 #endif /* CONFIG_SWSCALE_ALPHA */
2139 *yuv2packedX = yuv2rgbx64le_full_X_c;
2140 *yuv2packed2 = yuv2rgbx64le_full_2_c;
2141 *yuv2packed1 = yuv2rgbx64le_full_1_c;
2144 case AV_PIX_FMT_RGBA64BE:
2145 #if CONFIG_SWSCALE_ALPHA
2147 *yuv2packedX = yuv2rgba64be_full_X_c;
2148 *yuv2packed2 = yuv2rgba64be_full_2_c;
2149 *yuv2packed1 = yuv2rgba64be_full_1_c;
2151 #endif /* CONFIG_SWSCALE_ALPHA */
2153 *yuv2packedX = yuv2rgbx64be_full_X_c;
2154 *yuv2packed2 = yuv2rgbx64be_full_2_c;
2155 *yuv2packed1 = yuv2rgbx64be_full_1_c;
2158 case AV_PIX_FMT_BGRA64LE:
2159 #if CONFIG_SWSCALE_ALPHA
2161 *yuv2packedX = yuv2bgra64le_full_X_c;
2162 *yuv2packed2 = yuv2bgra64le_full_2_c;
2163 *yuv2packed1 = yuv2bgra64le_full_1_c;
2165 #endif /* CONFIG_SWSCALE_ALPHA */
2167 *yuv2packedX = yuv2bgrx64le_full_X_c;
2168 *yuv2packed2 = yuv2bgrx64le_full_2_c;
2169 *yuv2packed1 = yuv2bgrx64le_full_1_c;
2172 case AV_PIX_FMT_BGRA64BE:
2173 #if CONFIG_SWSCALE_ALPHA
2175 *yuv2packedX = yuv2bgra64be_full_X_c;
2176 *yuv2packed2 = yuv2bgra64be_full_2_c;
2177 *yuv2packed1 = yuv2bgra64be_full_1_c;
2179 #endif /* CONFIG_SWSCALE_ALPHA */
2181 *yuv2packedX = yuv2bgrx64be_full_X_c;
2182 *yuv2packed2 = yuv2bgrx64be_full_2_c;
2183 *yuv2packed1 = yuv2bgrx64be_full_1_c;
2187 case AV_PIX_FMT_RGB24:
2188 *yuv2packedX = yuv2rgb24_full_X_c;
2189 *yuv2packed2 = yuv2rgb24_full_2_c;
2190 *yuv2packed1 = yuv2rgb24_full_1_c;
2192 case AV_PIX_FMT_BGR24:
2193 *yuv2packedX = yuv2bgr24_full_X_c;
2194 *yuv2packed2 = yuv2bgr24_full_2_c;
2195 *yuv2packed1 = yuv2bgr24_full_1_c;
2197 case AV_PIX_FMT_RGB48LE:
2198 *yuv2packedX = yuv2rgb48le_full_X_c;
2199 *yuv2packed2 = yuv2rgb48le_full_2_c;
2200 *yuv2packed1 = yuv2rgb48le_full_1_c;
2202 case AV_PIX_FMT_BGR48LE:
2203 *yuv2packedX = yuv2bgr48le_full_X_c;
2204 *yuv2packed2 = yuv2bgr48le_full_2_c;
2205 *yuv2packed1 = yuv2bgr48le_full_1_c;
2207 case AV_PIX_FMT_RGB48BE:
2208 *yuv2packedX = yuv2rgb48be_full_X_c;
2209 *yuv2packed2 = yuv2rgb48be_full_2_c;
2210 *yuv2packed1 = yuv2rgb48be_full_1_c;
2212 case AV_PIX_FMT_BGR48BE:
2213 *yuv2packedX = yuv2bgr48be_full_X_c;
2214 *yuv2packed2 = yuv2bgr48be_full_2_c;
2215 *yuv2packed1 = yuv2bgr48be_full_1_c;
2217 case AV_PIX_FMT_BGR4_BYTE:
2218 *yuv2packedX = yuv2bgr4_byte_full_X_c;
2219 *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2220 *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2222 case AV_PIX_FMT_RGB4_BYTE:
2223 *yuv2packedX = yuv2rgb4_byte_full_X_c;
2224 *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2225 *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2227 case AV_PIX_FMT_BGR8:
2228 *yuv2packedX = yuv2bgr8_full_X_c;
2229 *yuv2packed2 = yuv2bgr8_full_2_c;
2230 *yuv2packed1 = yuv2bgr8_full_1_c;
2232 case AV_PIX_FMT_RGB8:
2233 *yuv2packedX = yuv2rgb8_full_X_c;
2234 *yuv2packed2 = yuv2rgb8_full_2_c;
2235 *yuv2packed1 = yuv2rgb8_full_1_c;
2237 case AV_PIX_FMT_GBRP:
2238 case AV_PIX_FMT_GBRP9BE:
2239 case AV_PIX_FMT_GBRP9LE:
2240 case AV_PIX_FMT_GBRP10BE:
2241 case AV_PIX_FMT_GBRP10LE:
2242 case AV_PIX_FMT_GBRP12BE:
2243 case AV_PIX_FMT_GBRP12LE:
2244 case AV_PIX_FMT_GBRP14BE:
2245 case AV_PIX_FMT_GBRP14LE:
2246 case AV_PIX_FMT_GBRP16BE:
2247 case AV_PIX_FMT_GBRP16LE:
2248 case AV_PIX_FMT_GBRAP:
2249 *yuv2anyX = yuv2gbrp_full_X_c;
2252 if (!*yuv2packedX && !*yuv2anyX)
2256 switch (dstFormat) {
2257 case AV_PIX_FMT_RGBA64LE:
2258 #if CONFIG_SWSCALE_ALPHA
2260 *yuv2packed1 = yuv2rgba64le_1_c;
2261 *yuv2packed2 = yuv2rgba64le_2_c;
2262 *yuv2packedX = yuv2rgba64le_X_c;
2264 #endif /* CONFIG_SWSCALE_ALPHA */
2266 *yuv2packed1 = yuv2rgbx64le_1_c;
2267 *yuv2packed2 = yuv2rgbx64le_2_c;
2268 *yuv2packedX = yuv2rgbx64le_X_c;
2271 case AV_PIX_FMT_RGBA64BE:
2272 #if CONFIG_SWSCALE_ALPHA
2274 *yuv2packed1 = yuv2rgba64be_1_c;
2275 *yuv2packed2 = yuv2rgba64be_2_c;
2276 *yuv2packedX = yuv2rgba64be_X_c;
2278 #endif /* CONFIG_SWSCALE_ALPHA */
2280 *yuv2packed1 = yuv2rgbx64be_1_c;
2281 *yuv2packed2 = yuv2rgbx64be_2_c;
2282 *yuv2packedX = yuv2rgbx64be_X_c;
2285 case AV_PIX_FMT_BGRA64LE:
2286 #if CONFIG_SWSCALE_ALPHA
2288 *yuv2packed1 = yuv2bgra64le_1_c;
2289 *yuv2packed2 = yuv2bgra64le_2_c;
2290 *yuv2packedX = yuv2bgra64le_X_c;
2292 #endif /* CONFIG_SWSCALE_ALPHA */
2294 *yuv2packed1 = yuv2bgrx64le_1_c;
2295 *yuv2packed2 = yuv2bgrx64le_2_c;
2296 *yuv2packedX = yuv2bgrx64le_X_c;
2299 case AV_PIX_FMT_BGRA64BE:
2300 #if CONFIG_SWSCALE_ALPHA
2302 *yuv2packed1 = yuv2bgra64be_1_c;
2303 *yuv2packed2 = yuv2bgra64be_2_c;
2304 *yuv2packedX = yuv2bgra64be_X_c;
2306 #endif /* CONFIG_SWSCALE_ALPHA */
2308 *yuv2packed1 = yuv2bgrx64be_1_c;
2309 *yuv2packed2 = yuv2bgrx64be_2_c;
2310 *yuv2packedX = yuv2bgrx64be_X_c;
2313 case AV_PIX_FMT_RGB48LE:
2314 *yuv2packed1 = yuv2rgb48le_1_c;
2315 *yuv2packed2 = yuv2rgb48le_2_c;
2316 *yuv2packedX = yuv2rgb48le_X_c;
2318 case AV_PIX_FMT_RGB48BE:
2319 *yuv2packed1 = yuv2rgb48be_1_c;
2320 *yuv2packed2 = yuv2rgb48be_2_c;
2321 *yuv2packedX = yuv2rgb48be_X_c;
2323 case AV_PIX_FMT_BGR48LE:
2324 *yuv2packed1 = yuv2bgr48le_1_c;
2325 *yuv2packed2 = yuv2bgr48le_2_c;
2326 *yuv2packedX = yuv2bgr48le_X_c;
2328 case AV_PIX_FMT_BGR48BE:
2329 *yuv2packed1 = yuv2bgr48be_1_c;
2330 *yuv2packed2 = yuv2bgr48be_2_c;
2331 *yuv2packedX = yuv2bgr48be_X_c;
2333 case AV_PIX_FMT_RGB32:
2334 case AV_PIX_FMT_BGR32:
2336 *yuv2packed1 = yuv2rgb32_1_c;
2337 *yuv2packed2 = yuv2rgb32_2_c;
2338 *yuv2packedX = yuv2rgb32_X_c;
2340 #if CONFIG_SWSCALE_ALPHA
2342 *yuv2packed1 = yuv2rgba32_1_c;
2343 *yuv2packed2 = yuv2rgba32_2_c;
2344 *yuv2packedX = yuv2rgba32_X_c;
2346 #endif /* CONFIG_SWSCALE_ALPHA */
2348 *yuv2packed1 = yuv2rgbx32_1_c;
2349 *yuv2packed2 = yuv2rgbx32_2_c;
2350 *yuv2packedX = yuv2rgbx32_X_c;
2352 #endif /* !CONFIG_SMALL */
2354 case AV_PIX_FMT_RGB32_1:
2355 case AV_PIX_FMT_BGR32_1:
2357 *yuv2packed1 = yuv2rgb32_1_1_c;
2358 *yuv2packed2 = yuv2rgb32_1_2_c;
2359 *yuv2packedX = yuv2rgb32_1_X_c;
2361 #if CONFIG_SWSCALE_ALPHA
2363 *yuv2packed1 = yuv2rgba32_1_1_c;
2364 *yuv2packed2 = yuv2rgba32_1_2_c;
2365 *yuv2packedX = yuv2rgba32_1_X_c;
2367 #endif /* CONFIG_SWSCALE_ALPHA */
2369 *yuv2packed1 = yuv2rgbx32_1_1_c;
2370 *yuv2packed2 = yuv2rgbx32_1_2_c;
2371 *yuv2packedX = yuv2rgbx32_1_X_c;
2373 #endif /* !CONFIG_SMALL */
2375 case AV_PIX_FMT_RGB24:
2376 *yuv2packed1 = yuv2rgb24_1_c;
2377 *yuv2packed2 = yuv2rgb24_2_c;
2378 *yuv2packedX = yuv2rgb24_X_c;
2380 case AV_PIX_FMT_BGR24:
2381 *yuv2packed1 = yuv2bgr24_1_c;
2382 *yuv2packed2 = yuv2bgr24_2_c;
2383 *yuv2packedX = yuv2bgr24_X_c;
2385 case AV_PIX_FMT_RGB565LE:
2386 case AV_PIX_FMT_RGB565BE:
2387 case AV_PIX_FMT_BGR565LE:
2388 case AV_PIX_FMT_BGR565BE:
2389 *yuv2packed1 = yuv2rgb16_1_c;
2390 *yuv2packed2 = yuv2rgb16_2_c;
2391 *yuv2packedX = yuv2rgb16_X_c;
2393 case AV_PIX_FMT_RGB555LE:
2394 case AV_PIX_FMT_RGB555BE:
2395 case AV_PIX_FMT_BGR555LE:
2396 case AV_PIX_FMT_BGR555BE:
2397 *yuv2packed1 = yuv2rgb15_1_c;
2398 *yuv2packed2 = yuv2rgb15_2_c;
2399 *yuv2packedX = yuv2rgb15_X_c;
2401 case AV_PIX_FMT_RGB444LE:
2402 case AV_PIX_FMT_RGB444BE:
2403 case AV_PIX_FMT_BGR444LE:
2404 case AV_PIX_FMT_BGR444BE:
2405 *yuv2packed1 = yuv2rgb12_1_c;
2406 *yuv2packed2 = yuv2rgb12_2_c;
2407 *yuv2packedX = yuv2rgb12_X_c;
2409 case AV_PIX_FMT_RGB8:
2410 case AV_PIX_FMT_BGR8:
2411 *yuv2packed1 = yuv2rgb8_1_c;
2412 *yuv2packed2 = yuv2rgb8_2_c;
2413 *yuv2packedX = yuv2rgb8_X_c;
2415 case AV_PIX_FMT_RGB4:
2416 case AV_PIX_FMT_BGR4:
2417 *yuv2packed1 = yuv2rgb4_1_c;
2418 *yuv2packed2 = yuv2rgb4_2_c;
2419 *yuv2packedX = yuv2rgb4_X_c;
2421 case AV_PIX_FMT_RGB4_BYTE:
2422 case AV_PIX_FMT_BGR4_BYTE:
2423 *yuv2packed1 = yuv2rgb4b_1_c;
2424 *yuv2packed2 = yuv2rgb4b_2_c;
2425 *yuv2packedX = yuv2rgb4b_X_c;
2429 switch (dstFormat) {
2430 case AV_PIX_FMT_MONOWHITE:
2431 *yuv2packed1 = yuv2monowhite_1_c;
2432 *yuv2packed2 = yuv2monowhite_2_c;
2433 *yuv2packedX = yuv2monowhite_X_c;
2435 case AV_PIX_FMT_MONOBLACK:
2436 *yuv2packed1 = yuv2monoblack_1_c;
2437 *yuv2packed2 = yuv2monoblack_2_c;
2438 *yuv2packedX = yuv2monoblack_X_c;
2440 case AV_PIX_FMT_YUYV422:
2441 *yuv2packed1 = yuv2yuyv422_1_c;
2442 *yuv2packed2 = yuv2yuyv422_2_c;
2443 *yuv2packedX = yuv2yuyv422_X_c;
2445 case AV_PIX_FMT_YVYU422:
2446 *yuv2packed1 = yuv2yvyu422_1_c;
2447 *yuv2packed2 = yuv2yvyu422_2_c;
2448 *yuv2packedX = yuv2yvyu422_X_c;
2450 case AV_PIX_FMT_UYVY422:
2451 *yuv2packed1 = yuv2uyvy422_1_c;
2452 *yuv2packed2 = yuv2uyvy422_2_c;
2453 *yuv2packedX = yuv2uyvy422_X_c;
2455 case AV_PIX_FMT_YA8:
2456 *yuv2packed1 = yuv2ya8_1_c;
2457 *yuv2packed2 = yuv2ya8_2_c;
2458 *yuv2packedX = yuv2ya8_X_c;