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);
183 static void yuv2p016cX_c(enum AVPixelFormat dstFormat, const uint8_t *chrDither, const int16_t *chrFilter, int chrFilterSize,
184 const int16_t **chrUSrc, const int16_t **chrVSrc,
185 uint8_t *dest8, int chrDstW)
187 uint16_t *dest = (uint16_t*)dest8;
188 const int32_t **uSrc = (const int32_t **)chrUSrc;
189 const int32_t **vSrc = (const int32_t **)chrVSrc;
191 int big_endian = dstFormat == AV_PIX_FMT_P016BE;
194 for (i = 0; i < chrDstW; i++) {
195 int u = 1 << (shift - 1);
196 int v = 1 << (shift - 1);
198 /* See yuv2planeX_16_c_template for details. */
201 for (j = 0; j < chrFilterSize; j++) {
202 u += uSrc[j][i] * (unsigned)chrFilter[j];
203 v += vSrc[j][i] * (unsigned)chrFilter[j];
206 output_pixel(&dest[2*i] , u, 0x8000, int);
207 output_pixel(&dest[2*i+1], v, 0x8000, int);
211 static av_always_inline void
212 yuv2plane1_float_c_template(const int32_t *src, float *dest, int dstW)
214 static const int big_endian = HAVE_BIGENDIAN;
215 static const int shift = 3;
216 static const float float_mult = 1.0f / 65535.0f;
220 for (i = 0; i < dstW; ++i){
221 val = src[i] + (1 << (shift - 1));
222 output_pixel(&val_uint, val, 0, uint);
223 dest[i] = float_mult * (float)val_uint;
227 static av_always_inline void
228 yuv2plane1_float_bswap_c_template(const int32_t *src, uint32_t *dest, int dstW)
230 static const int big_endian = HAVE_BIGENDIAN;
231 static const int shift = 3;
232 static const float float_mult = 1.0f / 65535.0f;
236 for (i = 0; i < dstW; ++i){
237 val = src[i] + (1 << (shift - 1));
238 output_pixel(&val_uint, val, 0, uint);
239 dest[i] = av_bswap32(av_float2int(float_mult * (float)val_uint));
243 static av_always_inline void
244 yuv2planeX_float_c_template(const int16_t *filter, int filterSize, const int32_t **src,
245 float *dest, int dstW)
247 static const int big_endian = HAVE_BIGENDIAN;
248 static const int shift = 15;
249 static const float float_mult = 1.0f / 65535.0f;
253 for (i = 0; i < dstW; ++i){
254 val = (1 << (shift - 1)) - 0x40000000;
255 for (j = 0; j < filterSize; ++j){
256 val += src[j][i] * (unsigned)filter[j];
258 output_pixel(&val_uint, val, 0x8000, int);
259 dest[i] = float_mult * (float)val_uint;
263 static av_always_inline void
264 yuv2planeX_float_bswap_c_template(const int16_t *filter, int filterSize, const int32_t **src,
265 uint32_t *dest, int dstW)
267 static const int big_endian = HAVE_BIGENDIAN;
268 static const int shift = 15;
269 static const float float_mult = 1.0f / 65535.0f;
273 for (i = 0; i < dstW; ++i){
274 val = (1 << (shift - 1)) - 0x40000000;
275 for (j = 0; j < filterSize; ++j){
276 val += src[j][i] * (unsigned)filter[j];
278 output_pixel(&val_uint, val, 0x8000, int);
279 dest[i] = av_bswap32(av_float2int(float_mult * (float)val_uint));
283 #define yuv2plane1_float(template, dest_type, BE_LE) \
284 static void yuv2plane1_float ## BE_LE ## _c(const int16_t *src, uint8_t *dest, int dstW, \
285 const uint8_t *dither, int offset) \
287 template((const int32_t *)src, (dest_type *)dest, dstW); \
290 #define yuv2planeX_float(template, dest_type, BE_LE) \
291 static void yuv2planeX_float ## BE_LE ## _c(const int16_t *filter, int filterSize, \
292 const int16_t **src, uint8_t *dest, int dstW, \
293 const uint8_t *dither, int offset) \
295 template(filter, filterSize, (const int32_t **)src, (dest_type *)dest, dstW); \
299 yuv2plane1_float(yuv2plane1_float_c_template, float, BE)
300 yuv2plane1_float(yuv2plane1_float_bswap_c_template, uint32_t, LE)
301 yuv2planeX_float(yuv2planeX_float_c_template, float, BE)
302 yuv2planeX_float(yuv2planeX_float_bswap_c_template, uint32_t, LE)
304 yuv2plane1_float(yuv2plane1_float_c_template, float, LE)
305 yuv2plane1_float(yuv2plane1_float_bswap_c_template, uint32_t, BE)
306 yuv2planeX_float(yuv2planeX_float_c_template, float, LE)
307 yuv2planeX_float(yuv2planeX_float_bswap_c_template, uint32_t, BE)
312 #define output_pixel(pos, val) \
314 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
316 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
319 static av_always_inline void
320 yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
321 int big_endian, int output_bits)
324 int shift = 15 - output_bits;
326 for (i = 0; i < dstW; i++) {
327 int val = src[i] + (1 << (shift - 1));
328 output_pixel(&dest[i], val);
332 static av_always_inline void
333 yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
334 const int16_t **src, uint16_t *dest, int dstW,
335 int big_endian, int output_bits)
338 int shift = 11 + 16 - output_bits;
340 for (i = 0; i < dstW; i++) {
341 int val = 1 << (shift - 1);
344 for (j = 0; j < filterSize; j++)
345 val += src[j][i] * filter[j];
347 output_pixel(&dest[i], val);
353 #define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
354 static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
355 uint8_t *dest, int dstW, \
356 const uint8_t *dither, int offset)\
358 yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
359 (uint16_t *) dest, dstW, is_be, bits); \
361 static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
362 const int16_t **src, uint8_t *dest, int dstW, \
363 const uint8_t *dither, int offset)\
365 yuv2planeX_## template_size ## _c_template(filter, \
366 filterSize, (const typeX_t **) src, \
367 (uint16_t *) dest, dstW, is_be, bits); \
369 yuv2NBPS( 9, BE, 1, 10, int16_t)
370 yuv2NBPS( 9, LE, 0, 10, int16_t)
371 yuv2NBPS(10, BE, 1, 10, int16_t)
372 yuv2NBPS(10, LE, 0, 10, int16_t)
373 yuv2NBPS(12, BE, 1, 10, int16_t)
374 yuv2NBPS(12, LE, 0, 10, int16_t)
375 yuv2NBPS(14, BE, 1, 10, int16_t)
376 yuv2NBPS(14, LE, 0, 10, int16_t)
377 yuv2NBPS(16, BE, 1, 16, int32_t)
378 yuv2NBPS(16, LE, 0, 16, int32_t)
380 static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
381 const int16_t **src, uint8_t *dest, int dstW,
382 const uint8_t *dither, int offset)
385 for (i=0; i<dstW; i++) {
386 int val = dither[(i + offset) & 7] << 12;
388 for (j=0; j<filterSize; j++)
389 val += src[j][i] * filter[j];
391 dest[i]= av_clip_uint8(val>>19);
395 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
396 const uint8_t *dither, int offset)
399 for (i=0; i<dstW; i++) {
400 int val = (src[i] + dither[(i + offset) & 7]) >> 7;
401 dest[i]= av_clip_uint8(val);
405 static void yuv2nv12cX_c(enum AVPixelFormat dstFormat, const uint8_t *chrDither, const int16_t *chrFilter, int chrFilterSize,
406 const int16_t **chrUSrc, const int16_t **chrVSrc,
407 uint8_t *dest, int chrDstW)
411 if (dstFormat == AV_PIX_FMT_NV12 ||
412 dstFormat == AV_PIX_FMT_NV24)
413 for (i=0; i<chrDstW; i++) {
414 int u = chrDither[i & 7] << 12;
415 int v = chrDither[(i + 3) & 7] << 12;
417 for (j=0; j<chrFilterSize; j++) {
418 u += chrUSrc[j][i] * chrFilter[j];
419 v += chrVSrc[j][i] * chrFilter[j];
422 dest[2*i]= av_clip_uint8(u>>19);
423 dest[2*i+1]= av_clip_uint8(v>>19);
426 for (i=0; i<chrDstW; i++) {
427 int u = chrDither[i & 7] << 12;
428 int v = chrDither[(i + 3) & 7] << 12;
430 for (j=0; j<chrFilterSize; j++) {
431 u += chrUSrc[j][i] * chrFilter[j];
432 v += chrVSrc[j][i] * chrFilter[j];
435 dest[2*i]= av_clip_uint8(v>>19);
436 dest[2*i+1]= av_clip_uint8(u>>19);
441 #define output_pixel(pos, val) \
443 AV_WB16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
445 AV_WL16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
448 static void yuv2p010l1_c(const int16_t *src,
449 uint16_t *dest, int dstW,
455 for (i = 0; i < dstW; i++) {
456 int val = src[i] + (1 << (shift - 1));
457 output_pixel(&dest[i], val);
461 static void yuv2p010lX_c(const int16_t *filter, int filterSize,
462 const int16_t **src, uint16_t *dest, int dstW,
468 for (i = 0; i < dstW; i++) {
469 int val = 1 << (shift - 1);
471 for (j = 0; j < filterSize; j++)
472 val += src[j][i] * filter[j];
474 output_pixel(&dest[i], val);
478 static void yuv2p010cX_c(enum AVPixelFormat dstFormat, const uint8_t *chrDither, const int16_t *chrFilter, int chrFilterSize,
479 const int16_t **chrUSrc, const int16_t **chrVSrc,
480 uint8_t *dest8, int chrDstW)
482 uint16_t *dest = (uint16_t*)dest8;
484 int big_endian = dstFormat == AV_PIX_FMT_P010BE;
487 for (i = 0; i < chrDstW; i++) {
488 int u = 1 << (shift - 1);
489 int v = 1 << (shift - 1);
491 for (j = 0; j < chrFilterSize; j++) {
492 u += chrUSrc[j][i] * chrFilter[j];
493 v += chrVSrc[j][i] * chrFilter[j];
496 output_pixel(&dest[2*i] , u);
497 output_pixel(&dest[2*i+1], v);
501 static void yuv2p010l1_LE_c(const int16_t *src,
502 uint8_t *dest, int dstW,
503 const uint8_t *dither, int offset)
505 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 0);
508 static void yuv2p010l1_BE_c(const int16_t *src,
509 uint8_t *dest, int dstW,
510 const uint8_t *dither, int offset)
512 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 1);
515 static void yuv2p010lX_LE_c(const int16_t *filter, int filterSize,
516 const int16_t **src, uint8_t *dest, int dstW,
517 const uint8_t *dither, int offset)
519 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 0);
522 static void yuv2p010lX_BE_c(const int16_t *filter, int filterSize,
523 const int16_t **src, uint8_t *dest, int dstW,
524 const uint8_t *dither, int offset)
526 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 1);
532 #define accumulate_bit(acc, val) \
535 #define output_pixel(pos, acc) \
536 if (target == AV_PIX_FMT_MONOBLACK) { \
542 static av_always_inline void
543 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
544 const int16_t **lumSrc, int lumFilterSize,
545 const int16_t *chrFilter, const int16_t **chrUSrc,
546 const int16_t **chrVSrc, int chrFilterSize,
547 const int16_t **alpSrc, uint8_t *dest, int dstW,
548 int y, enum AVPixelFormat target)
550 const uint8_t * const d128 = ff_dither_8x8_220[y&7];
555 for (i = 0; i < dstW; i += 2) {
560 for (j = 0; j < lumFilterSize; j++) {
561 Y1 += lumSrc[j][i] * lumFilter[j];
562 Y2 += lumSrc[j][i+1] * lumFilter[j];
566 if ((Y1 | Y2) & 0x100) {
567 Y1 = av_clip_uint8(Y1);
568 Y2 = av_clip_uint8(Y2);
570 if (c->dither == SWS_DITHER_ED) {
571 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;
572 c->dither_error[0][i] = err;
573 acc = 2*acc + (Y1 >= 128);
576 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);
577 c->dither_error[0][i+1] = Y1;
578 acc = 2*acc + (err >= 128);
581 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
582 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
585 output_pixel(*dest++, acc);
588 c->dither_error[0][i] = err;
591 output_pixel(*dest, acc);
595 static av_always_inline void
596 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
597 const int16_t *ubuf[2], const int16_t *vbuf[2],
598 const int16_t *abuf[2], uint8_t *dest, int dstW,
599 int yalpha, int uvalpha, int y,
600 enum AVPixelFormat target)
602 const int16_t *buf0 = buf[0], *buf1 = buf[1];
603 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
604 int yalpha1 = 4096 - yalpha;
606 av_assert2(yalpha <= 4096U);
608 if (c->dither == SWS_DITHER_ED) {
611 for (i = 0; i < dstW; i +=2) {
614 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
615 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;
616 c->dither_error[0][i] = err;
617 acc = 2*acc + (Y >= 128);
620 err = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
621 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;
622 c->dither_error[0][i+1] = Y;
623 acc = 2*acc + (err >= 128);
627 output_pixel(*dest++, acc);
629 c->dither_error[0][i] = err;
631 for (i = 0; i < dstW; i += 8) {
634 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
635 accumulate_bit(acc, Y + d128[0]);
636 Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
637 accumulate_bit(acc, Y + d128[1]);
638 Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
639 accumulate_bit(acc, Y + d128[2]);
640 Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
641 accumulate_bit(acc, Y + d128[3]);
642 Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
643 accumulate_bit(acc, Y + d128[4]);
644 Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
645 accumulate_bit(acc, Y + d128[5]);
646 Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
647 accumulate_bit(acc, Y + d128[6]);
648 Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
649 accumulate_bit(acc, Y + d128[7]);
651 output_pixel(*dest++, acc);
656 static av_always_inline void
657 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
658 const int16_t *ubuf[2], const int16_t *vbuf[2],
659 const int16_t *abuf0, uint8_t *dest, int dstW,
660 int uvalpha, int y, enum AVPixelFormat target)
662 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
665 if (c->dither == SWS_DITHER_ED) {
668 for (i = 0; i < dstW; i +=2) {
671 Y = ((buf0[i + 0] + 64) >> 7);
672 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;
673 c->dither_error[0][i] = err;
674 acc = 2*acc + (Y >= 128);
677 err = ((buf0[i + 1] + 64) >> 7);
678 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;
679 c->dither_error[0][i+1] = Y;
680 acc = 2*acc + (err >= 128);
684 output_pixel(*dest++, acc);
686 c->dither_error[0][i] = err;
688 for (i = 0; i < dstW; i += 8) {
690 accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
691 accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
692 accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
693 accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
694 accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
695 accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
696 accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
697 accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
699 output_pixel(*dest++, acc);
705 #undef accumulate_bit
707 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
708 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
709 const int16_t **lumSrc, int lumFilterSize, \
710 const int16_t *chrFilter, const int16_t **chrUSrc, \
711 const int16_t **chrVSrc, int chrFilterSize, \
712 const int16_t **alpSrc, uint8_t *dest, int dstW, \
715 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
716 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
717 alpSrc, dest, dstW, y, fmt); \
720 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
721 const int16_t *ubuf[2], const int16_t *vbuf[2], \
722 const int16_t *abuf[2], uint8_t *dest, int dstW, \
723 int yalpha, int uvalpha, int y) \
725 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
726 dest, dstW, yalpha, uvalpha, y, fmt); \
729 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
730 const int16_t *ubuf[2], const int16_t *vbuf[2], \
731 const int16_t *abuf0, uint8_t *dest, int dstW, \
732 int uvalpha, int y) \
734 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
735 abuf0, dest, dstW, uvalpha, \
739 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
740 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
742 #define output_pixels(pos, Y1, U, Y2, V) \
743 if (target == AV_PIX_FMT_YUYV422) { \
744 dest[pos + 0] = Y1; \
746 dest[pos + 2] = Y2; \
748 } else if (target == AV_PIX_FMT_YVYU422) { \
749 dest[pos + 0] = Y1; \
751 dest[pos + 2] = Y2; \
753 } else { /* AV_PIX_FMT_UYVY422 */ \
755 dest[pos + 1] = Y1; \
757 dest[pos + 3] = Y2; \
760 static av_always_inline void
761 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
762 const int16_t **lumSrc, int lumFilterSize,
763 const int16_t *chrFilter, const int16_t **chrUSrc,
764 const int16_t **chrVSrc, int chrFilterSize,
765 const int16_t **alpSrc, uint8_t *dest, int dstW,
766 int y, enum AVPixelFormat target)
770 for (i = 0; i < ((dstW + 1) >> 1); i++) {
777 for (j = 0; j < lumFilterSize; j++) {
778 Y1 += lumSrc[j][i * 2] * lumFilter[j];
779 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
781 for (j = 0; j < chrFilterSize; j++) {
782 U += chrUSrc[j][i] * chrFilter[j];
783 V += chrVSrc[j][i] * chrFilter[j];
789 if ((Y1 | Y2 | U | V) & 0x100) {
790 Y1 = av_clip_uint8(Y1);
791 Y2 = av_clip_uint8(Y2);
792 U = av_clip_uint8(U);
793 V = av_clip_uint8(V);
795 output_pixels(4*i, Y1, U, Y2, V);
799 static av_always_inline void
800 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
801 const int16_t *ubuf[2], const int16_t *vbuf[2],
802 const int16_t *abuf[2], uint8_t *dest, int dstW,
803 int yalpha, int uvalpha, int y,
804 enum AVPixelFormat target)
806 const int16_t *buf0 = buf[0], *buf1 = buf[1],
807 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
808 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
809 int yalpha1 = 4096 - yalpha;
810 int uvalpha1 = 4096 - uvalpha;
812 av_assert2(yalpha <= 4096U);
813 av_assert2(uvalpha <= 4096U);
815 for (i = 0; i < ((dstW + 1) >> 1); i++) {
816 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
817 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
818 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
819 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
821 if ((Y1 | Y2 | U | V) & 0x100) {
822 Y1 = av_clip_uint8(Y1);
823 Y2 = av_clip_uint8(Y2);
824 U = av_clip_uint8(U);
825 V = av_clip_uint8(V);
828 output_pixels(i * 4, Y1, U, Y2, V);
832 static av_always_inline void
833 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
834 const int16_t *ubuf[2], const int16_t *vbuf[2],
835 const int16_t *abuf0, uint8_t *dest, int dstW,
836 int uvalpha, int y, enum AVPixelFormat target)
838 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
841 if (uvalpha < 2048) {
842 for (i = 0; i < ((dstW + 1) >> 1); i++) {
843 int Y1 = (buf0[i * 2 ]+64) >> 7;
844 int Y2 = (buf0[i * 2 + 1]+64) >> 7;
845 int U = (ubuf0[i] +64) >> 7;
846 int V = (vbuf0[i] +64) >> 7;
848 if ((Y1 | Y2 | U | V) & 0x100) {
849 Y1 = av_clip_uint8(Y1);
850 Y2 = av_clip_uint8(Y2);
851 U = av_clip_uint8(U);
852 V = av_clip_uint8(V);
855 output_pixels(i * 4, Y1, U, Y2, V);
858 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
859 for (i = 0; i < ((dstW + 1) >> 1); i++) {
860 int Y1 = (buf0[i * 2 ] + 64) >> 7;
861 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
862 int U = (ubuf0[i] + ubuf1[i]+128) >> 8;
863 int V = (vbuf0[i] + vbuf1[i]+128) >> 8;
865 if ((Y1 | Y2 | U | V) & 0x100) {
866 Y1 = av_clip_uint8(Y1);
867 Y2 = av_clip_uint8(Y2);
868 U = av_clip_uint8(U);
869 V = av_clip_uint8(V);
872 output_pixels(i * 4, Y1, U, Y2, V);
879 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422)
880 YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422)
881 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422)
883 #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)
884 #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)
885 #define output_pixel(pos, val) \
886 if (isBE(target)) { \
892 static av_always_inline void
893 yuv2ya16_X_c_template(SwsContext *c, const int16_t *lumFilter,
894 const int32_t **lumSrc, int lumFilterSize,
895 const int16_t *chrFilter, const int32_t **unused_chrUSrc,
896 const int32_t **unused_chrVSrc, int unused_chrFilterSize,
897 const int32_t **alpSrc, uint16_t *dest, int dstW,
898 int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
900 int hasAlpha = !!alpSrc;
903 for (i = 0; i < dstW; i++) {
908 for (j = 0; j < lumFilterSize; j++)
909 Y += lumSrc[j][i] * lumFilter[j];
912 Y += (1<<3) + 0x8000;
913 Y = av_clip_uint16(Y);
916 A = -0x40000000 + (1<<14);
917 for (j = 0; j < lumFilterSize; j++)
918 A += alpSrc[j][i] * lumFilter[j];
922 A = av_clip_uint16(A);
925 output_pixel(&dest[2 * i ], Y);
926 output_pixel(&dest[2 * i + 1], A);
930 static av_always_inline void
931 yuv2ya16_2_c_template(SwsContext *c, const int32_t *buf[2],
932 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
933 const int32_t *abuf[2], uint16_t *dest, int dstW,
934 int yalpha, int unused_uvalpha, int y,
935 enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
937 int hasAlpha = abuf && abuf[0] && abuf[1];
938 const int32_t *buf0 = buf[0], *buf1 = buf[1],
939 *abuf0 = hasAlpha ? abuf[0] : NULL,
940 *abuf1 = hasAlpha ? abuf[1] : NULL;
941 int yalpha1 = 4096 - yalpha;
944 av_assert2(yalpha <= 4096U);
946 for (i = 0; i < dstW; i++) {
947 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 15;
950 Y = av_clip_uint16(Y);
953 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 15;
954 A = av_clip_uint16(A);
957 output_pixel(&dest[2 * i ], Y);
958 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
962 static av_always_inline void
963 yuv2ya16_1_c_template(SwsContext *c, const int32_t *buf0,
964 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
965 const int32_t *abuf0, uint16_t *dest, int dstW,
966 int unused_uvalpha, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
968 int hasAlpha = !!abuf0;
971 for (i = 0; i < dstW; i++) {
972 int Y = buf0[i] >> 3;/* 19 - 16 */
975 Y = av_clip_uint16(Y);
980 A = av_clip_uint16(A);
983 output_pixel(&dest[2 * i ], Y);
984 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
988 static av_always_inline void
989 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
990 const int32_t **lumSrc, int lumFilterSize,
991 const int16_t *chrFilter, const int32_t **chrUSrc,
992 const int32_t **chrVSrc, int chrFilterSize,
993 const int32_t **alpSrc, uint16_t *dest, int dstW,
994 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
997 int A1 = 0xffff<<14, A2 = 0xffff<<14;
999 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1001 int Y1 = -0x40000000;
1002 int Y2 = -0x40000000;
1003 int U = -(128 << 23); // 19
1004 int V = -(128 << 23);
1007 for (j = 0; j < lumFilterSize; j++) {
1008 Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
1009 Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1011 for (j = 0; j < chrFilterSize; j++) {;
1012 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1013 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1019 for (j = 0; j < lumFilterSize; j++) {
1020 A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
1021 A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1029 // 8 bits: 12+15=27; 16 bits: 12+19=31
1037 // 8 bits: 27 -> 17 bits, 16 bits: 31 - 14 = 17 bits
1038 Y1 -= c->yuv2rgb_y_offset;
1039 Y2 -= c->yuv2rgb_y_offset;
1040 Y1 *= c->yuv2rgb_y_coeff;
1041 Y2 *= c->yuv2rgb_y_coeff;
1042 Y1 += 1 << 13; // 21
1044 // 8 bits: 17 + 13 bits = 30 bits, 16 bits: 17 + 13 bits = 30 bits
1046 R = V * c->yuv2rgb_v2r_coeff;
1047 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1048 B = U * c->yuv2rgb_u2b_coeff;
1050 // 8 bits: 30 - 22 = 8 bits, 16 bits: 30 bits - 14 = 16 bits
1051 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1052 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1053 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1055 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1056 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1057 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1058 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1059 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1062 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1063 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1064 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1070 static av_always_inline void
1071 yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
1072 const int32_t *ubuf[2], const int32_t *vbuf[2],
1073 const int32_t *abuf[2], uint16_t *dest, int dstW,
1074 int yalpha, int uvalpha, int y,
1075 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1077 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1078 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1079 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1080 *abuf0 = hasAlpha ? abuf[0] : NULL,
1081 *abuf1 = hasAlpha ? abuf[1] : NULL;
1082 int yalpha1 = 4096 - yalpha;
1083 int uvalpha1 = 4096 - uvalpha;
1085 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1087 av_assert2(yalpha <= 4096U);
1088 av_assert2(uvalpha <= 4096U);
1090 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1091 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
1092 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
1093 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1094 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1097 Y1 -= c->yuv2rgb_y_offset;
1098 Y2 -= c->yuv2rgb_y_offset;
1099 Y1 *= c->yuv2rgb_y_coeff;
1100 Y2 *= c->yuv2rgb_y_coeff;
1104 R = V * c->yuv2rgb_v2r_coeff;
1105 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1106 B = U * c->yuv2rgb_u2b_coeff;
1109 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
1110 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
1116 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1117 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1118 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1120 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1121 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1122 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1123 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1124 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1127 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1128 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1129 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1135 static av_always_inline void
1136 yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
1137 const int32_t *ubuf[2], const int32_t *vbuf[2],
1138 const int32_t *abuf0, uint16_t *dest, int dstW,
1139 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1141 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1143 int A1 = 0xffff<<14, A2= 0xffff<<14;
1145 if (uvalpha < 2048) {
1146 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1147 int Y1 = (buf0[i * 2] ) >> 2;
1148 int Y2 = (buf0[i * 2 + 1]) >> 2;
1149 int U = (ubuf0[i] - (128 << 11)) >> 2;
1150 int V = (vbuf0[i] - (128 << 11)) >> 2;
1153 Y1 -= c->yuv2rgb_y_offset;
1154 Y2 -= c->yuv2rgb_y_offset;
1155 Y1 *= c->yuv2rgb_y_coeff;
1156 Y2 *= c->yuv2rgb_y_coeff;
1161 A1 = abuf0[i * 2 ] << 11;
1162 A2 = abuf0[i * 2 + 1] << 11;
1168 R = V * c->yuv2rgb_v2r_coeff;
1169 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1170 B = U * c->yuv2rgb_u2b_coeff;
1172 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1173 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1174 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1176 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1177 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1178 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1179 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1180 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1183 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1184 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1185 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1190 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1191 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1192 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1193 int Y1 = (buf0[i * 2] ) >> 2;
1194 int Y2 = (buf0[i * 2 + 1]) >> 2;
1195 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1196 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1199 Y1 -= c->yuv2rgb_y_offset;
1200 Y2 -= c->yuv2rgb_y_offset;
1201 Y1 *= c->yuv2rgb_y_coeff;
1202 Y2 *= c->yuv2rgb_y_coeff;
1207 A1 = abuf0[i * 2 ] << 11;
1208 A2 = abuf0[i * 2 + 1] << 11;
1214 R = V * c->yuv2rgb_v2r_coeff;
1215 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1216 B = U * c->yuv2rgb_u2b_coeff;
1218 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1219 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1220 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1222 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1223 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1224 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1225 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1226 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1229 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1230 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1231 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1238 static av_always_inline void
1239 yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1240 const int32_t **lumSrc, int lumFilterSize,
1241 const int16_t *chrFilter, const int32_t **chrUSrc,
1242 const int32_t **chrVSrc, int chrFilterSize,
1243 const int32_t **alpSrc, uint16_t *dest, int dstW,
1244 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1249 for (i = 0; i < dstW; i++) {
1251 int Y = -0x40000000;
1252 int U = -(128 << 23); // 19
1253 int V = -(128 << 23);
1256 for (j = 0; j < lumFilterSize; j++) {
1257 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
1259 for (j = 0; j < chrFilterSize; j++) {;
1260 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1261 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1266 for (j = 0; j < lumFilterSize; j++) {
1267 A += alpSrc[j][i] * (unsigned)lumFilter[j];
1273 // 8bit: 12+15=27; 16-bit: 12+19=31
1279 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
1280 Y -= c->yuv2rgb_y_offset;
1281 Y *= c->yuv2rgb_y_coeff;
1283 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
1285 R = V * c->yuv2rgb_v2r_coeff;
1286 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1287 B = U * c->yuv2rgb_u2b_coeff;
1289 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
1290 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1291 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1292 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1294 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1302 static av_always_inline void
1303 yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
1304 const int32_t *ubuf[2], const int32_t *vbuf[2],
1305 const int32_t *abuf[2], uint16_t *dest, int dstW,
1306 int yalpha, int uvalpha, int y,
1307 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1309 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1310 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1311 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1312 *abuf0 = hasAlpha ? abuf[0] : NULL,
1313 *abuf1 = hasAlpha ? abuf[1] : NULL;
1314 int yalpha1 = 4096 - yalpha;
1315 int uvalpha1 = 4096 - uvalpha;
1319 av_assert2(yalpha <= 4096U);
1320 av_assert2(uvalpha <= 4096U);
1322 for (i = 0; i < dstW; i++) {
1323 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 14;
1324 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1325 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1328 Y -= c->yuv2rgb_y_offset;
1329 Y *= c->yuv2rgb_y_coeff;
1332 R = V * c->yuv2rgb_v2r_coeff;
1333 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1334 B = U * c->yuv2rgb_u2b_coeff;
1337 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1342 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1343 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1344 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1346 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1354 static av_always_inline void
1355 yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1356 const int32_t *ubuf[2], const int32_t *vbuf[2],
1357 const int32_t *abuf0, uint16_t *dest, int dstW,
1358 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1360 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1364 if (uvalpha < 2048) {
1365 for (i = 0; i < dstW; i++) {
1366 int Y = (buf0[i]) >> 2;
1367 int U = (ubuf0[i] - (128 << 11)) >> 2;
1368 int V = (vbuf0[i] - (128 << 11)) >> 2;
1371 Y -= c->yuv2rgb_y_offset;
1372 Y *= c->yuv2rgb_y_coeff;
1381 R = V * c->yuv2rgb_v2r_coeff;
1382 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1383 B = U * c->yuv2rgb_u2b_coeff;
1385 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1386 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1387 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1389 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1396 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1398 for (i = 0; i < dstW; i++) {
1399 int Y = (buf0[i] ) >> 2;
1400 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1401 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1404 Y -= c->yuv2rgb_y_offset;
1405 Y *= c->yuv2rgb_y_coeff;
1414 R = V * c->yuv2rgb_v2r_coeff;
1415 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1416 B = U * c->yuv2rgb_u2b_coeff;
1418 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1419 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1420 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1422 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1435 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1436 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1437 const int16_t **_lumSrc, int lumFilterSize, \
1438 const int16_t *chrFilter, const int16_t **_chrUSrc, \
1439 const int16_t **_chrVSrc, int chrFilterSize, \
1440 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1443 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1444 **chrUSrc = (const int32_t **) _chrUSrc, \
1445 **chrVSrc = (const int32_t **) _chrVSrc, \
1446 **alpSrc = (const int32_t **) _alpSrc; \
1447 uint16_t *dest = (uint16_t *) _dest; \
1448 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1449 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1450 alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1453 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1454 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1455 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1456 int yalpha, int uvalpha, int y) \
1458 const int32_t **buf = (const int32_t **) _buf, \
1459 **ubuf = (const int32_t **) _ubuf, \
1460 **vbuf = (const int32_t **) _vbuf, \
1461 **abuf = (const int32_t **) _abuf; \
1462 uint16_t *dest = (uint16_t *) _dest; \
1463 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1464 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1467 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1468 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1469 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1470 int uvalpha, int y) \
1472 const int32_t *buf0 = (const int32_t *) _buf0, \
1473 **ubuf = (const int32_t **) _ubuf, \
1474 **vbuf = (const int32_t **) _vbuf, \
1475 *abuf0 = (const int32_t *) _abuf0; \
1476 uint16_t *dest = (uint16_t *) _dest; \
1477 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1478 dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1481 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1482 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1483 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1484 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1485 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1486 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1487 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1488 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1489 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1490 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1491 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1492 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1493 YUV2PACKED16WRAPPER(yuv2, ya16, ya16be, AV_PIX_FMT_YA16BE, 1, 0)
1494 YUV2PACKED16WRAPPER(yuv2, ya16, ya16le, AV_PIX_FMT_YA16LE, 1, 0)
1496 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1497 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1498 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1499 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1500 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1501 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1502 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1503 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1504 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1505 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1506 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1507 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1510 * Write out 2 RGB pixels in the target pixel format. This function takes a
1511 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1512 * things like endianness conversion and shifting. The caller takes care of
1513 * setting the correct offset in these tables from the chroma (U/V) values.
1514 * This function then uses the luminance (Y1/Y2) values to write out the
1515 * correct RGB values into the destination buffer.
1517 static av_always_inline void
1518 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1519 unsigned A1, unsigned A2,
1520 const void *_r, const void *_g, const void *_b, int y,
1521 enum AVPixelFormat target, int hasAlpha)
1523 if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1524 target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1525 uint32_t *dest = (uint32_t *) _dest;
1526 const uint32_t *r = (const uint32_t *) _r;
1527 const uint32_t *g = (const uint32_t *) _g;
1528 const uint32_t *b = (const uint32_t *) _b;
1531 int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1533 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1534 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1537 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1539 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1540 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1541 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1543 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1544 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1546 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1548 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1549 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1552 } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1553 uint8_t *dest = (uint8_t *) _dest;
1554 const uint8_t *r = (const uint8_t *) _r;
1555 const uint8_t *g = (const uint8_t *) _g;
1556 const uint8_t *b = (const uint8_t *) _b;
1558 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1559 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1561 dest[i * 6 + 0] = r_b[Y1];
1562 dest[i * 6 + 1] = g[Y1];
1563 dest[i * 6 + 2] = b_r[Y1];
1564 dest[i * 6 + 3] = r_b[Y2];
1565 dest[i * 6 + 4] = g[Y2];
1566 dest[i * 6 + 5] = b_r[Y2];
1569 } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1570 target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1571 target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1572 uint16_t *dest = (uint16_t *) _dest;
1573 const uint16_t *r = (const uint16_t *) _r;
1574 const uint16_t *g = (const uint16_t *) _g;
1575 const uint16_t *b = (const uint16_t *) _b;
1576 int dr1, dg1, db1, dr2, dg2, db2;
1578 if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1579 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1580 dg1 = ff_dither_2x2_4[ y & 1 ][0];
1581 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1582 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1583 dg2 = ff_dither_2x2_4[ y & 1 ][1];
1584 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1585 } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1586 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1587 dg1 = ff_dither_2x2_8[ y & 1 ][1];
1588 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1589 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1590 dg2 = ff_dither_2x2_8[ y & 1 ][0];
1591 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1593 dr1 = ff_dither_4x4_16[ y & 3 ][0];
1594 dg1 = ff_dither_4x4_16[ y & 3 ][1];
1595 db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1596 dr2 = ff_dither_4x4_16[ y & 3 ][1];
1597 dg2 = ff_dither_4x4_16[ y & 3 ][0];
1598 db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1601 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1602 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1603 } else if (target == AV_PIX_FMT_X2RGB10) {
1604 uint32_t *dest = (uint32_t *) _dest;
1605 const uint32_t *r = (const uint32_t *) _r;
1606 const uint32_t *g = (const uint32_t *) _g;
1607 const uint32_t *b = (const uint32_t *) _b;
1608 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1609 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1610 } else /* 8/4 bits */ {
1611 uint8_t *dest = (uint8_t *) _dest;
1612 const uint8_t *r = (const uint8_t *) _r;
1613 const uint8_t *g = (const uint8_t *) _g;
1614 const uint8_t *b = (const uint8_t *) _b;
1615 int dr1, dg1, db1, dr2, dg2, db2;
1617 if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1618 const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1619 const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1620 dr1 = dg1 = d32[(i * 2 + 0) & 7];
1621 db1 = d64[(i * 2 + 0) & 7];
1622 dr2 = dg2 = d32[(i * 2 + 1) & 7];
1623 db2 = d64[(i * 2 + 1) & 7];
1625 const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1626 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1627 dr1 = db1 = d128[(i * 2 + 0) & 7];
1628 dg1 = d64[(i * 2 + 0) & 7];
1629 dr2 = db2 = d128[(i * 2 + 1) & 7];
1630 dg2 = d64[(i * 2 + 1) & 7];
1633 if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1634 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1635 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1637 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1638 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1643 static av_always_inline void
1644 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1645 const int16_t **lumSrc, int lumFilterSize,
1646 const int16_t *chrFilter, const int16_t **chrUSrc,
1647 const int16_t **chrVSrc, int chrFilterSize,
1648 const int16_t **alpSrc, uint8_t *dest, int dstW,
1649 int y, enum AVPixelFormat target, int hasAlpha)
1653 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1659 const void *r, *g, *b;
1661 for (j = 0; j < lumFilterSize; j++) {
1662 Y1 += lumSrc[j][i * 2] * lumFilter[j];
1663 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1665 for (j = 0; j < chrFilterSize; j++) {
1666 U += chrUSrc[j][i] * chrFilter[j];
1667 V += chrVSrc[j][i] * chrFilter[j];
1676 for (j = 0; j < lumFilterSize; j++) {
1677 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1678 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1682 if ((A1 | A2) & 0x100) {
1683 A1 = av_clip_uint8(A1);
1684 A2 = av_clip_uint8(A2);
1688 r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1689 g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1690 b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1692 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1693 r, g, b, y, target, hasAlpha);
1697 static av_always_inline void
1698 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1699 const int16_t *ubuf[2], const int16_t *vbuf[2],
1700 const int16_t *abuf[2], uint8_t *dest, int dstW,
1701 int yalpha, int uvalpha, int y,
1702 enum AVPixelFormat target, int hasAlpha)
1704 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1705 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1706 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1707 *abuf0 = hasAlpha ? abuf[0] : NULL,
1708 *abuf1 = hasAlpha ? abuf[1] : NULL;
1709 int yalpha1 = 4096 - yalpha;
1710 int uvalpha1 = 4096 - uvalpha;
1712 av_assert2(yalpha <= 4096U);
1713 av_assert2(uvalpha <= 4096U);
1715 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1716 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1717 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1718 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1719 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1721 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1722 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1723 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1726 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1727 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1728 A1 = av_clip_uint8(A1);
1729 A2 = av_clip_uint8(A2);
1732 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1733 r, g, b, y, target, hasAlpha);
1737 static av_always_inline void
1738 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1739 const int16_t *ubuf[2], const int16_t *vbuf[2],
1740 const int16_t *abuf0, uint8_t *dest, int dstW,
1741 int uvalpha, int y, enum AVPixelFormat target,
1744 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1747 if (uvalpha < 2048) {
1748 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1749 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1750 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1751 int U = (ubuf0[i] + 64) >> 7;
1752 int V = (vbuf0[i] + 64) >> 7;
1754 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1755 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1756 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1759 A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1760 A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1761 A1 = av_clip_uint8(A1);
1762 A2 = av_clip_uint8(A2);
1765 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1766 r, g, b, y, target, hasAlpha);
1769 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1770 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1771 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1772 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1773 int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1774 int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1776 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1777 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1778 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1781 A1 = (abuf0[i * 2 ] + 64) >> 7;
1782 A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1783 A1 = av_clip_uint8(A1);
1784 A2 = av_clip_uint8(A2);
1787 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1788 r, g, b, y, target, hasAlpha);
1793 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1794 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1795 const int16_t **lumSrc, int lumFilterSize, \
1796 const int16_t *chrFilter, const int16_t **chrUSrc, \
1797 const int16_t **chrVSrc, int chrFilterSize, \
1798 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1801 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1802 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1803 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1806 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1807 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1808 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1809 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1810 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1811 int yalpha, int uvalpha, int y) \
1813 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1814 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1817 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1818 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1819 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1820 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1821 const int16_t *abuf0, uint8_t *dest, int dstW, \
1822 int uvalpha, int y) \
1824 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1825 dstW, uvalpha, y, fmt, hasAlpha); \
1829 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1830 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1832 #if CONFIG_SWSCALE_ALPHA
1833 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1834 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1836 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1837 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1839 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1840 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1841 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1842 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1843 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1844 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1845 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1846 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1847 YUV2RGBWRAPPER(yuv2, rgb, x2rgb10, AV_PIX_FMT_X2RGB10, 0)
1849 static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1850 uint8_t *dest, int i, int Y, int A, int U, int V,
1851 int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1854 int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1856 Y -= c->yuv2rgb_y_offset;
1857 Y *= c->yuv2rgb_y_coeff;
1859 R = (unsigned)Y + V*c->yuv2rgb_v2r_coeff;
1860 G = (unsigned)Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1861 B = (unsigned)Y + U*c->yuv2rgb_u2b_coeff;
1862 if ((R | G | B) & 0xC0000000) {
1863 R = av_clip_uintp2(R, 30);
1864 G = av_clip_uintp2(G, 30);
1865 B = av_clip_uintp2(B, 30);
1869 case AV_PIX_FMT_ARGB:
1870 dest[0] = hasAlpha ? A : 255;
1875 case AV_PIX_FMT_RGB24:
1880 case AV_PIX_FMT_RGBA:
1884 dest[3] = hasAlpha ? A : 255;
1886 case AV_PIX_FMT_ABGR:
1887 dest[0] = hasAlpha ? A : 255;
1892 case AV_PIX_FMT_BGR24:
1897 case AV_PIX_FMT_BGRA:
1901 dest[3] = hasAlpha ? A : 255;
1903 case AV_PIX_FMT_BGR4_BYTE:
1904 case AV_PIX_FMT_RGB4_BYTE:
1905 case AV_PIX_FMT_BGR8:
1906 case AV_PIX_FMT_RGB8:
1910 switch (c->dither) {
1912 case SWS_DITHER_AUTO:
1917 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;
1918 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;
1919 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;
1920 c->dither_error[0][i] = err[0];
1921 c->dither_error[1][i] = err[1];
1922 c->dither_error[2][i] = err[2];
1923 r = R >> (isrgb8 ? 5 : 7);
1924 g = G >> (isrgb8 ? 5 : 6);
1925 b = B >> (isrgb8 ? 6 : 7);
1926 r = av_clip(r, 0, isrgb8 ? 7 : 1);
1927 g = av_clip(g, 0, isrgb8 ? 7 : 3);
1928 b = av_clip(b, 0, isrgb8 ? 3 : 1);
1929 err[0] = R - r*(isrgb8 ? 36 : 255);
1930 err[1] = G - g*(isrgb8 ? 36 : 85);
1931 err[2] = B - b*(isrgb8 ? 85 : 255);
1933 case SWS_DITHER_A_DITHER:
1935 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1936 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1937 r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1938 g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1939 b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1940 r = av_clip_uintp2(r, 3);
1941 g = av_clip_uintp2(g, 3);
1942 b = av_clip_uintp2(b, 2);
1944 r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1945 g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1946 b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1947 r = av_clip_uintp2(r, 1);
1948 g = av_clip_uintp2(g, 2);
1949 b = av_clip_uintp2(b, 1);
1952 case SWS_DITHER_X_DITHER:
1954 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1955 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1956 r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1957 g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1958 b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1959 r = av_clip_uintp2(r, 3);
1960 g = av_clip_uintp2(g, 3);
1961 b = av_clip_uintp2(b, 2);
1963 r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1964 g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1965 b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1966 r = av_clip_uintp2(r, 1);
1967 g = av_clip_uintp2(g, 2);
1968 b = av_clip_uintp2(b, 1);
1974 if(target == AV_PIX_FMT_BGR4_BYTE) {
1975 dest[0] = r + 2*g + 8*b;
1976 } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1977 dest[0] = b + 2*g + 8*r;
1978 } else if(target == AV_PIX_FMT_BGR8) {
1979 dest[0] = r + 8*g + 64*b;
1980 } else if(target == AV_PIX_FMT_RGB8) {
1981 dest[0] = b + 4*g + 32*r;
1988 static av_always_inline void
1989 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1990 const int16_t **lumSrc, int lumFilterSize,
1991 const int16_t *chrFilter, const int16_t **chrUSrc,
1992 const int16_t **chrVSrc, int chrFilterSize,
1993 const int16_t **alpSrc, uint8_t *dest,
1994 int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1997 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1999 int A = 0; //init to silence warning
2001 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2002 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2005 for (i = 0; i < dstW; i++) {
2008 int U = (1<<9)-(128 << 19);
2009 int V = (1<<9)-(128 << 19);
2011 for (j = 0; j < lumFilterSize; j++) {
2012 Y += lumSrc[j][i] * lumFilter[j];
2014 for (j = 0; j < chrFilterSize; j++) {
2015 U += chrUSrc[j][i] * chrFilter[j];
2016 V += chrVSrc[j][i] * chrFilter[j];
2023 for (j = 0; j < lumFilterSize; j++) {
2024 A += alpSrc[j][i] * lumFilter[j];
2028 A = av_clip_uint8(A);
2030 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2033 c->dither_error[0][i] = err[0];
2034 c->dither_error[1][i] = err[1];
2035 c->dither_error[2][i] = err[2];
2038 static av_always_inline void
2039 yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
2040 const int16_t *ubuf[2], const int16_t *vbuf[2],
2041 const int16_t *abuf[2], uint8_t *dest, int dstW,
2042 int yalpha, int uvalpha, int y,
2043 enum AVPixelFormat target, int hasAlpha)
2045 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2046 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
2047 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
2048 *abuf0 = hasAlpha ? abuf[0] : NULL,
2049 *abuf1 = hasAlpha ? abuf[1] : NULL;
2050 int yalpha1 = 4096 - yalpha;
2051 int uvalpha1 = 4096 - uvalpha;
2053 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2055 int A = 0; // init to silcene warning
2057 av_assert2(yalpha <= 4096U);
2058 av_assert2(uvalpha <= 4096U);
2060 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2061 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2064 for (i = 0; i < dstW; i++) {
2065 int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
2066 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
2067 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
2070 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
2072 A = av_clip_uint8(A);
2075 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2078 c->dither_error[0][i] = err[0];
2079 c->dither_error[1][i] = err[1];
2080 c->dither_error[2][i] = err[2];
2083 static av_always_inline void
2084 yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
2085 const int16_t *ubuf[2], const int16_t *vbuf[2],
2086 const int16_t *abuf0, uint8_t *dest, int dstW,
2087 int uvalpha, int y, enum AVPixelFormat target,
2090 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
2092 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2095 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2096 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2099 if (uvalpha < 2048) {
2100 int A = 0; //init to silence warning
2101 for (i = 0; i < dstW; i++) {
2102 int Y = buf0[i] * 4;
2103 int U = (ubuf0[i] - (128<<7)) * 4;
2104 int V = (vbuf0[i] - (128<<7)) * 4;
2107 A = (abuf0[i] + 64) >> 7;
2109 A = av_clip_uint8(A);
2112 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2116 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
2117 int A = 0; //init to silence warning
2118 for (i = 0; i < dstW; i++) {
2119 int Y = buf0[i] * 4;
2120 int U = (ubuf0[i] + ubuf1[i] - (128<<8)) * 2;
2121 int V = (vbuf0[i] + vbuf1[i] - (128<<8)) * 2;
2124 A = (abuf0[i] + 64) >> 7;
2126 A = av_clip_uint8(A);
2129 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2134 c->dither_error[0][i] = err[0];
2135 c->dither_error[1][i] = err[1];
2136 c->dither_error[2][i] = err[2];
2140 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2141 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2142 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2143 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2145 #if CONFIG_SWSCALE_ALPHA
2146 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
2147 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
2148 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
2149 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
2151 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
2152 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
2153 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
2154 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
2156 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
2157 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
2159 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
2160 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
2161 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
2162 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
2165 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
2166 const int16_t **lumSrc, int lumFilterSize,
2167 const int16_t *chrFilter, const int16_t **chrUSrc,
2168 const int16_t **chrVSrc, int chrFilterSize,
2169 const int16_t **alpSrc, uint8_t **dest,
2172 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2174 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
2175 uint16_t **dest16 = (uint16_t**)dest;
2176 int SH = 22 + 8 - desc->comp[0].depth;
2177 int A = 0; // init to silence warning
2179 for (i = 0; i < dstW; i++) {
2182 int U = (1 << 9) - (128 << 19);
2183 int V = (1 << 9) - (128 << 19);
2186 for (j = 0; j < lumFilterSize; j++)
2187 Y += lumSrc[j][i] * lumFilter[j];
2189 for (j = 0; j < chrFilterSize; j++) {
2190 U += chrUSrc[j][i] * chrFilter[j];
2191 V += chrVSrc[j][i] * chrFilter[j];
2201 for (j = 0; j < lumFilterSize; j++)
2202 A += alpSrc[j][i] * lumFilter[j];
2205 A = av_clip_uintp2(A, 27);
2208 Y -= c->yuv2rgb_y_offset;
2209 Y *= c->yuv2rgb_y_coeff;
2211 R = Y + V * c->yuv2rgb_v2r_coeff;
2212 G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2213 B = Y + U * c->yuv2rgb_u2b_coeff;
2215 if ((R | G | B) & 0xC0000000) {
2216 R = av_clip_uintp2(R, 30);
2217 G = av_clip_uintp2(G, 30);
2218 B = av_clip_uintp2(B, 30);
2222 dest16[0][i] = G >> SH;
2223 dest16[1][i] = B >> SH;
2224 dest16[2][i] = R >> SH;
2226 dest16[3][i] = A >> (SH - 3);
2228 dest[0][i] = G >> 22;
2229 dest[1][i] = B >> 22;
2230 dest[2][i] = R >> 22;
2232 dest[3][i] = A >> 19;
2235 if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2236 for (i = 0; i < dstW; i++) {
2237 dest16[0][i] = av_bswap16(dest16[0][i]);
2238 dest16[1][i] = av_bswap16(dest16[1][i]);
2239 dest16[2][i] = av_bswap16(dest16[2][i]);
2241 dest16[3][i] = av_bswap16(dest16[3][i]);
2247 yuv2gbrp16_full_X_c(SwsContext *c, const int16_t *lumFilter,
2248 const int16_t **lumSrcx, int lumFilterSize,
2249 const int16_t *chrFilter, const int16_t **chrUSrcx,
2250 const int16_t **chrVSrcx, int chrFilterSize,
2251 const int16_t **alpSrcx, uint8_t **dest,
2254 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2256 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2257 uint16_t **dest16 = (uint16_t**)dest;
2258 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2259 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2260 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2261 const int32_t **alpSrc = (const int32_t**)alpSrcx;
2263 for (i = 0; i < dstW; i++) {
2265 int Y = -0x40000000;
2266 int U = -(128 << 23);
2267 int V = -(128 << 23);
2270 for (j = 0; j < lumFilterSize; j++)
2271 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2273 for (j = 0; j < chrFilterSize; j++) {
2274 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2275 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2286 for (j = 0; j < lumFilterSize; j++)
2287 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2293 Y -= c->yuv2rgb_y_offset;
2294 Y *= c->yuv2rgb_y_coeff;
2296 R = V * c->yuv2rgb_v2r_coeff;
2297 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2298 B = U * c->yuv2rgb_u2b_coeff;
2300 R = av_clip_uintp2(Y + R, 30);
2301 G = av_clip_uintp2(Y + G, 30);
2302 B = av_clip_uintp2(Y + B, 30);
2304 dest16[0][i] = G >> 14;
2305 dest16[1][i] = B >> 14;
2306 dest16[2][i] = R >> 14;
2308 dest16[3][i] = av_clip_uintp2(A, 30) >> 14;
2310 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2311 for (i = 0; i < dstW; i++) {
2312 dest16[0][i] = av_bswap16(dest16[0][i]);
2313 dest16[1][i] = av_bswap16(dest16[1][i]);
2314 dest16[2][i] = av_bswap16(dest16[2][i]);
2316 dest16[3][i] = av_bswap16(dest16[3][i]);
2322 yuv2gbrpf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
2323 const int16_t **lumSrcx, int lumFilterSize,
2324 const int16_t *chrFilter, const int16_t **chrUSrcx,
2325 const int16_t **chrVSrcx, int chrFilterSize,
2326 const int16_t **alpSrcx, uint8_t **dest,
2329 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2331 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2332 uint32_t **dest32 = (uint32_t**)dest;
2333 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2334 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2335 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2336 const int32_t **alpSrc = (const int32_t**)alpSrcx;
2337 static const float float_mult = 1.0f / 65535.0f;
2339 for (i = 0; i < dstW; i++) {
2341 int Y = -0x40000000;
2342 int U = -(128 << 23);
2343 int V = -(128 << 23);
2346 for (j = 0; j < lumFilterSize; j++)
2347 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2349 for (j = 0; j < chrFilterSize; j++) {
2350 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2351 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2362 for (j = 0; j < lumFilterSize; j++)
2363 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2369 Y -= c->yuv2rgb_y_offset;
2370 Y *= c->yuv2rgb_y_coeff;
2372 R = V * c->yuv2rgb_v2r_coeff;
2373 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2374 B = U * c->yuv2rgb_u2b_coeff;
2376 R = av_clip_uintp2(Y + R, 30);
2377 G = av_clip_uintp2(Y + G, 30);
2378 B = av_clip_uintp2(Y + B, 30);
2380 dest32[0][i] = av_float2int(float_mult * (float)(G >> 14));
2381 dest32[1][i] = av_float2int(float_mult * (float)(B >> 14));
2382 dest32[2][i] = av_float2int(float_mult * (float)(R >> 14));
2384 dest32[3][i] = av_float2int(float_mult * (float)(av_clip_uintp2(A, 30) >> 14));
2386 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2387 for (i = 0; i < dstW; i++) {
2388 dest32[0][i] = av_bswap32(dest32[0][i]);
2389 dest32[1][i] = av_bswap32(dest32[1][i]);
2390 dest32[2][i] = av_bswap32(dest32[2][i]);
2392 dest32[3][i] = av_bswap32(dest32[3][i]);
2398 yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
2399 const int16_t *ubuf[2], const int16_t *vbuf[2],
2400 const int16_t *abuf0, uint8_t *dest, int dstW,
2403 int hasAlpha = !!abuf0;
2406 for (i = 0; i < dstW; i++) {
2407 int Y = (buf0[i] + 64) >> 7;
2410 Y = av_clip_uint8(Y);
2413 A = (abuf0[i] + 64) >> 7;
2415 A = av_clip_uint8(A);
2419 dest[i * 2 + 1] = hasAlpha ? A : 255;
2424 yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
2425 const int16_t *ubuf[2], const int16_t *vbuf[2],
2426 const int16_t *abuf[2], uint8_t *dest, int dstW,
2427 int yalpha, int uvalpha, int y)
2429 int hasAlpha = abuf && abuf[0] && abuf[1];
2430 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2431 *abuf0 = hasAlpha ? abuf[0] : NULL,
2432 *abuf1 = hasAlpha ? abuf[1] : NULL;
2433 int yalpha1 = 4096 - yalpha;
2436 av_assert2(yalpha <= 4096U);
2438 for (i = 0; i < dstW; i++) {
2439 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 19;
2442 Y = av_clip_uint8(Y);
2445 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 19;
2446 A = av_clip_uint8(A);
2450 dest[i * 2 + 1] = hasAlpha ? A : 255;
2455 yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
2456 const int16_t **lumSrc, int lumFilterSize,
2457 const int16_t *chrFilter, const int16_t **chrUSrc,
2458 const int16_t **chrVSrc, int chrFilterSize,
2459 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
2461 int hasAlpha = !!alpSrc;
2464 for (i = 0; i < dstW; i++) {
2466 int Y = 1 << 18, A = 1 << 18;
2468 for (j = 0; j < lumFilterSize; j++)
2469 Y += lumSrc[j][i] * lumFilter[j];
2473 Y = av_clip_uint8(Y);
2476 for (j = 0; j < lumFilterSize; j++)
2477 A += alpSrc[j][i] * lumFilter[j];
2482 A = av_clip_uint8(A);
2486 dest[2 * i + 1] = hasAlpha ? A : 255;
2491 yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter,
2492 const int16_t **_lumSrc, int lumFilterSize,
2493 const int16_t *chrFilter, const int16_t **_chrUSrc,
2494 const int16_t **_chrVSrc, int chrFilterSize,
2495 const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
2497 const int32_t **lumSrc = (const int32_t **) _lumSrc,
2498 **chrUSrc = (const int32_t **) _chrUSrc,
2499 **chrVSrc = (const int32_t **) _chrVSrc,
2500 **alpSrc = (const int32_t **) _alpSrc;
2501 int hasAlpha = !!alpSrc;
2504 for (i = 0; i < dstW; i++) {
2505 int Y = 1 << 14, U = 1 << 14;
2506 int V = 1 << 14, A = 1 << 14;
2514 for (j = 0; j < lumFilterSize; j++)
2515 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2517 for (j = 0; j < chrFilterSize; j++)
2518 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2520 for (j = 0; j < chrFilterSize; j++)
2521 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2524 for (j = 0; j < lumFilterSize; j++)
2525 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2527 Y = 0x8000 + av_clip_int16(Y >> 15);
2528 U = 0x8000 + av_clip_int16(U >> 15);
2529 V = 0x8000 + av_clip_int16(V >> 15);
2530 A = 0x8000 + av_clip_int16(A >> 15);
2532 AV_WL16(dest + 8 * i, hasAlpha ? A : 65535);
2533 AV_WL16(dest + 8 * i + 2, Y);
2534 AV_WL16(dest + 8 * i + 4, U);
2535 AV_WL16(dest + 8 * i + 6, V);
2539 av_cold void ff_sws_init_output_funcs(SwsContext *c,
2540 yuv2planar1_fn *yuv2plane1,
2541 yuv2planarX_fn *yuv2planeX,
2542 yuv2interleavedX_fn *yuv2nv12cX,
2543 yuv2packed1_fn *yuv2packed1,
2544 yuv2packed2_fn *yuv2packed2,
2545 yuv2packedX_fn *yuv2packedX,
2546 yuv2anyX_fn *yuv2anyX)
2548 enum AVPixelFormat dstFormat = c->dstFormat;
2549 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2551 if (dstFormat == AV_PIX_FMT_P010LE || dstFormat == AV_PIX_FMT_P010BE) {
2552 *yuv2plane1 = isBE(dstFormat) ? yuv2p010l1_BE_c : yuv2p010l1_LE_c;
2553 *yuv2planeX = isBE(dstFormat) ? yuv2p010lX_BE_c : yuv2p010lX_LE_c;
2554 *yuv2nv12cX = yuv2p010cX_c;
2555 } else if (is16BPS(dstFormat)) {
2556 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
2557 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
2558 if (dstFormat == AV_PIX_FMT_P016LE || dstFormat == AV_PIX_FMT_P016BE) {
2559 *yuv2nv12cX = yuv2p016cX_c;
2561 } else if (isNBPS(dstFormat)) {
2562 if (desc->comp[0].depth == 9) {
2563 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
2564 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
2565 } else if (desc->comp[0].depth == 10) {
2566 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
2567 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
2568 } else if (desc->comp[0].depth == 12) {
2569 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
2570 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
2571 } else if (desc->comp[0].depth == 14) {
2572 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
2573 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
2576 } else if (dstFormat == AV_PIX_FMT_GRAYF32BE) {
2577 *yuv2planeX = yuv2planeX_floatBE_c;
2578 *yuv2plane1 = yuv2plane1_floatBE_c;
2579 } else if (dstFormat == AV_PIX_FMT_GRAYF32LE) {
2580 *yuv2planeX = yuv2planeX_floatLE_c;
2581 *yuv2plane1 = yuv2plane1_floatLE_c;
2583 *yuv2plane1 = yuv2plane1_8_c;
2584 *yuv2planeX = yuv2planeX_8_c;
2585 if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21 ||
2586 dstFormat == AV_PIX_FMT_NV24 || dstFormat == AV_PIX_FMT_NV42)
2587 *yuv2nv12cX = yuv2nv12cX_c;
2590 if(c->flags & SWS_FULL_CHR_H_INT) {
2591 switch (dstFormat) {
2592 case AV_PIX_FMT_RGBA:
2594 *yuv2packedX = yuv2rgba32_full_X_c;
2595 *yuv2packed2 = yuv2rgba32_full_2_c;
2596 *yuv2packed1 = yuv2rgba32_full_1_c;
2598 #if CONFIG_SWSCALE_ALPHA
2600 *yuv2packedX = yuv2rgba32_full_X_c;
2601 *yuv2packed2 = yuv2rgba32_full_2_c;
2602 *yuv2packed1 = yuv2rgba32_full_1_c;
2604 #endif /* CONFIG_SWSCALE_ALPHA */
2606 *yuv2packedX = yuv2rgbx32_full_X_c;
2607 *yuv2packed2 = yuv2rgbx32_full_2_c;
2608 *yuv2packed1 = yuv2rgbx32_full_1_c;
2610 #endif /* !CONFIG_SMALL */
2612 case AV_PIX_FMT_ARGB:
2614 *yuv2packedX = yuv2argb32_full_X_c;
2615 *yuv2packed2 = yuv2argb32_full_2_c;
2616 *yuv2packed1 = yuv2argb32_full_1_c;
2618 #if CONFIG_SWSCALE_ALPHA
2620 *yuv2packedX = yuv2argb32_full_X_c;
2621 *yuv2packed2 = yuv2argb32_full_2_c;
2622 *yuv2packed1 = yuv2argb32_full_1_c;
2624 #endif /* CONFIG_SWSCALE_ALPHA */
2626 *yuv2packedX = yuv2xrgb32_full_X_c;
2627 *yuv2packed2 = yuv2xrgb32_full_2_c;
2628 *yuv2packed1 = yuv2xrgb32_full_1_c;
2630 #endif /* !CONFIG_SMALL */
2632 case AV_PIX_FMT_BGRA:
2634 *yuv2packedX = yuv2bgra32_full_X_c;
2635 *yuv2packed2 = yuv2bgra32_full_2_c;
2636 *yuv2packed1 = yuv2bgra32_full_1_c;
2638 #if CONFIG_SWSCALE_ALPHA
2640 *yuv2packedX = yuv2bgra32_full_X_c;
2641 *yuv2packed2 = yuv2bgra32_full_2_c;
2642 *yuv2packed1 = yuv2bgra32_full_1_c;
2644 #endif /* CONFIG_SWSCALE_ALPHA */
2646 *yuv2packedX = yuv2bgrx32_full_X_c;
2647 *yuv2packed2 = yuv2bgrx32_full_2_c;
2648 *yuv2packed1 = yuv2bgrx32_full_1_c;
2650 #endif /* !CONFIG_SMALL */
2652 case AV_PIX_FMT_ABGR:
2654 *yuv2packedX = yuv2abgr32_full_X_c;
2655 *yuv2packed2 = yuv2abgr32_full_2_c;
2656 *yuv2packed1 = yuv2abgr32_full_1_c;
2658 #if CONFIG_SWSCALE_ALPHA
2660 *yuv2packedX = yuv2abgr32_full_X_c;
2661 *yuv2packed2 = yuv2abgr32_full_2_c;
2662 *yuv2packed1 = yuv2abgr32_full_1_c;
2664 #endif /* CONFIG_SWSCALE_ALPHA */
2666 *yuv2packedX = yuv2xbgr32_full_X_c;
2667 *yuv2packed2 = yuv2xbgr32_full_2_c;
2668 *yuv2packed1 = yuv2xbgr32_full_1_c;
2670 #endif /* !CONFIG_SMALL */
2672 case AV_PIX_FMT_RGBA64LE:
2673 #if CONFIG_SWSCALE_ALPHA
2675 *yuv2packedX = yuv2rgba64le_full_X_c;
2676 *yuv2packed2 = yuv2rgba64le_full_2_c;
2677 *yuv2packed1 = yuv2rgba64le_full_1_c;
2679 #endif /* CONFIG_SWSCALE_ALPHA */
2681 *yuv2packedX = yuv2rgbx64le_full_X_c;
2682 *yuv2packed2 = yuv2rgbx64le_full_2_c;
2683 *yuv2packed1 = yuv2rgbx64le_full_1_c;
2686 case AV_PIX_FMT_RGBA64BE:
2687 #if CONFIG_SWSCALE_ALPHA
2689 *yuv2packedX = yuv2rgba64be_full_X_c;
2690 *yuv2packed2 = yuv2rgba64be_full_2_c;
2691 *yuv2packed1 = yuv2rgba64be_full_1_c;
2693 #endif /* CONFIG_SWSCALE_ALPHA */
2695 *yuv2packedX = yuv2rgbx64be_full_X_c;
2696 *yuv2packed2 = yuv2rgbx64be_full_2_c;
2697 *yuv2packed1 = yuv2rgbx64be_full_1_c;
2700 case AV_PIX_FMT_BGRA64LE:
2701 #if CONFIG_SWSCALE_ALPHA
2703 *yuv2packedX = yuv2bgra64le_full_X_c;
2704 *yuv2packed2 = yuv2bgra64le_full_2_c;
2705 *yuv2packed1 = yuv2bgra64le_full_1_c;
2707 #endif /* CONFIG_SWSCALE_ALPHA */
2709 *yuv2packedX = yuv2bgrx64le_full_X_c;
2710 *yuv2packed2 = yuv2bgrx64le_full_2_c;
2711 *yuv2packed1 = yuv2bgrx64le_full_1_c;
2714 case AV_PIX_FMT_BGRA64BE:
2715 #if CONFIG_SWSCALE_ALPHA
2717 *yuv2packedX = yuv2bgra64be_full_X_c;
2718 *yuv2packed2 = yuv2bgra64be_full_2_c;
2719 *yuv2packed1 = yuv2bgra64be_full_1_c;
2721 #endif /* CONFIG_SWSCALE_ALPHA */
2723 *yuv2packedX = yuv2bgrx64be_full_X_c;
2724 *yuv2packed2 = yuv2bgrx64be_full_2_c;
2725 *yuv2packed1 = yuv2bgrx64be_full_1_c;
2729 case AV_PIX_FMT_RGB24:
2730 *yuv2packedX = yuv2rgb24_full_X_c;
2731 *yuv2packed2 = yuv2rgb24_full_2_c;
2732 *yuv2packed1 = yuv2rgb24_full_1_c;
2734 case AV_PIX_FMT_BGR24:
2735 *yuv2packedX = yuv2bgr24_full_X_c;
2736 *yuv2packed2 = yuv2bgr24_full_2_c;
2737 *yuv2packed1 = yuv2bgr24_full_1_c;
2739 case AV_PIX_FMT_RGB48LE:
2740 *yuv2packedX = yuv2rgb48le_full_X_c;
2741 *yuv2packed2 = yuv2rgb48le_full_2_c;
2742 *yuv2packed1 = yuv2rgb48le_full_1_c;
2744 case AV_PIX_FMT_BGR48LE:
2745 *yuv2packedX = yuv2bgr48le_full_X_c;
2746 *yuv2packed2 = yuv2bgr48le_full_2_c;
2747 *yuv2packed1 = yuv2bgr48le_full_1_c;
2749 case AV_PIX_FMT_RGB48BE:
2750 *yuv2packedX = yuv2rgb48be_full_X_c;
2751 *yuv2packed2 = yuv2rgb48be_full_2_c;
2752 *yuv2packed1 = yuv2rgb48be_full_1_c;
2754 case AV_PIX_FMT_BGR48BE:
2755 *yuv2packedX = yuv2bgr48be_full_X_c;
2756 *yuv2packed2 = yuv2bgr48be_full_2_c;
2757 *yuv2packed1 = yuv2bgr48be_full_1_c;
2759 case AV_PIX_FMT_BGR4_BYTE:
2760 *yuv2packedX = yuv2bgr4_byte_full_X_c;
2761 *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2762 *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2764 case AV_PIX_FMT_RGB4_BYTE:
2765 *yuv2packedX = yuv2rgb4_byte_full_X_c;
2766 *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2767 *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2769 case AV_PIX_FMT_BGR8:
2770 *yuv2packedX = yuv2bgr8_full_X_c;
2771 *yuv2packed2 = yuv2bgr8_full_2_c;
2772 *yuv2packed1 = yuv2bgr8_full_1_c;
2774 case AV_PIX_FMT_RGB8:
2775 *yuv2packedX = yuv2rgb8_full_X_c;
2776 *yuv2packed2 = yuv2rgb8_full_2_c;
2777 *yuv2packed1 = yuv2rgb8_full_1_c;
2779 case AV_PIX_FMT_GBRP:
2780 case AV_PIX_FMT_GBRP9BE:
2781 case AV_PIX_FMT_GBRP9LE:
2782 case AV_PIX_FMT_GBRP10BE:
2783 case AV_PIX_FMT_GBRP10LE:
2784 case AV_PIX_FMT_GBRP12BE:
2785 case AV_PIX_FMT_GBRP12LE:
2786 case AV_PIX_FMT_GBRP14BE:
2787 case AV_PIX_FMT_GBRP14LE:
2788 case AV_PIX_FMT_GBRAP:
2789 case AV_PIX_FMT_GBRAP10BE:
2790 case AV_PIX_FMT_GBRAP10LE:
2791 case AV_PIX_FMT_GBRAP12BE:
2792 case AV_PIX_FMT_GBRAP12LE:
2793 *yuv2anyX = yuv2gbrp_full_X_c;
2795 case AV_PIX_FMT_GBRP16BE:
2796 case AV_PIX_FMT_GBRP16LE:
2797 case AV_PIX_FMT_GBRAP16BE:
2798 case AV_PIX_FMT_GBRAP16LE:
2799 *yuv2anyX = yuv2gbrp16_full_X_c;
2801 case AV_PIX_FMT_GBRPF32BE:
2802 case AV_PIX_FMT_GBRPF32LE:
2803 case AV_PIX_FMT_GBRAPF32BE:
2804 case AV_PIX_FMT_GBRAPF32LE:
2805 *yuv2anyX = yuv2gbrpf32_full_X_c;
2808 if (!*yuv2packedX && !*yuv2anyX)
2812 switch (dstFormat) {
2813 case AV_PIX_FMT_RGBA64LE:
2814 #if CONFIG_SWSCALE_ALPHA
2816 *yuv2packed1 = yuv2rgba64le_1_c;
2817 *yuv2packed2 = yuv2rgba64le_2_c;
2818 *yuv2packedX = yuv2rgba64le_X_c;
2820 #endif /* CONFIG_SWSCALE_ALPHA */
2822 *yuv2packed1 = yuv2rgbx64le_1_c;
2823 *yuv2packed2 = yuv2rgbx64le_2_c;
2824 *yuv2packedX = yuv2rgbx64le_X_c;
2827 case AV_PIX_FMT_RGBA64BE:
2828 #if CONFIG_SWSCALE_ALPHA
2830 *yuv2packed1 = yuv2rgba64be_1_c;
2831 *yuv2packed2 = yuv2rgba64be_2_c;
2832 *yuv2packedX = yuv2rgba64be_X_c;
2834 #endif /* CONFIG_SWSCALE_ALPHA */
2836 *yuv2packed1 = yuv2rgbx64be_1_c;
2837 *yuv2packed2 = yuv2rgbx64be_2_c;
2838 *yuv2packedX = yuv2rgbx64be_X_c;
2841 case AV_PIX_FMT_BGRA64LE:
2842 #if CONFIG_SWSCALE_ALPHA
2844 *yuv2packed1 = yuv2bgra64le_1_c;
2845 *yuv2packed2 = yuv2bgra64le_2_c;
2846 *yuv2packedX = yuv2bgra64le_X_c;
2848 #endif /* CONFIG_SWSCALE_ALPHA */
2850 *yuv2packed1 = yuv2bgrx64le_1_c;
2851 *yuv2packed2 = yuv2bgrx64le_2_c;
2852 *yuv2packedX = yuv2bgrx64le_X_c;
2855 case AV_PIX_FMT_BGRA64BE:
2856 #if CONFIG_SWSCALE_ALPHA
2858 *yuv2packed1 = yuv2bgra64be_1_c;
2859 *yuv2packed2 = yuv2bgra64be_2_c;
2860 *yuv2packedX = yuv2bgra64be_X_c;
2862 #endif /* CONFIG_SWSCALE_ALPHA */
2864 *yuv2packed1 = yuv2bgrx64be_1_c;
2865 *yuv2packed2 = yuv2bgrx64be_2_c;
2866 *yuv2packedX = yuv2bgrx64be_X_c;
2869 case AV_PIX_FMT_RGB48LE:
2870 *yuv2packed1 = yuv2rgb48le_1_c;
2871 *yuv2packed2 = yuv2rgb48le_2_c;
2872 *yuv2packedX = yuv2rgb48le_X_c;
2874 case AV_PIX_FMT_RGB48BE:
2875 *yuv2packed1 = yuv2rgb48be_1_c;
2876 *yuv2packed2 = yuv2rgb48be_2_c;
2877 *yuv2packedX = yuv2rgb48be_X_c;
2879 case AV_PIX_FMT_BGR48LE:
2880 *yuv2packed1 = yuv2bgr48le_1_c;
2881 *yuv2packed2 = yuv2bgr48le_2_c;
2882 *yuv2packedX = yuv2bgr48le_X_c;
2884 case AV_PIX_FMT_BGR48BE:
2885 *yuv2packed1 = yuv2bgr48be_1_c;
2886 *yuv2packed2 = yuv2bgr48be_2_c;
2887 *yuv2packedX = yuv2bgr48be_X_c;
2889 case AV_PIX_FMT_RGB32:
2890 case AV_PIX_FMT_BGR32:
2892 *yuv2packed1 = yuv2rgb32_1_c;
2893 *yuv2packed2 = yuv2rgb32_2_c;
2894 *yuv2packedX = yuv2rgb32_X_c;
2896 #if CONFIG_SWSCALE_ALPHA
2898 *yuv2packed1 = yuv2rgba32_1_c;
2899 *yuv2packed2 = yuv2rgba32_2_c;
2900 *yuv2packedX = yuv2rgba32_X_c;
2902 #endif /* CONFIG_SWSCALE_ALPHA */
2904 *yuv2packed1 = yuv2rgbx32_1_c;
2905 *yuv2packed2 = yuv2rgbx32_2_c;
2906 *yuv2packedX = yuv2rgbx32_X_c;
2908 #endif /* !CONFIG_SMALL */
2910 case AV_PIX_FMT_RGB32_1:
2911 case AV_PIX_FMT_BGR32_1:
2913 *yuv2packed1 = yuv2rgb32_1_1_c;
2914 *yuv2packed2 = yuv2rgb32_1_2_c;
2915 *yuv2packedX = yuv2rgb32_1_X_c;
2917 #if CONFIG_SWSCALE_ALPHA
2919 *yuv2packed1 = yuv2rgba32_1_1_c;
2920 *yuv2packed2 = yuv2rgba32_1_2_c;
2921 *yuv2packedX = yuv2rgba32_1_X_c;
2923 #endif /* CONFIG_SWSCALE_ALPHA */
2925 *yuv2packed1 = yuv2rgbx32_1_1_c;
2926 *yuv2packed2 = yuv2rgbx32_1_2_c;
2927 *yuv2packedX = yuv2rgbx32_1_X_c;
2929 #endif /* !CONFIG_SMALL */
2931 case AV_PIX_FMT_RGB24:
2932 *yuv2packed1 = yuv2rgb24_1_c;
2933 *yuv2packed2 = yuv2rgb24_2_c;
2934 *yuv2packedX = yuv2rgb24_X_c;
2936 case AV_PIX_FMT_BGR24:
2937 *yuv2packed1 = yuv2bgr24_1_c;
2938 *yuv2packed2 = yuv2bgr24_2_c;
2939 *yuv2packedX = yuv2bgr24_X_c;
2941 case AV_PIX_FMT_RGB565LE:
2942 case AV_PIX_FMT_RGB565BE:
2943 case AV_PIX_FMT_BGR565LE:
2944 case AV_PIX_FMT_BGR565BE:
2945 *yuv2packed1 = yuv2rgb16_1_c;
2946 *yuv2packed2 = yuv2rgb16_2_c;
2947 *yuv2packedX = yuv2rgb16_X_c;
2949 case AV_PIX_FMT_RGB555LE:
2950 case AV_PIX_FMT_RGB555BE:
2951 case AV_PIX_FMT_BGR555LE:
2952 case AV_PIX_FMT_BGR555BE:
2953 *yuv2packed1 = yuv2rgb15_1_c;
2954 *yuv2packed2 = yuv2rgb15_2_c;
2955 *yuv2packedX = yuv2rgb15_X_c;
2957 case AV_PIX_FMT_RGB444LE:
2958 case AV_PIX_FMT_RGB444BE:
2959 case AV_PIX_FMT_BGR444LE:
2960 case AV_PIX_FMT_BGR444BE:
2961 *yuv2packed1 = yuv2rgb12_1_c;
2962 *yuv2packed2 = yuv2rgb12_2_c;
2963 *yuv2packedX = yuv2rgb12_X_c;
2965 case AV_PIX_FMT_RGB8:
2966 case AV_PIX_FMT_BGR8:
2967 *yuv2packed1 = yuv2rgb8_1_c;
2968 *yuv2packed2 = yuv2rgb8_2_c;
2969 *yuv2packedX = yuv2rgb8_X_c;
2971 case AV_PIX_FMT_RGB4:
2972 case AV_PIX_FMT_BGR4:
2973 *yuv2packed1 = yuv2rgb4_1_c;
2974 *yuv2packed2 = yuv2rgb4_2_c;
2975 *yuv2packedX = yuv2rgb4_X_c;
2977 case AV_PIX_FMT_RGB4_BYTE:
2978 case AV_PIX_FMT_BGR4_BYTE:
2979 *yuv2packed1 = yuv2rgb4b_1_c;
2980 *yuv2packed2 = yuv2rgb4b_2_c;
2981 *yuv2packedX = yuv2rgb4b_X_c;
2983 case AV_PIX_FMT_X2RGB10LE:
2984 case AV_PIX_FMT_X2RGB10BE:
2985 *yuv2packed1 = yuv2x2rgb10_1_c;
2986 *yuv2packed2 = yuv2x2rgb10_2_c;
2987 *yuv2packedX = yuv2x2rgb10_X_c;
2991 switch (dstFormat) {
2992 case AV_PIX_FMT_MONOWHITE:
2993 *yuv2packed1 = yuv2monowhite_1_c;
2994 *yuv2packed2 = yuv2monowhite_2_c;
2995 *yuv2packedX = yuv2monowhite_X_c;
2997 case AV_PIX_FMT_MONOBLACK:
2998 *yuv2packed1 = yuv2monoblack_1_c;
2999 *yuv2packed2 = yuv2monoblack_2_c;
3000 *yuv2packedX = yuv2monoblack_X_c;
3002 case AV_PIX_FMT_YUYV422:
3003 *yuv2packed1 = yuv2yuyv422_1_c;
3004 *yuv2packed2 = yuv2yuyv422_2_c;
3005 *yuv2packedX = yuv2yuyv422_X_c;
3007 case AV_PIX_FMT_YVYU422:
3008 *yuv2packed1 = yuv2yvyu422_1_c;
3009 *yuv2packed2 = yuv2yvyu422_2_c;
3010 *yuv2packedX = yuv2yvyu422_X_c;
3012 case AV_PIX_FMT_UYVY422:
3013 *yuv2packed1 = yuv2uyvy422_1_c;
3014 *yuv2packed2 = yuv2uyvy422_2_c;
3015 *yuv2packedX = yuv2uyvy422_X_c;
3017 case AV_PIX_FMT_YA8:
3018 *yuv2packed1 = yuv2ya8_1_c;
3019 *yuv2packed2 = yuv2ya8_2_c;
3020 *yuv2packedX = yuv2ya8_X_c;
3022 case AV_PIX_FMT_YA16LE:
3023 *yuv2packed1 = yuv2ya16le_1_c;
3024 *yuv2packed2 = yuv2ya16le_2_c;
3025 *yuv2packedX = yuv2ya16le_X_c;
3027 case AV_PIX_FMT_YA16BE:
3028 *yuv2packed1 = yuv2ya16be_1_c;
3029 *yuv2packed2 = yuv2ya16be_2_c;
3030 *yuv2packedX = yuv2ya16be_X_c;
3032 case AV_PIX_FMT_AYUV64LE:
3033 *yuv2packedX = yuv2ayuv64le_X_c;