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(SwsContext *c, 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 = c->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(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
406 const int16_t **chrUSrc, const int16_t **chrVSrc,
407 uint8_t *dest, int chrDstW)
409 enum AVPixelFormat dstFormat = c->dstFormat;
410 const uint8_t *chrDither = c->chrDither8;
413 if (dstFormat == AV_PIX_FMT_NV12 ||
414 dstFormat == AV_PIX_FMT_NV24)
415 for (i=0; i<chrDstW; i++) {
416 int u = chrDither[i & 7] << 12;
417 int v = chrDither[(i + 3) & 7] << 12;
419 for (j=0; j<chrFilterSize; j++) {
420 u += chrUSrc[j][i] * chrFilter[j];
421 v += chrVSrc[j][i] * chrFilter[j];
424 dest[2*i]= av_clip_uint8(u>>19);
425 dest[2*i+1]= av_clip_uint8(v>>19);
428 for (i=0; i<chrDstW; i++) {
429 int u = chrDither[i & 7] << 12;
430 int v = chrDither[(i + 3) & 7] << 12;
432 for (j=0; j<chrFilterSize; j++) {
433 u += chrUSrc[j][i] * chrFilter[j];
434 v += chrVSrc[j][i] * chrFilter[j];
437 dest[2*i]= av_clip_uint8(v>>19);
438 dest[2*i+1]= av_clip_uint8(u>>19);
443 #define output_pixel(pos, val) \
445 AV_WB16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
447 AV_WL16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
450 static void yuv2p010l1_c(const int16_t *src,
451 uint16_t *dest, int dstW,
457 for (i = 0; i < dstW; i++) {
458 int val = src[i] + (1 << (shift - 1));
459 output_pixel(&dest[i], val);
463 static void yuv2p010lX_c(const int16_t *filter, int filterSize,
464 const int16_t **src, uint16_t *dest, int dstW,
470 for (i = 0; i < dstW; i++) {
471 int val = 1 << (shift - 1);
473 for (j = 0; j < filterSize; j++)
474 val += src[j][i] * filter[j];
476 output_pixel(&dest[i], val);
480 static void yuv2p010cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
481 const int16_t **chrUSrc, const int16_t **chrVSrc,
482 uint8_t *dest8, int chrDstW)
484 uint16_t *dest = (uint16_t*)dest8;
486 int big_endian = c->dstFormat == AV_PIX_FMT_P010BE;
489 for (i = 0; i < chrDstW; i++) {
490 int u = 1 << (shift - 1);
491 int v = 1 << (shift - 1);
493 for (j = 0; j < chrFilterSize; j++) {
494 u += chrUSrc[j][i] * chrFilter[j];
495 v += chrVSrc[j][i] * chrFilter[j];
498 output_pixel(&dest[2*i] , u);
499 output_pixel(&dest[2*i+1], v);
503 static void yuv2p010l1_LE_c(const int16_t *src,
504 uint8_t *dest, int dstW,
505 const uint8_t *dither, int offset)
507 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 0);
510 static void yuv2p010l1_BE_c(const int16_t *src,
511 uint8_t *dest, int dstW,
512 const uint8_t *dither, int offset)
514 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 1);
517 static void yuv2p010lX_LE_c(const int16_t *filter, int filterSize,
518 const int16_t **src, uint8_t *dest, int dstW,
519 const uint8_t *dither, int offset)
521 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 0);
524 static void yuv2p010lX_BE_c(const int16_t *filter, int filterSize,
525 const int16_t **src, uint8_t *dest, int dstW,
526 const uint8_t *dither, int offset)
528 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 1);
534 #define accumulate_bit(acc, val) \
537 #define output_pixel(pos, acc) \
538 if (target == AV_PIX_FMT_MONOBLACK) { \
544 static av_always_inline void
545 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
546 const int16_t **lumSrc, int lumFilterSize,
547 const int16_t *chrFilter, const int16_t **chrUSrc,
548 const int16_t **chrVSrc, int chrFilterSize,
549 const int16_t **alpSrc, uint8_t *dest, int dstW,
550 int y, enum AVPixelFormat target)
552 const uint8_t * const d128 = ff_dither_8x8_220[y&7];
557 for (i = 0; i < dstW; i += 2) {
562 for (j = 0; j < lumFilterSize; j++) {
563 Y1 += lumSrc[j][i] * lumFilter[j];
564 Y2 += lumSrc[j][i+1] * lumFilter[j];
568 if ((Y1 | Y2) & 0x100) {
569 Y1 = av_clip_uint8(Y1);
570 Y2 = av_clip_uint8(Y2);
572 if (c->dither == SWS_DITHER_ED) {
573 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;
574 c->dither_error[0][i] = err;
575 acc = 2*acc + (Y1 >= 128);
578 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);
579 c->dither_error[0][i+1] = Y1;
580 acc = 2*acc + (err >= 128);
583 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
584 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
587 output_pixel(*dest++, acc);
590 c->dither_error[0][i] = err;
593 output_pixel(*dest, acc);
597 static av_always_inline void
598 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
599 const int16_t *ubuf[2], const int16_t *vbuf[2],
600 const int16_t *abuf[2], uint8_t *dest, int dstW,
601 int yalpha, int uvalpha, int y,
602 enum AVPixelFormat target)
604 const int16_t *buf0 = buf[0], *buf1 = buf[1];
605 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
606 int yalpha1 = 4096 - yalpha;
608 av_assert2(yalpha <= 4096U);
610 if (c->dither == SWS_DITHER_ED) {
613 for (i = 0; i < dstW; i +=2) {
616 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
617 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;
618 c->dither_error[0][i] = err;
619 acc = 2*acc + (Y >= 128);
622 err = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
623 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;
624 c->dither_error[0][i+1] = Y;
625 acc = 2*acc + (err >= 128);
629 output_pixel(*dest++, acc);
631 c->dither_error[0][i] = err;
633 for (i = 0; i < dstW; i += 8) {
636 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
637 accumulate_bit(acc, Y + d128[0]);
638 Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
639 accumulate_bit(acc, Y + d128[1]);
640 Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
641 accumulate_bit(acc, Y + d128[2]);
642 Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
643 accumulate_bit(acc, Y + d128[3]);
644 Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
645 accumulate_bit(acc, Y + d128[4]);
646 Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
647 accumulate_bit(acc, Y + d128[5]);
648 Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
649 accumulate_bit(acc, Y + d128[6]);
650 Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
651 accumulate_bit(acc, Y + d128[7]);
653 output_pixel(*dest++, acc);
658 static av_always_inline void
659 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
660 const int16_t *ubuf[2], const int16_t *vbuf[2],
661 const int16_t *abuf0, uint8_t *dest, int dstW,
662 int uvalpha, int y, enum AVPixelFormat target)
664 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
667 if (c->dither == SWS_DITHER_ED) {
670 for (i = 0; i < dstW; i +=2) {
673 Y = ((buf0[i + 0] + 64) >> 7);
674 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;
675 c->dither_error[0][i] = err;
676 acc = 2*acc + (Y >= 128);
679 err = ((buf0[i + 1] + 64) >> 7);
680 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;
681 c->dither_error[0][i+1] = Y;
682 acc = 2*acc + (err >= 128);
686 output_pixel(*dest++, acc);
688 c->dither_error[0][i] = err;
690 for (i = 0; i < dstW; i += 8) {
692 accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
693 accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
694 accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
695 accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
696 accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
697 accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
698 accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
699 accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
701 output_pixel(*dest++, acc);
707 #undef accumulate_bit
709 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
710 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
711 const int16_t **lumSrc, int lumFilterSize, \
712 const int16_t *chrFilter, const int16_t **chrUSrc, \
713 const int16_t **chrVSrc, int chrFilterSize, \
714 const int16_t **alpSrc, uint8_t *dest, int dstW, \
717 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
718 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
719 alpSrc, dest, dstW, y, fmt); \
722 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
723 const int16_t *ubuf[2], const int16_t *vbuf[2], \
724 const int16_t *abuf[2], uint8_t *dest, int dstW, \
725 int yalpha, int uvalpha, int y) \
727 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
728 dest, dstW, yalpha, uvalpha, y, fmt); \
731 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
732 const int16_t *ubuf[2], const int16_t *vbuf[2], \
733 const int16_t *abuf0, uint8_t *dest, int dstW, \
734 int uvalpha, int y) \
736 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
737 abuf0, dest, dstW, uvalpha, \
741 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
742 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
744 #define output_pixels(pos, Y1, U, Y2, V) \
745 if (target == AV_PIX_FMT_YUYV422) { \
746 dest[pos + 0] = Y1; \
748 dest[pos + 2] = Y2; \
750 } else if (target == AV_PIX_FMT_YVYU422) { \
751 dest[pos + 0] = Y1; \
753 dest[pos + 2] = Y2; \
755 } else { /* AV_PIX_FMT_UYVY422 */ \
757 dest[pos + 1] = Y1; \
759 dest[pos + 3] = Y2; \
762 static av_always_inline void
763 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
764 const int16_t **lumSrc, int lumFilterSize,
765 const int16_t *chrFilter, const int16_t **chrUSrc,
766 const int16_t **chrVSrc, int chrFilterSize,
767 const int16_t **alpSrc, uint8_t *dest, int dstW,
768 int y, enum AVPixelFormat target)
772 for (i = 0; i < ((dstW + 1) >> 1); i++) {
779 for (j = 0; j < lumFilterSize; j++) {
780 Y1 += lumSrc[j][i * 2] * lumFilter[j];
781 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
783 for (j = 0; j < chrFilterSize; j++) {
784 U += chrUSrc[j][i] * chrFilter[j];
785 V += chrVSrc[j][i] * chrFilter[j];
791 if ((Y1 | Y2 | U | V) & 0x100) {
792 Y1 = av_clip_uint8(Y1);
793 Y2 = av_clip_uint8(Y2);
794 U = av_clip_uint8(U);
795 V = av_clip_uint8(V);
797 output_pixels(4*i, Y1, U, Y2, V);
801 static av_always_inline void
802 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
803 const int16_t *ubuf[2], const int16_t *vbuf[2],
804 const int16_t *abuf[2], uint8_t *dest, int dstW,
805 int yalpha, int uvalpha, int y,
806 enum AVPixelFormat target)
808 const int16_t *buf0 = buf[0], *buf1 = buf[1],
809 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
810 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
811 int yalpha1 = 4096 - yalpha;
812 int uvalpha1 = 4096 - uvalpha;
814 av_assert2(yalpha <= 4096U);
815 av_assert2(uvalpha <= 4096U);
817 for (i = 0; i < ((dstW + 1) >> 1); i++) {
818 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
819 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
820 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
821 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
823 if ((Y1 | Y2 | U | V) & 0x100) {
824 Y1 = av_clip_uint8(Y1);
825 Y2 = av_clip_uint8(Y2);
826 U = av_clip_uint8(U);
827 V = av_clip_uint8(V);
830 output_pixels(i * 4, Y1, U, Y2, V);
834 static av_always_inline void
835 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
836 const int16_t *ubuf[2], const int16_t *vbuf[2],
837 const int16_t *abuf0, uint8_t *dest, int dstW,
838 int uvalpha, int y, enum AVPixelFormat target)
840 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
843 if (uvalpha < 2048) {
844 for (i = 0; i < ((dstW + 1) >> 1); i++) {
845 int Y1 = (buf0[i * 2 ]+64) >> 7;
846 int Y2 = (buf0[i * 2 + 1]+64) >> 7;
847 int U = (ubuf0[i] +64) >> 7;
848 int V = (vbuf0[i] +64) >> 7;
850 if ((Y1 | Y2 | U | V) & 0x100) {
851 Y1 = av_clip_uint8(Y1);
852 Y2 = av_clip_uint8(Y2);
853 U = av_clip_uint8(U);
854 V = av_clip_uint8(V);
857 output_pixels(i * 4, Y1, U, Y2, V);
860 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
861 for (i = 0; i < ((dstW + 1) >> 1); i++) {
862 int Y1 = (buf0[i * 2 ] + 64) >> 7;
863 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
864 int U = (ubuf0[i] + ubuf1[i]+128) >> 8;
865 int V = (vbuf0[i] + vbuf1[i]+128) >> 8;
867 if ((Y1 | Y2 | U | V) & 0x100) {
868 Y1 = av_clip_uint8(Y1);
869 Y2 = av_clip_uint8(Y2);
870 U = av_clip_uint8(U);
871 V = av_clip_uint8(V);
874 output_pixels(i * 4, Y1, U, Y2, V);
881 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422)
882 YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422)
883 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422)
885 #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)
886 #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)
887 #define output_pixel(pos, val) \
888 if (isBE(target)) { \
894 static av_always_inline void
895 yuv2ya16_X_c_template(SwsContext *c, const int16_t *lumFilter,
896 const int32_t **lumSrc, int lumFilterSize,
897 const int16_t *chrFilter, const int32_t **unused_chrUSrc,
898 const int32_t **unused_chrVSrc, int unused_chrFilterSize,
899 const int32_t **alpSrc, uint16_t *dest, int dstW,
900 int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
902 int hasAlpha = !!alpSrc;
905 for (i = 0; i < dstW; i++) {
908 int64_t A = 0xffff<<14;
910 for (j = 0; j < lumFilterSize; j++)
911 Y += lumSrc[j][i] * lumFilter[j];
914 Y = av_clip_uint16(Y);
917 for (j = 0; j < lumFilterSize; j++)
918 A += alpSrc[j][i] * lumFilter[j];
921 A = av_clip_uint16(A);
924 output_pixel(&dest[2 * i ], Y);
925 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
929 static av_always_inline void
930 yuv2ya16_2_c_template(SwsContext *c, const int32_t *buf[2],
931 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
932 const int32_t *abuf[2], uint16_t *dest, int dstW,
933 int yalpha, int unused_uvalpha, int y,
934 enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
936 int hasAlpha = abuf && abuf[0] && abuf[1];
937 const int32_t *buf0 = buf[0], *buf1 = buf[1],
938 *abuf0 = hasAlpha ? abuf[0] : NULL,
939 *abuf1 = hasAlpha ? abuf[1] : NULL;
940 int yalpha1 = 4096 - yalpha;
943 av_assert2(yalpha <= 4096U);
945 for (i = 0; i < dstW; i++) {
946 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 15;
949 Y = av_clip_uint16(Y);
952 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 15;
953 A = av_clip_uint16(A);
956 output_pixel(&dest[2 * i ], Y);
957 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
961 static av_always_inline void
962 yuv2ya16_1_c_template(SwsContext *c, const int32_t *buf0,
963 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
964 const int32_t *abuf0, uint16_t *dest, int dstW,
965 int unused_uvalpha, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
967 int hasAlpha = !!abuf0;
970 for (i = 0; i < dstW; i++) {
971 int Y = buf0[i] >> 3;/* 19 - 16 */
974 Y = av_clip_uint16(Y);
979 A = av_clip_uint16(A);
982 output_pixel(&dest[2 * i ], Y);
983 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
987 static av_always_inline void
988 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
989 const int32_t **lumSrc, int lumFilterSize,
990 const int16_t *chrFilter, const int32_t **chrUSrc,
991 const int32_t **chrVSrc, int chrFilterSize,
992 const int32_t **alpSrc, uint16_t *dest, int dstW,
993 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
996 int A1 = 0xffff<<14, A2 = 0xffff<<14;
998 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1000 int Y1 = -0x40000000;
1001 int Y2 = -0x40000000;
1002 int U = -(128 << 23); // 19
1003 int V = -(128 << 23);
1006 for (j = 0; j < lumFilterSize; j++) {
1007 Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
1008 Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1010 for (j = 0; j < chrFilterSize; j++) {;
1011 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1012 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1018 for (j = 0; j < lumFilterSize; j++) {
1019 A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
1020 A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1028 // 8 bits: 12+15=27; 16 bits: 12+19=31
1036 // 8 bits: 27 -> 17 bits, 16 bits: 31 - 14 = 17 bits
1037 Y1 -= c->yuv2rgb_y_offset;
1038 Y2 -= c->yuv2rgb_y_offset;
1039 Y1 *= c->yuv2rgb_y_coeff;
1040 Y2 *= c->yuv2rgb_y_coeff;
1041 Y1 += 1 << 13; // 21
1043 // 8 bits: 17 + 13 bits = 30 bits, 16 bits: 17 + 13 bits = 30 bits
1045 R = V * c->yuv2rgb_v2r_coeff;
1046 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1047 B = U * c->yuv2rgb_u2b_coeff;
1049 // 8 bits: 30 - 22 = 8 bits, 16 bits: 30 bits - 14 = 16 bits
1050 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1051 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1052 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1054 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1055 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1056 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1057 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1058 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1061 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1062 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1063 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1069 static av_always_inline void
1070 yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
1071 const int32_t *ubuf[2], const int32_t *vbuf[2],
1072 const int32_t *abuf[2], uint16_t *dest, int dstW,
1073 int yalpha, int uvalpha, int y,
1074 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1076 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1077 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1078 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1079 *abuf0 = hasAlpha ? abuf[0] : NULL,
1080 *abuf1 = hasAlpha ? abuf[1] : NULL;
1081 int yalpha1 = 4096 - yalpha;
1082 int uvalpha1 = 4096 - uvalpha;
1084 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1086 av_assert2(yalpha <= 4096U);
1087 av_assert2(uvalpha <= 4096U);
1089 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1090 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
1091 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
1092 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1093 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1096 Y1 -= c->yuv2rgb_y_offset;
1097 Y2 -= c->yuv2rgb_y_offset;
1098 Y1 *= c->yuv2rgb_y_coeff;
1099 Y2 *= c->yuv2rgb_y_coeff;
1103 R = V * c->yuv2rgb_v2r_coeff;
1104 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1105 B = U * c->yuv2rgb_u2b_coeff;
1108 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
1109 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
1115 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1116 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1117 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1119 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1120 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1121 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1122 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1123 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1126 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1127 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1128 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1134 static av_always_inline void
1135 yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
1136 const int32_t *ubuf[2], const int32_t *vbuf[2],
1137 const int32_t *abuf0, uint16_t *dest, int dstW,
1138 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1140 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1142 int A1 = 0xffff<<14, A2= 0xffff<<14;
1144 if (uvalpha < 2048) {
1145 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1146 int Y1 = (buf0[i * 2] ) >> 2;
1147 int Y2 = (buf0[i * 2 + 1]) >> 2;
1148 int U = (ubuf0[i] - (128 << 11)) >> 2;
1149 int V = (vbuf0[i] - (128 << 11)) >> 2;
1152 Y1 -= c->yuv2rgb_y_offset;
1153 Y2 -= c->yuv2rgb_y_offset;
1154 Y1 *= c->yuv2rgb_y_coeff;
1155 Y2 *= c->yuv2rgb_y_coeff;
1160 A1 = abuf0[i * 2 ] << 11;
1161 A2 = abuf0[i * 2 + 1] << 11;
1167 R = V * c->yuv2rgb_v2r_coeff;
1168 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1169 B = U * c->yuv2rgb_u2b_coeff;
1171 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1172 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1173 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1175 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1176 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1177 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1178 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1179 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1182 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1183 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1184 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1189 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1190 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1191 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1192 int Y1 = (buf0[i * 2] ) >> 2;
1193 int Y2 = (buf0[i * 2 + 1]) >> 2;
1194 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1195 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1198 Y1 -= c->yuv2rgb_y_offset;
1199 Y2 -= c->yuv2rgb_y_offset;
1200 Y1 *= c->yuv2rgb_y_coeff;
1201 Y2 *= c->yuv2rgb_y_coeff;
1206 A1 = abuf0[i * 2 ] << 11;
1207 A2 = abuf0[i * 2 + 1] << 11;
1213 R = V * c->yuv2rgb_v2r_coeff;
1214 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1215 B = U * c->yuv2rgb_u2b_coeff;
1217 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1218 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1219 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1221 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1222 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1223 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1224 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1225 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1228 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1229 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1230 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1237 static av_always_inline void
1238 yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1239 const int32_t **lumSrc, int lumFilterSize,
1240 const int16_t *chrFilter, const int32_t **chrUSrc,
1241 const int32_t **chrVSrc, int chrFilterSize,
1242 const int32_t **alpSrc, uint16_t *dest, int dstW,
1243 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1248 for (i = 0; i < dstW; i++) {
1250 int Y = -0x40000000;
1251 int U = -(128 << 23); // 19
1252 int V = -(128 << 23);
1255 for (j = 0; j < lumFilterSize; j++) {
1256 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
1258 for (j = 0; j < chrFilterSize; j++) {;
1259 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1260 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1265 for (j = 0; j < lumFilterSize; j++) {
1266 A += alpSrc[j][i] * (unsigned)lumFilter[j];
1272 // 8bit: 12+15=27; 16-bit: 12+19=31
1278 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
1279 Y -= c->yuv2rgb_y_offset;
1280 Y *= c->yuv2rgb_y_coeff;
1282 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
1284 R = V * c->yuv2rgb_v2r_coeff;
1285 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1286 B = U * c->yuv2rgb_u2b_coeff;
1288 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
1289 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1290 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1291 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1293 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1301 static av_always_inline void
1302 yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
1303 const int32_t *ubuf[2], const int32_t *vbuf[2],
1304 const int32_t *abuf[2], uint16_t *dest, int dstW,
1305 int yalpha, int uvalpha, int y,
1306 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1308 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1309 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1310 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1311 *abuf0 = hasAlpha ? abuf[0] : NULL,
1312 *abuf1 = hasAlpha ? abuf[1] : NULL;
1313 int yalpha1 = 4096 - yalpha;
1314 int uvalpha1 = 4096 - uvalpha;
1318 av_assert2(yalpha <= 4096U);
1319 av_assert2(uvalpha <= 4096U);
1321 for (i = 0; i < dstW; i++) {
1322 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 14;
1323 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1324 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1327 Y -= c->yuv2rgb_y_offset;
1328 Y *= c->yuv2rgb_y_coeff;
1331 R = V * c->yuv2rgb_v2r_coeff;
1332 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1333 B = U * c->yuv2rgb_u2b_coeff;
1336 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1341 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1342 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1343 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1345 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1353 static av_always_inline void
1354 yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1355 const int32_t *ubuf[2], const int32_t *vbuf[2],
1356 const int32_t *abuf0, uint16_t *dest, int dstW,
1357 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1359 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1363 if (uvalpha < 2048) {
1364 for (i = 0; i < dstW; i++) {
1365 int Y = (buf0[i]) >> 2;
1366 int U = (ubuf0[i] - (128 << 11)) >> 2;
1367 int V = (vbuf0[i] - (128 << 11)) >> 2;
1370 Y -= c->yuv2rgb_y_offset;
1371 Y *= c->yuv2rgb_y_coeff;
1380 R = V * c->yuv2rgb_v2r_coeff;
1381 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1382 B = U * c->yuv2rgb_u2b_coeff;
1384 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1385 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1386 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1388 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1395 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1397 for (i = 0; i < dstW; i++) {
1398 int Y = (buf0[i] ) >> 2;
1399 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1400 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1403 Y -= c->yuv2rgb_y_offset;
1404 Y *= c->yuv2rgb_y_coeff;
1413 R = V * c->yuv2rgb_v2r_coeff;
1414 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1415 B = U * c->yuv2rgb_u2b_coeff;
1417 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1418 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1419 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1421 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1434 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1435 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1436 const int16_t **_lumSrc, int lumFilterSize, \
1437 const int16_t *chrFilter, const int16_t **_chrUSrc, \
1438 const int16_t **_chrVSrc, int chrFilterSize, \
1439 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1442 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1443 **chrUSrc = (const int32_t **) _chrUSrc, \
1444 **chrVSrc = (const int32_t **) _chrVSrc, \
1445 **alpSrc = (const int32_t **) _alpSrc; \
1446 uint16_t *dest = (uint16_t *) _dest; \
1447 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1448 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1449 alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1452 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1453 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1454 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1455 int yalpha, int uvalpha, int y) \
1457 const int32_t **buf = (const int32_t **) _buf, \
1458 **ubuf = (const int32_t **) _ubuf, \
1459 **vbuf = (const int32_t **) _vbuf, \
1460 **abuf = (const int32_t **) _abuf; \
1461 uint16_t *dest = (uint16_t *) _dest; \
1462 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1463 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1466 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1467 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1468 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1469 int uvalpha, int y) \
1471 const int32_t *buf0 = (const int32_t *) _buf0, \
1472 **ubuf = (const int32_t **) _ubuf, \
1473 **vbuf = (const int32_t **) _vbuf, \
1474 *abuf0 = (const int32_t *) _abuf0; \
1475 uint16_t *dest = (uint16_t *) _dest; \
1476 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1477 dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1480 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1481 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1482 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1483 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1484 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1485 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1486 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1487 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1488 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1489 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1490 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1491 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1492 YUV2PACKED16WRAPPER(yuv2, ya16, ya16be, AV_PIX_FMT_YA16BE, 1, 0)
1493 YUV2PACKED16WRAPPER(yuv2, ya16, ya16le, AV_PIX_FMT_YA16LE, 1, 0)
1495 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1496 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1497 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1498 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1499 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1500 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1501 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1502 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1503 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1504 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1505 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1506 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1509 * Write out 2 RGB pixels in the target pixel format. This function takes a
1510 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1511 * things like endianness conversion and shifting. The caller takes care of
1512 * setting the correct offset in these tables from the chroma (U/V) values.
1513 * This function then uses the luminance (Y1/Y2) values to write out the
1514 * correct RGB values into the destination buffer.
1516 static av_always_inline void
1517 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1518 unsigned A1, unsigned A2,
1519 const void *_r, const void *_g, const void *_b, int y,
1520 enum AVPixelFormat target, int hasAlpha)
1522 if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1523 target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1524 uint32_t *dest = (uint32_t *) _dest;
1525 const uint32_t *r = (const uint32_t *) _r;
1526 const uint32_t *g = (const uint32_t *) _g;
1527 const uint32_t *b = (const uint32_t *) _b;
1530 int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1532 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1533 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1536 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1538 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1539 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1540 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1542 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1543 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1545 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1547 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1548 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1551 } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1552 uint8_t *dest = (uint8_t *) _dest;
1553 const uint8_t *r = (const uint8_t *) _r;
1554 const uint8_t *g = (const uint8_t *) _g;
1555 const uint8_t *b = (const uint8_t *) _b;
1557 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1558 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1560 dest[i * 6 + 0] = r_b[Y1];
1561 dest[i * 6 + 1] = g[Y1];
1562 dest[i * 6 + 2] = b_r[Y1];
1563 dest[i * 6 + 3] = r_b[Y2];
1564 dest[i * 6 + 4] = g[Y2];
1565 dest[i * 6 + 5] = b_r[Y2];
1568 } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1569 target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1570 target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1571 uint16_t *dest = (uint16_t *) _dest;
1572 const uint16_t *r = (const uint16_t *) _r;
1573 const uint16_t *g = (const uint16_t *) _g;
1574 const uint16_t *b = (const uint16_t *) _b;
1575 int dr1, dg1, db1, dr2, dg2, db2;
1577 if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1578 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1579 dg1 = ff_dither_2x2_4[ y & 1 ][0];
1580 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1581 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1582 dg2 = ff_dither_2x2_4[ y & 1 ][1];
1583 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1584 } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1585 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1586 dg1 = ff_dither_2x2_8[ y & 1 ][1];
1587 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1588 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1589 dg2 = ff_dither_2x2_8[ y & 1 ][0];
1590 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1592 dr1 = ff_dither_4x4_16[ y & 3 ][0];
1593 dg1 = ff_dither_4x4_16[ y & 3 ][1];
1594 db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1595 dr2 = ff_dither_4x4_16[ y & 3 ][1];
1596 dg2 = ff_dither_4x4_16[ y & 3 ][0];
1597 db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1600 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1601 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1602 } else /* 8/4 bits */ {
1603 uint8_t *dest = (uint8_t *) _dest;
1604 const uint8_t *r = (const uint8_t *) _r;
1605 const uint8_t *g = (const uint8_t *) _g;
1606 const uint8_t *b = (const uint8_t *) _b;
1607 int dr1, dg1, db1, dr2, dg2, db2;
1609 if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1610 const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1611 const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1612 dr1 = dg1 = d32[(i * 2 + 0) & 7];
1613 db1 = d64[(i * 2 + 0) & 7];
1614 dr2 = dg2 = d32[(i * 2 + 1) & 7];
1615 db2 = d64[(i * 2 + 1) & 7];
1617 const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1618 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1619 dr1 = db1 = d128[(i * 2 + 0) & 7];
1620 dg1 = d64[(i * 2 + 0) & 7];
1621 dr2 = db2 = d128[(i * 2 + 1) & 7];
1622 dg2 = d64[(i * 2 + 1) & 7];
1625 if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1626 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1627 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1629 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1630 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1635 static av_always_inline void
1636 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1637 const int16_t **lumSrc, int lumFilterSize,
1638 const int16_t *chrFilter, const int16_t **chrUSrc,
1639 const int16_t **chrVSrc, int chrFilterSize,
1640 const int16_t **alpSrc, uint8_t *dest, int dstW,
1641 int y, enum AVPixelFormat target, int hasAlpha)
1645 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1651 const void *r, *g, *b;
1653 for (j = 0; j < lumFilterSize; j++) {
1654 Y1 += lumSrc[j][i * 2] * lumFilter[j];
1655 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1657 for (j = 0; j < chrFilterSize; j++) {
1658 U += chrUSrc[j][i] * chrFilter[j];
1659 V += chrVSrc[j][i] * chrFilter[j];
1668 for (j = 0; j < lumFilterSize; j++) {
1669 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1670 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1674 if ((A1 | A2) & 0x100) {
1675 A1 = av_clip_uint8(A1);
1676 A2 = av_clip_uint8(A2);
1680 r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1681 g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1682 b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1684 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1685 r, g, b, y, target, hasAlpha);
1689 static av_always_inline void
1690 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1691 const int16_t *ubuf[2], const int16_t *vbuf[2],
1692 const int16_t *abuf[2], uint8_t *dest, int dstW,
1693 int yalpha, int uvalpha, int y,
1694 enum AVPixelFormat target, int hasAlpha)
1696 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1697 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1698 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1699 *abuf0 = hasAlpha ? abuf[0] : NULL,
1700 *abuf1 = hasAlpha ? abuf[1] : NULL;
1701 int yalpha1 = 4096 - yalpha;
1702 int uvalpha1 = 4096 - uvalpha;
1704 av_assert2(yalpha <= 4096U);
1705 av_assert2(uvalpha <= 4096U);
1707 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1708 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1709 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1710 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1711 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1713 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1714 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1715 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1718 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1719 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1720 A1 = av_clip_uint8(A1);
1721 A2 = av_clip_uint8(A2);
1724 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1725 r, g, b, y, target, hasAlpha);
1729 static av_always_inline void
1730 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1731 const int16_t *ubuf[2], const int16_t *vbuf[2],
1732 const int16_t *abuf0, uint8_t *dest, int dstW,
1733 int uvalpha, int y, enum AVPixelFormat target,
1736 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1739 if (uvalpha < 2048) {
1740 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1741 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1742 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1743 int U = (ubuf0[i] + 64) >> 7;
1744 int V = (vbuf0[i] + 64) >> 7;
1746 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1747 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1748 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1751 A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1752 A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1753 A1 = av_clip_uint8(A1);
1754 A2 = av_clip_uint8(A2);
1757 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1758 r, g, b, y, target, hasAlpha);
1761 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1762 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1763 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1764 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1765 int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1766 int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1768 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1769 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1770 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1773 A1 = (abuf0[i * 2 ] + 64) >> 7;
1774 A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1775 A1 = av_clip_uint8(A1);
1776 A2 = av_clip_uint8(A2);
1779 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1780 r, g, b, y, target, hasAlpha);
1785 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1786 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1787 const int16_t **lumSrc, int lumFilterSize, \
1788 const int16_t *chrFilter, const int16_t **chrUSrc, \
1789 const int16_t **chrVSrc, int chrFilterSize, \
1790 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1793 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1794 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1795 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1798 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1799 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1800 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1801 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1802 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1803 int yalpha, int uvalpha, int y) \
1805 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1806 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1809 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1810 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1811 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1812 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1813 const int16_t *abuf0, uint8_t *dest, int dstW, \
1814 int uvalpha, int y) \
1816 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1817 dstW, uvalpha, y, fmt, hasAlpha); \
1821 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1822 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1824 #if CONFIG_SWSCALE_ALPHA
1825 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1826 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1828 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1829 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1831 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1832 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1833 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1834 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1835 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1836 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1837 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1838 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1840 static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1841 uint8_t *dest, int i, int Y, int A, int U, int V,
1842 int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1845 int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1847 Y -= c->yuv2rgb_y_offset;
1848 Y *= c->yuv2rgb_y_coeff;
1850 R = Y + V*c->yuv2rgb_v2r_coeff;
1851 G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1852 B = Y + U*c->yuv2rgb_u2b_coeff;
1853 if ((R | G | B) & 0xC0000000) {
1854 R = av_clip_uintp2(R, 30);
1855 G = av_clip_uintp2(G, 30);
1856 B = av_clip_uintp2(B, 30);
1860 case AV_PIX_FMT_ARGB:
1861 dest[0] = hasAlpha ? A : 255;
1866 case AV_PIX_FMT_RGB24:
1871 case AV_PIX_FMT_RGBA:
1875 dest[3] = hasAlpha ? A : 255;
1877 case AV_PIX_FMT_ABGR:
1878 dest[0] = hasAlpha ? A : 255;
1883 case AV_PIX_FMT_BGR24:
1888 case AV_PIX_FMT_BGRA:
1892 dest[3] = hasAlpha ? A : 255;
1894 case AV_PIX_FMT_BGR4_BYTE:
1895 case AV_PIX_FMT_RGB4_BYTE:
1896 case AV_PIX_FMT_BGR8:
1897 case AV_PIX_FMT_RGB8:
1901 switch (c->dither) {
1903 case SWS_DITHER_AUTO:
1908 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;
1909 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;
1910 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;
1911 c->dither_error[0][i] = err[0];
1912 c->dither_error[1][i] = err[1];
1913 c->dither_error[2][i] = err[2];
1914 r = R >> (isrgb8 ? 5 : 7);
1915 g = G >> (isrgb8 ? 5 : 6);
1916 b = B >> (isrgb8 ? 6 : 7);
1917 r = av_clip(r, 0, isrgb8 ? 7 : 1);
1918 g = av_clip(g, 0, isrgb8 ? 7 : 3);
1919 b = av_clip(b, 0, isrgb8 ? 3 : 1);
1920 err[0] = R - r*(isrgb8 ? 36 : 255);
1921 err[1] = G - g*(isrgb8 ? 36 : 85);
1922 err[2] = B - b*(isrgb8 ? 85 : 255);
1924 case SWS_DITHER_A_DITHER:
1926 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1927 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1928 r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1929 g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1930 b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1931 r = av_clip_uintp2(r, 3);
1932 g = av_clip_uintp2(g, 3);
1933 b = av_clip_uintp2(b, 2);
1935 r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1936 g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1937 b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1938 r = av_clip_uintp2(r, 1);
1939 g = av_clip_uintp2(g, 2);
1940 b = av_clip_uintp2(b, 1);
1943 case SWS_DITHER_X_DITHER:
1945 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1946 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1947 r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1948 g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1949 b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1950 r = av_clip_uintp2(r, 3);
1951 g = av_clip_uintp2(g, 3);
1952 b = av_clip_uintp2(b, 2);
1954 r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1955 g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1956 b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1957 r = av_clip_uintp2(r, 1);
1958 g = av_clip_uintp2(g, 2);
1959 b = av_clip_uintp2(b, 1);
1965 if(target == AV_PIX_FMT_BGR4_BYTE) {
1966 dest[0] = r + 2*g + 8*b;
1967 } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1968 dest[0] = b + 2*g + 8*r;
1969 } else if(target == AV_PIX_FMT_BGR8) {
1970 dest[0] = r + 8*g + 64*b;
1971 } else if(target == AV_PIX_FMT_RGB8) {
1972 dest[0] = b + 4*g + 32*r;
1979 static av_always_inline void
1980 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1981 const int16_t **lumSrc, int lumFilterSize,
1982 const int16_t *chrFilter, const int16_t **chrUSrc,
1983 const int16_t **chrVSrc, int chrFilterSize,
1984 const int16_t **alpSrc, uint8_t *dest,
1985 int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1988 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1990 int A = 0; //init to silence warning
1992 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1993 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1996 for (i = 0; i < dstW; i++) {
1999 int U = (1<<9)-(128 << 19);
2000 int V = (1<<9)-(128 << 19);
2002 for (j = 0; j < lumFilterSize; j++) {
2003 Y += lumSrc[j][i] * lumFilter[j];
2005 for (j = 0; j < chrFilterSize; j++) {
2006 U += chrUSrc[j][i] * chrFilter[j];
2007 V += chrVSrc[j][i] * chrFilter[j];
2014 for (j = 0; j < lumFilterSize; j++) {
2015 A += alpSrc[j][i] * lumFilter[j];
2019 A = av_clip_uint8(A);
2021 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2024 c->dither_error[0][i] = err[0];
2025 c->dither_error[1][i] = err[1];
2026 c->dither_error[2][i] = err[2];
2029 static av_always_inline void
2030 yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
2031 const int16_t *ubuf[2], const int16_t *vbuf[2],
2032 const int16_t *abuf[2], uint8_t *dest, int dstW,
2033 int yalpha, int uvalpha, int y,
2034 enum AVPixelFormat target, int hasAlpha)
2036 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2037 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
2038 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
2039 *abuf0 = hasAlpha ? abuf[0] : NULL,
2040 *abuf1 = hasAlpha ? abuf[1] : NULL;
2041 int yalpha1 = 4096 - yalpha;
2042 int uvalpha1 = 4096 - uvalpha;
2044 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2046 int A = 0; // init to silcene warning
2048 av_assert2(yalpha <= 4096U);
2049 av_assert2(uvalpha <= 4096U);
2051 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2052 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2055 for (i = 0; i < dstW; i++) {
2056 int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
2057 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
2058 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
2061 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
2063 A = av_clip_uint8(A);
2066 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2069 c->dither_error[0][i] = err[0];
2070 c->dither_error[1][i] = err[1];
2071 c->dither_error[2][i] = err[2];
2074 static av_always_inline void
2075 yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
2076 const int16_t *ubuf[2], const int16_t *vbuf[2],
2077 const int16_t *abuf0, uint8_t *dest, int dstW,
2078 int uvalpha, int y, enum AVPixelFormat target,
2081 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
2083 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2086 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2087 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2090 if (uvalpha < 2048) {
2091 int A = 0; //init to silence warning
2092 for (i = 0; i < dstW; i++) {
2093 int Y = buf0[i] << 2;
2094 int U = (ubuf0[i] - (128<<7)) * 4;
2095 int V = (vbuf0[i] - (128<<7)) * 4;
2098 A = (abuf0[i] + 64) >> 7;
2100 A = av_clip_uint8(A);
2103 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2107 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
2108 int A = 0; //init to silence warning
2109 for (i = 0; i < dstW; i++) {
2110 int Y = buf0[i] << 2;
2111 int U = (ubuf0[i] + ubuf1[i] - (128<<8)) << 1;
2112 int V = (vbuf0[i] + vbuf1[i] - (128<<8)) << 1;
2115 A = (abuf0[i] + 64) >> 7;
2117 A = av_clip_uint8(A);
2120 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2125 c->dither_error[0][i] = err[0];
2126 c->dither_error[1][i] = err[1];
2127 c->dither_error[2][i] = err[2];
2131 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2132 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2133 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2134 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2136 #if CONFIG_SWSCALE_ALPHA
2137 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
2138 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
2139 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
2140 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
2142 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
2143 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
2144 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
2145 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
2147 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
2148 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
2150 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
2151 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
2152 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
2153 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
2156 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
2157 const int16_t **lumSrc, int lumFilterSize,
2158 const int16_t *chrFilter, const int16_t **chrUSrc,
2159 const int16_t **chrVSrc, int chrFilterSize,
2160 const int16_t **alpSrc, uint8_t **dest,
2163 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2165 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
2166 uint16_t **dest16 = (uint16_t**)dest;
2167 int SH = 22 + 8 - desc->comp[0].depth;
2168 int A = 0; // init to silence warning
2170 for (i = 0; i < dstW; i++) {
2173 int U = (1 << 9) - (128 << 19);
2174 int V = (1 << 9) - (128 << 19);
2177 for (j = 0; j < lumFilterSize; j++)
2178 Y += lumSrc[j][i] * lumFilter[j];
2180 for (j = 0; j < chrFilterSize; j++) {
2181 U += chrUSrc[j][i] * chrFilter[j];
2182 V += chrVSrc[j][i] * chrFilter[j];
2192 for (j = 0; j < lumFilterSize; j++)
2193 A += alpSrc[j][i] * lumFilter[j];
2196 A = av_clip_uintp2(A, 27);
2199 Y -= c->yuv2rgb_y_offset;
2200 Y *= c->yuv2rgb_y_coeff;
2202 R = Y + V * c->yuv2rgb_v2r_coeff;
2203 G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2204 B = Y + U * c->yuv2rgb_u2b_coeff;
2206 if ((R | G | B) & 0xC0000000) {
2207 R = av_clip_uintp2(R, 30);
2208 G = av_clip_uintp2(G, 30);
2209 B = av_clip_uintp2(B, 30);
2213 dest16[0][i] = G >> SH;
2214 dest16[1][i] = B >> SH;
2215 dest16[2][i] = R >> SH;
2217 dest16[3][i] = A >> (SH - 3);
2219 dest[0][i] = G >> 22;
2220 dest[1][i] = B >> 22;
2221 dest[2][i] = R >> 22;
2223 dest[3][i] = A >> 19;
2226 if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2227 for (i = 0; i < dstW; i++) {
2228 dest16[0][i] = av_bswap16(dest16[0][i]);
2229 dest16[1][i] = av_bswap16(dest16[1][i]);
2230 dest16[2][i] = av_bswap16(dest16[2][i]);
2232 dest16[3][i] = av_bswap16(dest16[3][i]);
2238 yuv2gbrp16_full_X_c(SwsContext *c, const int16_t *lumFilter,
2239 const int16_t **lumSrcx, int lumFilterSize,
2240 const int16_t *chrFilter, const int16_t **chrUSrcx,
2241 const int16_t **chrVSrcx, int chrFilterSize,
2242 const int16_t **alpSrcx, uint8_t **dest,
2245 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2247 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2248 uint16_t **dest16 = (uint16_t**)dest;
2249 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2250 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2251 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2252 const int32_t **alpSrc = (const int32_t**)alpSrcx;
2254 for (i = 0; i < dstW; i++) {
2256 int Y = -0x40000000;
2257 int U = -(128 << 23);
2258 int V = -(128 << 23);
2261 for (j = 0; j < lumFilterSize; j++)
2262 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2264 for (j = 0; j < chrFilterSize; j++) {
2265 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2266 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2277 for (j = 0; j < lumFilterSize; j++)
2278 A += alpSrc[j][i] * lumFilter[j];
2284 Y -= c->yuv2rgb_y_offset;
2285 Y *= c->yuv2rgb_y_coeff;
2287 R = V * c->yuv2rgb_v2r_coeff;
2288 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2289 B = U * c->yuv2rgb_u2b_coeff;
2291 R = av_clip_uintp2(Y + R, 30);
2292 G = av_clip_uintp2(Y + G, 30);
2293 B = av_clip_uintp2(Y + B, 30);
2295 dest16[0][i] = G >> 14;
2296 dest16[1][i] = B >> 14;
2297 dest16[2][i] = R >> 14;
2299 dest16[3][i] = av_clip_uintp2(A, 30) >> 14;
2301 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2302 for (i = 0; i < dstW; i++) {
2303 dest16[0][i] = av_bswap16(dest16[0][i]);
2304 dest16[1][i] = av_bswap16(dest16[1][i]);
2305 dest16[2][i] = av_bswap16(dest16[2][i]);
2307 dest16[3][i] = av_bswap16(dest16[3][i]);
2313 yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
2314 const int16_t *ubuf[2], const int16_t *vbuf[2],
2315 const int16_t *abuf0, uint8_t *dest, int dstW,
2318 int hasAlpha = !!abuf0;
2321 for (i = 0; i < dstW; i++) {
2322 int Y = (buf0[i] + 64) >> 7;
2325 Y = av_clip_uint8(Y);
2328 A = (abuf0[i] + 64) >> 7;
2330 A = av_clip_uint8(A);
2334 dest[i * 2 + 1] = hasAlpha ? A : 255;
2339 yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
2340 const int16_t *ubuf[2], const int16_t *vbuf[2],
2341 const int16_t *abuf[2], uint8_t *dest, int dstW,
2342 int yalpha, int uvalpha, int y)
2344 int hasAlpha = abuf && abuf[0] && abuf[1];
2345 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2346 *abuf0 = hasAlpha ? abuf[0] : NULL,
2347 *abuf1 = hasAlpha ? abuf[1] : NULL;
2348 int yalpha1 = 4096 - yalpha;
2351 av_assert2(yalpha <= 4096U);
2353 for (i = 0; i < dstW; i++) {
2354 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 19;
2357 Y = av_clip_uint8(Y);
2360 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 19;
2361 A = av_clip_uint8(A);
2365 dest[i * 2 + 1] = hasAlpha ? A : 255;
2370 yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
2371 const int16_t **lumSrc, int lumFilterSize,
2372 const int16_t *chrFilter, const int16_t **chrUSrc,
2373 const int16_t **chrVSrc, int chrFilterSize,
2374 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
2376 int hasAlpha = !!alpSrc;
2379 for (i = 0; i < dstW; i++) {
2381 int Y = 1 << 18, A = 1 << 18;
2383 for (j = 0; j < lumFilterSize; j++)
2384 Y += lumSrc[j][i] * lumFilter[j];
2388 Y = av_clip_uint8(Y);
2391 for (j = 0; j < lumFilterSize; j++)
2392 A += alpSrc[j][i] * lumFilter[j];
2397 A = av_clip_uint8(A);
2401 dest[2 * i + 1] = hasAlpha ? A : 255;
2406 yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter,
2407 const int16_t **_lumSrc, int lumFilterSize,
2408 const int16_t *chrFilter, const int16_t **_chrUSrc,
2409 const int16_t **_chrVSrc, int chrFilterSize,
2410 const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
2412 const int32_t **lumSrc = (const int32_t **) _lumSrc,
2413 **chrUSrc = (const int32_t **) _chrUSrc,
2414 **chrVSrc = (const int32_t **) _chrVSrc,
2415 **alpSrc = (const int32_t **) _alpSrc;
2416 int hasAlpha = !!alpSrc;
2419 for (i = 0; i < dstW; i++) {
2420 int Y = 1 << 14, U = 1 << 14;
2421 int V = 1 << 14, A = 1 << 14;
2429 for (j = 0; j < lumFilterSize; j++)
2430 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2432 for (j = 0; j < chrFilterSize; j++)
2433 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2435 for (j = 0; j < chrFilterSize; j++)
2436 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2439 for (j = 0; j < lumFilterSize; j++)
2440 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2442 Y = 0x8000 + av_clip_int16(Y >> 15);
2443 U = 0x8000 + av_clip_int16(U >> 15);
2444 V = 0x8000 + av_clip_int16(V >> 15);
2445 A = 0x8000 + av_clip_int16(A >> 15);
2447 AV_WL16(dest + 8 * i, hasAlpha ? A : 65535);
2448 AV_WL16(dest + 8 * i + 2, Y);
2449 AV_WL16(dest + 8 * i + 4, U);
2450 AV_WL16(dest + 8 * i + 6, V);
2454 av_cold void ff_sws_init_output_funcs(SwsContext *c,
2455 yuv2planar1_fn *yuv2plane1,
2456 yuv2planarX_fn *yuv2planeX,
2457 yuv2interleavedX_fn *yuv2nv12cX,
2458 yuv2packed1_fn *yuv2packed1,
2459 yuv2packed2_fn *yuv2packed2,
2460 yuv2packedX_fn *yuv2packedX,
2461 yuv2anyX_fn *yuv2anyX)
2463 enum AVPixelFormat dstFormat = c->dstFormat;
2464 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2466 if (dstFormat == AV_PIX_FMT_P010LE || dstFormat == AV_PIX_FMT_P010BE) {
2467 *yuv2plane1 = isBE(dstFormat) ? yuv2p010l1_BE_c : yuv2p010l1_LE_c;
2468 *yuv2planeX = isBE(dstFormat) ? yuv2p010lX_BE_c : yuv2p010lX_LE_c;
2469 *yuv2nv12cX = yuv2p010cX_c;
2470 } else if (is16BPS(dstFormat)) {
2471 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
2472 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
2473 if (dstFormat == AV_PIX_FMT_P016LE || dstFormat == AV_PIX_FMT_P016BE) {
2474 *yuv2nv12cX = yuv2p016cX_c;
2476 } else if (isNBPS(dstFormat)) {
2477 if (desc->comp[0].depth == 9) {
2478 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
2479 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
2480 } else if (desc->comp[0].depth == 10) {
2481 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
2482 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
2483 } else if (desc->comp[0].depth == 12) {
2484 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
2485 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
2486 } else if (desc->comp[0].depth == 14) {
2487 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
2488 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
2491 } else if (dstFormat == AV_PIX_FMT_GRAYF32BE) {
2492 *yuv2planeX = yuv2planeX_floatBE_c;
2493 *yuv2plane1 = yuv2plane1_floatBE_c;
2494 } else if (dstFormat == AV_PIX_FMT_GRAYF32LE) {
2495 *yuv2planeX = yuv2planeX_floatLE_c;
2496 *yuv2plane1 = yuv2plane1_floatLE_c;
2498 *yuv2plane1 = yuv2plane1_8_c;
2499 *yuv2planeX = yuv2planeX_8_c;
2500 if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21 ||
2501 dstFormat == AV_PIX_FMT_NV24 || dstFormat == AV_PIX_FMT_NV42)
2502 *yuv2nv12cX = yuv2nv12cX_c;
2505 if(c->flags & SWS_FULL_CHR_H_INT) {
2506 switch (dstFormat) {
2507 case AV_PIX_FMT_RGBA:
2509 *yuv2packedX = yuv2rgba32_full_X_c;
2510 *yuv2packed2 = yuv2rgba32_full_2_c;
2511 *yuv2packed1 = yuv2rgba32_full_1_c;
2513 #if CONFIG_SWSCALE_ALPHA
2515 *yuv2packedX = yuv2rgba32_full_X_c;
2516 *yuv2packed2 = yuv2rgba32_full_2_c;
2517 *yuv2packed1 = yuv2rgba32_full_1_c;
2519 #endif /* CONFIG_SWSCALE_ALPHA */
2521 *yuv2packedX = yuv2rgbx32_full_X_c;
2522 *yuv2packed2 = yuv2rgbx32_full_2_c;
2523 *yuv2packed1 = yuv2rgbx32_full_1_c;
2525 #endif /* !CONFIG_SMALL */
2527 case AV_PIX_FMT_ARGB:
2529 *yuv2packedX = yuv2argb32_full_X_c;
2530 *yuv2packed2 = yuv2argb32_full_2_c;
2531 *yuv2packed1 = yuv2argb32_full_1_c;
2533 #if CONFIG_SWSCALE_ALPHA
2535 *yuv2packedX = yuv2argb32_full_X_c;
2536 *yuv2packed2 = yuv2argb32_full_2_c;
2537 *yuv2packed1 = yuv2argb32_full_1_c;
2539 #endif /* CONFIG_SWSCALE_ALPHA */
2541 *yuv2packedX = yuv2xrgb32_full_X_c;
2542 *yuv2packed2 = yuv2xrgb32_full_2_c;
2543 *yuv2packed1 = yuv2xrgb32_full_1_c;
2545 #endif /* !CONFIG_SMALL */
2547 case AV_PIX_FMT_BGRA:
2549 *yuv2packedX = yuv2bgra32_full_X_c;
2550 *yuv2packed2 = yuv2bgra32_full_2_c;
2551 *yuv2packed1 = yuv2bgra32_full_1_c;
2553 #if CONFIG_SWSCALE_ALPHA
2555 *yuv2packedX = yuv2bgra32_full_X_c;
2556 *yuv2packed2 = yuv2bgra32_full_2_c;
2557 *yuv2packed1 = yuv2bgra32_full_1_c;
2559 #endif /* CONFIG_SWSCALE_ALPHA */
2561 *yuv2packedX = yuv2bgrx32_full_X_c;
2562 *yuv2packed2 = yuv2bgrx32_full_2_c;
2563 *yuv2packed1 = yuv2bgrx32_full_1_c;
2565 #endif /* !CONFIG_SMALL */
2567 case AV_PIX_FMT_ABGR:
2569 *yuv2packedX = yuv2abgr32_full_X_c;
2570 *yuv2packed2 = yuv2abgr32_full_2_c;
2571 *yuv2packed1 = yuv2abgr32_full_1_c;
2573 #if CONFIG_SWSCALE_ALPHA
2575 *yuv2packedX = yuv2abgr32_full_X_c;
2576 *yuv2packed2 = yuv2abgr32_full_2_c;
2577 *yuv2packed1 = yuv2abgr32_full_1_c;
2579 #endif /* CONFIG_SWSCALE_ALPHA */
2581 *yuv2packedX = yuv2xbgr32_full_X_c;
2582 *yuv2packed2 = yuv2xbgr32_full_2_c;
2583 *yuv2packed1 = yuv2xbgr32_full_1_c;
2585 #endif /* !CONFIG_SMALL */
2587 case AV_PIX_FMT_RGBA64LE:
2588 #if CONFIG_SWSCALE_ALPHA
2590 *yuv2packedX = yuv2rgba64le_full_X_c;
2591 *yuv2packed2 = yuv2rgba64le_full_2_c;
2592 *yuv2packed1 = yuv2rgba64le_full_1_c;
2594 #endif /* CONFIG_SWSCALE_ALPHA */
2596 *yuv2packedX = yuv2rgbx64le_full_X_c;
2597 *yuv2packed2 = yuv2rgbx64le_full_2_c;
2598 *yuv2packed1 = yuv2rgbx64le_full_1_c;
2601 case AV_PIX_FMT_RGBA64BE:
2602 #if CONFIG_SWSCALE_ALPHA
2604 *yuv2packedX = yuv2rgba64be_full_X_c;
2605 *yuv2packed2 = yuv2rgba64be_full_2_c;
2606 *yuv2packed1 = yuv2rgba64be_full_1_c;
2608 #endif /* CONFIG_SWSCALE_ALPHA */
2610 *yuv2packedX = yuv2rgbx64be_full_X_c;
2611 *yuv2packed2 = yuv2rgbx64be_full_2_c;
2612 *yuv2packed1 = yuv2rgbx64be_full_1_c;
2615 case AV_PIX_FMT_BGRA64LE:
2616 #if CONFIG_SWSCALE_ALPHA
2618 *yuv2packedX = yuv2bgra64le_full_X_c;
2619 *yuv2packed2 = yuv2bgra64le_full_2_c;
2620 *yuv2packed1 = yuv2bgra64le_full_1_c;
2622 #endif /* CONFIG_SWSCALE_ALPHA */
2624 *yuv2packedX = yuv2bgrx64le_full_X_c;
2625 *yuv2packed2 = yuv2bgrx64le_full_2_c;
2626 *yuv2packed1 = yuv2bgrx64le_full_1_c;
2629 case AV_PIX_FMT_BGRA64BE:
2630 #if CONFIG_SWSCALE_ALPHA
2632 *yuv2packedX = yuv2bgra64be_full_X_c;
2633 *yuv2packed2 = yuv2bgra64be_full_2_c;
2634 *yuv2packed1 = yuv2bgra64be_full_1_c;
2636 #endif /* CONFIG_SWSCALE_ALPHA */
2638 *yuv2packedX = yuv2bgrx64be_full_X_c;
2639 *yuv2packed2 = yuv2bgrx64be_full_2_c;
2640 *yuv2packed1 = yuv2bgrx64be_full_1_c;
2644 case AV_PIX_FMT_RGB24:
2645 *yuv2packedX = yuv2rgb24_full_X_c;
2646 *yuv2packed2 = yuv2rgb24_full_2_c;
2647 *yuv2packed1 = yuv2rgb24_full_1_c;
2649 case AV_PIX_FMT_BGR24:
2650 *yuv2packedX = yuv2bgr24_full_X_c;
2651 *yuv2packed2 = yuv2bgr24_full_2_c;
2652 *yuv2packed1 = yuv2bgr24_full_1_c;
2654 case AV_PIX_FMT_RGB48LE:
2655 *yuv2packedX = yuv2rgb48le_full_X_c;
2656 *yuv2packed2 = yuv2rgb48le_full_2_c;
2657 *yuv2packed1 = yuv2rgb48le_full_1_c;
2659 case AV_PIX_FMT_BGR48LE:
2660 *yuv2packedX = yuv2bgr48le_full_X_c;
2661 *yuv2packed2 = yuv2bgr48le_full_2_c;
2662 *yuv2packed1 = yuv2bgr48le_full_1_c;
2664 case AV_PIX_FMT_RGB48BE:
2665 *yuv2packedX = yuv2rgb48be_full_X_c;
2666 *yuv2packed2 = yuv2rgb48be_full_2_c;
2667 *yuv2packed1 = yuv2rgb48be_full_1_c;
2669 case AV_PIX_FMT_BGR48BE:
2670 *yuv2packedX = yuv2bgr48be_full_X_c;
2671 *yuv2packed2 = yuv2bgr48be_full_2_c;
2672 *yuv2packed1 = yuv2bgr48be_full_1_c;
2674 case AV_PIX_FMT_BGR4_BYTE:
2675 *yuv2packedX = yuv2bgr4_byte_full_X_c;
2676 *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2677 *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2679 case AV_PIX_FMT_RGB4_BYTE:
2680 *yuv2packedX = yuv2rgb4_byte_full_X_c;
2681 *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2682 *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2684 case AV_PIX_FMT_BGR8:
2685 *yuv2packedX = yuv2bgr8_full_X_c;
2686 *yuv2packed2 = yuv2bgr8_full_2_c;
2687 *yuv2packed1 = yuv2bgr8_full_1_c;
2689 case AV_PIX_FMT_RGB8:
2690 *yuv2packedX = yuv2rgb8_full_X_c;
2691 *yuv2packed2 = yuv2rgb8_full_2_c;
2692 *yuv2packed1 = yuv2rgb8_full_1_c;
2694 case AV_PIX_FMT_GBRP:
2695 case AV_PIX_FMT_GBRP9BE:
2696 case AV_PIX_FMT_GBRP9LE:
2697 case AV_PIX_FMT_GBRP10BE:
2698 case AV_PIX_FMT_GBRP10LE:
2699 case AV_PIX_FMT_GBRP12BE:
2700 case AV_PIX_FMT_GBRP12LE:
2701 case AV_PIX_FMT_GBRP14BE:
2702 case AV_PIX_FMT_GBRP14LE:
2703 case AV_PIX_FMT_GBRAP:
2704 case AV_PIX_FMT_GBRAP10BE:
2705 case AV_PIX_FMT_GBRAP10LE:
2706 case AV_PIX_FMT_GBRAP12BE:
2707 case AV_PIX_FMT_GBRAP12LE:
2708 *yuv2anyX = yuv2gbrp_full_X_c;
2710 case AV_PIX_FMT_GBRP16BE:
2711 case AV_PIX_FMT_GBRP16LE:
2712 case AV_PIX_FMT_GBRAP16BE:
2713 case AV_PIX_FMT_GBRAP16LE:
2714 *yuv2anyX = yuv2gbrp16_full_X_c;
2717 if (!*yuv2packedX && !*yuv2anyX)
2721 switch (dstFormat) {
2722 case AV_PIX_FMT_RGBA64LE:
2723 #if CONFIG_SWSCALE_ALPHA
2725 *yuv2packed1 = yuv2rgba64le_1_c;
2726 *yuv2packed2 = yuv2rgba64le_2_c;
2727 *yuv2packedX = yuv2rgba64le_X_c;
2729 #endif /* CONFIG_SWSCALE_ALPHA */
2731 *yuv2packed1 = yuv2rgbx64le_1_c;
2732 *yuv2packed2 = yuv2rgbx64le_2_c;
2733 *yuv2packedX = yuv2rgbx64le_X_c;
2736 case AV_PIX_FMT_RGBA64BE:
2737 #if CONFIG_SWSCALE_ALPHA
2739 *yuv2packed1 = yuv2rgba64be_1_c;
2740 *yuv2packed2 = yuv2rgba64be_2_c;
2741 *yuv2packedX = yuv2rgba64be_X_c;
2743 #endif /* CONFIG_SWSCALE_ALPHA */
2745 *yuv2packed1 = yuv2rgbx64be_1_c;
2746 *yuv2packed2 = yuv2rgbx64be_2_c;
2747 *yuv2packedX = yuv2rgbx64be_X_c;
2750 case AV_PIX_FMT_BGRA64LE:
2751 #if CONFIG_SWSCALE_ALPHA
2753 *yuv2packed1 = yuv2bgra64le_1_c;
2754 *yuv2packed2 = yuv2bgra64le_2_c;
2755 *yuv2packedX = yuv2bgra64le_X_c;
2757 #endif /* CONFIG_SWSCALE_ALPHA */
2759 *yuv2packed1 = yuv2bgrx64le_1_c;
2760 *yuv2packed2 = yuv2bgrx64le_2_c;
2761 *yuv2packedX = yuv2bgrx64le_X_c;
2764 case AV_PIX_FMT_BGRA64BE:
2765 #if CONFIG_SWSCALE_ALPHA
2767 *yuv2packed1 = yuv2bgra64be_1_c;
2768 *yuv2packed2 = yuv2bgra64be_2_c;
2769 *yuv2packedX = yuv2bgra64be_X_c;
2771 #endif /* CONFIG_SWSCALE_ALPHA */
2773 *yuv2packed1 = yuv2bgrx64be_1_c;
2774 *yuv2packed2 = yuv2bgrx64be_2_c;
2775 *yuv2packedX = yuv2bgrx64be_X_c;
2778 case AV_PIX_FMT_RGB48LE:
2779 *yuv2packed1 = yuv2rgb48le_1_c;
2780 *yuv2packed2 = yuv2rgb48le_2_c;
2781 *yuv2packedX = yuv2rgb48le_X_c;
2783 case AV_PIX_FMT_RGB48BE:
2784 *yuv2packed1 = yuv2rgb48be_1_c;
2785 *yuv2packed2 = yuv2rgb48be_2_c;
2786 *yuv2packedX = yuv2rgb48be_X_c;
2788 case AV_PIX_FMT_BGR48LE:
2789 *yuv2packed1 = yuv2bgr48le_1_c;
2790 *yuv2packed2 = yuv2bgr48le_2_c;
2791 *yuv2packedX = yuv2bgr48le_X_c;
2793 case AV_PIX_FMT_BGR48BE:
2794 *yuv2packed1 = yuv2bgr48be_1_c;
2795 *yuv2packed2 = yuv2bgr48be_2_c;
2796 *yuv2packedX = yuv2bgr48be_X_c;
2798 case AV_PIX_FMT_RGB32:
2799 case AV_PIX_FMT_BGR32:
2801 *yuv2packed1 = yuv2rgb32_1_c;
2802 *yuv2packed2 = yuv2rgb32_2_c;
2803 *yuv2packedX = yuv2rgb32_X_c;
2805 #if CONFIG_SWSCALE_ALPHA
2807 *yuv2packed1 = yuv2rgba32_1_c;
2808 *yuv2packed2 = yuv2rgba32_2_c;
2809 *yuv2packedX = yuv2rgba32_X_c;
2811 #endif /* CONFIG_SWSCALE_ALPHA */
2813 *yuv2packed1 = yuv2rgbx32_1_c;
2814 *yuv2packed2 = yuv2rgbx32_2_c;
2815 *yuv2packedX = yuv2rgbx32_X_c;
2817 #endif /* !CONFIG_SMALL */
2819 case AV_PIX_FMT_RGB32_1:
2820 case AV_PIX_FMT_BGR32_1:
2822 *yuv2packed1 = yuv2rgb32_1_1_c;
2823 *yuv2packed2 = yuv2rgb32_1_2_c;
2824 *yuv2packedX = yuv2rgb32_1_X_c;
2826 #if CONFIG_SWSCALE_ALPHA
2828 *yuv2packed1 = yuv2rgba32_1_1_c;
2829 *yuv2packed2 = yuv2rgba32_1_2_c;
2830 *yuv2packedX = yuv2rgba32_1_X_c;
2832 #endif /* CONFIG_SWSCALE_ALPHA */
2834 *yuv2packed1 = yuv2rgbx32_1_1_c;
2835 *yuv2packed2 = yuv2rgbx32_1_2_c;
2836 *yuv2packedX = yuv2rgbx32_1_X_c;
2838 #endif /* !CONFIG_SMALL */
2840 case AV_PIX_FMT_RGB24:
2841 *yuv2packed1 = yuv2rgb24_1_c;
2842 *yuv2packed2 = yuv2rgb24_2_c;
2843 *yuv2packedX = yuv2rgb24_X_c;
2845 case AV_PIX_FMT_BGR24:
2846 *yuv2packed1 = yuv2bgr24_1_c;
2847 *yuv2packed2 = yuv2bgr24_2_c;
2848 *yuv2packedX = yuv2bgr24_X_c;
2850 case AV_PIX_FMT_RGB565LE:
2851 case AV_PIX_FMT_RGB565BE:
2852 case AV_PIX_FMT_BGR565LE:
2853 case AV_PIX_FMT_BGR565BE:
2854 *yuv2packed1 = yuv2rgb16_1_c;
2855 *yuv2packed2 = yuv2rgb16_2_c;
2856 *yuv2packedX = yuv2rgb16_X_c;
2858 case AV_PIX_FMT_RGB555LE:
2859 case AV_PIX_FMT_RGB555BE:
2860 case AV_PIX_FMT_BGR555LE:
2861 case AV_PIX_FMT_BGR555BE:
2862 *yuv2packed1 = yuv2rgb15_1_c;
2863 *yuv2packed2 = yuv2rgb15_2_c;
2864 *yuv2packedX = yuv2rgb15_X_c;
2866 case AV_PIX_FMT_RGB444LE:
2867 case AV_PIX_FMT_RGB444BE:
2868 case AV_PIX_FMT_BGR444LE:
2869 case AV_PIX_FMT_BGR444BE:
2870 *yuv2packed1 = yuv2rgb12_1_c;
2871 *yuv2packed2 = yuv2rgb12_2_c;
2872 *yuv2packedX = yuv2rgb12_X_c;
2874 case AV_PIX_FMT_RGB8:
2875 case AV_PIX_FMT_BGR8:
2876 *yuv2packed1 = yuv2rgb8_1_c;
2877 *yuv2packed2 = yuv2rgb8_2_c;
2878 *yuv2packedX = yuv2rgb8_X_c;
2880 case AV_PIX_FMT_RGB4:
2881 case AV_PIX_FMT_BGR4:
2882 *yuv2packed1 = yuv2rgb4_1_c;
2883 *yuv2packed2 = yuv2rgb4_2_c;
2884 *yuv2packedX = yuv2rgb4_X_c;
2886 case AV_PIX_FMT_RGB4_BYTE:
2887 case AV_PIX_FMT_BGR4_BYTE:
2888 *yuv2packed1 = yuv2rgb4b_1_c;
2889 *yuv2packed2 = yuv2rgb4b_2_c;
2890 *yuv2packedX = yuv2rgb4b_X_c;
2894 switch (dstFormat) {
2895 case AV_PIX_FMT_MONOWHITE:
2896 *yuv2packed1 = yuv2monowhite_1_c;
2897 *yuv2packed2 = yuv2monowhite_2_c;
2898 *yuv2packedX = yuv2monowhite_X_c;
2900 case AV_PIX_FMT_MONOBLACK:
2901 *yuv2packed1 = yuv2monoblack_1_c;
2902 *yuv2packed2 = yuv2monoblack_2_c;
2903 *yuv2packedX = yuv2monoblack_X_c;
2905 case AV_PIX_FMT_YUYV422:
2906 *yuv2packed1 = yuv2yuyv422_1_c;
2907 *yuv2packed2 = yuv2yuyv422_2_c;
2908 *yuv2packedX = yuv2yuyv422_X_c;
2910 case AV_PIX_FMT_YVYU422:
2911 *yuv2packed1 = yuv2yvyu422_1_c;
2912 *yuv2packed2 = yuv2yvyu422_2_c;
2913 *yuv2packedX = yuv2yvyu422_X_c;
2915 case AV_PIX_FMT_UYVY422:
2916 *yuv2packed1 = yuv2uyvy422_1_c;
2917 *yuv2packed2 = yuv2uyvy422_2_c;
2918 *yuv2packedX = yuv2uyvy422_X_c;
2920 case AV_PIX_FMT_YA8:
2921 *yuv2packed1 = yuv2ya8_1_c;
2922 *yuv2packed2 = yuv2ya8_2_c;
2923 *yuv2packedX = yuv2ya8_X_c;
2925 case AV_PIX_FMT_YA16LE:
2926 *yuv2packed1 = yuv2ya16le_1_c;
2927 *yuv2packed2 = yuv2ya16le_2_c;
2928 *yuv2packedX = yuv2ya16le_X_c;
2930 case AV_PIX_FMT_YA16BE:
2931 *yuv2packed1 = yuv2ya16be_1_c;
2932 *yuv2packed2 = yuv2ya16be_2_c;
2933 *yuv2packedX = yuv2ya16be_X_c;
2935 case AV_PIX_FMT_AYUV64LE:
2936 *yuv2packedX = yuv2ayuv64le_X_c;