FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
output.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2001-2012 Michael Niedermayer <michaelni@gmx.at>
3  *
4  * This file is part of FFmpeg.
5  *
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.
10  *
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.
15  *
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
19  */
20 
21 #include <math.h>
22 #include <stdint.h>
23 #include <stdio.h>
24 #include <string.h>
25 
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"
34 #include "config.h"
35 #include "rgb2rgb.h"
36 #include "swscale.h"
37 #include "swscale_internal.h"
38 
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, },
43 };
44 
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, },
49 };
50 
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, },
57 };
58 
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, },
69 };
70 
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, },
81 };
82 
83 #if 1
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, },
94 };
95 #elif 1
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, },
107 };
108 #elif 1
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, },
120 };
121 #else
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, },
133 };
134 #endif
135 
136 #define output_pixel(pos, val, bias, signedness) \
137  if (big_endian) { \
138  AV_WB16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
139  } else { \
140  AV_WL16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
141  }
142 
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)
146 {
147  int i;
148  int shift = 3;
149  av_assert0(output_bits == 16);
150 
151  for (i = 0; i < dstW; i++) {
152  int val = src[i] + (1 << (shift - 1));
153  output_pixel(&dest[i], val, 0, uint);
154  }
155 }
156 
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)
161 {
162  int i;
163  int shift = 15;
164  av_assert0(output_bits == 16);
165 
166  for (i = 0; i < dstW; i++) {
167  int val = 1 << (shift - 1);
168  int j;
169 
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. */
175  val -= 0x40000000;
176  for (j = 0; j < filterSize; j++)
177  val += src[j][i] * (unsigned)filter[j];
178 
179  output_pixel(&dest[i], val, 0x8000, int);
180  }
181 }
182 
183 #undef output_pixel
184 
185 #define output_pixel(pos, val) \
186  if (big_endian) { \
187  AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
188  } else { \
189  AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
190  }
191 
192 static av_always_inline void
193 yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
194  int big_endian, int output_bits)
195 {
196  int i;
197  int shift = 15 - output_bits;
198 
199  for (i = 0; i < dstW; i++) {
200  int val = src[i] + (1 << (shift - 1));
201  output_pixel(&dest[i], val);
202  }
203 }
204 
205 static av_always_inline void
206 yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
207  const int16_t **src, uint16_t *dest, int dstW,
208  int big_endian, int output_bits)
209 {
210  int i;
211  int shift = 11 + 16 - output_bits;
212 
213  for (i = 0; i < dstW; i++) {
214  int val = 1 << (shift - 1);
215  int j;
216 
217  for (j = 0; j < filterSize; j++)
218  val += src[j][i] * filter[j];
219 
220  output_pixel(&dest[i], val);
221  }
222 }
223 
224 #undef output_pixel
225 
226 #define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
227 static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
228  uint8_t *dest, int dstW, \
229  const uint8_t *dither, int offset)\
230 { \
231  yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
232  (uint16_t *) dest, dstW, is_be, bits); \
233 }\
234 static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
235  const int16_t **src, uint8_t *dest, int dstW, \
236  const uint8_t *dither, int offset)\
237 { \
238  yuv2planeX_## template_size ## _c_template(filter, \
239  filterSize, (const typeX_t **) src, \
240  (uint16_t *) dest, dstW, is_be, bits); \
241 }
242 yuv2NBPS( 9, BE, 1, 10, int16_t)
243 yuv2NBPS( 9, LE, 0, 10, int16_t)
244 yuv2NBPS(10, BE, 1, 10, int16_t)
245 yuv2NBPS(10, LE, 0, 10, int16_t)
246 yuv2NBPS(12, BE, 1, 10, int16_t)
247 yuv2NBPS(12, LE, 0, 10, int16_t)
248 yuv2NBPS(14, BE, 1, 10, int16_t)
249 yuv2NBPS(14, LE, 0, 10, int16_t)
250 yuv2NBPS(16, BE, 1, 16, int32_t)
251 yuv2NBPS(16, LE, 0, 16, int32_t)
252 
253 static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
254  const int16_t **src, uint8_t *dest, int dstW,
255  const uint8_t *dither, int offset)
256 {
257  int i;
258  for (i=0; i<dstW; i++) {
259  int val = dither[(i + offset) & 7] << 12;
260  int j;
261  for (j=0; j<filterSize; j++)
262  val += src[j][i] * filter[j];
263 
264  dest[i]= av_clip_uint8(val>>19);
265  }
266 }
267 
268 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
269  const uint8_t *dither, int offset)
270 {
271  int i;
272  for (i=0; i<dstW; i++) {
273  int val = (src[i] + dither[(i + offset) & 7]) >> 7;
274  dest[i]= av_clip_uint8(val);
275  }
276 }
277 
278 static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
279  const int16_t **chrUSrc, const int16_t **chrVSrc,
280  uint8_t *dest, int chrDstW)
281 {
282  enum AVPixelFormat dstFormat = c->dstFormat;
283  const uint8_t *chrDither = c->chrDither8;
284  int i;
285 
286  if (dstFormat == AV_PIX_FMT_NV12)
287  for (i=0; i<chrDstW; i++) {
288  int u = chrDither[i & 7] << 12;
289  int v = chrDither[(i + 3) & 7] << 12;
290  int j;
291  for (j=0; j<chrFilterSize; j++) {
292  u += chrUSrc[j][i] * chrFilter[j];
293  v += chrVSrc[j][i] * chrFilter[j];
294  }
295 
296  dest[2*i]= av_clip_uint8(u>>19);
297  dest[2*i+1]= av_clip_uint8(v>>19);
298  }
299  else
300  for (i=0; i<chrDstW; i++) {
301  int u = chrDither[i & 7] << 12;
302  int v = chrDither[(i + 3) & 7] << 12;
303  int j;
304  for (j=0; j<chrFilterSize; j++) {
305  u += chrUSrc[j][i] * chrFilter[j];
306  v += chrVSrc[j][i] * chrFilter[j];
307  }
308 
309  dest[2*i]= av_clip_uint8(v>>19);
310  dest[2*i+1]= av_clip_uint8(u>>19);
311  }
312 }
313 
314 #define accumulate_bit(acc, val) \
315  acc <<= 1; \
316  acc |= (val) >= (128 + 110)
317 #define output_pixel(pos, acc) \
318  if (target == AV_PIX_FMT_MONOBLACK) { \
319  pos = acc; \
320  } else { \
321  pos = ~acc; \
322  }
323 
324 static av_always_inline void
325 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
326  const int16_t **lumSrc, int lumFilterSize,
327  const int16_t *chrFilter, const int16_t **chrUSrc,
328  const int16_t **chrVSrc, int chrFilterSize,
329  const int16_t **alpSrc, uint8_t *dest, int dstW,
330  int y, enum AVPixelFormat target)
331 {
332  const uint8_t * const d128 = ff_dither_8x8_220[y&7];
333  int i;
334  unsigned acc = 0;
335  int err = 0;
336 
337  for (i = 0; i < dstW; i += 2) {
338  int j;
339  int Y1 = 1 << 18;
340  int Y2 = 1 << 18;
341 
342  for (j = 0; j < lumFilterSize; j++) {
343  Y1 += lumSrc[j][i] * lumFilter[j];
344  Y2 += lumSrc[j][i+1] * lumFilter[j];
345  }
346  Y1 >>= 19;
347  Y2 >>= 19;
348  if ((Y1 | Y2) & 0x100) {
349  Y1 = av_clip_uint8(Y1);
350  Y2 = av_clip_uint8(Y2);
351  }
352  if (c->dither == SWS_DITHER_ED) {
353  Y1 += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
354  c->dither_error[0][i] = err;
355  acc = 2*acc + (Y1 >= 128);
356  Y1 -= 220*(acc&1);
357 
358  err = Y2 + ((7*Y1 + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4);
359  c->dither_error[0][i+1] = Y1;
360  acc = 2*acc + (err >= 128);
361  err -= 220*(acc&1);
362  } else {
363  accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
364  accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
365  }
366  if ((i & 7) == 6) {
367  output_pixel(*dest++, acc);
368  }
369  }
370  c->dither_error[0][i] = err;
371 
372  if (i & 6) {
373  output_pixel(*dest, acc);
374  }
375 }
376 
377 static av_always_inline void
378 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
379  const int16_t *ubuf[2], const int16_t *vbuf[2],
380  const int16_t *abuf[2], uint8_t *dest, int dstW,
381  int yalpha, int uvalpha, int y,
382  enum AVPixelFormat target)
383 {
384  const int16_t *buf0 = buf[0], *buf1 = buf[1];
385  const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
386  int yalpha1 = 4096 - yalpha;
387  int i;
388 
389  if (c->dither == SWS_DITHER_ED) {
390  int err = 0;
391  int acc = 0;
392  for (i = 0; i < dstW; i +=2) {
393  int Y;
394 
395  Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
396  Y += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
397  c->dither_error[0][i] = err;
398  acc = 2*acc + (Y >= 128);
399  Y -= 220*(acc&1);
400 
401  err = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
402  err += (7*Y + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4;
403  c->dither_error[0][i+1] = Y;
404  acc = 2*acc + (err >= 128);
405  err -= 220*(acc&1);
406 
407  if ((i & 7) == 6)
408  output_pixel(*dest++, acc);
409  }
410  c->dither_error[0][i] = err;
411  } else {
412  for (i = 0; i < dstW; i += 8) {
413  int Y, acc = 0;
414 
415  Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
416  accumulate_bit(acc, Y + d128[0]);
417  Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
418  accumulate_bit(acc, Y + d128[1]);
419  Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
420  accumulate_bit(acc, Y + d128[2]);
421  Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
422  accumulate_bit(acc, Y + d128[3]);
423  Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
424  accumulate_bit(acc, Y + d128[4]);
425  Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
426  accumulate_bit(acc, Y + d128[5]);
427  Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
428  accumulate_bit(acc, Y + d128[6]);
429  Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
430  accumulate_bit(acc, Y + d128[7]);
431 
432  output_pixel(*dest++, acc);
433  }
434  }
435 }
436 
437 static av_always_inline void
438 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
439  const int16_t *ubuf[2], const int16_t *vbuf[2],
440  const int16_t *abuf0, uint8_t *dest, int dstW,
441  int uvalpha, int y, enum AVPixelFormat target)
442 {
443  const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
444  int i;
445 
446  if (c->dither == SWS_DITHER_ED) {
447  int err = 0;
448  int acc = 0;
449  for (i = 0; i < dstW; i +=2) {
450  int Y;
451 
452  Y = ((buf0[i + 0] + 64) >> 7);
453  Y += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
454  c->dither_error[0][i] = err;
455  acc = 2*acc + (Y >= 128);
456  Y -= 220*(acc&1);
457 
458  err = ((buf0[i + 1] + 64) >> 7);
459  err += (7*Y + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4;
460  c->dither_error[0][i+1] = Y;
461  acc = 2*acc + (err >= 128);
462  err -= 220*(acc&1);
463 
464  if ((i & 7) == 6)
465  output_pixel(*dest++, acc);
466  }
467  c->dither_error[0][i] = err;
468  } else {
469  for (i = 0; i < dstW; i += 8) {
470  int acc = 0;
471  accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
472  accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
473  accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
474  accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
475  accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
476  accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
477  accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
478  accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
479 
480  output_pixel(*dest++, acc);
481  }
482  }
483 }
484 
485 #undef output_pixel
486 #undef accumulate_bit
487 
488 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
489 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
490  const int16_t **lumSrc, int lumFilterSize, \
491  const int16_t *chrFilter, const int16_t **chrUSrc, \
492  const int16_t **chrVSrc, int chrFilterSize, \
493  const int16_t **alpSrc, uint8_t *dest, int dstW, \
494  int y) \
495 { \
496  name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
497  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
498  alpSrc, dest, dstW, y, fmt); \
499 } \
500  \
501 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
502  const int16_t *ubuf[2], const int16_t *vbuf[2], \
503  const int16_t *abuf[2], uint8_t *dest, int dstW, \
504  int yalpha, int uvalpha, int y) \
505 { \
506  name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
507  dest, dstW, yalpha, uvalpha, y, fmt); \
508 } \
509  \
510 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
511  const int16_t *ubuf[2], const int16_t *vbuf[2], \
512  const int16_t *abuf0, uint8_t *dest, int dstW, \
513  int uvalpha, int y) \
514 { \
515  name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
516  abuf0, dest, dstW, uvalpha, \
517  y, fmt); \
518 }
519 
520 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
521 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
522 
523 #define output_pixels(pos, Y1, U, Y2, V) \
524  if (target == AV_PIX_FMT_YUYV422) { \
525  dest[pos + 0] = Y1; \
526  dest[pos + 1] = U; \
527  dest[pos + 2] = Y2; \
528  dest[pos + 3] = V; \
529  } else if (target == AV_PIX_FMT_YVYU422) { \
530  dest[pos + 0] = Y1; \
531  dest[pos + 1] = V; \
532  dest[pos + 2] = Y2; \
533  dest[pos + 3] = U; \
534  } else { /* AV_PIX_FMT_UYVY422 */ \
535  dest[pos + 0] = U; \
536  dest[pos + 1] = Y1; \
537  dest[pos + 2] = V; \
538  dest[pos + 3] = Y2; \
539  }
540 
541 static av_always_inline void
542 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
543  const int16_t **lumSrc, int lumFilterSize,
544  const int16_t *chrFilter, const int16_t **chrUSrc,
545  const int16_t **chrVSrc, int chrFilterSize,
546  const int16_t **alpSrc, uint8_t *dest, int dstW,
547  int y, enum AVPixelFormat target)
548 {
549  int i;
550 
551  for (i = 0; i < ((dstW + 1) >> 1); i++) {
552  int j;
553  int Y1 = 1 << 18;
554  int Y2 = 1 << 18;
555  int U = 1 << 18;
556  int V = 1 << 18;
557 
558  for (j = 0; j < lumFilterSize; j++) {
559  Y1 += lumSrc[j][i * 2] * lumFilter[j];
560  Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
561  }
562  for (j = 0; j < chrFilterSize; j++) {
563  U += chrUSrc[j][i] * chrFilter[j];
564  V += chrVSrc[j][i] * chrFilter[j];
565  }
566  Y1 >>= 19;
567  Y2 >>= 19;
568  U >>= 19;
569  V >>= 19;
570  if ((Y1 | Y2 | U | V) & 0x100) {
571  Y1 = av_clip_uint8(Y1);
572  Y2 = av_clip_uint8(Y2);
573  U = av_clip_uint8(U);
574  V = av_clip_uint8(V);
575  }
576  output_pixels(4*i, Y1, U, Y2, V);
577  }
578 }
579 
580 static av_always_inline void
581 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
582  const int16_t *ubuf[2], const int16_t *vbuf[2],
583  const int16_t *abuf[2], uint8_t *dest, int dstW,
584  int yalpha, int uvalpha, int y,
585  enum AVPixelFormat target)
586 {
587  const int16_t *buf0 = buf[0], *buf1 = buf[1],
588  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
589  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
590  int yalpha1 = 4096 - yalpha;
591  int uvalpha1 = 4096 - uvalpha;
592  int i;
593 
594  for (i = 0; i < ((dstW + 1) >> 1); i++) {
595  int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
596  int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
597  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
598  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
599 
600  if ((Y1 | Y2 | U | V) & 0x100) {
601  Y1 = av_clip_uint8(Y1);
602  Y2 = av_clip_uint8(Y2);
603  U = av_clip_uint8(U);
604  V = av_clip_uint8(V);
605  }
606 
607  output_pixels(i * 4, Y1, U, Y2, V);
608  }
609 }
610 
611 static av_always_inline void
612 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
613  const int16_t *ubuf[2], const int16_t *vbuf[2],
614  const int16_t *abuf0, uint8_t *dest, int dstW,
615  int uvalpha, int y, enum AVPixelFormat target)
616 {
617  const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
618  int i;
619 
620  if (uvalpha < 2048) {
621  for (i = 0; i < ((dstW + 1) >> 1); i++) {
622  int Y1 = (buf0[i * 2 ]+64) >> 7;
623  int Y2 = (buf0[i * 2 + 1]+64) >> 7;
624  int U = (ubuf0[i] +64) >> 7;
625  int V = (vbuf0[i] +64) >> 7;
626 
627  if ((Y1 | Y2 | U | V) & 0x100) {
628  Y1 = av_clip_uint8(Y1);
629  Y2 = av_clip_uint8(Y2);
630  U = av_clip_uint8(U);
631  V = av_clip_uint8(V);
632  }
633 
634  Y1 = av_clip_uint8(Y1);
635  Y2 = av_clip_uint8(Y2);
636  U = av_clip_uint8(U);
637  V = av_clip_uint8(V);
638 
639  output_pixels(i * 4, Y1, U, Y2, V);
640  }
641  } else {
642  const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
643  for (i = 0; i < ((dstW + 1) >> 1); i++) {
644  int Y1 = (buf0[i * 2 ] + 64) >> 7;
645  int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
646  int U = (ubuf0[i] + ubuf1[i]+128) >> 8;
647  int V = (vbuf0[i] + vbuf1[i]+128) >> 8;
648 
649  if ((Y1 | Y2 | U | V) & 0x100) {
650  Y1 = av_clip_uint8(Y1);
651  Y2 = av_clip_uint8(Y2);
652  U = av_clip_uint8(U);
653  V = av_clip_uint8(V);
654  }
655 
656  Y1 = av_clip_uint8(Y1);
657  Y2 = av_clip_uint8(Y2);
658  U = av_clip_uint8(U);
659  V = av_clip_uint8(V);
660 
661  output_pixels(i * 4, Y1, U, Y2, V);
662  }
663  }
664 }
665 
666 #undef output_pixels
667 
668 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422)
669 YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422)
670 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422)
671 
672 #define R_B ((target == AV_PIX_FMT_RGB48LE || target == AV_PIX_FMT_RGB48BE || target == AV_PIX_FMT_RGBA64LE || target == AV_PIX_FMT_RGBA64BE) ? R : B)
673 #define B_R ((target == AV_PIX_FMT_RGB48LE || target == AV_PIX_FMT_RGB48BE || target == AV_PIX_FMT_RGBA64LE || target == AV_PIX_FMT_RGBA64BE) ? B : R)
674 #define output_pixel(pos, val) \
675  if (isBE(target)) { \
676  AV_WB16(pos, val); \
677  } else { \
678  AV_WL16(pos, val); \
679  }
680 
681 static av_always_inline void
682 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
683  const int32_t **lumSrc, int lumFilterSize,
684  const int16_t *chrFilter, const int32_t **chrUSrc,
685  const int32_t **chrVSrc, int chrFilterSize,
686  const int32_t **alpSrc, uint16_t *dest, int dstW,
687  int y, enum AVPixelFormat target, int hasAlpha)
688 {
689  int i;
690  int A1 = 0xffff<<14, A2 = 0xffff<<14;
691 
692  for (i = 0; i < ((dstW + 1) >> 1); i++) {
693  int j;
694  int Y1 = -0x40000000;
695  int Y2 = -0x40000000;
696  int U = -128 << 23; // 19
697  int V = -128 << 23;
698  int R, G, B;
699 
700  for (j = 0; j < lumFilterSize; j++) {
701  Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
702  Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
703  }
704  for (j = 0; j < chrFilterSize; j++) {;
705  U += chrUSrc[j][i] * (unsigned)chrFilter[j];
706  V += chrVSrc[j][i] * (unsigned)chrFilter[j];
707  }
708 
709  if (hasAlpha) {
710  A1 = -0x40000000;
711  A2 = -0x40000000;
712  for (j = 0; j < lumFilterSize; j++) {
713  A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
714  A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
715  }
716  A1 >>= 1;
717  A1 += 0x20002000;
718  A2 >>= 1;
719  A2 += 0x20002000;
720  }
721 
722  // 8bit: 12+15=27; 16-bit: 12+19=31
723  Y1 >>= 14; // 10
724  Y1 += 0x10000;
725  Y2 >>= 14;
726  Y2 += 0x10000;
727  U >>= 14;
728  V >>= 14;
729 
730  // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
731  Y1 -= c->yuv2rgb_y_offset;
732  Y2 -= c->yuv2rgb_y_offset;
733  Y1 *= c->yuv2rgb_y_coeff;
734  Y2 *= c->yuv2rgb_y_coeff;
735  Y1 += 1 << 13; // 21
736  Y2 += 1 << 13;
737  // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
738 
739  R = V * c->yuv2rgb_v2r_coeff;
740  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
741  B = U * c->yuv2rgb_u2b_coeff;
742 
743  // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
744  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
745  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
746  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
747  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
748  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
749  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
750  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
751  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
752  dest += 8;
753  }
754 }
755 
756 static av_always_inline void
758  const int32_t *ubuf[2], const int32_t *vbuf[2],
759  const int32_t *abuf[2], uint16_t *dest, int dstW,
760  int yalpha, int uvalpha, int y,
761  enum AVPixelFormat target, int hasAlpha)
762 {
763  const int32_t *buf0 = buf[0], *buf1 = buf[1],
764  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
765  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
766  *abuf0 = hasAlpha ? abuf[0] : NULL,
767  *abuf1 = hasAlpha ? abuf[1] : NULL;
768  int yalpha1 = 4096 - yalpha;
769  int uvalpha1 = 4096 - uvalpha;
770  int i;
771  int A1 = 0xffff<<14, A2 = 0xffff<<14;
772 
773  for (i = 0; i < ((dstW + 1) >> 1); i++) {
774  int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
775  int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
776  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha + (-128 << 23)) >> 14;
777  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha + (-128 << 23)) >> 14;
778  int R, G, B;
779 
780  Y1 -= c->yuv2rgb_y_offset;
781  Y2 -= c->yuv2rgb_y_offset;
782  Y1 *= c->yuv2rgb_y_coeff;
783  Y2 *= c->yuv2rgb_y_coeff;
784  Y1 += 1 << 13;
785  Y2 += 1 << 13;
786 
787  R = V * c->yuv2rgb_v2r_coeff;
788  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
789  B = U * c->yuv2rgb_u2b_coeff;
790 
791  if (hasAlpha) {
792  A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
793  A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
794 
795  A1 += 1 << 13;
796  A2 += 1 << 13;
797  }
798 
799  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
800  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
801  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
802  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
803  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
804  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
805  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
806  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
807  dest += 8;
808  }
809 }
810 
811 static av_always_inline void
813  const int32_t *ubuf[2], const int32_t *vbuf[2],
814  const int32_t *abuf0, uint16_t *dest, int dstW,
815  int uvalpha, int y, enum AVPixelFormat target, int hasAlpha)
816 {
817  const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
818  int i;
819  int A1 = 0xffff<<14, A2= 0xffff<<14;
820 
821  if (uvalpha < 2048) {
822  for (i = 0; i < ((dstW + 1) >> 1); i++) {
823  int Y1 = (buf0[i * 2] ) >> 2;
824  int Y2 = (buf0[i * 2 + 1]) >> 2;
825  int U = (ubuf0[i] + (-128 << 11)) >> 2;
826  int V = (vbuf0[i] + (-128 << 11)) >> 2;
827  int R, G, B;
828 
829  Y1 -= c->yuv2rgb_y_offset;
830  Y2 -= c->yuv2rgb_y_offset;
831  Y1 *= c->yuv2rgb_y_coeff;
832  Y2 *= c->yuv2rgb_y_coeff;
833  Y1 += 1 << 13;
834  Y2 += 1 << 13;
835 
836  if (hasAlpha) {
837  A1 = abuf0[i * 2 ] << 11;
838  A2 = abuf0[i * 2 + 1] << 11;
839 
840  A1 += 1 << 13;
841  A2 += 1 << 13;
842  }
843 
844  R = V * c->yuv2rgb_v2r_coeff;
845  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
846  B = U * c->yuv2rgb_u2b_coeff;
847 
848  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
849  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
850  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
851  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
852  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
853  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
854  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
855  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
856  dest += 8;
857  }
858  } else {
859  const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
860  int A1 = 0xffff<<14, A2 = 0xffff<<14;
861  for (i = 0; i < ((dstW + 1) >> 1); i++) {
862  int Y1 = (buf0[i * 2] ) >> 2;
863  int Y2 = (buf0[i * 2 + 1]) >> 2;
864  int U = (ubuf0[i] + ubuf1[i] + (-128 << 12)) >> 3;
865  int V = (vbuf0[i] + vbuf1[i] + (-128 << 12)) >> 3;
866  int R, G, B;
867 
868  Y1 -= c->yuv2rgb_y_offset;
869  Y2 -= c->yuv2rgb_y_offset;
870  Y1 *= c->yuv2rgb_y_coeff;
871  Y2 *= c->yuv2rgb_y_coeff;
872  Y1 += 1 << 13;
873  Y2 += 1 << 13;
874 
875  if (hasAlpha) {
876  A1 = abuf0[i * 2 ] << 11;
877  A2 = abuf0[i * 2 + 1] << 11;
878 
879  A1 += 1 << 13;
880  A2 += 1 << 13;
881  }
882 
883  R = V * c->yuv2rgb_v2r_coeff;
884  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
885  B = U * c->yuv2rgb_u2b_coeff;
886 
887  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
888  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
889  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
890  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
891  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
892  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
893  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
894  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
895  dest += 8;
896  }
897  }
898 }
899 
900 static av_always_inline void
901 yuv2rgb48_X_c_template(SwsContext *c, const int16_t *lumFilter,
902  const int32_t **lumSrc, int lumFilterSize,
903  const int16_t *chrFilter, const int32_t **chrUSrc,
904  const int32_t **chrVSrc, int chrFilterSize,
905  const int32_t **alpSrc, uint16_t *dest, int dstW,
906  int y, enum AVPixelFormat target, int hasAlpha)
907 {
908  int i;
909 
910  for (i = 0; i < ((dstW + 1) >> 1); i++) {
911  int j;
912  int Y1 = -0x40000000;
913  int Y2 = -0x40000000;
914  int U = -128 << 23; // 19
915  int V = -128 << 23;
916  int R, G, B;
917 
918  for (j = 0; j < lumFilterSize; j++) {
919  Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
920  Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
921  }
922  for (j = 0; j < chrFilterSize; j++) {;
923  U += chrUSrc[j][i] * (unsigned)chrFilter[j];
924  V += chrVSrc[j][i] * (unsigned)chrFilter[j];
925  }
926 
927  // 8bit: 12+15=27; 16-bit: 12+19=31
928  Y1 >>= 14; // 10
929  Y1 += 0x10000;
930  Y2 >>= 14;
931  Y2 += 0x10000;
932  U >>= 14;
933  V >>= 14;
934 
935  // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
936  Y1 -= c->yuv2rgb_y_offset;
937  Y2 -= c->yuv2rgb_y_offset;
938  Y1 *= c->yuv2rgb_y_coeff;
939  Y2 *= c->yuv2rgb_y_coeff;
940  Y1 += 1 << 13; // 21
941  Y2 += 1 << 13;
942  // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
943 
944  R = V * c->yuv2rgb_v2r_coeff;
945  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
946  B = U * c->yuv2rgb_u2b_coeff;
947 
948  // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
949  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
950  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
951  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
952  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
953  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
954  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
955  dest += 6;
956  }
957 }
958 
959 static av_always_inline void
961  const int32_t *ubuf[2], const int32_t *vbuf[2],
962  const int32_t *abuf[2], uint16_t *dest, int dstW,
963  int yalpha, int uvalpha, int y,
964  enum AVPixelFormat target, int hasAlpha)
965 {
966  const int32_t *buf0 = buf[0], *buf1 = buf[1],
967  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
968  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
969  int yalpha1 = 4096 - yalpha;
970  int uvalpha1 = 4096 - uvalpha;
971  int i;
972 
973  for (i = 0; i < ((dstW + 1) >> 1); i++) {
974  int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
975  int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
976  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha + (-128 << 23)) >> 14;
977  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha + (-128 << 23)) >> 14;
978  int R, G, B;
979 
980  Y1 -= c->yuv2rgb_y_offset;
981  Y2 -= c->yuv2rgb_y_offset;
982  Y1 *= c->yuv2rgb_y_coeff;
983  Y2 *= c->yuv2rgb_y_coeff;
984  Y1 += 1 << 13;
985  Y2 += 1 << 13;
986 
987  R = V * c->yuv2rgb_v2r_coeff;
988  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
989  B = U * c->yuv2rgb_u2b_coeff;
990 
991  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
992  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
993  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
994  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
995  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
996  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
997  dest += 6;
998  }
999 }
1000 
1001 static av_always_inline void
1003  const int32_t *ubuf[2], const int32_t *vbuf[2],
1004  const int32_t *abuf0, uint16_t *dest, int dstW,
1005  int uvalpha, int y, enum AVPixelFormat target, int hasAlpha)
1006 {
1007  const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1008  int i;
1009 
1010  if (uvalpha < 2048) {
1011  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1012  int Y1 = (buf0[i * 2] ) >> 2;
1013  int Y2 = (buf0[i * 2 + 1]) >> 2;
1014  int U = (ubuf0[i] + (-128 << 11)) >> 2;
1015  int V = (vbuf0[i] + (-128 << 11)) >> 2;
1016  int R, G, B;
1017 
1018  Y1 -= c->yuv2rgb_y_offset;
1019  Y2 -= c->yuv2rgb_y_offset;
1020  Y1 *= c->yuv2rgb_y_coeff;
1021  Y2 *= c->yuv2rgb_y_coeff;
1022  Y1 += 1 << 13;
1023  Y2 += 1 << 13;
1024 
1025  R = V * c->yuv2rgb_v2r_coeff;
1026  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1027  B = U * c->yuv2rgb_u2b_coeff;
1028 
1029  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1030  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1031  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1032  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1033  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1034  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1035  dest += 6;
1036  }
1037  } else {
1038  const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1039  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1040  int Y1 = (buf0[i * 2] ) >> 2;
1041  int Y2 = (buf0[i * 2 + 1]) >> 2;
1042  int U = (ubuf0[i] + ubuf1[i] + (-128 << 12)) >> 3;
1043  int V = (vbuf0[i] + vbuf1[i] + (-128 << 12)) >> 3;
1044  int R, G, B;
1045 
1046  Y1 -= c->yuv2rgb_y_offset;
1047  Y2 -= c->yuv2rgb_y_offset;
1048  Y1 *= c->yuv2rgb_y_coeff;
1049  Y2 *= c->yuv2rgb_y_coeff;
1050  Y1 += 1 << 13;
1051  Y2 += 1 << 13;
1052 
1053  R = V * c->yuv2rgb_v2r_coeff;
1054  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1055  B = U * c->yuv2rgb_u2b_coeff;
1056 
1057  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1058  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1059  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1060  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1061  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1062  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1063  dest += 6;
1064  }
1065  }
1066 }
1067 
1068 #undef output_pixel
1069 #undef r_b
1070 #undef b_r
1071 
1072 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha) \
1073 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1074  const int16_t **_lumSrc, int lumFilterSize, \
1075  const int16_t *chrFilter, const int16_t **_chrUSrc, \
1076  const int16_t **_chrVSrc, int chrFilterSize, \
1077  const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1078  int y) \
1079 { \
1080  const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1081  **chrUSrc = (const int32_t **) _chrUSrc, \
1082  **chrVSrc = (const int32_t **) _chrVSrc, \
1083  **alpSrc = (const int32_t **) _alpSrc; \
1084  uint16_t *dest = (uint16_t *) _dest; \
1085  name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1086  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1087  alpSrc, dest, dstW, y, fmt, hasAlpha); \
1088 } \
1089  \
1090 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1091  const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1092  const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1093  int yalpha, int uvalpha, int y) \
1094 { \
1095  const int32_t **buf = (const int32_t **) _buf, \
1096  **ubuf = (const int32_t **) _ubuf, \
1097  **vbuf = (const int32_t **) _vbuf, \
1098  **abuf = (const int32_t **) _abuf; \
1099  uint16_t *dest = (uint16_t *) _dest; \
1100  name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1101  dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1102 } \
1103  \
1104 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1105  const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1106  const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1107  int uvalpha, int y) \
1108 { \
1109  const int32_t *buf0 = (const int32_t *) _buf0, \
1110  **ubuf = (const int32_t **) _ubuf, \
1111  **vbuf = (const int32_t **) _vbuf, \
1112  *abuf0 = (const int32_t *) _abuf0; \
1113  uint16_t *dest = (uint16_t *) _dest; \
1114  name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1115  dstW, uvalpha, y, fmt, hasAlpha); \
1116 }
1117 
1118 YUV2PACKED16WRAPPER(yuv2, rgb48, rgb48be, AV_PIX_FMT_RGB48BE, 0)
1119 YUV2PACKED16WRAPPER(yuv2, rgb48, rgb48le, AV_PIX_FMT_RGB48LE, 0)
1120 YUV2PACKED16WRAPPER(yuv2, rgb48, bgr48be, AV_PIX_FMT_BGR48BE, 0)
1121 YUV2PACKED16WRAPPER(yuv2, rgb48, bgr48le, AV_PIX_FMT_BGR48LE, 0)
1122 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1)
1123 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1)
1124 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0)
1125 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0)
1126 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1)
1127 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1)
1128 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0)
1129 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0)
1130 
1131 /*
1132  * Write out 2 RGB pixels in the target pixel format. This function takes a
1133  * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1134  * things like endianness conversion and shifting. The caller takes care of
1135  * setting the correct offset in these tables from the chroma (U/V) values.
1136  * This function then uses the luminance (Y1/Y2) values to write out the
1137  * correct RGB values into the destination buffer.
1138  */
1139 static av_always_inline void
1140 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1141  unsigned A1, unsigned A2,
1142  const void *_r, const void *_g, const void *_b, int y,
1143  enum AVPixelFormat target, int hasAlpha)
1144 {
1145  if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1146  target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1147  uint32_t *dest = (uint32_t *) _dest;
1148  const uint32_t *r = (const uint32_t *) _r;
1149  const uint32_t *g = (const uint32_t *) _g;
1150  const uint32_t *b = (const uint32_t *) _b;
1151 
1152 #if CONFIG_SMALL
1153  int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1154 
1155  dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1156  dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1157 #else
1158  if (hasAlpha) {
1159  int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1160 
1161  av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1162  dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1163  dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1164  } else {
1165 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1166  int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1167 
1168  av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1169 #endif
1170  dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1171  dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1172  }
1173 #endif
1174  } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1175  uint8_t *dest = (uint8_t *) _dest;
1176  const uint8_t *r = (const uint8_t *) _r;
1177  const uint8_t *g = (const uint8_t *) _g;
1178  const uint8_t *b = (const uint8_t *) _b;
1179 
1180 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1181 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1182 
1183  dest[i * 6 + 0] = r_b[Y1];
1184  dest[i * 6 + 1] = g[Y1];
1185  dest[i * 6 + 2] = b_r[Y1];
1186  dest[i * 6 + 3] = r_b[Y2];
1187  dest[i * 6 + 4] = g[Y2];
1188  dest[i * 6 + 5] = b_r[Y2];
1189 #undef r_b
1190 #undef b_r
1191  } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1192  target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1193  target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1194  uint16_t *dest = (uint16_t *) _dest;
1195  const uint16_t *r = (const uint16_t *) _r;
1196  const uint16_t *g = (const uint16_t *) _g;
1197  const uint16_t *b = (const uint16_t *) _b;
1198  int dr1, dg1, db1, dr2, dg2, db2;
1199 
1200  if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1201  dr1 = ff_dither_2x2_8[ y & 1 ][0];
1202  dg1 = ff_dither_2x2_4[ y & 1 ][0];
1203  db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1204  dr2 = ff_dither_2x2_8[ y & 1 ][1];
1205  dg2 = ff_dither_2x2_4[ y & 1 ][1];
1206  db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1207  } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1208  dr1 = ff_dither_2x2_8[ y & 1 ][0];
1209  dg1 = ff_dither_2x2_8[ y & 1 ][1];
1210  db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1211  dr2 = ff_dither_2x2_8[ y & 1 ][1];
1212  dg2 = ff_dither_2x2_8[ y & 1 ][0];
1213  db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1214  } else {
1215  dr1 = ff_dither_4x4_16[ y & 3 ][0];
1216  dg1 = ff_dither_4x4_16[ y & 3 ][1];
1217  db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1218  dr2 = ff_dither_4x4_16[ y & 3 ][1];
1219  dg2 = ff_dither_4x4_16[ y & 3 ][0];
1220  db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1221  }
1222 
1223  dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1224  dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1225  } else /* 8/4-bit */ {
1226  uint8_t *dest = (uint8_t *) _dest;
1227  const uint8_t *r = (const uint8_t *) _r;
1228  const uint8_t *g = (const uint8_t *) _g;
1229  const uint8_t *b = (const uint8_t *) _b;
1230  int dr1, dg1, db1, dr2, dg2, db2;
1231 
1232  if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1233  const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1234  const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1235  dr1 = dg1 = d32[(i * 2 + 0) & 7];
1236  db1 = d64[(i * 2 + 0) & 7];
1237  dr2 = dg2 = d32[(i * 2 + 1) & 7];
1238  db2 = d64[(i * 2 + 1) & 7];
1239  } else {
1240  const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1241  const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1242  dr1 = db1 = d128[(i * 2 + 0) & 7];
1243  dg1 = d64[(i * 2 + 0) & 7];
1244  dr2 = db2 = d128[(i * 2 + 1) & 7];
1245  dg2 = d64[(i * 2 + 1) & 7];
1246  }
1247 
1248  if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1249  dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1250  ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1251  } else {
1252  dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1253  dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1254  }
1255  }
1256 }
1257 
1258 static av_always_inline void
1259 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1260  const int16_t **lumSrc, int lumFilterSize,
1261  const int16_t *chrFilter, const int16_t **chrUSrc,
1262  const int16_t **chrVSrc, int chrFilterSize,
1263  const int16_t **alpSrc, uint8_t *dest, int dstW,
1264  int y, enum AVPixelFormat target, int hasAlpha)
1265 {
1266  int i;
1267 
1268  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1269  int j, A1, A2;
1270  int Y1 = 1 << 18;
1271  int Y2 = 1 << 18;
1272  int U = 1 << 18;
1273  int V = 1 << 18;
1274  const void *r, *g, *b;
1275 
1276  for (j = 0; j < lumFilterSize; j++) {
1277  Y1 += lumSrc[j][i * 2] * lumFilter[j];
1278  Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1279  }
1280  for (j = 0; j < chrFilterSize; j++) {
1281  U += chrUSrc[j][i] * chrFilter[j];
1282  V += chrVSrc[j][i] * chrFilter[j];
1283  }
1284  Y1 >>= 19;
1285  Y2 >>= 19;
1286  U >>= 19;
1287  V >>= 19;
1288  if (hasAlpha) {
1289  A1 = 1 << 18;
1290  A2 = 1 << 18;
1291  for (j = 0; j < lumFilterSize; j++) {
1292  A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1293  A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1294  }
1295  A1 >>= 19;
1296  A2 >>= 19;
1297  if ((A1 | A2) & 0x100) {
1298  A1 = av_clip_uint8(A1);
1299  A2 = av_clip_uint8(A2);
1300  }
1301  }
1302 
1303  r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1305  b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1306 
1307  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1308  r, g, b, y, target, hasAlpha);
1309  }
1310 }
1311 
1312 static av_always_inline void
1313 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1314  const int16_t *ubuf[2], const int16_t *vbuf[2],
1315  const int16_t *abuf[2], uint8_t *dest, int dstW,
1316  int yalpha, int uvalpha, int y,
1317  enum AVPixelFormat target, int hasAlpha)
1318 {
1319  const int16_t *buf0 = buf[0], *buf1 = buf[1],
1320  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1321  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1322  *abuf0 = hasAlpha ? abuf[0] : NULL,
1323  *abuf1 = hasAlpha ? abuf[1] : NULL;
1324  int yalpha1 = 4096 - yalpha;
1325  int uvalpha1 = 4096 - uvalpha;
1326  int i;
1327 
1328  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1329  int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1330  int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1331  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1332  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1333  int A1, A2;
1334  const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1336  *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1337 
1338  if (hasAlpha) {
1339  A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1340  A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1341  A1 = av_clip_uint8(A1);
1342  A2 = av_clip_uint8(A2);
1343  }
1344 
1345  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1346  r, g, b, y, target, hasAlpha);
1347  }
1348 }
1349 
1350 static av_always_inline void
1351 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1352  const int16_t *ubuf[2], const int16_t *vbuf[2],
1353  const int16_t *abuf0, uint8_t *dest, int dstW,
1354  int uvalpha, int y, enum AVPixelFormat target,
1355  int hasAlpha)
1356 {
1357  const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1358  int i;
1359 
1360  if (uvalpha < 2048) {
1361  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1362  int Y1 = (buf0[i * 2 ] + 64) >> 7;
1363  int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1364  int U = (ubuf0[i] + 64) >> 7;
1365  int V = (vbuf0[i] + 64) >> 7;
1366  int A1, A2;
1367  const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1369  *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1370 
1371  if (hasAlpha) {
1372  A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1373  A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1374  A1 = av_clip_uint8(A1);
1375  A2 = av_clip_uint8(A2);
1376  }
1377 
1378  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1379  r, g, b, y, target, hasAlpha);
1380  }
1381  } else {
1382  const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1383  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1384  int Y1 = (buf0[i * 2 ] + 64) >> 7;
1385  int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1386  int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1387  int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1388  int A1, A2;
1389  const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1391  *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1392 
1393  if (hasAlpha) {
1394  A1 = (abuf0[i * 2 ] + 64) >> 7;
1395  A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1396  A1 = av_clip_uint8(A1);
1397  A2 = av_clip_uint8(A2);
1398  }
1399 
1400  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1401  r, g, b, y, target, hasAlpha);
1402  }
1403  }
1404 }
1405 
1406 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1407 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1408  const int16_t **lumSrc, int lumFilterSize, \
1409  const int16_t *chrFilter, const int16_t **chrUSrc, \
1410  const int16_t **chrVSrc, int chrFilterSize, \
1411  const int16_t **alpSrc, uint8_t *dest, int dstW, \
1412  int y) \
1413 { \
1414  name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1415  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1416  alpSrc, dest, dstW, y, fmt, hasAlpha); \
1417 }
1418 
1419 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1420 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1421 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1422  const int16_t *ubuf[2], const int16_t *vbuf[2], \
1423  const int16_t *abuf[2], uint8_t *dest, int dstW, \
1424  int yalpha, int uvalpha, int y) \
1425 { \
1426  name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1427  dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1428 }
1429 
1430 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1431 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1432 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1433  const int16_t *ubuf[2], const int16_t *vbuf[2], \
1434  const int16_t *abuf0, uint8_t *dest, int dstW, \
1435  int uvalpha, int y) \
1436 { \
1437  name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1438  dstW, uvalpha, y, fmt, hasAlpha); \
1439 }
1440 
1441 #if CONFIG_SMALL
1442 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1443 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1444 #else
1445 #if CONFIG_SWSCALE_ALPHA
1446 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1447 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1448 #endif
1449 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1450 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1451 #endif
1452 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1453 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1454 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1455 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1456 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1457 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1458 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1459 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1460 
1462  uint8_t *dest, int i, int Y, int A, int U, int V,
1463  int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1464 {
1465  int R, G, B;
1466  int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1467 
1468  Y -= c->yuv2rgb_y_offset;
1469  Y *= c->yuv2rgb_y_coeff;
1470  Y += 1 << 21;
1471  R = Y + V*c->yuv2rgb_v2r_coeff;
1472  G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1473  B = Y + U*c->yuv2rgb_u2b_coeff;
1474  if ((R | G | B) & 0xC0000000) {
1475  R = av_clip_uintp2(R, 30);
1476  G = av_clip_uintp2(G, 30);
1477  B = av_clip_uintp2(B, 30);
1478  }
1479 
1480  switch(target) {
1481  case AV_PIX_FMT_ARGB:
1482  dest[0] = hasAlpha ? A : 255;
1483  dest[1] = R >> 22;
1484  dest[2] = G >> 22;
1485  dest[3] = B >> 22;
1486  break;
1487  case AV_PIX_FMT_RGB24:
1488  dest[0] = R >> 22;
1489  dest[1] = G >> 22;
1490  dest[2] = B >> 22;
1491  break;
1492  case AV_PIX_FMT_RGBA:
1493  dest[0] = R >> 22;
1494  dest[1] = G >> 22;
1495  dest[2] = B >> 22;
1496  dest[3] = hasAlpha ? A : 255;
1497  break;
1498  case AV_PIX_FMT_ABGR:
1499  dest[0] = hasAlpha ? A : 255;
1500  dest[1] = B >> 22;
1501  dest[2] = G >> 22;
1502  dest[3] = R >> 22;
1503  break;
1504  case AV_PIX_FMT_BGR24:
1505  dest[0] = B >> 22;
1506  dest[1] = G >> 22;
1507  dest[2] = R >> 22;
1508  break;
1509  case AV_PIX_FMT_BGRA:
1510  dest[0] = B >> 22;
1511  dest[1] = G >> 22;
1512  dest[2] = R >> 22;
1513  dest[3] = hasAlpha ? A : 255;
1514  break;
1515  case AV_PIX_FMT_BGR4_BYTE:
1516  case AV_PIX_FMT_RGB4_BYTE:
1517  case AV_PIX_FMT_BGR8:
1518  case AV_PIX_FMT_RGB8:
1519  {
1520  int r,g,b;
1521 
1522  switch (c->dither) {
1523  default:
1524  case SWS_DITHER_AUTO:
1525  case SWS_DITHER_ED:
1526  R >>= 22;
1527  G >>= 22;
1528  B >>= 22;
1529  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;
1530  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;
1531  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;
1532  c->dither_error[0][i] = err[0];
1533  c->dither_error[1][i] = err[1];
1534  c->dither_error[2][i] = err[2];
1535  r = R >> (isrgb8 ? 5 : 7);
1536  g = G >> (isrgb8 ? 5 : 6);
1537  b = B >> (isrgb8 ? 6 : 7);
1538  r = av_clip(r, 0, isrgb8 ? 7 : 1);
1539  g = av_clip(g, 0, isrgb8 ? 7 : 3);
1540  b = av_clip(b, 0, isrgb8 ? 3 : 1);
1541  err[0] = R - r*(isrgb8 ? 36 : 255);
1542  err[1] = G - g*(isrgb8 ? 36 : 85);
1543  err[2] = B - b*(isrgb8 ? 85 : 255);
1544  break;
1545  case SWS_DITHER_A_DITHER:
1546  if (isrgb8) {
1547  /* see http://pippin.gimp.org/a_dither/ for details/origin */
1548 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1549  r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1550  g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1551  b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1552  r = av_clip(r, 0, 7);
1553  g = av_clip(g, 0, 7);
1554  b = av_clip(b, 0, 3);
1555  } else {
1556  r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1557  g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1558  b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1559  r = av_clip(r, 0, 1);
1560  g = av_clip(g, 0, 3);
1561  b = av_clip(b, 0, 1);
1562  }
1563  break;
1564  case SWS_DITHER_X_DITHER:
1565  if (isrgb8) {
1566  /* see http://pippin.gimp.org/a_dither/ for details/origin */
1567 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1568  r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1569  g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1570  b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1571  r = av_clip(r, 0, 7);
1572  g = av_clip(g, 0, 7);
1573  b = av_clip(b, 0, 3);
1574  } else {
1575  r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1576  g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1577  b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1578  r = av_clip(r, 0, 1);
1579  g = av_clip(g, 0, 3);
1580  b = av_clip(b, 0, 1);
1581  }
1582 
1583  break;
1584  }
1585 
1586  if(target == AV_PIX_FMT_BGR4_BYTE) {
1587  dest[0] = r + 2*g + 8*b;
1588  } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1589  dest[0] = b + 2*g + 8*r;
1590  } else if(target == AV_PIX_FMT_BGR8) {
1591  dest[0] = r + 8*g + 64*b;
1592  } else if(target == AV_PIX_FMT_RGB8) {
1593  dest[0] = b + 4*g + 32*r;
1594  } else
1595  av_assert2(0);
1596  break;}
1597  }
1598 }
1599 
1600 static av_always_inline void
1601 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1602  const int16_t **lumSrc, int lumFilterSize,
1603  const int16_t *chrFilter, const int16_t **chrUSrc,
1604  const int16_t **chrVSrc, int chrFilterSize,
1605  const int16_t **alpSrc, uint8_t *dest,
1606  int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1607 {
1608  int i;
1609  int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1610  int err[4] = {0};
1611  int A = 0; //init to silence warning
1612 
1613  if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1614  || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1615  step = 1;
1616 
1617  for (i = 0; i < dstW; i++) {
1618  int j;
1619  int Y = 1<<9;
1620  int U = (1<<9)-(128 << 19);
1621  int V = (1<<9)-(128 << 19);
1622 
1623  for (j = 0; j < lumFilterSize; j++) {
1624  Y += lumSrc[j][i] * lumFilter[j];
1625  }
1626  for (j = 0; j < chrFilterSize; j++) {
1627  U += chrUSrc[j][i] * chrFilter[j];
1628  V += chrVSrc[j][i] * chrFilter[j];
1629  }
1630  Y >>= 10;
1631  U >>= 10;
1632  V >>= 10;
1633  if (hasAlpha) {
1634  A = 1 << 18;
1635  for (j = 0; j < lumFilterSize; j++) {
1636  A += alpSrc[j][i] * lumFilter[j];
1637  }
1638  A >>= 19;
1639  if (A & 0x100)
1640  A = av_clip_uint8(A);
1641  }
1642  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1643  dest += step;
1644  }
1645  c->dither_error[0][i] = err[0];
1646  c->dither_error[1][i] = err[1];
1647  c->dither_error[2][i] = err[2];
1648 }
1649 
1650 static av_always_inline void
1652  const int16_t *ubuf[2], const int16_t *vbuf[2],
1653  const int16_t *abuf[2], uint8_t *dest, int dstW,
1654  int yalpha, int uvalpha, int y,
1655  enum AVPixelFormat target, int hasAlpha)
1656 {
1657  const int16_t *buf0 = buf[0], *buf1 = buf[1],
1658  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1659  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1660  *abuf0 = hasAlpha ? abuf[0] : NULL,
1661  *abuf1 = hasAlpha ? abuf[1] : NULL;
1662  int yalpha1 = 4096 - yalpha;
1663  int uvalpha1 = 4096 - uvalpha;
1664  int i;
1665  int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1666  int err[4] = {0};
1667  int A = 0; // init to silcene warning
1668 
1669  if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1670  || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1671  step = 1;
1672 
1673  for (i = 0; i < dstW; i++) {
1674  int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
1675  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
1676  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
1677 
1678  if (hasAlpha) {
1679  A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
1680  if (A & 0x100)
1681  A = av_clip_uint8(A);
1682  }
1683 
1684  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1685  dest += step;
1686  }
1687  c->dither_error[0][i] = err[0];
1688  c->dither_error[1][i] = err[1];
1689  c->dither_error[2][i] = err[2];
1690 }
1691 
1692 static av_always_inline void
1694  const int16_t *ubuf[2], const int16_t *vbuf[2],
1695  const int16_t *abuf0, uint8_t *dest, int dstW,
1696  int uvalpha, int y, enum AVPixelFormat target,
1697  int hasAlpha)
1698 {
1699  const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1700  int i;
1701  int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1702  int err[4] = {0};
1703 
1704  if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1705  || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1706  step = 1;
1707 
1708  if (uvalpha < 2048) {
1709  int A = 0; //init to silence warning
1710  for (i = 0; i < dstW; i++) {
1711  int Y = buf0[i] << 2;
1712  int U = (ubuf0[i] - (128<<7)) << 2;
1713  int V = (vbuf0[i] - (128<<7)) << 2;
1714 
1715  if (hasAlpha) {
1716  A = (abuf0[i] + 64) >> 7;
1717  if (A & 0x100)
1718  A = av_clip_uint8(A);
1719  }
1720 
1721  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1722  dest += step;
1723  }
1724  } else {
1725  const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1726  int A = 0; //init to silence warning
1727  for (i = 0; i < dstW; i++) {
1728  int Y = buf0[i] << 2;
1729  int U = (ubuf0[i] + ubuf1[i] - (128<<8)) << 1;
1730  int V = (vbuf0[i] + vbuf1[i] - (128<<8)) << 1;
1731 
1732  if (hasAlpha) {
1733  A = (abuf0[i] + 64) >> 7;
1734  if (A & 0x100)
1735  A = av_clip_uint8(A);
1736  }
1737 
1738  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1739  dest += step;
1740  }
1741  }
1742 
1743  c->dither_error[0][i] = err[0];
1744  c->dither_error[1][i] = err[1];
1745  c->dither_error[2][i] = err[2];
1746 }
1747 
1748 #if CONFIG_SMALL
1749 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1750 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1751 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1752 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1753 #else
1754 #if CONFIG_SWSCALE_ALPHA
1755 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
1756 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
1757 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
1758 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
1759 #endif
1760 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
1761 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
1762 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
1763 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
1764 #endif
1765 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
1766 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
1767 
1768 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
1769 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
1770 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
1771 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
1772 
1773 static void
1774 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
1775  const int16_t **lumSrc, int lumFilterSize,
1776  const int16_t *chrFilter, const int16_t **chrUSrc,
1777  const int16_t **chrVSrc, int chrFilterSize,
1778  const int16_t **alpSrc, uint8_t **dest,
1779  int dstW, int y)
1780 {
1781  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
1782  int i;
1783  int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
1784  uint16_t **dest16 = (uint16_t**)dest;
1785  int SH = 22 + 7 - desc->comp[0].depth_minus1;
1786  int A = 0; // init to silence warning
1787 
1788  for (i = 0; i < dstW; i++) {
1789  int j;
1790  int Y = 1 << 9;
1791  int U = (1 << 9) - (128 << 19);
1792  int V = (1 << 9) - (128 << 19);
1793  int R, G, B;
1794 
1795  for (j = 0; j < lumFilterSize; j++)
1796  Y += lumSrc[j][i] * lumFilter[j];
1797 
1798  for (j = 0; j < chrFilterSize; j++) {
1799  U += chrUSrc[j][i] * chrFilter[j];
1800  V += chrVSrc[j][i] * chrFilter[j];
1801  }
1802 
1803  Y >>= 10;
1804  U >>= 10;
1805  V >>= 10;
1806 
1807  if (hasAlpha) {
1808  A = 1 << 18;
1809 
1810  for (j = 0; j < lumFilterSize; j++)
1811  A += alpSrc[j][i] * lumFilter[j];
1812 
1813  A >>= 19;
1814 
1815  if (A & 0x100)
1816  A = av_clip_uint8(A);
1817  }
1818 
1819  Y -= c->yuv2rgb_y_offset;
1820  Y *= c->yuv2rgb_y_coeff;
1821  Y += 1 << 21;
1822  R = Y + V * c->yuv2rgb_v2r_coeff;
1823  G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1824  B = Y + U * c->yuv2rgb_u2b_coeff;
1825 
1826  if ((R | G | B) & 0xC0000000) {
1827  R = av_clip_uintp2(R, 30);
1828  G = av_clip_uintp2(G, 30);
1829  B = av_clip_uintp2(B, 30);
1830  }
1831 
1832  if (SH != 22) {
1833  dest16[0][i] = G >> SH;
1834  dest16[1][i] = B >> SH;
1835  dest16[2][i] = R >> SH;
1836  if (hasAlpha)
1837  dest16[3][i] = A;
1838  } else {
1839  dest[0][i] = G >> 22;
1840  dest[1][i] = B >> 22;
1841  dest[2][i] = R >> 22;
1842  if (hasAlpha)
1843  dest[3][i] = A;
1844  }
1845  }
1846  if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
1847  for (i = 0; i < dstW; i++) {
1848  dest16[0][i] = av_bswap16(dest16[0][i]);
1849  dest16[1][i] = av_bswap16(dest16[1][i]);
1850  dest16[2][i] = av_bswap16(dest16[2][i]);
1851  if (hasAlpha)
1852  dest16[3][i] = av_bswap16(dest16[3][i]);
1853  }
1854  }
1855 }
1856 
1858  yuv2planar1_fn *yuv2plane1,
1859  yuv2planarX_fn *yuv2planeX,
1860  yuv2interleavedX_fn *yuv2nv12cX,
1861  yuv2packed1_fn *yuv2packed1,
1862  yuv2packed2_fn *yuv2packed2,
1863  yuv2packedX_fn *yuv2packedX,
1864  yuv2anyX_fn *yuv2anyX)
1865 {
1866  enum AVPixelFormat dstFormat = c->dstFormat;
1867  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
1868 
1869  if (is16BPS(dstFormat)) {
1870  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
1871  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
1872  } else if (is9_OR_10BPS(dstFormat)) {
1873  if (desc->comp[0].depth_minus1 == 8) {
1874  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
1875  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
1876  } else if (desc->comp[0].depth_minus1 == 9) {
1877  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
1878  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
1879  } else if (desc->comp[0].depth_minus1 == 11) {
1880  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
1881  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
1882  } else if (desc->comp[0].depth_minus1 == 13) {
1883  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
1884  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
1885  } else
1886  av_assert0(0);
1887  } else {
1888  *yuv2plane1 = yuv2plane1_8_c;
1889  *yuv2planeX = yuv2planeX_8_c;
1890  if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21)
1891  *yuv2nv12cX = yuv2nv12cX_c;
1892  }
1893 
1894  if(c->flags & SWS_FULL_CHR_H_INT) {
1895  switch (dstFormat) {
1896  case AV_PIX_FMT_RGBA:
1897 #if CONFIG_SMALL
1898  *yuv2packedX = yuv2rgba32_full_X_c;
1899  *yuv2packed2 = yuv2rgba32_full_2_c;
1900  *yuv2packed1 = yuv2rgba32_full_1_c;
1901 #else
1902 #if CONFIG_SWSCALE_ALPHA
1903  if (c->alpPixBuf) {
1904  *yuv2packedX = yuv2rgba32_full_X_c;
1905  *yuv2packed2 = yuv2rgba32_full_2_c;
1906  *yuv2packed1 = yuv2rgba32_full_1_c;
1907  } else
1908 #endif /* CONFIG_SWSCALE_ALPHA */
1909  {
1910  *yuv2packedX = yuv2rgbx32_full_X_c;
1911  *yuv2packed2 = yuv2rgbx32_full_2_c;
1912  *yuv2packed1 = yuv2rgbx32_full_1_c;
1913  }
1914 #endif /* !CONFIG_SMALL */
1915  break;
1916  case AV_PIX_FMT_ARGB:
1917 #if CONFIG_SMALL
1918  *yuv2packedX = yuv2argb32_full_X_c;
1919  *yuv2packed2 = yuv2argb32_full_2_c;
1920  *yuv2packed1 = yuv2argb32_full_1_c;
1921 #else
1922 #if CONFIG_SWSCALE_ALPHA
1923  if (c->alpPixBuf) {
1924  *yuv2packedX = yuv2argb32_full_X_c;
1925  *yuv2packed2 = yuv2argb32_full_2_c;
1926  *yuv2packed1 = yuv2argb32_full_1_c;
1927  } else
1928 #endif /* CONFIG_SWSCALE_ALPHA */
1929  {
1930  *yuv2packedX = yuv2xrgb32_full_X_c;
1931  *yuv2packed2 = yuv2xrgb32_full_2_c;
1932  *yuv2packed1 = yuv2xrgb32_full_1_c;
1933  }
1934 #endif /* !CONFIG_SMALL */
1935  break;
1936  case AV_PIX_FMT_BGRA:
1937 #if CONFIG_SMALL
1938  *yuv2packedX = yuv2bgra32_full_X_c;
1939  *yuv2packed2 = yuv2bgra32_full_2_c;
1940  *yuv2packed1 = yuv2bgra32_full_1_c;
1941 #else
1942 #if CONFIG_SWSCALE_ALPHA
1943  if (c->alpPixBuf) {
1944  *yuv2packedX = yuv2bgra32_full_X_c;
1945  *yuv2packed2 = yuv2bgra32_full_2_c;
1946  *yuv2packed1 = yuv2bgra32_full_1_c;
1947  } else
1948 #endif /* CONFIG_SWSCALE_ALPHA */
1949  {
1950  *yuv2packedX = yuv2bgrx32_full_X_c;
1951  *yuv2packed2 = yuv2bgrx32_full_2_c;
1952  *yuv2packed1 = yuv2bgrx32_full_1_c;
1953  }
1954 #endif /* !CONFIG_SMALL */
1955  break;
1956  case AV_PIX_FMT_ABGR:
1957 #if CONFIG_SMALL
1958  *yuv2packedX = yuv2abgr32_full_X_c;
1959  *yuv2packed2 = yuv2abgr32_full_2_c;
1960  *yuv2packed1 = yuv2abgr32_full_1_c;
1961 #else
1962 #if CONFIG_SWSCALE_ALPHA
1963  if (c->alpPixBuf) {
1964  *yuv2packedX = yuv2abgr32_full_X_c;
1965  *yuv2packed2 = yuv2abgr32_full_2_c;
1966  *yuv2packed1 = yuv2abgr32_full_1_c;
1967  } else
1968 #endif /* CONFIG_SWSCALE_ALPHA */
1969  {
1970  *yuv2packedX = yuv2xbgr32_full_X_c;
1971  *yuv2packed2 = yuv2xbgr32_full_2_c;
1972  *yuv2packed1 = yuv2xbgr32_full_1_c;
1973  }
1974 #endif /* !CONFIG_SMALL */
1975  break;
1976  case AV_PIX_FMT_RGB24:
1977  *yuv2packedX = yuv2rgb24_full_X_c;
1978  *yuv2packed2 = yuv2rgb24_full_2_c;
1979  *yuv2packed1 = yuv2rgb24_full_1_c;
1980  break;
1981  case AV_PIX_FMT_BGR24:
1982  *yuv2packedX = yuv2bgr24_full_X_c;
1983  *yuv2packed2 = yuv2bgr24_full_2_c;
1984  *yuv2packed1 = yuv2bgr24_full_1_c;
1985  break;
1986  case AV_PIX_FMT_BGR4_BYTE:
1987  *yuv2packedX = yuv2bgr4_byte_full_X_c;
1988  *yuv2packed2 = yuv2bgr4_byte_full_2_c;
1989  *yuv2packed1 = yuv2bgr4_byte_full_1_c;
1990  break;
1991  case AV_PIX_FMT_RGB4_BYTE:
1992  *yuv2packedX = yuv2rgb4_byte_full_X_c;
1993  *yuv2packed2 = yuv2rgb4_byte_full_2_c;
1994  *yuv2packed1 = yuv2rgb4_byte_full_1_c;
1995  break;
1996  case AV_PIX_FMT_BGR8:
1997  *yuv2packedX = yuv2bgr8_full_X_c;
1998  *yuv2packed2 = yuv2bgr8_full_2_c;
1999  *yuv2packed1 = yuv2bgr8_full_1_c;
2000  break;
2001  case AV_PIX_FMT_RGB8:
2002  *yuv2packedX = yuv2rgb8_full_X_c;
2003  *yuv2packed2 = yuv2rgb8_full_2_c;
2004  *yuv2packed1 = yuv2rgb8_full_1_c;
2005  break;
2006  case AV_PIX_FMT_GBRP:
2007  case AV_PIX_FMT_GBRP9BE:
2008  case AV_PIX_FMT_GBRP9LE:
2009  case AV_PIX_FMT_GBRP10BE:
2010  case AV_PIX_FMT_GBRP10LE:
2011  case AV_PIX_FMT_GBRP12BE:
2012  case AV_PIX_FMT_GBRP12LE:
2013  case AV_PIX_FMT_GBRP14BE:
2014  case AV_PIX_FMT_GBRP14LE:
2015  case AV_PIX_FMT_GBRP16BE:
2016  case AV_PIX_FMT_GBRP16LE:
2017  case AV_PIX_FMT_GBRAP:
2018  *yuv2anyX = yuv2gbrp_full_X_c;
2019  break;
2020  }
2021  if (!*yuv2packedX && !*yuv2anyX)
2022  goto YUV_PACKED;
2023  } else {
2024  YUV_PACKED:
2025  switch (dstFormat) {
2026  case AV_PIX_FMT_RGBA64LE:
2027 #if CONFIG_SWSCALE_ALPHA
2028  if (c->alpPixBuf) {
2029  *yuv2packed1 = yuv2rgba64le_1_c;
2030  *yuv2packed2 = yuv2rgba64le_2_c;
2031  *yuv2packedX = yuv2rgba64le_X_c;
2032  } else
2033 #endif /* CONFIG_SWSCALE_ALPHA */
2034  {
2035  *yuv2packed1 = yuv2rgbx64le_1_c;
2036  *yuv2packed2 = yuv2rgbx64le_2_c;
2037  *yuv2packedX = yuv2rgbx64le_X_c;
2038  }
2039  break;
2040  case AV_PIX_FMT_RGBA64BE:
2041 #if CONFIG_SWSCALE_ALPHA
2042  if (c->alpPixBuf) {
2043  *yuv2packed1 = yuv2rgba64be_1_c;
2044  *yuv2packed2 = yuv2rgba64be_2_c;
2045  *yuv2packedX = yuv2rgba64be_X_c;
2046  } else
2047 #endif /* CONFIG_SWSCALE_ALPHA */
2048  {
2049  *yuv2packed1 = yuv2rgbx64be_1_c;
2050  *yuv2packed2 = yuv2rgbx64be_2_c;
2051  *yuv2packedX = yuv2rgbx64be_X_c;
2052  }
2053  break;
2054  case AV_PIX_FMT_BGRA64LE:
2055 #if CONFIG_SWSCALE_ALPHA
2056  if (c->alpPixBuf) {
2057  *yuv2packed1 = yuv2bgra64le_1_c;
2058  *yuv2packed2 = yuv2bgra64le_2_c;
2059  *yuv2packedX = yuv2bgra64le_X_c;
2060  } else
2061 #endif /* CONFIG_SWSCALE_ALPHA */
2062  {
2063  *yuv2packed1 = yuv2bgrx64le_1_c;
2064  *yuv2packed2 = yuv2bgrx64le_2_c;
2065  *yuv2packedX = yuv2bgrx64le_X_c;
2066  }
2067  break;
2068  case AV_PIX_FMT_BGRA64BE:
2069 #if CONFIG_SWSCALE_ALPHA
2070  if (c->alpPixBuf) {
2071  *yuv2packed1 = yuv2bgra64be_1_c;
2072  *yuv2packed2 = yuv2bgra64be_2_c;
2073  *yuv2packedX = yuv2bgra64be_X_c;
2074  } else
2075 #endif /* CONFIG_SWSCALE_ALPHA */
2076  {
2077  *yuv2packed1 = yuv2bgrx64be_1_c;
2078  *yuv2packed2 = yuv2bgrx64be_2_c;
2079  *yuv2packedX = yuv2bgrx64be_X_c;
2080  }
2081  break;
2082  case AV_PIX_FMT_RGB48LE:
2083  *yuv2packed1 = yuv2rgb48le_1_c;
2084  *yuv2packed2 = yuv2rgb48le_2_c;
2085  *yuv2packedX = yuv2rgb48le_X_c;
2086  break;
2087  case AV_PIX_FMT_RGB48BE:
2088  *yuv2packed1 = yuv2rgb48be_1_c;
2089  *yuv2packed2 = yuv2rgb48be_2_c;
2090  *yuv2packedX = yuv2rgb48be_X_c;
2091  break;
2092  case AV_PIX_FMT_BGR48LE:
2093  *yuv2packed1 = yuv2bgr48le_1_c;
2094  *yuv2packed2 = yuv2bgr48le_2_c;
2095  *yuv2packedX = yuv2bgr48le_X_c;
2096  break;
2097  case AV_PIX_FMT_BGR48BE:
2098  *yuv2packed1 = yuv2bgr48be_1_c;
2099  *yuv2packed2 = yuv2bgr48be_2_c;
2100  *yuv2packedX = yuv2bgr48be_X_c;
2101  break;
2102  case AV_PIX_FMT_RGB32:
2103  case AV_PIX_FMT_BGR32:
2104 #if CONFIG_SMALL
2105  *yuv2packed1 = yuv2rgb32_1_c;
2106  *yuv2packed2 = yuv2rgb32_2_c;
2107  *yuv2packedX = yuv2rgb32_X_c;
2108 #else
2109 #if CONFIG_SWSCALE_ALPHA
2110  if (c->alpPixBuf) {
2111  *yuv2packed1 = yuv2rgba32_1_c;
2112  *yuv2packed2 = yuv2rgba32_2_c;
2113  *yuv2packedX = yuv2rgba32_X_c;
2114  } else
2115 #endif /* CONFIG_SWSCALE_ALPHA */
2116  {
2117  *yuv2packed1 = yuv2rgbx32_1_c;
2118  *yuv2packed2 = yuv2rgbx32_2_c;
2119  *yuv2packedX = yuv2rgbx32_X_c;
2120  }
2121 #endif /* !CONFIG_SMALL */
2122  break;
2123  case AV_PIX_FMT_RGB32_1:
2124  case AV_PIX_FMT_BGR32_1:
2125 #if CONFIG_SMALL
2126  *yuv2packed1 = yuv2rgb32_1_1_c;
2127  *yuv2packed2 = yuv2rgb32_1_2_c;
2128  *yuv2packedX = yuv2rgb32_1_X_c;
2129 #else
2130 #if CONFIG_SWSCALE_ALPHA
2131  if (c->alpPixBuf) {
2132  *yuv2packed1 = yuv2rgba32_1_1_c;
2133  *yuv2packed2 = yuv2rgba32_1_2_c;
2134  *yuv2packedX = yuv2rgba32_1_X_c;
2135  } else
2136 #endif /* CONFIG_SWSCALE_ALPHA */
2137  {
2138  *yuv2packed1 = yuv2rgbx32_1_1_c;
2139  *yuv2packed2 = yuv2rgbx32_1_2_c;
2140  *yuv2packedX = yuv2rgbx32_1_X_c;
2141  }
2142 #endif /* !CONFIG_SMALL */
2143  break;
2144  case AV_PIX_FMT_RGB24:
2145  *yuv2packed1 = yuv2rgb24_1_c;
2146  *yuv2packed2 = yuv2rgb24_2_c;
2147  *yuv2packedX = yuv2rgb24_X_c;
2148  break;
2149  case AV_PIX_FMT_BGR24:
2150  *yuv2packed1 = yuv2bgr24_1_c;
2151  *yuv2packed2 = yuv2bgr24_2_c;
2152  *yuv2packedX = yuv2bgr24_X_c;
2153  break;
2154  case AV_PIX_FMT_RGB565LE:
2155  case AV_PIX_FMT_RGB565BE:
2156  case AV_PIX_FMT_BGR565LE:
2157  case AV_PIX_FMT_BGR565BE:
2158  *yuv2packed1 = yuv2rgb16_1_c;
2159  *yuv2packed2 = yuv2rgb16_2_c;
2160  *yuv2packedX = yuv2rgb16_X_c;
2161  break;
2162  case AV_PIX_FMT_RGB555LE:
2163  case AV_PIX_FMT_RGB555BE:
2164  case AV_PIX_FMT_BGR555LE:
2165  case AV_PIX_FMT_BGR555BE:
2166  *yuv2packed1 = yuv2rgb15_1_c;
2167  *yuv2packed2 = yuv2rgb15_2_c;
2168  *yuv2packedX = yuv2rgb15_X_c;
2169  break;
2170  case AV_PIX_FMT_RGB444LE:
2171  case AV_PIX_FMT_RGB444BE:
2172  case AV_PIX_FMT_BGR444LE:
2173  case AV_PIX_FMT_BGR444BE:
2174  *yuv2packed1 = yuv2rgb12_1_c;
2175  *yuv2packed2 = yuv2rgb12_2_c;
2176  *yuv2packedX = yuv2rgb12_X_c;
2177  break;
2178  case AV_PIX_FMT_RGB8:
2179  case AV_PIX_FMT_BGR8:
2180  *yuv2packed1 = yuv2rgb8_1_c;
2181  *yuv2packed2 = yuv2rgb8_2_c;
2182  *yuv2packedX = yuv2rgb8_X_c;
2183  break;
2184  case AV_PIX_FMT_RGB4:
2185  case AV_PIX_FMT_BGR4:
2186  *yuv2packed1 = yuv2rgb4_1_c;
2187  *yuv2packed2 = yuv2rgb4_2_c;
2188  *yuv2packedX = yuv2rgb4_X_c;
2189  break;
2190  case AV_PIX_FMT_RGB4_BYTE:
2191  case AV_PIX_FMT_BGR4_BYTE:
2192  *yuv2packed1 = yuv2rgb4b_1_c;
2193  *yuv2packed2 = yuv2rgb4b_2_c;
2194  *yuv2packedX = yuv2rgb4b_X_c;
2195  break;
2196  }
2197  }
2198  switch (dstFormat) {
2199  case AV_PIX_FMT_MONOWHITE:
2200  *yuv2packed1 = yuv2monowhite_1_c;
2201  *yuv2packed2 = yuv2monowhite_2_c;
2202  *yuv2packedX = yuv2monowhite_X_c;
2203  break;
2204  case AV_PIX_FMT_MONOBLACK:
2205  *yuv2packed1 = yuv2monoblack_1_c;
2206  *yuv2packed2 = yuv2monoblack_2_c;
2207  *yuv2packedX = yuv2monoblack_X_c;
2208  break;
2209  case AV_PIX_FMT_YUYV422:
2210  *yuv2packed1 = yuv2yuyv422_1_c;
2211  *yuv2packed2 = yuv2yuyv422_2_c;
2212  *yuv2packedX = yuv2yuyv422_X_c;
2213  break;
2214  case AV_PIX_FMT_YVYU422:
2215  *yuv2packed1 = yuv2yvyu422_1_c;
2216  *yuv2packed2 = yuv2yvyu422_2_c;
2217  *yuv2packedX = yuv2yvyu422_X_c;
2218  break;
2219  case AV_PIX_FMT_UYVY422:
2220  *yuv2packed1 = yuv2uyvy422_1_c;
2221  *yuv2packed2 = yuv2uyvy422_2_c;
2222  *yuv2packedX = yuv2uyvy422_X_c;
2223  break;
2224  }
2225 }