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
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, },
43 };
44
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, },
49 };
50
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, },
57 };
58
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, },
69 };
70
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, },
81 };
82
83 #if 1
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, },
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
yuv2plane1_16_c_template(const int32_t * src,uint16_t * dest,int dstW,int big_endian,int output_bits)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
yuv2planeX_16_c_template(const int16_t * filter,int filterSize,const int32_t ** src,uint16_t * dest,int dstW,int big_endian,int output_bits)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
yuv2p016cX_c(SwsContext * c,const int16_t * chrFilter,int chrFilterSize,const int16_t ** chrUSrc,const int16_t ** chrVSrc,uint8_t * dest8,int chrDstW)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)
186 {
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;
190 int shift = 15;
191 int big_endian = c->dstFormat == AV_PIX_FMT_P016BE;
192 int i, j;
193
194 for (i = 0; i < chrDstW; i++) {
195 int u = 1 << (shift - 1);
196 int v = 1 << (shift - 1);
197
198 /* See yuv2planeX_16_c_template for details. */
199 u -= 0x40000000;
200 v -= 0x40000000;
201 for (j = 0; j < chrFilterSize; j++) {
202 u += uSrc[j][i] * (unsigned)chrFilter[j];
203 v += vSrc[j][i] * (unsigned)chrFilter[j];
204 }
205
206 output_pixel(&dest[2*i] , u, 0x8000, int);
207 output_pixel(&dest[2*i+1], v, 0x8000, int);
208 }
209 }
210
211 static av_always_inline void
yuv2plane1_float_c_template(const int32_t * src,float * dest,int dstW)212 yuv2plane1_float_c_template(const int32_t *src, float *dest, int dstW)
213 {
214 static const int big_endian = HAVE_BIGENDIAN;
215 static const int shift = 3;
216 static const float float_mult = 1.0f / 65535.0f;
217 int i, val;
218 uint16_t val_uint;
219
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;
224 }
225 }
226
227 static av_always_inline void
yuv2plane1_float_bswap_c_template(const int32_t * src,uint32_t * dest,int dstW)228 yuv2plane1_float_bswap_c_template(const int32_t *src, uint32_t *dest, int dstW)
229 {
230 static const int big_endian = HAVE_BIGENDIAN;
231 static const int shift = 3;
232 static const float float_mult = 1.0f / 65535.0f;
233 int i, val;
234 uint16_t val_uint;
235
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));
240 }
241 }
242
243 static av_always_inline void
yuv2planeX_float_c_template(const int16_t * filter,int filterSize,const int32_t ** src,float * dest,int dstW)244 yuv2planeX_float_c_template(const int16_t *filter, int filterSize, const int32_t **src,
245 float *dest, int dstW)
246 {
247 static const int big_endian = HAVE_BIGENDIAN;
248 static const int shift = 15;
249 static const float float_mult = 1.0f / 65535.0f;
250 int i, j, val;
251 uint16_t val_uint;
252
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];
257 }
258 output_pixel(&val_uint, val, 0x8000, int);
259 dest[i] = float_mult * (float)val_uint;
260 }
261 }
262
263 static av_always_inline void
yuv2planeX_float_bswap_c_template(const int16_t * filter,int filterSize,const int32_t ** src,uint32_t * dest,int dstW)264 yuv2planeX_float_bswap_c_template(const int16_t *filter, int filterSize, const int32_t **src,
265 uint32_t *dest, int dstW)
266 {
267 static const int big_endian = HAVE_BIGENDIAN;
268 static const int shift = 15;
269 static const float float_mult = 1.0f / 65535.0f;
270 int i, j, val;
271 uint16_t val_uint;
272
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];
277 }
278 output_pixel(&val_uint, val, 0x8000, int);
279 dest[i] = av_bswap32(av_float2int(float_mult * (float)val_uint));
280 }
281 }
282
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) \
286 { \
287 template((const int32_t *)src, (dest_type *)dest, dstW); \
288 }
289
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) \
294 { \
295 template(filter, filterSize, (const int32_t **)src, (dest_type *)dest, dstW); \
296 }
297
298 #if HAVE_BIGENDIAN
yuv2plane1_float(yuv2plane1_float_c_template,float,BE)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)
303 #else
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)
308 #endif
309
310 #undef output_pixel
311
312 #define output_pixel(pos, val) \
313 if (big_endian) { \
314 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
315 } else { \
316 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
317 }
318
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)
322 {
323 int i;
324 int shift = 15 - output_bits;
325
326 for (i = 0; i < dstW; i++) {
327 int val = src[i] + (1 << (shift - 1));
328 output_pixel(&dest[i], val);
329 }
330 }
331
332 static av_always_inline void
yuv2planeX_10_c_template(const int16_t * filter,int filterSize,const int16_t ** src,uint16_t * dest,int dstW,int big_endian,int output_bits)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)
336 {
337 int i;
338 int shift = 11 + 16 - output_bits;
339
340 for (i = 0; i < dstW; i++) {
341 int val = 1 << (shift - 1);
342 int j;
343
344 for (j = 0; j < filterSize; j++)
345 val += src[j][i] * filter[j];
346
347 output_pixel(&dest[i], val);
348 }
349 }
350
351 #undef output_pixel
352
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)\
357 { \
358 yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
359 (uint16_t *) dest, dstW, is_be, bits); \
360 }\
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)\
364 { \
365 yuv2planeX_## template_size ## _c_template(filter, \
366 filterSize, (const typeX_t **) src, \
367 (uint16_t *) dest, dstW, is_be, bits); \
368 }
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)
379
yuv2planeX_8_c(const int16_t * filter,int filterSize,const int16_t ** src,uint8_t * dest,int dstW,const uint8_t * dither,int offset)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)
383 {
384 int i;
385 for (i=0; i<dstW; i++) {
386 int val = dither[(i + offset) & 7] << 12;
387 int j;
388 for (j=0; j<filterSize; j++)
389 val += src[j][i] * filter[j];
390
391 dest[i]= av_clip_uint8(val>>19);
392 }
393 }
394
yuv2plane1_8_c(const int16_t * src,uint8_t * dest,int dstW,const uint8_t * dither,int offset)395 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
396 const uint8_t *dither, int offset)
397 {
398 int i;
399 for (i=0; i<dstW; i++) {
400 int val = (src[i] + dither[(i + offset) & 7]) >> 7;
401 dest[i]= av_clip_uint8(val);
402 }
403 }
404
yuv2nv12cX_c(SwsContext * c,const int16_t * chrFilter,int chrFilterSize,const int16_t ** chrUSrc,const int16_t ** chrVSrc,uint8_t * dest,int chrDstW)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)
408 {
409 enum AVPixelFormat dstFormat = c->dstFormat;
410 const uint8_t *chrDither = c->chrDither8;
411 int i;
412
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;
418 int j;
419 for (j=0; j<chrFilterSize; j++) {
420 u += chrUSrc[j][i] * chrFilter[j];
421 v += chrVSrc[j][i] * chrFilter[j];
422 }
423
424 dest[2*i]= av_clip_uint8(u>>19);
425 dest[2*i+1]= av_clip_uint8(v>>19);
426 }
427 else
428 for (i=0; i<chrDstW; i++) {
429 int u = chrDither[i & 7] << 12;
430 int v = chrDither[(i + 3) & 7] << 12;
431 int j;
432 for (j=0; j<chrFilterSize; j++) {
433 u += chrUSrc[j][i] * chrFilter[j];
434 v += chrVSrc[j][i] * chrFilter[j];
435 }
436
437 dest[2*i]= av_clip_uint8(v>>19);
438 dest[2*i+1]= av_clip_uint8(u>>19);
439 }
440 }
441
442
443 #define output_pixel(pos, val) \
444 if (big_endian) { \
445 AV_WB16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
446 } else { \
447 AV_WL16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
448 }
449
yuv2p010l1_c(const int16_t * src,uint16_t * dest,int dstW,int big_endian)450 static void yuv2p010l1_c(const int16_t *src,
451 uint16_t *dest, int dstW,
452 int big_endian)
453 {
454 int i;
455 int shift = 5;
456
457 for (i = 0; i < dstW; i++) {
458 int val = src[i] + (1 << (shift - 1));
459 output_pixel(&dest[i], val);
460 }
461 }
462
yuv2p010lX_c(const int16_t * filter,int filterSize,const int16_t ** src,uint16_t * dest,int dstW,int big_endian)463 static void yuv2p010lX_c(const int16_t *filter, int filterSize,
464 const int16_t **src, uint16_t *dest, int dstW,
465 int big_endian)
466 {
467 int i, j;
468 int shift = 17;
469
470 for (i = 0; i < dstW; i++) {
471 int val = 1 << (shift - 1);
472
473 for (j = 0; j < filterSize; j++)
474 val += src[j][i] * filter[j];
475
476 output_pixel(&dest[i], val);
477 }
478 }
479
yuv2p010cX_c(SwsContext * c,const int16_t * chrFilter,int chrFilterSize,const int16_t ** chrUSrc,const int16_t ** chrVSrc,uint8_t * dest8,int chrDstW)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)
483 {
484 uint16_t *dest = (uint16_t*)dest8;
485 int shift = 17;
486 int big_endian = c->dstFormat == AV_PIX_FMT_P010BE;
487 int i, j;
488
489 for (i = 0; i < chrDstW; i++) {
490 int u = 1 << (shift - 1);
491 int v = 1 << (shift - 1);
492
493 for (j = 0; j < chrFilterSize; j++) {
494 u += chrUSrc[j][i] * chrFilter[j];
495 v += chrVSrc[j][i] * chrFilter[j];
496 }
497
498 output_pixel(&dest[2*i] , u);
499 output_pixel(&dest[2*i+1], v);
500 }
501 }
502
yuv2p010l1_LE_c(const int16_t * src,uint8_t * dest,int dstW,const uint8_t * dither,int offset)503 static void yuv2p010l1_LE_c(const int16_t *src,
504 uint8_t *dest, int dstW,
505 const uint8_t *dither, int offset)
506 {
507 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 0);
508 }
509
yuv2p010l1_BE_c(const int16_t * src,uint8_t * dest,int dstW,const uint8_t * dither,int offset)510 static void yuv2p010l1_BE_c(const int16_t *src,
511 uint8_t *dest, int dstW,
512 const uint8_t *dither, int offset)
513 {
514 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 1);
515 }
516
yuv2p010lX_LE_c(const int16_t * filter,int filterSize,const int16_t ** src,uint8_t * dest,int dstW,const uint8_t * dither,int offset)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)
520 {
521 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 0);
522 }
523
yuv2p010lX_BE_c(const int16_t * filter,int filterSize,const int16_t ** src,uint8_t * dest,int dstW,const uint8_t * dither,int offset)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)
527 {
528 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 1);
529 }
530
531 #undef output_pixel
532
533
534 #define accumulate_bit(acc, val) \
535 acc <<= 1; \
536 acc |= (val) >= 234
537 #define output_pixel(pos, acc) \
538 if (target == AV_PIX_FMT_MONOBLACK) { \
539 pos = acc; \
540 } else { \
541 pos = ~acc; \
542 }
543
544 static av_always_inline void
yuv2mono_X_c_template(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrc,const int16_t ** chrVSrc,int chrFilterSize,const int16_t ** alpSrc,uint8_t * dest,int dstW,int y,enum AVPixelFormat target)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)
551 {
552 const uint8_t * const d128 = ff_dither_8x8_220[y&7];
553 int i;
554 unsigned acc = 0;
555 int err = 0;
556
557 for (i = 0; i < dstW; i += 2) {
558 int j;
559 int Y1 = 1 << 18;
560 int Y2 = 1 << 18;
561
562 for (j = 0; j < lumFilterSize; j++) {
563 Y1 += lumSrc[j][i] * lumFilter[j];
564 Y2 += lumSrc[j][i+1] * lumFilter[j];
565 }
566 Y1 >>= 19;
567 Y2 >>= 19;
568 if ((Y1 | Y2) & 0x100) {
569 Y1 = av_clip_uint8(Y1);
570 Y2 = av_clip_uint8(Y2);
571 }
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);
576 Y1 -= 220*(acc&1);
577
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);
581 err -= 220*(acc&1);
582 } else {
583 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
584 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
585 }
586 if ((i & 7) == 6) {
587 output_pixel(*dest++, acc);
588 }
589 }
590 c->dither_error[0][i] = err;
591
592 if (i & 6) {
593 output_pixel(*dest, acc);
594 }
595 }
596
597 static av_always_inline void
yuv2mono_2_c_template(SwsContext * c,const int16_t * buf[2],const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf[2],uint8_t * dest,int dstW,int yalpha,int uvalpha,int y,enum AVPixelFormat target)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)
603 {
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;
607 int i;
608 av_assert2(yalpha <= 4096U);
609
610 if (c->dither == SWS_DITHER_ED) {
611 int err = 0;
612 int acc = 0;
613 for (i = 0; i < dstW; i +=2) {
614 int Y;
615
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);
620 Y -= 220*(acc&1);
621
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);
626 err -= 220*(acc&1);
627
628 if ((i & 7) == 6)
629 output_pixel(*dest++, acc);
630 }
631 c->dither_error[0][i] = err;
632 } else {
633 for (i = 0; i < dstW; i += 8) {
634 int Y, acc = 0;
635
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]);
652
653 output_pixel(*dest++, acc);
654 }
655 }
656 }
657
658 static av_always_inline void
yuv2mono_1_c_template(SwsContext * c,const int16_t * buf0,const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf0,uint8_t * dest,int dstW,int uvalpha,int y,enum AVPixelFormat target)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)
663 {
664 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
665 int i;
666
667 if (c->dither == SWS_DITHER_ED) {
668 int err = 0;
669 int acc = 0;
670 for (i = 0; i < dstW; i +=2) {
671 int Y;
672
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);
677 Y -= 220*(acc&1);
678
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);
683 err -= 220*(acc&1);
684
685 if ((i & 7) == 6)
686 output_pixel(*dest++, acc);
687 }
688 c->dither_error[0][i] = err;
689 } else {
690 for (i = 0; i < dstW; i += 8) {
691 int acc = 0;
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]);
700
701 output_pixel(*dest++, acc);
702 }
703 }
704 }
705
706 #undef output_pixel
707 #undef accumulate_bit
708
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, \
715 int y) \
716 { \
717 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
718 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
719 alpSrc, dest, dstW, y, fmt); \
720 } \
721 \
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) \
726 { \
727 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
728 dest, dstW, yalpha, uvalpha, y, fmt); \
729 } \
730 \
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) \
735 { \
736 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
737 abuf0, dest, dstW, uvalpha, \
738 y, fmt); \
739 }
740
741 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
742 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
743
744 #define output_pixels(pos, Y1, U, Y2, V) \
745 if (target == AV_PIX_FMT_YUYV422) { \
746 dest[pos + 0] = Y1; \
747 dest[pos + 1] = U; \
748 dest[pos + 2] = Y2; \
749 dest[pos + 3] = V; \
750 } else if (target == AV_PIX_FMT_YVYU422) { \
751 dest[pos + 0] = Y1; \
752 dest[pos + 1] = V; \
753 dest[pos + 2] = Y2; \
754 dest[pos + 3] = U; \
755 } else { /* AV_PIX_FMT_UYVY422 */ \
756 dest[pos + 0] = U; \
757 dest[pos + 1] = Y1; \
758 dest[pos + 2] = V; \
759 dest[pos + 3] = Y2; \
760 }
761
762 static av_always_inline void
yuv2422_X_c_template(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrc,const int16_t ** chrVSrc,int chrFilterSize,const int16_t ** alpSrc,uint8_t * dest,int dstW,int y,enum AVPixelFormat target)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)
769 {
770 int i;
771
772 for (i = 0; i < ((dstW + 1) >> 1); i++) {
773 int j;
774 int Y1 = 1 << 18;
775 int Y2 = 1 << 18;
776 int U = 1 << 18;
777 int V = 1 << 18;
778
779 for (j = 0; j < lumFilterSize; j++) {
780 Y1 += lumSrc[j][i * 2] * lumFilter[j];
781 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
782 }
783 for (j = 0; j < chrFilterSize; j++) {
784 U += chrUSrc[j][i] * chrFilter[j];
785 V += chrVSrc[j][i] * chrFilter[j];
786 }
787 Y1 >>= 19;
788 Y2 >>= 19;
789 U >>= 19;
790 V >>= 19;
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);
796 }
797 output_pixels(4*i, Y1, U, Y2, V);
798 }
799 }
800
801 static av_always_inline void
yuv2422_2_c_template(SwsContext * c,const int16_t * buf[2],const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf[2],uint8_t * dest,int dstW,int yalpha,int uvalpha,int y,enum AVPixelFormat target)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)
807 {
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;
813 int i;
814 av_assert2(yalpha <= 4096U);
815 av_assert2(uvalpha <= 4096U);
816
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;
822
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);
828 }
829
830 output_pixels(i * 4, Y1, U, Y2, V);
831 }
832 }
833
834 static av_always_inline void
yuv2422_1_c_template(SwsContext * c,const int16_t * buf0,const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf0,uint8_t * dest,int dstW,int uvalpha,int y,enum AVPixelFormat target)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)
839 {
840 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
841 int i;
842
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;
849
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);
855 }
856
857 output_pixels(i * 4, Y1, U, Y2, V);
858 }
859 } else {
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;
866
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);
872 }
873
874 output_pixels(i * 4, Y1, U, Y2, V);
875 }
876 }
877 }
878
879 #undef output_pixels
880
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)
884
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)) { \
889 AV_WB16(pos, val); \
890 } else { \
891 AV_WL16(pos, val); \
892 }
893
894 static av_always_inline void
yuv2ya16_X_c_template(SwsContext * c,const int16_t * lumFilter,const int32_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int32_t ** unused_chrUSrc,const int32_t ** unused_chrVSrc,int unused_chrFilterSize,const int32_t ** alpSrc,uint16_t * dest,int dstW,int y,enum AVPixelFormat target,int unused_hasAlpha,int unused_eightbytes)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)
901 {
902 int hasAlpha = !!alpSrc;
903 int i;
904
905 for (i = 0; i < dstW; i++) {
906 int j;
907 int Y = -0x40000000;
908 int A = 0xffff;
909
910 for (j = 0; j < lumFilterSize; j++)
911 Y += lumSrc[j][i] * lumFilter[j];
912
913 Y >>= 15;
914 Y += (1<<3) + 0x8000;
915 Y = av_clip_uint16(Y);
916
917 if (hasAlpha) {
918 A = -0x40000000 + (1<<14);
919 for (j = 0; j < lumFilterSize; j++)
920 A += alpSrc[j][i] * lumFilter[j];
921
922 A >>= 15;
923 A += 0x8000;
924 A = av_clip_uint16(A);
925 }
926
927 output_pixel(&dest[2 * i ], Y);
928 output_pixel(&dest[2 * i + 1], A);
929 }
930 }
931
932 static av_always_inline void
yuv2ya16_2_c_template(SwsContext * c,const int32_t * buf[2],const int32_t * unused_ubuf[2],const int32_t * unused_vbuf[2],const int32_t * abuf[2],uint16_t * dest,int dstW,int yalpha,int unused_uvalpha,int y,enum AVPixelFormat target,int unused_hasAlpha,int unused_eightbytes)933 yuv2ya16_2_c_template(SwsContext *c, const int32_t *buf[2],
934 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
935 const int32_t *abuf[2], uint16_t *dest, int dstW,
936 int yalpha, int unused_uvalpha, int y,
937 enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
938 {
939 int hasAlpha = abuf && abuf[0] && abuf[1];
940 const int32_t *buf0 = buf[0], *buf1 = buf[1],
941 *abuf0 = hasAlpha ? abuf[0] : NULL,
942 *abuf1 = hasAlpha ? abuf[1] : NULL;
943 int yalpha1 = 4096 - yalpha;
944 int i;
945
946 av_assert2(yalpha <= 4096U);
947
948 for (i = 0; i < dstW; i++) {
949 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 15;
950 int A;
951
952 Y = av_clip_uint16(Y);
953
954 if (hasAlpha) {
955 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 15;
956 A = av_clip_uint16(A);
957 }
958
959 output_pixel(&dest[2 * i ], Y);
960 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
961 }
962 }
963
964 static av_always_inline void
yuv2ya16_1_c_template(SwsContext * c,const int32_t * buf0,const int32_t * unused_ubuf[2],const int32_t * unused_vbuf[2],const int32_t * abuf0,uint16_t * dest,int dstW,int unused_uvalpha,int y,enum AVPixelFormat target,int unused_hasAlpha,int unused_eightbytes)965 yuv2ya16_1_c_template(SwsContext *c, const int32_t *buf0,
966 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
967 const int32_t *abuf0, uint16_t *dest, int dstW,
968 int unused_uvalpha, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
969 {
970 int hasAlpha = !!abuf0;
971 int i;
972
973 for (i = 0; i < dstW; i++) {
974 int Y = buf0[i] >> 3;/* 19 - 16 */
975 int A;
976
977 Y = av_clip_uint16(Y);
978
979 if (hasAlpha) {
980 A = abuf0[i] >> 3;
981 if (A & 0x100)
982 A = av_clip_uint16(A);
983 }
984
985 output_pixel(&dest[2 * i ], Y);
986 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
987 }
988 }
989
990 static av_always_inline void
yuv2rgba64_X_c_template(SwsContext * c,const int16_t * lumFilter,const int32_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int32_t ** chrUSrc,const int32_t ** chrVSrc,int chrFilterSize,const int32_t ** alpSrc,uint16_t * dest,int dstW,int y,enum AVPixelFormat target,int hasAlpha,int eightbytes)991 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
992 const int32_t **lumSrc, int lumFilterSize,
993 const int16_t *chrFilter, const int32_t **chrUSrc,
994 const int32_t **chrVSrc, int chrFilterSize,
995 const int32_t **alpSrc, uint16_t *dest, int dstW,
996 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
997 {
998 int i;
999 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1000
1001 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1002 int j;
1003 int Y1 = -0x40000000;
1004 int Y2 = -0x40000000;
1005 int U = -(128 << 23); // 19
1006 int V = -(128 << 23);
1007 int R, G, B;
1008
1009 for (j = 0; j < lumFilterSize; j++) {
1010 Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
1011 Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1012 }
1013 for (j = 0; j < chrFilterSize; j++) {;
1014 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1015 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1016 }
1017
1018 if (hasAlpha) {
1019 A1 = -0x40000000;
1020 A2 = -0x40000000;
1021 for (j = 0; j < lumFilterSize; j++) {
1022 A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
1023 A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1024 }
1025 A1 >>= 1;
1026 A1 += 0x20002000;
1027 A2 >>= 1;
1028 A2 += 0x20002000;
1029 }
1030
1031 // 8 bits: 12+15=27; 16 bits: 12+19=31
1032 Y1 >>= 14; // 10
1033 Y1 += 0x10000;
1034 Y2 >>= 14;
1035 Y2 += 0x10000;
1036 U >>= 14;
1037 V >>= 14;
1038
1039 // 8 bits: 27 -> 17 bits, 16 bits: 31 - 14 = 17 bits
1040 Y1 -= c->yuv2rgb_y_offset;
1041 Y2 -= c->yuv2rgb_y_offset;
1042 Y1 *= c->yuv2rgb_y_coeff;
1043 Y2 *= c->yuv2rgb_y_coeff;
1044 Y1 += 1 << 13; // 21
1045 Y2 += 1 << 13;
1046 // 8 bits: 17 + 13 bits = 30 bits, 16 bits: 17 + 13 bits = 30 bits
1047
1048 R = V * c->yuv2rgb_v2r_coeff;
1049 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1050 B = U * c->yuv2rgb_u2b_coeff;
1051
1052 // 8 bits: 30 - 22 = 8 bits, 16 bits: 30 bits - 14 = 16 bits
1053 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1054 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1055 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1056 if (eightbytes) {
1057 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1058 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1059 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1060 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1061 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1062 dest += 8;
1063 } else {
1064 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1065 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1066 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1067 dest += 6;
1068 }
1069 }
1070 }
1071
1072 static av_always_inline void
yuv2rgba64_2_c_template(SwsContext * c,const int32_t * buf[2],const int32_t * ubuf[2],const int32_t * vbuf[2],const int32_t * abuf[2],uint16_t * dest,int dstW,int yalpha,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha,int eightbytes)1073 yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
1074 const int32_t *ubuf[2], const int32_t *vbuf[2],
1075 const int32_t *abuf[2], uint16_t *dest, int dstW,
1076 int yalpha, int uvalpha, int y,
1077 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1078 {
1079 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1080 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1081 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1082 *abuf0 = hasAlpha ? abuf[0] : NULL,
1083 *abuf1 = hasAlpha ? abuf[1] : NULL;
1084 int yalpha1 = 4096 - yalpha;
1085 int uvalpha1 = 4096 - uvalpha;
1086 int i;
1087 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1088
1089 av_assert2(yalpha <= 4096U);
1090 av_assert2(uvalpha <= 4096U);
1091
1092 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1093 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
1094 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
1095 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1096 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1097 int R, G, B;
1098
1099 Y1 -= c->yuv2rgb_y_offset;
1100 Y2 -= c->yuv2rgb_y_offset;
1101 Y1 *= c->yuv2rgb_y_coeff;
1102 Y2 *= c->yuv2rgb_y_coeff;
1103 Y1 += 1 << 13;
1104 Y2 += 1 << 13;
1105
1106 R = V * c->yuv2rgb_v2r_coeff;
1107 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1108 B = U * c->yuv2rgb_u2b_coeff;
1109
1110 if (hasAlpha) {
1111 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
1112 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
1113
1114 A1 += 1 << 13;
1115 A2 += 1 << 13;
1116 }
1117
1118 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1119 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1120 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1121 if (eightbytes) {
1122 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1123 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1124 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1125 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1126 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1127 dest += 8;
1128 } else {
1129 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1130 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1131 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1132 dest += 6;
1133 }
1134 }
1135 }
1136
1137 static av_always_inline void
yuv2rgba64_1_c_template(SwsContext * c,const int32_t * buf0,const int32_t * ubuf[2],const int32_t * vbuf[2],const int32_t * abuf0,uint16_t * dest,int dstW,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha,int eightbytes)1138 yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
1139 const int32_t *ubuf[2], const int32_t *vbuf[2],
1140 const int32_t *abuf0, uint16_t *dest, int dstW,
1141 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1142 {
1143 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1144 int i;
1145 int A1 = 0xffff<<14, A2= 0xffff<<14;
1146
1147 if (uvalpha < 2048) {
1148 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1149 int Y1 = (buf0[i * 2] ) >> 2;
1150 int Y2 = (buf0[i * 2 + 1]) >> 2;
1151 int U = (ubuf0[i] - (128 << 11)) >> 2;
1152 int V = (vbuf0[i] - (128 << 11)) >> 2;
1153 int R, G, B;
1154
1155 Y1 -= c->yuv2rgb_y_offset;
1156 Y2 -= c->yuv2rgb_y_offset;
1157 Y1 *= c->yuv2rgb_y_coeff;
1158 Y2 *= c->yuv2rgb_y_coeff;
1159 Y1 += 1 << 13;
1160 Y2 += 1 << 13;
1161
1162 if (hasAlpha) {
1163 A1 = abuf0[i * 2 ] << 11;
1164 A2 = abuf0[i * 2 + 1] << 11;
1165
1166 A1 += 1 << 13;
1167 A2 += 1 << 13;
1168 }
1169
1170 R = V * c->yuv2rgb_v2r_coeff;
1171 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1172 B = U * c->yuv2rgb_u2b_coeff;
1173
1174 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1175 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1176 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1177 if (eightbytes) {
1178 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1179 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1180 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1181 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1182 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1183 dest += 8;
1184 } else {
1185 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1186 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1187 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1188 dest += 6;
1189 }
1190 }
1191 } else {
1192 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1193 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1194 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1195 int Y1 = (buf0[i * 2] ) >> 2;
1196 int Y2 = (buf0[i * 2 + 1]) >> 2;
1197 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1198 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1199 int R, G, B;
1200
1201 Y1 -= c->yuv2rgb_y_offset;
1202 Y2 -= c->yuv2rgb_y_offset;
1203 Y1 *= c->yuv2rgb_y_coeff;
1204 Y2 *= c->yuv2rgb_y_coeff;
1205 Y1 += 1 << 13;
1206 Y2 += 1 << 13;
1207
1208 if (hasAlpha) {
1209 A1 = abuf0[i * 2 ] << 11;
1210 A2 = abuf0[i * 2 + 1] << 11;
1211
1212 A1 += 1 << 13;
1213 A2 += 1 << 13;
1214 }
1215
1216 R = V * c->yuv2rgb_v2r_coeff;
1217 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1218 B = U * c->yuv2rgb_u2b_coeff;
1219
1220 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1221 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1222 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1223 if (eightbytes) {
1224 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1225 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1226 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1227 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1228 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1229 dest += 8;
1230 } else {
1231 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1232 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1233 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1234 dest += 6;
1235 }
1236 }
1237 }
1238 }
1239
1240 static av_always_inline void
yuv2rgba64_full_X_c_template(SwsContext * c,const int16_t * lumFilter,const int32_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int32_t ** chrUSrc,const int32_t ** chrVSrc,int chrFilterSize,const int32_t ** alpSrc,uint16_t * dest,int dstW,int y,enum AVPixelFormat target,int hasAlpha,int eightbytes)1241 yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1242 const int32_t **lumSrc, int lumFilterSize,
1243 const int16_t *chrFilter, const int32_t **chrUSrc,
1244 const int32_t **chrVSrc, int chrFilterSize,
1245 const int32_t **alpSrc, uint16_t *dest, int dstW,
1246 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1247 {
1248 int i;
1249 int A = 0xffff<<14;
1250
1251 for (i = 0; i < dstW; i++) {
1252 int j;
1253 int Y = -0x40000000;
1254 int U = -(128 << 23); // 19
1255 int V = -(128 << 23);
1256 int R, G, B;
1257
1258 for (j = 0; j < lumFilterSize; j++) {
1259 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
1260 }
1261 for (j = 0; j < chrFilterSize; j++) {;
1262 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1263 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1264 }
1265
1266 if (hasAlpha) {
1267 A = -0x40000000;
1268 for (j = 0; j < lumFilterSize; j++) {
1269 A += alpSrc[j][i] * (unsigned)lumFilter[j];
1270 }
1271 A >>= 1;
1272 A += 0x20002000;
1273 }
1274
1275 // 8bit: 12+15=27; 16-bit: 12+19=31
1276 Y >>= 14; // 10
1277 Y += 0x10000;
1278 U >>= 14;
1279 V >>= 14;
1280
1281 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
1282 Y -= c->yuv2rgb_y_offset;
1283 Y *= c->yuv2rgb_y_coeff;
1284 Y += 1 << 13; // 21
1285 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
1286
1287 R = V * c->yuv2rgb_v2r_coeff;
1288 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1289 B = U * c->yuv2rgb_u2b_coeff;
1290
1291 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
1292 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1293 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1294 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1295 if (eightbytes) {
1296 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1297 dest += 4;
1298 } else {
1299 dest += 3;
1300 }
1301 }
1302 }
1303
1304 static av_always_inline void
yuv2rgba64_full_2_c_template(SwsContext * c,const int32_t * buf[2],const int32_t * ubuf[2],const int32_t * vbuf[2],const int32_t * abuf[2],uint16_t * dest,int dstW,int yalpha,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha,int eightbytes)1305 yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
1306 const int32_t *ubuf[2], const int32_t *vbuf[2],
1307 const int32_t *abuf[2], uint16_t *dest, int dstW,
1308 int yalpha, int uvalpha, int y,
1309 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1310 {
1311 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1312 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1313 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1314 *abuf0 = hasAlpha ? abuf[0] : NULL,
1315 *abuf1 = hasAlpha ? abuf[1] : NULL;
1316 int yalpha1 = 4096 - yalpha;
1317 int uvalpha1 = 4096 - uvalpha;
1318 int i;
1319 int A = 0xffff<<14;
1320
1321 av_assert2(yalpha <= 4096U);
1322 av_assert2(uvalpha <= 4096U);
1323
1324 for (i = 0; i < dstW; i++) {
1325 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 14;
1326 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1327 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1328 int R, G, B;
1329
1330 Y -= c->yuv2rgb_y_offset;
1331 Y *= c->yuv2rgb_y_coeff;
1332 Y += 1 << 13;
1333
1334 R = V * c->yuv2rgb_v2r_coeff;
1335 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1336 B = U * c->yuv2rgb_u2b_coeff;
1337
1338 if (hasAlpha) {
1339 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1340
1341 A += 1 << 13;
1342 }
1343
1344 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1345 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1346 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1347 if (eightbytes) {
1348 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1349 dest += 4;
1350 } else {
1351 dest += 3;
1352 }
1353 }
1354 }
1355
1356 static av_always_inline void
yuv2rgba64_full_1_c_template(SwsContext * c,const int32_t * buf0,const int32_t * ubuf[2],const int32_t * vbuf[2],const int32_t * abuf0,uint16_t * dest,int dstW,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha,int eightbytes)1357 yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1358 const int32_t *ubuf[2], const int32_t *vbuf[2],
1359 const int32_t *abuf0, uint16_t *dest, int dstW,
1360 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1361 {
1362 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1363 int i;
1364 int A = 0xffff<<14;
1365
1366 if (uvalpha < 2048) {
1367 for (i = 0; i < dstW; i++) {
1368 int Y = (buf0[i]) >> 2;
1369 int U = (ubuf0[i] - (128 << 11)) >> 2;
1370 int V = (vbuf0[i] - (128 << 11)) >> 2;
1371 int R, G, B;
1372
1373 Y -= c->yuv2rgb_y_offset;
1374 Y *= c->yuv2rgb_y_coeff;
1375 Y += 1 << 13;
1376
1377 if (hasAlpha) {
1378 A = abuf0[i] << 11;
1379
1380 A += 1 << 13;
1381 }
1382
1383 R = V * c->yuv2rgb_v2r_coeff;
1384 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1385 B = U * c->yuv2rgb_u2b_coeff;
1386
1387 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1388 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1389 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1390 if (eightbytes) {
1391 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1392 dest += 4;
1393 } else {
1394 dest += 3;
1395 }
1396 }
1397 } else {
1398 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1399 int A = 0xffff<<14;
1400 for (i = 0; i < dstW; i++) {
1401 int Y = (buf0[i] ) >> 2;
1402 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1403 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1404 int R, G, B;
1405
1406 Y -= c->yuv2rgb_y_offset;
1407 Y *= c->yuv2rgb_y_coeff;
1408 Y += 1 << 13;
1409
1410 if (hasAlpha) {
1411 A = abuf0[i] << 11;
1412
1413 A += 1 << 13;
1414 }
1415
1416 R = V * c->yuv2rgb_v2r_coeff;
1417 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1418 B = U * c->yuv2rgb_u2b_coeff;
1419
1420 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1421 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1422 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1423 if (eightbytes) {
1424 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1425 dest += 4;
1426 } else {
1427 dest += 3;
1428 }
1429 }
1430 }
1431 }
1432
1433 #undef output_pixel
1434 #undef r_b
1435 #undef b_r
1436
1437 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1438 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1439 const int16_t **_lumSrc, int lumFilterSize, \
1440 const int16_t *chrFilter, const int16_t **_chrUSrc, \
1441 const int16_t **_chrVSrc, int chrFilterSize, \
1442 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1443 int y) \
1444 { \
1445 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1446 **chrUSrc = (const int32_t **) _chrUSrc, \
1447 **chrVSrc = (const int32_t **) _chrVSrc, \
1448 **alpSrc = (const int32_t **) _alpSrc; \
1449 uint16_t *dest = (uint16_t *) _dest; \
1450 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1451 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1452 alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1453 } \
1454 \
1455 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1456 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1457 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1458 int yalpha, int uvalpha, int y) \
1459 { \
1460 const int32_t **buf = (const int32_t **) _buf, \
1461 **ubuf = (const int32_t **) _ubuf, \
1462 **vbuf = (const int32_t **) _vbuf, \
1463 **abuf = (const int32_t **) _abuf; \
1464 uint16_t *dest = (uint16_t *) _dest; \
1465 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1466 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1467 } \
1468 \
1469 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1470 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1471 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1472 int uvalpha, int y) \
1473 { \
1474 const int32_t *buf0 = (const int32_t *) _buf0, \
1475 **ubuf = (const int32_t **) _ubuf, \
1476 **vbuf = (const int32_t **) _vbuf, \
1477 *abuf0 = (const int32_t *) _abuf0; \
1478 uint16_t *dest = (uint16_t *) _dest; \
1479 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1480 dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1481 }
1482
1483 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1484 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1485 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1486 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1487 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1488 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1489 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1490 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1491 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1492 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1493 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1494 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1495 YUV2PACKED16WRAPPER(yuv2, ya16, ya16be, AV_PIX_FMT_YA16BE, 1, 0)
1496 YUV2PACKED16WRAPPER(yuv2, ya16, ya16le, AV_PIX_FMT_YA16LE, 1, 0)
1497
1498 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1499 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1500 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1501 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1502 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1503 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1504 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1505 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1506 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1507 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1508 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1509 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1510
1511 /*
1512 * Write out 2 RGB pixels in the target pixel format. This function takes a
1513 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1514 * things like endianness conversion and shifting. The caller takes care of
1515 * setting the correct offset in these tables from the chroma (U/V) values.
1516 * This function then uses the luminance (Y1/Y2) values to write out the
1517 * correct RGB values into the destination buffer.
1518 */
1519 static av_always_inline void
yuv2rgb_write(uint8_t * _dest,int i,int Y1,int Y2,unsigned A1,unsigned A2,const void * _r,const void * _g,const void * _b,int y,enum AVPixelFormat target,int hasAlpha)1520 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1521 unsigned A1, unsigned A2,
1522 const void *_r, const void *_g, const void *_b, int y,
1523 enum AVPixelFormat target, int hasAlpha)
1524 {
1525 if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1526 target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1527 uint32_t *dest = (uint32_t *) _dest;
1528 const uint32_t *r = (const uint32_t *) _r;
1529 const uint32_t *g = (const uint32_t *) _g;
1530 const uint32_t *b = (const uint32_t *) _b;
1531
1532 #if CONFIG_SMALL
1533 int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1534
1535 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1536 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1537 #else
1538 if (hasAlpha) {
1539 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1540
1541 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1542 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1543 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1544 } else {
1545 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1546 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1547
1548 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1549 #endif
1550 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1551 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1552 }
1553 #endif
1554 } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1555 uint8_t *dest = (uint8_t *) _dest;
1556 const uint8_t *r = (const uint8_t *) _r;
1557 const uint8_t *g = (const uint8_t *) _g;
1558 const uint8_t *b = (const uint8_t *) _b;
1559
1560 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1561 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1562
1563 dest[i * 6 + 0] = r_b[Y1];
1564 dest[i * 6 + 1] = g[Y1];
1565 dest[i * 6 + 2] = b_r[Y1];
1566 dest[i * 6 + 3] = r_b[Y2];
1567 dest[i * 6 + 4] = g[Y2];
1568 dest[i * 6 + 5] = b_r[Y2];
1569 #undef r_b
1570 #undef b_r
1571 } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1572 target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1573 target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1574 uint16_t *dest = (uint16_t *) _dest;
1575 const uint16_t *r = (const uint16_t *) _r;
1576 const uint16_t *g = (const uint16_t *) _g;
1577 const uint16_t *b = (const uint16_t *) _b;
1578 int dr1, dg1, db1, dr2, dg2, db2;
1579
1580 if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1581 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1582 dg1 = ff_dither_2x2_4[ y & 1 ][0];
1583 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1584 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1585 dg2 = ff_dither_2x2_4[ y & 1 ][1];
1586 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1587 } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1588 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1589 dg1 = ff_dither_2x2_8[ y & 1 ][1];
1590 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1591 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1592 dg2 = ff_dither_2x2_8[ y & 1 ][0];
1593 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1594 } else {
1595 dr1 = ff_dither_4x4_16[ y & 3 ][0];
1596 dg1 = ff_dither_4x4_16[ y & 3 ][1];
1597 db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1598 dr2 = ff_dither_4x4_16[ y & 3 ][1];
1599 dg2 = ff_dither_4x4_16[ y & 3 ][0];
1600 db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1601 }
1602
1603 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1604 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1605 } else /* 8/4 bits */ {
1606 uint8_t *dest = (uint8_t *) _dest;
1607 const uint8_t *r = (const uint8_t *) _r;
1608 const uint8_t *g = (const uint8_t *) _g;
1609 const uint8_t *b = (const uint8_t *) _b;
1610 int dr1, dg1, db1, dr2, dg2, db2;
1611
1612 if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1613 const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1614 const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1615 dr1 = dg1 = d32[(i * 2 + 0) & 7];
1616 db1 = d64[(i * 2 + 0) & 7];
1617 dr2 = dg2 = d32[(i * 2 + 1) & 7];
1618 db2 = d64[(i * 2 + 1) & 7];
1619 } else {
1620 const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1621 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1622 dr1 = db1 = d128[(i * 2 + 0) & 7];
1623 dg1 = d64[(i * 2 + 0) & 7];
1624 dr2 = db2 = d128[(i * 2 + 1) & 7];
1625 dg2 = d64[(i * 2 + 1) & 7];
1626 }
1627
1628 if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1629 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1630 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1631 } else {
1632 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1633 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1634 }
1635 }
1636 }
1637
1638 static av_always_inline void
yuv2rgb_X_c_template(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrc,const int16_t ** chrVSrc,int chrFilterSize,const int16_t ** alpSrc,uint8_t * dest,int dstW,int y,enum AVPixelFormat target,int hasAlpha)1639 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1640 const int16_t **lumSrc, int lumFilterSize,
1641 const int16_t *chrFilter, const int16_t **chrUSrc,
1642 const int16_t **chrVSrc, int chrFilterSize,
1643 const int16_t **alpSrc, uint8_t *dest, int dstW,
1644 int y, enum AVPixelFormat target, int hasAlpha)
1645 {
1646 int i;
1647
1648 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1649 int j, A1, A2;
1650 int Y1 = 1 << 18;
1651 int Y2 = 1 << 18;
1652 int U = 1 << 18;
1653 int V = 1 << 18;
1654 const void *r, *g, *b;
1655
1656 for (j = 0; j < lumFilterSize; j++) {
1657 Y1 += lumSrc[j][i * 2] * lumFilter[j];
1658 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1659 }
1660 for (j = 0; j < chrFilterSize; j++) {
1661 U += chrUSrc[j][i] * chrFilter[j];
1662 V += chrVSrc[j][i] * chrFilter[j];
1663 }
1664 Y1 >>= 19;
1665 Y2 >>= 19;
1666 U >>= 19;
1667 V >>= 19;
1668 if (hasAlpha) {
1669 A1 = 1 << 18;
1670 A2 = 1 << 18;
1671 for (j = 0; j < lumFilterSize; j++) {
1672 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1673 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1674 }
1675 A1 >>= 19;
1676 A2 >>= 19;
1677 if ((A1 | A2) & 0x100) {
1678 A1 = av_clip_uint8(A1);
1679 A2 = av_clip_uint8(A2);
1680 }
1681 }
1682
1683 r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1684 g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1685 b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1686
1687 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1688 r, g, b, y, target, hasAlpha);
1689 }
1690 }
1691
1692 static av_always_inline void
yuv2rgb_2_c_template(SwsContext * c,const int16_t * buf[2],const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf[2],uint8_t * dest,int dstW,int yalpha,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha)1693 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1694 const int16_t *ubuf[2], const int16_t *vbuf[2],
1695 const int16_t *abuf[2], uint8_t *dest, int dstW,
1696 int yalpha, int uvalpha, int y,
1697 enum AVPixelFormat target, int hasAlpha)
1698 {
1699 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1700 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1701 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1702 *abuf0 = hasAlpha ? abuf[0] : NULL,
1703 *abuf1 = hasAlpha ? abuf[1] : NULL;
1704 int yalpha1 = 4096 - yalpha;
1705 int uvalpha1 = 4096 - uvalpha;
1706 int i;
1707 av_assert2(yalpha <= 4096U);
1708 av_assert2(uvalpha <= 4096U);
1709
1710 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1711 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1712 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1713 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1714 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1715 int A1, A2;
1716 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1717 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1718 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1719
1720 if (hasAlpha) {
1721 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1722 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1723 A1 = av_clip_uint8(A1);
1724 A2 = av_clip_uint8(A2);
1725 }
1726
1727 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1728 r, g, b, y, target, hasAlpha);
1729 }
1730 }
1731
1732 static av_always_inline void
yuv2rgb_1_c_template(SwsContext * c,const int16_t * buf0,const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf0,uint8_t * dest,int dstW,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha)1733 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1734 const int16_t *ubuf[2], const int16_t *vbuf[2],
1735 const int16_t *abuf0, uint8_t *dest, int dstW,
1736 int uvalpha, int y, enum AVPixelFormat target,
1737 int hasAlpha)
1738 {
1739 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1740 int i;
1741
1742 if (uvalpha < 2048) {
1743 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1744 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1745 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1746 int U = (ubuf0[i] + 64) >> 7;
1747 int V = (vbuf0[i] + 64) >> 7;
1748 int A1, A2;
1749 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1750 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1751 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1752
1753 if (hasAlpha) {
1754 A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1755 A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1756 A1 = av_clip_uint8(A1);
1757 A2 = av_clip_uint8(A2);
1758 }
1759
1760 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1761 r, g, b, y, target, hasAlpha);
1762 }
1763 } else {
1764 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1765 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1766 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1767 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1768 int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1769 int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1770 int A1, A2;
1771 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1772 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1773 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1774
1775 if (hasAlpha) {
1776 A1 = (abuf0[i * 2 ] + 64) >> 7;
1777 A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1778 A1 = av_clip_uint8(A1);
1779 A2 = av_clip_uint8(A2);
1780 }
1781
1782 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1783 r, g, b, y, target, hasAlpha);
1784 }
1785 }
1786 }
1787
1788 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1789 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1790 const int16_t **lumSrc, int lumFilterSize, \
1791 const int16_t *chrFilter, const int16_t **chrUSrc, \
1792 const int16_t **chrVSrc, int chrFilterSize, \
1793 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1794 int y) \
1795 { \
1796 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1797 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1798 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1799 }
1800
1801 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1802 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1803 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1804 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1805 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1806 int yalpha, int uvalpha, int y) \
1807 { \
1808 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1809 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1810 }
1811
1812 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1813 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1814 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1815 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1816 const int16_t *abuf0, uint8_t *dest, int dstW, \
1817 int uvalpha, int y) \
1818 { \
1819 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1820 dstW, uvalpha, y, fmt, hasAlpha); \
1821 }
1822
1823 #if CONFIG_SMALL
1824 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1825 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1826 #else
1827 #if CONFIG_SWSCALE_ALPHA
1828 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1829 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1830 #endif
1831 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1832 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1833 #endif
1834 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1835 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1836 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1837 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1838 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1839 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1840 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1841 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1842
yuv2rgb_write_full(SwsContext * c,uint8_t * dest,int i,int Y,int A,int U,int V,int y,enum AVPixelFormat target,int hasAlpha,int err[4])1843 static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1844 uint8_t *dest, int i, int Y, int A, int U, int V,
1845 int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1846 {
1847 int R, G, B;
1848 int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1849
1850 Y -= c->yuv2rgb_y_offset;
1851 Y *= c->yuv2rgb_y_coeff;
1852 Y += 1 << 21;
1853 R = (unsigned)Y + V*c->yuv2rgb_v2r_coeff;
1854 G = (unsigned)Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1855 B = (unsigned)Y + U*c->yuv2rgb_u2b_coeff;
1856 if ((R | G | B) & 0xC0000000) {
1857 R = av_clip_uintp2(R, 30);
1858 G = av_clip_uintp2(G, 30);
1859 B = av_clip_uintp2(B, 30);
1860 }
1861
1862 switch(target) {
1863 case AV_PIX_FMT_ARGB:
1864 dest[0] = hasAlpha ? A : 255;
1865 dest[1] = R >> 22;
1866 dest[2] = G >> 22;
1867 dest[3] = B >> 22;
1868 break;
1869 case AV_PIX_FMT_RGB24:
1870 dest[0] = R >> 22;
1871 dest[1] = G >> 22;
1872 dest[2] = B >> 22;
1873 break;
1874 case AV_PIX_FMT_RGBA:
1875 dest[0] = R >> 22;
1876 dest[1] = G >> 22;
1877 dest[2] = B >> 22;
1878 dest[3] = hasAlpha ? A : 255;
1879 break;
1880 case AV_PIX_FMT_ABGR:
1881 dest[0] = hasAlpha ? A : 255;
1882 dest[1] = B >> 22;
1883 dest[2] = G >> 22;
1884 dest[3] = R >> 22;
1885 break;
1886 case AV_PIX_FMT_BGR24:
1887 dest[0] = B >> 22;
1888 dest[1] = G >> 22;
1889 dest[2] = R >> 22;
1890 break;
1891 case AV_PIX_FMT_BGRA:
1892 dest[0] = B >> 22;
1893 dest[1] = G >> 22;
1894 dest[2] = R >> 22;
1895 dest[3] = hasAlpha ? A : 255;
1896 break;
1897 case AV_PIX_FMT_BGR4_BYTE:
1898 case AV_PIX_FMT_RGB4_BYTE:
1899 case AV_PIX_FMT_BGR8:
1900 case AV_PIX_FMT_RGB8:
1901 {
1902 int r,g,b;
1903
1904 switch (c->dither) {
1905 default:
1906 case SWS_DITHER_AUTO:
1907 case SWS_DITHER_ED:
1908 R >>= 22;
1909 G >>= 22;
1910 B >>= 22;
1911 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;
1912 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;
1913 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;
1914 c->dither_error[0][i] = err[0];
1915 c->dither_error[1][i] = err[1];
1916 c->dither_error[2][i] = err[2];
1917 r = R >> (isrgb8 ? 5 : 7);
1918 g = G >> (isrgb8 ? 5 : 6);
1919 b = B >> (isrgb8 ? 6 : 7);
1920 r = av_clip(r, 0, isrgb8 ? 7 : 1);
1921 g = av_clip(g, 0, isrgb8 ? 7 : 3);
1922 b = av_clip(b, 0, isrgb8 ? 3 : 1);
1923 err[0] = R - r*(isrgb8 ? 36 : 255);
1924 err[1] = G - g*(isrgb8 ? 36 : 85);
1925 err[2] = B - b*(isrgb8 ? 85 : 255);
1926 break;
1927 case SWS_DITHER_A_DITHER:
1928 if (isrgb8) {
1929 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1930 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1931 r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1932 g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1933 b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1934 r = av_clip_uintp2(r, 3);
1935 g = av_clip_uintp2(g, 3);
1936 b = av_clip_uintp2(b, 2);
1937 } else {
1938 r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1939 g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1940 b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1941 r = av_clip_uintp2(r, 1);
1942 g = av_clip_uintp2(g, 2);
1943 b = av_clip_uintp2(b, 1);
1944 }
1945 break;
1946 case SWS_DITHER_X_DITHER:
1947 if (isrgb8) {
1948 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1949 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1950 r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1951 g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1952 b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1953 r = av_clip_uintp2(r, 3);
1954 g = av_clip_uintp2(g, 3);
1955 b = av_clip_uintp2(b, 2);
1956 } else {
1957 r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1958 g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1959 b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1960 r = av_clip_uintp2(r, 1);
1961 g = av_clip_uintp2(g, 2);
1962 b = av_clip_uintp2(b, 1);
1963 }
1964
1965 break;
1966 }
1967
1968 if(target == AV_PIX_FMT_BGR4_BYTE) {
1969 dest[0] = r + 2*g + 8*b;
1970 } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1971 dest[0] = b + 2*g + 8*r;
1972 } else if(target == AV_PIX_FMT_BGR8) {
1973 dest[0] = r + 8*g + 64*b;
1974 } else if(target == AV_PIX_FMT_RGB8) {
1975 dest[0] = b + 4*g + 32*r;
1976 } else
1977 av_assert2(0);
1978 break;}
1979 }
1980 }
1981
1982 static av_always_inline void
yuv2rgb_full_X_c_template(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrc,const int16_t ** chrVSrc,int chrFilterSize,const int16_t ** alpSrc,uint8_t * dest,int dstW,int y,enum AVPixelFormat target,int hasAlpha)1983 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1984 const int16_t **lumSrc, int lumFilterSize,
1985 const int16_t *chrFilter, const int16_t **chrUSrc,
1986 const int16_t **chrVSrc, int chrFilterSize,
1987 const int16_t **alpSrc, uint8_t *dest,
1988 int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1989 {
1990 int i;
1991 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1992 int err[4] = {0};
1993 int A = 0; //init to silence warning
1994
1995 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1996 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1997 step = 1;
1998
1999 for (i = 0; i < dstW; i++) {
2000 int j;
2001 int Y = 1<<9;
2002 int U = (1<<9)-(128 << 19);
2003 int V = (1<<9)-(128 << 19);
2004
2005 for (j = 0; j < lumFilterSize; j++) {
2006 Y += lumSrc[j][i] * lumFilter[j];
2007 }
2008 for (j = 0; j < chrFilterSize; j++) {
2009 U += chrUSrc[j][i] * chrFilter[j];
2010 V += chrVSrc[j][i] * chrFilter[j];
2011 }
2012 Y >>= 10;
2013 U >>= 10;
2014 V >>= 10;
2015 if (hasAlpha) {
2016 A = 1 << 18;
2017 for (j = 0; j < lumFilterSize; j++) {
2018 A += alpSrc[j][i] * lumFilter[j];
2019 }
2020 A >>= 19;
2021 if (A & 0x100)
2022 A = av_clip_uint8(A);
2023 }
2024 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2025 dest += step;
2026 }
2027 c->dither_error[0][i] = err[0];
2028 c->dither_error[1][i] = err[1];
2029 c->dither_error[2][i] = err[2];
2030 }
2031
2032 static av_always_inline void
yuv2rgb_full_2_c_template(SwsContext * c,const int16_t * buf[2],const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf[2],uint8_t * dest,int dstW,int yalpha,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha)2033 yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
2034 const int16_t *ubuf[2], const int16_t *vbuf[2],
2035 const int16_t *abuf[2], uint8_t *dest, int dstW,
2036 int yalpha, int uvalpha, int y,
2037 enum AVPixelFormat target, int hasAlpha)
2038 {
2039 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2040 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
2041 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
2042 *abuf0 = hasAlpha ? abuf[0] : NULL,
2043 *abuf1 = hasAlpha ? abuf[1] : NULL;
2044 int yalpha1 = 4096 - yalpha;
2045 int uvalpha1 = 4096 - uvalpha;
2046 int i;
2047 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2048 int err[4] = {0};
2049 int A = 0; // init to silcene warning
2050
2051 av_assert2(yalpha <= 4096U);
2052 av_assert2(uvalpha <= 4096U);
2053
2054 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2055 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2056 step = 1;
2057
2058 for (i = 0; i < dstW; i++) {
2059 int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
2060 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
2061 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
2062
2063 if (hasAlpha) {
2064 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
2065 if (A & 0x100)
2066 A = av_clip_uint8(A);
2067 }
2068
2069 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2070 dest += step;
2071 }
2072 c->dither_error[0][i] = err[0];
2073 c->dither_error[1][i] = err[1];
2074 c->dither_error[2][i] = err[2];
2075 }
2076
2077 static av_always_inline void
yuv2rgb_full_1_c_template(SwsContext * c,const int16_t * buf0,const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf0,uint8_t * dest,int dstW,int uvalpha,int y,enum AVPixelFormat target,int hasAlpha)2078 yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
2079 const int16_t *ubuf[2], const int16_t *vbuf[2],
2080 const int16_t *abuf0, uint8_t *dest, int dstW,
2081 int uvalpha, int y, enum AVPixelFormat target,
2082 int hasAlpha)
2083 {
2084 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
2085 int i;
2086 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2087 int err[4] = {0};
2088
2089 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2090 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2091 step = 1;
2092
2093 if (uvalpha < 2048) {
2094 int A = 0; //init to silence warning
2095 for (i = 0; i < dstW; i++) {
2096 int Y = buf0[i] * 4;
2097 int U = (ubuf0[i] - (128<<7)) * 4;
2098 int V = (vbuf0[i] - (128<<7)) * 4;
2099
2100 if (hasAlpha) {
2101 A = (abuf0[i] + 64) >> 7;
2102 if (A & 0x100)
2103 A = av_clip_uint8(A);
2104 }
2105
2106 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2107 dest += step;
2108 }
2109 } else {
2110 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
2111 int A = 0; //init to silence warning
2112 for (i = 0; i < dstW; i++) {
2113 int Y = buf0[i] * 4;
2114 int U = (ubuf0[i] + ubuf1[i] - (128<<8)) * 2;
2115 int V = (vbuf0[i] + vbuf1[i] - (128<<8)) * 2;
2116
2117 if (hasAlpha) {
2118 A = (abuf0[i] + 64) >> 7;
2119 if (A & 0x100)
2120 A = av_clip_uint8(A);
2121 }
2122
2123 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2124 dest += step;
2125 }
2126 }
2127
2128 c->dither_error[0][i] = err[0];
2129 c->dither_error[1][i] = err[1];
2130 c->dither_error[2][i] = err[2];
2131 }
2132
2133 #if CONFIG_SMALL
2134 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2135 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2136 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2137 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2138 #else
2139 #if CONFIG_SWSCALE_ALPHA
2140 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
2141 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
2142 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
2143 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
2144 #endif
2145 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
2146 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
2147 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
2148 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
2149 #endif
2150 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
2151 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
2152
2153 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
2154 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
2155 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
2156 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
2157
2158 static void
yuv2gbrp_full_X_c(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrc,const int16_t ** chrVSrc,int chrFilterSize,const int16_t ** alpSrc,uint8_t ** dest,int dstW,int y)2159 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
2160 const int16_t **lumSrc, int lumFilterSize,
2161 const int16_t *chrFilter, const int16_t **chrUSrc,
2162 const int16_t **chrVSrc, int chrFilterSize,
2163 const int16_t **alpSrc, uint8_t **dest,
2164 int dstW, int y)
2165 {
2166 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2167 int i;
2168 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
2169 uint16_t **dest16 = (uint16_t**)dest;
2170 int SH = 22 + 8 - desc->comp[0].depth;
2171 int A = 0; // init to silence warning
2172
2173 for (i = 0; i < dstW; i++) {
2174 int j;
2175 int Y = 1 << 9;
2176 int U = (1 << 9) - (128 << 19);
2177 int V = (1 << 9) - (128 << 19);
2178 int R, G, B;
2179
2180 for (j = 0; j < lumFilterSize; j++)
2181 Y += lumSrc[j][i] * lumFilter[j];
2182
2183 for (j = 0; j < chrFilterSize; j++) {
2184 U += chrUSrc[j][i] * chrFilter[j];
2185 V += chrVSrc[j][i] * chrFilter[j];
2186 }
2187
2188 Y >>= 10;
2189 U >>= 10;
2190 V >>= 10;
2191
2192 if (hasAlpha) {
2193 A = 1 << 18;
2194
2195 for (j = 0; j < lumFilterSize; j++)
2196 A += alpSrc[j][i] * lumFilter[j];
2197
2198 if (A & 0xF8000000)
2199 A = av_clip_uintp2(A, 27);
2200 }
2201
2202 Y -= c->yuv2rgb_y_offset;
2203 Y *= c->yuv2rgb_y_coeff;
2204 Y += 1 << (SH-1);
2205 R = Y + V * c->yuv2rgb_v2r_coeff;
2206 G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2207 B = Y + U * c->yuv2rgb_u2b_coeff;
2208
2209 if ((R | G | B) & 0xC0000000) {
2210 R = av_clip_uintp2(R, 30);
2211 G = av_clip_uintp2(G, 30);
2212 B = av_clip_uintp2(B, 30);
2213 }
2214
2215 if (SH != 22) {
2216 dest16[0][i] = G >> SH;
2217 dest16[1][i] = B >> SH;
2218 dest16[2][i] = R >> SH;
2219 if (hasAlpha)
2220 dest16[3][i] = A >> (SH - 3);
2221 } else {
2222 dest[0][i] = G >> 22;
2223 dest[1][i] = B >> 22;
2224 dest[2][i] = R >> 22;
2225 if (hasAlpha)
2226 dest[3][i] = A >> 19;
2227 }
2228 }
2229 if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2230 for (i = 0; i < dstW; i++) {
2231 dest16[0][i] = av_bswap16(dest16[0][i]);
2232 dest16[1][i] = av_bswap16(dest16[1][i]);
2233 dest16[2][i] = av_bswap16(dest16[2][i]);
2234 if (hasAlpha)
2235 dest16[3][i] = av_bswap16(dest16[3][i]);
2236 }
2237 }
2238 }
2239
2240 static void
yuv2gbrp16_full_X_c(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrcx,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrcx,const int16_t ** chrVSrcx,int chrFilterSize,const int16_t ** alpSrcx,uint8_t ** dest,int dstW,int y)2241 yuv2gbrp16_full_X_c(SwsContext *c, const int16_t *lumFilter,
2242 const int16_t **lumSrcx, int lumFilterSize,
2243 const int16_t *chrFilter, const int16_t **chrUSrcx,
2244 const int16_t **chrVSrcx, int chrFilterSize,
2245 const int16_t **alpSrcx, uint8_t **dest,
2246 int dstW, int y)
2247 {
2248 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2249 int i;
2250 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2251 uint16_t **dest16 = (uint16_t**)dest;
2252 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2253 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2254 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2255 const int32_t **alpSrc = (const int32_t**)alpSrcx;
2256
2257 for (i = 0; i < dstW; i++) {
2258 int j;
2259 int Y = -0x40000000;
2260 int U = -(128 << 23);
2261 int V = -(128 << 23);
2262 int R, G, B, A;
2263
2264 for (j = 0; j < lumFilterSize; j++)
2265 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2266
2267 for (j = 0; j < chrFilterSize; j++) {
2268 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2269 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2270 }
2271
2272 Y >>= 14;
2273 Y += 0x10000;
2274 U >>= 14;
2275 V >>= 14;
2276
2277 if (hasAlpha) {
2278 A = -0x40000000;
2279
2280 for (j = 0; j < lumFilterSize; j++)
2281 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2282
2283 A >>= 1;
2284 A += 0x20002000;
2285 }
2286
2287 Y -= c->yuv2rgb_y_offset;
2288 Y *= c->yuv2rgb_y_coeff;
2289 Y += 1 << 13;
2290 R = V * c->yuv2rgb_v2r_coeff;
2291 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2292 B = U * c->yuv2rgb_u2b_coeff;
2293
2294 R = av_clip_uintp2(Y + R, 30);
2295 G = av_clip_uintp2(Y + G, 30);
2296 B = av_clip_uintp2(Y + B, 30);
2297
2298 dest16[0][i] = G >> 14;
2299 dest16[1][i] = B >> 14;
2300 dest16[2][i] = R >> 14;
2301 if (hasAlpha)
2302 dest16[3][i] = av_clip_uintp2(A, 30) >> 14;
2303 }
2304 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2305 for (i = 0; i < dstW; i++) {
2306 dest16[0][i] = av_bswap16(dest16[0][i]);
2307 dest16[1][i] = av_bswap16(dest16[1][i]);
2308 dest16[2][i] = av_bswap16(dest16[2][i]);
2309 if (hasAlpha)
2310 dest16[3][i] = av_bswap16(dest16[3][i]);
2311 }
2312 }
2313 }
2314
2315 static void
yuv2gbrpf32_full_X_c(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrcx,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrcx,const int16_t ** chrVSrcx,int chrFilterSize,const int16_t ** alpSrcx,uint8_t ** dest,int dstW,int y)2316 yuv2gbrpf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
2317 const int16_t **lumSrcx, int lumFilterSize,
2318 const int16_t *chrFilter, const int16_t **chrUSrcx,
2319 const int16_t **chrVSrcx, int chrFilterSize,
2320 const int16_t **alpSrcx, uint8_t **dest,
2321 int dstW, int y)
2322 {
2323 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2324 int i;
2325 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2326 uint32_t **dest32 = (uint32_t**)dest;
2327 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2328 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2329 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2330 const int32_t **alpSrc = (const int32_t**)alpSrcx;
2331 static const float float_mult = 1.0f / 65535.0f;
2332
2333 for (i = 0; i < dstW; i++) {
2334 int j;
2335 int Y = -0x40000000;
2336 int U = -(128 << 23);
2337 int V = -(128 << 23);
2338 int R, G, B, A;
2339
2340 for (j = 0; j < lumFilterSize; j++)
2341 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2342
2343 for (j = 0; j < chrFilterSize; j++) {
2344 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2345 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2346 }
2347
2348 Y >>= 14;
2349 Y += 0x10000;
2350 U >>= 14;
2351 V >>= 14;
2352
2353 if (hasAlpha) {
2354 A = -0x40000000;
2355
2356 for (j = 0; j < lumFilterSize; j++)
2357 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2358
2359 A >>= 1;
2360 A += 0x20002000;
2361 }
2362
2363 Y -= c->yuv2rgb_y_offset;
2364 Y *= c->yuv2rgb_y_coeff;
2365 Y += 1 << 13;
2366 R = V * c->yuv2rgb_v2r_coeff;
2367 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2368 B = U * c->yuv2rgb_u2b_coeff;
2369
2370 R = av_clip_uintp2(Y + R, 30);
2371 G = av_clip_uintp2(Y + G, 30);
2372 B = av_clip_uintp2(Y + B, 30);
2373
2374 dest32[0][i] = av_float2int(float_mult * (float)(G >> 14));
2375 dest32[1][i] = av_float2int(float_mult * (float)(B >> 14));
2376 dest32[2][i] = av_float2int(float_mult * (float)(R >> 14));
2377 if (hasAlpha)
2378 dest32[3][i] = av_float2int(float_mult * (float)(av_clip_uintp2(A, 30) >> 14));
2379 }
2380 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2381 for (i = 0; i < dstW; i++) {
2382 dest32[0][i] = av_bswap32(dest32[0][i]);
2383 dest32[1][i] = av_bswap32(dest32[1][i]);
2384 dest32[2][i] = av_bswap32(dest32[2][i]);
2385 if (hasAlpha)
2386 dest32[3][i] = av_bswap32(dest32[3][i]);
2387 }
2388 }
2389 }
2390
2391 static void
yuv2ya8_1_c(SwsContext * c,const int16_t * buf0,const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf0,uint8_t * dest,int dstW,int uvalpha,int y)2392 yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
2393 const int16_t *ubuf[2], const int16_t *vbuf[2],
2394 const int16_t *abuf0, uint8_t *dest, int dstW,
2395 int uvalpha, int y)
2396 {
2397 int hasAlpha = !!abuf0;
2398 int i;
2399
2400 for (i = 0; i < dstW; i++) {
2401 int Y = (buf0[i] + 64) >> 7;
2402 int A;
2403
2404 Y = av_clip_uint8(Y);
2405
2406 if (hasAlpha) {
2407 A = (abuf0[i] + 64) >> 7;
2408 if (A & 0x100)
2409 A = av_clip_uint8(A);
2410 }
2411
2412 dest[i * 2 ] = Y;
2413 dest[i * 2 + 1] = hasAlpha ? A : 255;
2414 }
2415 }
2416
2417 static void
yuv2ya8_2_c(SwsContext * c,const int16_t * buf[2],const int16_t * ubuf[2],const int16_t * vbuf[2],const int16_t * abuf[2],uint8_t * dest,int dstW,int yalpha,int uvalpha,int y)2418 yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
2419 const int16_t *ubuf[2], const int16_t *vbuf[2],
2420 const int16_t *abuf[2], uint8_t *dest, int dstW,
2421 int yalpha, int uvalpha, int y)
2422 {
2423 int hasAlpha = abuf && abuf[0] && abuf[1];
2424 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2425 *abuf0 = hasAlpha ? abuf[0] : NULL,
2426 *abuf1 = hasAlpha ? abuf[1] : NULL;
2427 int yalpha1 = 4096 - yalpha;
2428 int i;
2429
2430 av_assert2(yalpha <= 4096U);
2431
2432 for (i = 0; i < dstW; i++) {
2433 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 19;
2434 int A;
2435
2436 Y = av_clip_uint8(Y);
2437
2438 if (hasAlpha) {
2439 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 19;
2440 A = av_clip_uint8(A);
2441 }
2442
2443 dest[i * 2 ] = Y;
2444 dest[i * 2 + 1] = hasAlpha ? A : 255;
2445 }
2446 }
2447
2448 static void
yuv2ya8_X_c(SwsContext * c,const int16_t * lumFilter,const int16_t ** lumSrc,int lumFilterSize,const int16_t * chrFilter,const int16_t ** chrUSrc,const int16_t ** chrVSrc,int chrFilterSize,const int16_t ** alpSrc,uint8_t * dest,int dstW,int y)2449 yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
2450 const int16_t **lumSrc, int lumFilterSize,
2451 const int16_t *chrFilter, const int16_t **chrUSrc,
2452 const int16_t **chrVSrc, int chrFilterSize,
2453 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
2454 {
2455 int hasAlpha = !!alpSrc;
2456 int i;
2457
2458 for (i = 0; i < dstW; i++) {
2459 int j;
2460 int Y = 1 << 18, A = 1 << 18;
2461
2462 for (j = 0; j < lumFilterSize; j++)
2463 Y += lumSrc[j][i] * lumFilter[j];
2464
2465 Y >>= 19;
2466 if (Y & 0x100)
2467 Y = av_clip_uint8(Y);
2468
2469 if (hasAlpha) {
2470 for (j = 0; j < lumFilterSize; j++)
2471 A += alpSrc[j][i] * lumFilter[j];
2472
2473 A >>= 19;
2474
2475 if (A & 0x100)
2476 A = av_clip_uint8(A);
2477 }
2478
2479 dest[2 * i ] = Y;
2480 dest[2 * i + 1] = hasAlpha ? A : 255;
2481 }
2482 }
2483
2484 static void
yuv2ayuv64le_X_c(SwsContext * c,const int16_t * lumFilter,const int16_t ** _lumSrc,int lumFilterSize,const int16_t * chrFilter,const int16_t ** _chrUSrc,const int16_t ** _chrVSrc,int chrFilterSize,const int16_t ** _alpSrc,uint8_t * dest,int dstW,int y)2485 yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter,
2486 const int16_t **_lumSrc, int lumFilterSize,
2487 const int16_t *chrFilter, const int16_t **_chrUSrc,
2488 const int16_t **_chrVSrc, int chrFilterSize,
2489 const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
2490 {
2491 const int32_t **lumSrc = (const int32_t **) _lumSrc,
2492 **chrUSrc = (const int32_t **) _chrUSrc,
2493 **chrVSrc = (const int32_t **) _chrVSrc,
2494 **alpSrc = (const int32_t **) _alpSrc;
2495 int hasAlpha = !!alpSrc;
2496 int i;
2497
2498 for (i = 0; i < dstW; i++) {
2499 int Y = 1 << 14, U = 1 << 14;
2500 int V = 1 << 14, A = 1 << 14;
2501 int j;
2502
2503 Y -= 0x40000000;
2504 U -= 0x40000000;
2505 V -= 0x40000000;
2506 A -= 0x40000000;
2507
2508 for (j = 0; j < lumFilterSize; j++)
2509 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2510
2511 for (j = 0; j < chrFilterSize; j++)
2512 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2513
2514 for (j = 0; j < chrFilterSize; j++)
2515 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2516
2517 if (hasAlpha)
2518 for (j = 0; j < lumFilterSize; j++)
2519 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2520
2521 Y = 0x8000 + av_clip_int16(Y >> 15);
2522 U = 0x8000 + av_clip_int16(U >> 15);
2523 V = 0x8000 + av_clip_int16(V >> 15);
2524 A = 0x8000 + av_clip_int16(A >> 15);
2525
2526 AV_WL16(dest + 8 * i, hasAlpha ? A : 65535);
2527 AV_WL16(dest + 8 * i + 2, Y);
2528 AV_WL16(dest + 8 * i + 4, U);
2529 AV_WL16(dest + 8 * i + 6, V);
2530 }
2531 }
2532
ff_sws_init_output_funcs(SwsContext * c,yuv2planar1_fn * yuv2plane1,yuv2planarX_fn * yuv2planeX,yuv2interleavedX_fn * yuv2nv12cX,yuv2packed1_fn * yuv2packed1,yuv2packed2_fn * yuv2packed2,yuv2packedX_fn * yuv2packedX,yuv2anyX_fn * yuv2anyX)2533 av_cold void ff_sws_init_output_funcs(SwsContext *c,
2534 yuv2planar1_fn *yuv2plane1,
2535 yuv2planarX_fn *yuv2planeX,
2536 yuv2interleavedX_fn *yuv2nv12cX,
2537 yuv2packed1_fn *yuv2packed1,
2538 yuv2packed2_fn *yuv2packed2,
2539 yuv2packedX_fn *yuv2packedX,
2540 yuv2anyX_fn *yuv2anyX)
2541 {
2542 enum AVPixelFormat dstFormat = c->dstFormat;
2543 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2544
2545 if (dstFormat == AV_PIX_FMT_P010LE || dstFormat == AV_PIX_FMT_P010BE) {
2546 *yuv2plane1 = isBE(dstFormat) ? yuv2p010l1_BE_c : yuv2p010l1_LE_c;
2547 *yuv2planeX = isBE(dstFormat) ? yuv2p010lX_BE_c : yuv2p010lX_LE_c;
2548 *yuv2nv12cX = yuv2p010cX_c;
2549 } else if (is16BPS(dstFormat)) {
2550 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
2551 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
2552 if (dstFormat == AV_PIX_FMT_P016LE || dstFormat == AV_PIX_FMT_P016BE) {
2553 *yuv2nv12cX = yuv2p016cX_c;
2554 }
2555 } else if (isNBPS(dstFormat)) {
2556 if (desc->comp[0].depth == 9) {
2557 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
2558 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
2559 } else if (desc->comp[0].depth == 10) {
2560 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
2561 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
2562 } else if (desc->comp[0].depth == 12) {
2563 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
2564 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
2565 } else if (desc->comp[0].depth == 14) {
2566 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
2567 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
2568 } else
2569 av_assert0(0);
2570 } else if (dstFormat == AV_PIX_FMT_GRAYF32BE) {
2571 *yuv2planeX = yuv2planeX_floatBE_c;
2572 *yuv2plane1 = yuv2plane1_floatBE_c;
2573 } else if (dstFormat == AV_PIX_FMT_GRAYF32LE) {
2574 *yuv2planeX = yuv2planeX_floatLE_c;
2575 *yuv2plane1 = yuv2plane1_floatLE_c;
2576 } else {
2577 *yuv2plane1 = yuv2plane1_8_c;
2578 *yuv2planeX = yuv2planeX_8_c;
2579 if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21 ||
2580 dstFormat == AV_PIX_FMT_NV24 || dstFormat == AV_PIX_FMT_NV42)
2581 *yuv2nv12cX = yuv2nv12cX_c;
2582 }
2583
2584 if(c->flags & SWS_FULL_CHR_H_INT) {
2585 switch (dstFormat) {
2586 case AV_PIX_FMT_RGBA:
2587 #if CONFIG_SMALL
2588 *yuv2packedX = yuv2rgba32_full_X_c;
2589 *yuv2packed2 = yuv2rgba32_full_2_c;
2590 *yuv2packed1 = yuv2rgba32_full_1_c;
2591 #else
2592 #if CONFIG_SWSCALE_ALPHA
2593 if (c->needAlpha) {
2594 *yuv2packedX = yuv2rgba32_full_X_c;
2595 *yuv2packed2 = yuv2rgba32_full_2_c;
2596 *yuv2packed1 = yuv2rgba32_full_1_c;
2597 } else
2598 #endif /* CONFIG_SWSCALE_ALPHA */
2599 {
2600 *yuv2packedX = yuv2rgbx32_full_X_c;
2601 *yuv2packed2 = yuv2rgbx32_full_2_c;
2602 *yuv2packed1 = yuv2rgbx32_full_1_c;
2603 }
2604 #endif /* !CONFIG_SMALL */
2605 break;
2606 case AV_PIX_FMT_ARGB:
2607 #if CONFIG_SMALL
2608 *yuv2packedX = yuv2argb32_full_X_c;
2609 *yuv2packed2 = yuv2argb32_full_2_c;
2610 *yuv2packed1 = yuv2argb32_full_1_c;
2611 #else
2612 #if CONFIG_SWSCALE_ALPHA
2613 if (c->needAlpha) {
2614 *yuv2packedX = yuv2argb32_full_X_c;
2615 *yuv2packed2 = yuv2argb32_full_2_c;
2616 *yuv2packed1 = yuv2argb32_full_1_c;
2617 } else
2618 #endif /* CONFIG_SWSCALE_ALPHA */
2619 {
2620 *yuv2packedX = yuv2xrgb32_full_X_c;
2621 *yuv2packed2 = yuv2xrgb32_full_2_c;
2622 *yuv2packed1 = yuv2xrgb32_full_1_c;
2623 }
2624 #endif /* !CONFIG_SMALL */
2625 break;
2626 case AV_PIX_FMT_BGRA:
2627 #if CONFIG_SMALL
2628 *yuv2packedX = yuv2bgra32_full_X_c;
2629 *yuv2packed2 = yuv2bgra32_full_2_c;
2630 *yuv2packed1 = yuv2bgra32_full_1_c;
2631 #else
2632 #if CONFIG_SWSCALE_ALPHA
2633 if (c->needAlpha) {
2634 *yuv2packedX = yuv2bgra32_full_X_c;
2635 *yuv2packed2 = yuv2bgra32_full_2_c;
2636 *yuv2packed1 = yuv2bgra32_full_1_c;
2637 } else
2638 #endif /* CONFIG_SWSCALE_ALPHA */
2639 {
2640 *yuv2packedX = yuv2bgrx32_full_X_c;
2641 *yuv2packed2 = yuv2bgrx32_full_2_c;
2642 *yuv2packed1 = yuv2bgrx32_full_1_c;
2643 }
2644 #endif /* !CONFIG_SMALL */
2645 break;
2646 case AV_PIX_FMT_ABGR:
2647 #if CONFIG_SMALL
2648 *yuv2packedX = yuv2abgr32_full_X_c;
2649 *yuv2packed2 = yuv2abgr32_full_2_c;
2650 *yuv2packed1 = yuv2abgr32_full_1_c;
2651 #else
2652 #if CONFIG_SWSCALE_ALPHA
2653 if (c->needAlpha) {
2654 *yuv2packedX = yuv2abgr32_full_X_c;
2655 *yuv2packed2 = yuv2abgr32_full_2_c;
2656 *yuv2packed1 = yuv2abgr32_full_1_c;
2657 } else
2658 #endif /* CONFIG_SWSCALE_ALPHA */
2659 {
2660 *yuv2packedX = yuv2xbgr32_full_X_c;
2661 *yuv2packed2 = yuv2xbgr32_full_2_c;
2662 *yuv2packed1 = yuv2xbgr32_full_1_c;
2663 }
2664 #endif /* !CONFIG_SMALL */
2665 break;
2666 case AV_PIX_FMT_RGBA64LE:
2667 #if CONFIG_SWSCALE_ALPHA
2668 if (c->needAlpha) {
2669 *yuv2packedX = yuv2rgba64le_full_X_c;
2670 *yuv2packed2 = yuv2rgba64le_full_2_c;
2671 *yuv2packed1 = yuv2rgba64le_full_1_c;
2672 } else
2673 #endif /* CONFIG_SWSCALE_ALPHA */
2674 {
2675 *yuv2packedX = yuv2rgbx64le_full_X_c;
2676 *yuv2packed2 = yuv2rgbx64le_full_2_c;
2677 *yuv2packed1 = yuv2rgbx64le_full_1_c;
2678 }
2679 break;
2680 case AV_PIX_FMT_RGBA64BE:
2681 #if CONFIG_SWSCALE_ALPHA
2682 if (c->needAlpha) {
2683 *yuv2packedX = yuv2rgba64be_full_X_c;
2684 *yuv2packed2 = yuv2rgba64be_full_2_c;
2685 *yuv2packed1 = yuv2rgba64be_full_1_c;
2686 } else
2687 #endif /* CONFIG_SWSCALE_ALPHA */
2688 {
2689 *yuv2packedX = yuv2rgbx64be_full_X_c;
2690 *yuv2packed2 = yuv2rgbx64be_full_2_c;
2691 *yuv2packed1 = yuv2rgbx64be_full_1_c;
2692 }
2693 break;
2694 case AV_PIX_FMT_BGRA64LE:
2695 #if CONFIG_SWSCALE_ALPHA
2696 if (c->needAlpha) {
2697 *yuv2packedX = yuv2bgra64le_full_X_c;
2698 *yuv2packed2 = yuv2bgra64le_full_2_c;
2699 *yuv2packed1 = yuv2bgra64le_full_1_c;
2700 } else
2701 #endif /* CONFIG_SWSCALE_ALPHA */
2702 {
2703 *yuv2packedX = yuv2bgrx64le_full_X_c;
2704 *yuv2packed2 = yuv2bgrx64le_full_2_c;
2705 *yuv2packed1 = yuv2bgrx64le_full_1_c;
2706 }
2707 break;
2708 case AV_PIX_FMT_BGRA64BE:
2709 #if CONFIG_SWSCALE_ALPHA
2710 if (c->needAlpha) {
2711 *yuv2packedX = yuv2bgra64be_full_X_c;
2712 *yuv2packed2 = yuv2bgra64be_full_2_c;
2713 *yuv2packed1 = yuv2bgra64be_full_1_c;
2714 } else
2715 #endif /* CONFIG_SWSCALE_ALPHA */
2716 {
2717 *yuv2packedX = yuv2bgrx64be_full_X_c;
2718 *yuv2packed2 = yuv2bgrx64be_full_2_c;
2719 *yuv2packed1 = yuv2bgrx64be_full_1_c;
2720 }
2721 break;
2722
2723 case AV_PIX_FMT_RGB24:
2724 *yuv2packedX = yuv2rgb24_full_X_c;
2725 *yuv2packed2 = yuv2rgb24_full_2_c;
2726 *yuv2packed1 = yuv2rgb24_full_1_c;
2727 break;
2728 case AV_PIX_FMT_BGR24:
2729 *yuv2packedX = yuv2bgr24_full_X_c;
2730 *yuv2packed2 = yuv2bgr24_full_2_c;
2731 *yuv2packed1 = yuv2bgr24_full_1_c;
2732 break;
2733 case AV_PIX_FMT_RGB48LE:
2734 *yuv2packedX = yuv2rgb48le_full_X_c;
2735 *yuv2packed2 = yuv2rgb48le_full_2_c;
2736 *yuv2packed1 = yuv2rgb48le_full_1_c;
2737 break;
2738 case AV_PIX_FMT_BGR48LE:
2739 *yuv2packedX = yuv2bgr48le_full_X_c;
2740 *yuv2packed2 = yuv2bgr48le_full_2_c;
2741 *yuv2packed1 = yuv2bgr48le_full_1_c;
2742 break;
2743 case AV_PIX_FMT_RGB48BE:
2744 *yuv2packedX = yuv2rgb48be_full_X_c;
2745 *yuv2packed2 = yuv2rgb48be_full_2_c;
2746 *yuv2packed1 = yuv2rgb48be_full_1_c;
2747 break;
2748 case AV_PIX_FMT_BGR48BE:
2749 *yuv2packedX = yuv2bgr48be_full_X_c;
2750 *yuv2packed2 = yuv2bgr48be_full_2_c;
2751 *yuv2packed1 = yuv2bgr48be_full_1_c;
2752 break;
2753 case AV_PIX_FMT_BGR4_BYTE:
2754 *yuv2packedX = yuv2bgr4_byte_full_X_c;
2755 *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2756 *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2757 break;
2758 case AV_PIX_FMT_RGB4_BYTE:
2759 *yuv2packedX = yuv2rgb4_byte_full_X_c;
2760 *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2761 *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2762 break;
2763 case AV_PIX_FMT_BGR8:
2764 *yuv2packedX = yuv2bgr8_full_X_c;
2765 *yuv2packed2 = yuv2bgr8_full_2_c;
2766 *yuv2packed1 = yuv2bgr8_full_1_c;
2767 break;
2768 case AV_PIX_FMT_RGB8:
2769 *yuv2packedX = yuv2rgb8_full_X_c;
2770 *yuv2packed2 = yuv2rgb8_full_2_c;
2771 *yuv2packed1 = yuv2rgb8_full_1_c;
2772 break;
2773 case AV_PIX_FMT_GBRP:
2774 case AV_PIX_FMT_GBRP9BE:
2775 case AV_PIX_FMT_GBRP9LE:
2776 case AV_PIX_FMT_GBRP10BE:
2777 case AV_PIX_FMT_GBRP10LE:
2778 case AV_PIX_FMT_GBRP12BE:
2779 case AV_PIX_FMT_GBRP12LE:
2780 case AV_PIX_FMT_GBRP14BE:
2781 case AV_PIX_FMT_GBRP14LE:
2782 case AV_PIX_FMT_GBRAP:
2783 case AV_PIX_FMT_GBRAP10BE:
2784 case AV_PIX_FMT_GBRAP10LE:
2785 case AV_PIX_FMT_GBRAP12BE:
2786 case AV_PIX_FMT_GBRAP12LE:
2787 *yuv2anyX = yuv2gbrp_full_X_c;
2788 break;
2789 case AV_PIX_FMT_GBRP16BE:
2790 case AV_PIX_FMT_GBRP16LE:
2791 case AV_PIX_FMT_GBRAP16BE:
2792 case AV_PIX_FMT_GBRAP16LE:
2793 *yuv2anyX = yuv2gbrp16_full_X_c;
2794 break;
2795 case AV_PIX_FMT_GBRPF32BE:
2796 case AV_PIX_FMT_GBRPF32LE:
2797 case AV_PIX_FMT_GBRAPF32BE:
2798 case AV_PIX_FMT_GBRAPF32LE:
2799 *yuv2anyX = yuv2gbrpf32_full_X_c;
2800 break;
2801 }
2802 if (!*yuv2packedX && !*yuv2anyX)
2803 goto YUV_PACKED;
2804 } else {
2805 YUV_PACKED:
2806 switch (dstFormat) {
2807 case AV_PIX_FMT_RGBA64LE:
2808 #if CONFIG_SWSCALE_ALPHA
2809 if (c->needAlpha) {
2810 *yuv2packed1 = yuv2rgba64le_1_c;
2811 *yuv2packed2 = yuv2rgba64le_2_c;
2812 *yuv2packedX = yuv2rgba64le_X_c;
2813 } else
2814 #endif /* CONFIG_SWSCALE_ALPHA */
2815 {
2816 *yuv2packed1 = yuv2rgbx64le_1_c;
2817 *yuv2packed2 = yuv2rgbx64le_2_c;
2818 *yuv2packedX = yuv2rgbx64le_X_c;
2819 }
2820 break;
2821 case AV_PIX_FMT_RGBA64BE:
2822 #if CONFIG_SWSCALE_ALPHA
2823 if (c->needAlpha) {
2824 *yuv2packed1 = yuv2rgba64be_1_c;
2825 *yuv2packed2 = yuv2rgba64be_2_c;
2826 *yuv2packedX = yuv2rgba64be_X_c;
2827 } else
2828 #endif /* CONFIG_SWSCALE_ALPHA */
2829 {
2830 *yuv2packed1 = yuv2rgbx64be_1_c;
2831 *yuv2packed2 = yuv2rgbx64be_2_c;
2832 *yuv2packedX = yuv2rgbx64be_X_c;
2833 }
2834 break;
2835 case AV_PIX_FMT_BGRA64LE:
2836 #if CONFIG_SWSCALE_ALPHA
2837 if (c->needAlpha) {
2838 *yuv2packed1 = yuv2bgra64le_1_c;
2839 *yuv2packed2 = yuv2bgra64le_2_c;
2840 *yuv2packedX = yuv2bgra64le_X_c;
2841 } else
2842 #endif /* CONFIG_SWSCALE_ALPHA */
2843 {
2844 *yuv2packed1 = yuv2bgrx64le_1_c;
2845 *yuv2packed2 = yuv2bgrx64le_2_c;
2846 *yuv2packedX = yuv2bgrx64le_X_c;
2847 }
2848 break;
2849 case AV_PIX_FMT_BGRA64BE:
2850 #if CONFIG_SWSCALE_ALPHA
2851 if (c->needAlpha) {
2852 *yuv2packed1 = yuv2bgra64be_1_c;
2853 *yuv2packed2 = yuv2bgra64be_2_c;
2854 *yuv2packedX = yuv2bgra64be_X_c;
2855 } else
2856 #endif /* CONFIG_SWSCALE_ALPHA */
2857 {
2858 *yuv2packed1 = yuv2bgrx64be_1_c;
2859 *yuv2packed2 = yuv2bgrx64be_2_c;
2860 *yuv2packedX = yuv2bgrx64be_X_c;
2861 }
2862 break;
2863 case AV_PIX_FMT_RGB48LE:
2864 *yuv2packed1 = yuv2rgb48le_1_c;
2865 *yuv2packed2 = yuv2rgb48le_2_c;
2866 *yuv2packedX = yuv2rgb48le_X_c;
2867 break;
2868 case AV_PIX_FMT_RGB48BE:
2869 *yuv2packed1 = yuv2rgb48be_1_c;
2870 *yuv2packed2 = yuv2rgb48be_2_c;
2871 *yuv2packedX = yuv2rgb48be_X_c;
2872 break;
2873 case AV_PIX_FMT_BGR48LE:
2874 *yuv2packed1 = yuv2bgr48le_1_c;
2875 *yuv2packed2 = yuv2bgr48le_2_c;
2876 *yuv2packedX = yuv2bgr48le_X_c;
2877 break;
2878 case AV_PIX_FMT_BGR48BE:
2879 *yuv2packed1 = yuv2bgr48be_1_c;
2880 *yuv2packed2 = yuv2bgr48be_2_c;
2881 *yuv2packedX = yuv2bgr48be_X_c;
2882 break;
2883 case AV_PIX_FMT_RGB32:
2884 case AV_PIX_FMT_BGR32:
2885 #if CONFIG_SMALL
2886 *yuv2packed1 = yuv2rgb32_1_c;
2887 *yuv2packed2 = yuv2rgb32_2_c;
2888 *yuv2packedX = yuv2rgb32_X_c;
2889 #else
2890 #if CONFIG_SWSCALE_ALPHA
2891 if (c->needAlpha) {
2892 *yuv2packed1 = yuv2rgba32_1_c;
2893 *yuv2packed2 = yuv2rgba32_2_c;
2894 *yuv2packedX = yuv2rgba32_X_c;
2895 } else
2896 #endif /* CONFIG_SWSCALE_ALPHA */
2897 {
2898 *yuv2packed1 = yuv2rgbx32_1_c;
2899 *yuv2packed2 = yuv2rgbx32_2_c;
2900 *yuv2packedX = yuv2rgbx32_X_c;
2901 }
2902 #endif /* !CONFIG_SMALL */
2903 break;
2904 case AV_PIX_FMT_RGB32_1:
2905 case AV_PIX_FMT_BGR32_1:
2906 #if CONFIG_SMALL
2907 *yuv2packed1 = yuv2rgb32_1_1_c;
2908 *yuv2packed2 = yuv2rgb32_1_2_c;
2909 *yuv2packedX = yuv2rgb32_1_X_c;
2910 #else
2911 #if CONFIG_SWSCALE_ALPHA
2912 if (c->needAlpha) {
2913 *yuv2packed1 = yuv2rgba32_1_1_c;
2914 *yuv2packed2 = yuv2rgba32_1_2_c;
2915 *yuv2packedX = yuv2rgba32_1_X_c;
2916 } else
2917 #endif /* CONFIG_SWSCALE_ALPHA */
2918 {
2919 *yuv2packed1 = yuv2rgbx32_1_1_c;
2920 *yuv2packed2 = yuv2rgbx32_1_2_c;
2921 *yuv2packedX = yuv2rgbx32_1_X_c;
2922 }
2923 #endif /* !CONFIG_SMALL */
2924 break;
2925 case AV_PIX_FMT_RGB24:
2926 *yuv2packed1 = yuv2rgb24_1_c;
2927 *yuv2packed2 = yuv2rgb24_2_c;
2928 *yuv2packedX = yuv2rgb24_X_c;
2929 break;
2930 case AV_PIX_FMT_BGR24:
2931 *yuv2packed1 = yuv2bgr24_1_c;
2932 *yuv2packed2 = yuv2bgr24_2_c;
2933 *yuv2packedX = yuv2bgr24_X_c;
2934 break;
2935 case AV_PIX_FMT_RGB565LE:
2936 case AV_PIX_FMT_RGB565BE:
2937 case AV_PIX_FMT_BGR565LE:
2938 case AV_PIX_FMT_BGR565BE:
2939 *yuv2packed1 = yuv2rgb16_1_c;
2940 *yuv2packed2 = yuv2rgb16_2_c;
2941 *yuv2packedX = yuv2rgb16_X_c;
2942 break;
2943 case AV_PIX_FMT_RGB555LE:
2944 case AV_PIX_FMT_RGB555BE:
2945 case AV_PIX_FMT_BGR555LE:
2946 case AV_PIX_FMT_BGR555BE:
2947 *yuv2packed1 = yuv2rgb15_1_c;
2948 *yuv2packed2 = yuv2rgb15_2_c;
2949 *yuv2packedX = yuv2rgb15_X_c;
2950 break;
2951 case AV_PIX_FMT_RGB444LE:
2952 case AV_PIX_FMT_RGB444BE:
2953 case AV_PIX_FMT_BGR444LE:
2954 case AV_PIX_FMT_BGR444BE:
2955 *yuv2packed1 = yuv2rgb12_1_c;
2956 *yuv2packed2 = yuv2rgb12_2_c;
2957 *yuv2packedX = yuv2rgb12_X_c;
2958 break;
2959 case AV_PIX_FMT_RGB8:
2960 case AV_PIX_FMT_BGR8:
2961 *yuv2packed1 = yuv2rgb8_1_c;
2962 *yuv2packed2 = yuv2rgb8_2_c;
2963 *yuv2packedX = yuv2rgb8_X_c;
2964 break;
2965 case AV_PIX_FMT_RGB4:
2966 case AV_PIX_FMT_BGR4:
2967 *yuv2packed1 = yuv2rgb4_1_c;
2968 *yuv2packed2 = yuv2rgb4_2_c;
2969 *yuv2packedX = yuv2rgb4_X_c;
2970 break;
2971 case AV_PIX_FMT_RGB4_BYTE:
2972 case AV_PIX_FMT_BGR4_BYTE:
2973 *yuv2packed1 = yuv2rgb4b_1_c;
2974 *yuv2packed2 = yuv2rgb4b_2_c;
2975 *yuv2packedX = yuv2rgb4b_X_c;
2976 break;
2977 }
2978 }
2979 switch (dstFormat) {
2980 case AV_PIX_FMT_MONOWHITE:
2981 *yuv2packed1 = yuv2monowhite_1_c;
2982 *yuv2packed2 = yuv2monowhite_2_c;
2983 *yuv2packedX = yuv2monowhite_X_c;
2984 break;
2985 case AV_PIX_FMT_MONOBLACK:
2986 *yuv2packed1 = yuv2monoblack_1_c;
2987 *yuv2packed2 = yuv2monoblack_2_c;
2988 *yuv2packedX = yuv2monoblack_X_c;
2989 break;
2990 case AV_PIX_FMT_YUYV422:
2991 *yuv2packed1 = yuv2yuyv422_1_c;
2992 *yuv2packed2 = yuv2yuyv422_2_c;
2993 *yuv2packedX = yuv2yuyv422_X_c;
2994 break;
2995 case AV_PIX_FMT_YVYU422:
2996 *yuv2packed1 = yuv2yvyu422_1_c;
2997 *yuv2packed2 = yuv2yvyu422_2_c;
2998 *yuv2packedX = yuv2yvyu422_X_c;
2999 break;
3000 case AV_PIX_FMT_UYVY422:
3001 *yuv2packed1 = yuv2uyvy422_1_c;
3002 *yuv2packed2 = yuv2uyvy422_2_c;
3003 *yuv2packedX = yuv2uyvy422_X_c;
3004 break;
3005 case AV_PIX_FMT_YA8:
3006 *yuv2packed1 = yuv2ya8_1_c;
3007 *yuv2packed2 = yuv2ya8_2_c;
3008 *yuv2packedX = yuv2ya8_X_c;
3009 break;
3010 case AV_PIX_FMT_YA16LE:
3011 *yuv2packed1 = yuv2ya16le_1_c;
3012 *yuv2packed2 = yuv2ya16le_2_c;
3013 *yuv2packedX = yuv2ya16le_X_c;
3014 break;
3015 case AV_PIX_FMT_YA16BE:
3016 *yuv2packed1 = yuv2ya16be_1_c;
3017 *yuv2packed2 = yuv2ya16be_2_c;
3018 *yuv2packedX = yuv2ya16be_X_c;
3019 break;
3020 case AV_PIX_FMT_AYUV64LE:
3021 *yuv2packedX = yuv2ayuv64le_X_c;
3022 break;
3023 }
3024 }
3025