39
34
#include "swscale.h"
40
35
#include "swscale_internal.h"
42
#define DITHER1XBPP // only for mmx
44
const uint8_t __attribute__((aligned(8))) dither_2x2_4[2][8]={
45
{ 1, 3, 1, 3, 1, 3, 1, 3, },
46
{ 2, 0, 2, 0, 2, 0, 2, 0, },
49
const uint8_t __attribute__((aligned(8))) dither_2x2_8[2][8]={
50
{ 6, 2, 6, 2, 6, 2, 6, 2, },
51
{ 0, 4, 0, 4, 0, 4, 0, 4, },
54
const uint8_t __attribute__((aligned(8))) dither_8x8_32[8][8]={
55
{ 17, 9, 23, 15, 16, 8, 22, 14, },
56
{ 5, 29, 3, 27, 4, 28, 2, 26, },
57
{ 21, 13, 19, 11, 20, 12, 18, 10, },
58
{ 0, 24, 6, 30, 1, 25, 7, 31, },
59
{ 16, 8, 22, 14, 17, 9, 23, 15, },
60
{ 4, 28, 2, 26, 5, 29, 3, 27, },
61
{ 20, 12, 18, 10, 21, 13, 19, 11, },
62
{ 1, 25, 7, 31, 0, 24, 6, 30, },
66
const uint8_t __attribute__((aligned(8))) dither_8x8_64[8][8]={
67
{ 0, 48, 12, 60, 3, 51, 15, 63, },
68
{ 32, 16, 44, 28, 35, 19, 47, 31, },
69
{ 8, 56, 4, 52, 11, 59, 7, 55, },
70
{ 40, 24, 36, 20, 43, 27, 39, 23, },
71
{ 2, 50, 14, 62, 1, 49, 13, 61, },
72
{ 34, 18, 46, 30, 33, 17, 45, 29, },
73
{ 10, 58, 6, 54, 9, 57, 5, 53, },
74
{ 42, 26, 38, 22, 41, 25, 37, 21, },
78
const uint8_t __attribute__((aligned(8))) dither_8x8_73[8][8]={
79
{ 0, 55, 14, 68, 3, 58, 17, 72, },
80
{ 37, 18, 50, 32, 40, 22, 54, 35, },
81
{ 9, 64, 5, 59, 13, 67, 8, 63, },
82
{ 46, 27, 41, 23, 49, 31, 44, 26, },
83
{ 2, 57, 16, 71, 1, 56, 15, 70, },
84
{ 39, 21, 52, 34, 38, 19, 51, 33, },
85
{ 11, 66, 7, 62, 10, 65, 6, 60, },
86
{ 48, 30, 43, 25, 47, 29, 42, 24, },
90
const uint8_t __attribute__((aligned(8))) dither_8x8_128[8][8]={
91
{ 68, 36, 92, 60, 66, 34, 90, 58, },
92
{ 20, 116, 12, 108, 18, 114, 10, 106, },
93
{ 84, 52, 76, 44, 82, 50, 74, 42, },
94
{ 0, 96, 24, 120, 6, 102, 30, 126, },
95
{ 64, 32, 88, 56, 70, 38, 94, 62, },
96
{ 16, 112, 8, 104, 22, 118, 14, 110, },
97
{ 80, 48, 72, 40, 86, 54, 78, 46, },
98
{ 4, 100, 28, 124, 2, 98, 26, 122, },
103
const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
104
{117, 62, 158, 103, 113, 58, 155, 100, },
105
{ 34, 199, 21, 186, 31, 196, 17, 182, },
106
{144, 89, 131, 76, 141, 86, 127, 72, },
107
{ 0, 165, 41, 206, 10, 175, 52, 217, },
108
{110, 55, 151, 96, 120, 65, 162, 107, },
109
{ 28, 193, 14, 179, 38, 203, 24, 189, },
110
{138, 83, 124, 69, 148, 93, 134, 79, },
111
{ 7, 172, 48, 213, 3, 168, 45, 210, },
114
// tries to correct a gamma of 1.5
115
const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
116
{ 0, 143, 18, 200, 2, 156, 25, 215, },
117
{ 78, 28, 125, 64, 89, 36, 138, 74, },
118
{ 10, 180, 3, 161, 16, 195, 8, 175, },
119
{109, 51, 93, 38, 121, 60, 105, 47, },
120
{ 1, 152, 23, 210, 0, 147, 20, 205, },
121
{ 85, 33, 134, 71, 81, 30, 130, 67, },
122
{ 14, 190, 6, 171, 12, 185, 5, 166, },
123
{117, 57, 101, 44, 113, 54, 97, 41, },
126
// tries to correct a gamma of 2.0
127
const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
128
{ 0, 124, 8, 193, 0, 140, 12, 213, },
129
{ 55, 14, 104, 42, 66, 19, 119, 52, },
130
{ 3, 168, 1, 145, 6, 187, 3, 162, },
131
{ 86, 31, 70, 21, 99, 39, 82, 28, },
132
{ 0, 134, 11, 206, 0, 129, 9, 200, },
133
{ 62, 17, 114, 48, 58, 16, 109, 45, },
134
{ 5, 181, 2, 157, 4, 175, 1, 151, },
135
{ 95, 36, 78, 26, 90, 34, 74, 24, },
138
// tries to correct a gamma of 2.5
139
const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
140
{ 0, 107, 3, 187, 0, 125, 6, 212, },
141
{ 39, 7, 86, 28, 49, 11, 102, 36, },
142
{ 1, 158, 0, 131, 3, 180, 1, 151, },
143
{ 68, 19, 52, 12, 81, 25, 64, 17, },
144
{ 0, 119, 5, 203, 0, 113, 4, 195, },
145
{ 45, 9, 96, 33, 42, 8, 91, 30, },
146
{ 2, 172, 1, 144, 2, 165, 0, 137, },
147
{ 77, 23, 60, 15, 72, 21, 56, 14, },
37
#define DITHER1XBPP // only for MMX
39
extern const uint8_t dither_8x8_32[8][8];
40
extern const uint8_t dither_8x8_73[8][8];
41
extern const uint8_t dither_8x8_220[8][8];
43
#if HAVE_MMX && CONFIG_GPL
153
45
/* hope these constant values are cache line aligned */
154
46
DECLARE_ASM_CONST(8, uint64_t, mmx_00ffw) = 0x00ff00ff00ff00ffULL;
155
47
DECLARE_ASM_CONST(8, uint64_t, mmx_redmask) = 0xf8f8f8f8f8f8f8f8ULL;
156
48
DECLARE_ASM_CONST(8, uint64_t, mmx_grnmask) = 0xfcfcfcfcfcfcfcfcULL;
158
// the volatile is required because gcc otherwise optimizes some writes away not knowing that these
159
// are read in the asm block
160
static volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither;
161
static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither;
162
static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither;
163
static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither;
55
#define HAVE_AMD3DNOW 0
172
56
#define RENAME(a) a ## _MMX
173
57
#include "yuv2rgb_template.c"
180
63
#define RENAME(a) a ## _MMX2
181
64
#include "yuv2rgb_template.c"
183
#endif /* HAVE_MMX */
66
#endif /* HAVE_MMX && CONFIG_GPL */
185
const int32_t Inverse_Table_6_9[8][4] = {
68
const int32_t ff_yuv2rgb_coeffs[8][4] = {
186
69
{117504, 138453, 13954, 34903}, /* no sequence_display_extension */
187
70
{117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */
188
71
{104597, 132201, 25675, 53279}, /* unspecified */
270
135
if (c->dstW & 4) {\
271
136
int av_unused Y, U, V;\
138
#define ENDYUV2RGBFUNC()\
276
141
return srcSliceH;\
279
#define EPILOG(dst_delta)\
283
PROLOG(yuv2rgb_c_32, uint32_t)
309
PROLOG(yuv2rgb_c_24_rgb, uint8_t)
144
#define CLOSEYUV2RGBFUNC(dst_delta)\
145
ENDYUV2RGBLINE(dst_delta)\
148
YUV2RGBFUNC(yuv2rgb_c_32, uint32_t)
150
PUTRGB(dst_1,py_1,0,0);
151
PUTRGB(dst_2,py_2,0,1);
154
PUTRGB(dst_2,py_2,1,1);
155
PUTRGB(dst_1,py_1,1,0);
157
PUTRGB(dst_2,py_2,1,1);
158
PUTRGB(dst_1,py_1,1,0);
161
PUTRGB(dst_1,py_1,2,0);
162
PUTRGB(dst_2,py_2,2,1);
165
PUTRGB(dst_2,py_2,3,1);
166
PUTRGB(dst_1,py_1,3,0);
169
PUTRGB(dst_1,py_1,0,0);
170
PUTRGB(dst_2,py_2,0,1);
173
PUTRGB(dst_2,py_2,1,1);
174
PUTRGB(dst_1,py_1,1,0);
177
YUV2RGBFUNC(yuv2rgb_c_24_rgb, uint8_t)
179
PUTRGB24(dst_1,py_1,0);
180
PUTRGB24(dst_2,py_2,0);
183
PUTRGB24(dst_2,py_2,1);
184
PUTRGB24(dst_1,py_1,1);
187
PUTRGB24(dst_1,py_1,2);
188
PUTRGB24(dst_2,py_2,2);
191
PUTRGB24(dst_2,py_2,3);
192
PUTRGB24(dst_1,py_1,3);
195
PUTRGB24(dst_1,py_1,0);
196
PUTRGB24(dst_2,py_2,0);
199
PUTRGB24(dst_2,py_2,1);
200
PUTRGB24(dst_1,py_1,1);
335
203
// only trivial mods from yuv2rgb_c_24_rgb
336
PROLOG(yuv2rgb_c_24_bgr, uint8_t)
362
// This is exactly the same code as yuv2rgb_c_32 except for the types of
363
// r, g, b, dst_1, dst_2
364
PROLOG(yuv2rgb_c_16, uint16_t)
382
// This is exactly the same code as yuv2rgb_c_32 except for the types of
383
// r, g, b, dst_1, dst_2
384
PROLOG(yuv2rgb_c_8, uint8_t)
402
// r, g, b, dst_1, dst_2
403
PROLOG(yuv2rgb_c_8_ordered_dither, uint8_t)
404
const uint8_t *d32= dither_8x8_32[y&7];
405
const uint8_t *d64= dither_8x8_73[y&7];
406
#define DST1bpp8(i,o) \
408
dst_1[2*i] = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]]; \
410
dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
412
#define DST2bpp8(i,o) \
414
dst_2[2*i] = r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]]; \
416
dst_2[2*i+1] = r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
437
// This is exactly the same code as yuv2rgb_c_32 except for the types of
438
// r, g, b, dst_1, dst_2
439
PROLOG(yuv2rgb_c_4, uint8_t)
443
acc = r[Y] + g[Y] + b[Y]; \
445
acc |= (r[Y] + g[Y] + b[Y])<<4; \
450
acc = r[Y] + g[Y] + b[Y]; \
452
acc |= (r[Y] + g[Y] + b[Y])<<4; \
472
PROLOG(yuv2rgb_c_4_ordered_dither, uint8_t)
473
const uint8_t *d64= dither_8x8_73[y&7];
474
const uint8_t *d128=dither_8x8_220[y&7];
477
#define DST1bpp4(i,o) \
204
YUV2RGBFUNC(yuv2rgb_c_24_bgr, uint8_t)
206
PUTBGR24(dst_1,py_1,0);
207
PUTBGR24(dst_2,py_2,0);
210
PUTBGR24(dst_2,py_2,1);
211
PUTBGR24(dst_1,py_1,1);
214
PUTBGR24(dst_1,py_1,2);
215
PUTBGR24(dst_2,py_2,2);
218
PUTBGR24(dst_2,py_2,3);
219
PUTBGR24(dst_1,py_1,3);
222
PUTBGR24(dst_1,py_1,0);
223
PUTBGR24(dst_2,py_2,0);
226
PUTBGR24(dst_2,py_2,1);
227
PUTBGR24(dst_1,py_1,1);
230
// This is exactly the same code as yuv2rgb_c_32 except for the types of
231
// r, g, b, dst_1, dst_2
232
YUV2RGBFUNC(yuv2rgb_c_16, uint16_t)
234
PUTRGB(dst_1,py_1,0,0);
235
PUTRGB(dst_2,py_2,0,1);
238
PUTRGB(dst_2,py_2,1,1);
239
PUTRGB(dst_1,py_1,1,0);
242
PUTRGB(dst_1,py_1,2,0);
243
PUTRGB(dst_2,py_2,2,1);
246
PUTRGB(dst_2,py_2,3,1);
247
PUTRGB(dst_1,py_1,3,0);
250
// This is exactly the same code as yuv2rgb_c_32 except for the types of
251
// r, g, b, dst_1, dst_2
252
YUV2RGBFUNC(yuv2rgb_c_8, uint8_t)
254
PUTRGB(dst_1,py_1,0,0);
255
PUTRGB(dst_2,py_2,0,1);
258
PUTRGB(dst_2,py_2,1,1);
259
PUTRGB(dst_1,py_1,1,0);
262
PUTRGB(dst_1,py_1,2,0);
263
PUTRGB(dst_2,py_2,2,1);
266
PUTRGB(dst_2,py_2,3,1);
267
PUTRGB(dst_1,py_1,3,0);
270
// r, g, b, dst_1, dst_2
271
YUV2RGBFUNC(yuv2rgb_c_8_ordered_dither, uint8_t)
272
const uint8_t *d32 = dither_8x8_32[y&7];
273
const uint8_t *d64 = dither_8x8_73[y&7];
274
#define PUTRGB8(dst,src,i,o) \
276
dst[2*i] = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]]; \
278
dst[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
281
PUTRGB8(dst_1,py_1,0,0);
282
PUTRGB8(dst_2,py_2,0,0+8);
285
PUTRGB8(dst_2,py_2,1,2+8);
286
PUTRGB8(dst_1,py_1,1,2);
289
PUTRGB8(dst_1,py_1,2,4);
290
PUTRGB8(dst_2,py_2,2,4+8);
293
PUTRGB8(dst_2,py_2,3,6+8);
294
PUTRGB8(dst_1,py_1,3,6);
298
// This is exactly the same code as yuv2rgb_c_32 except for the types of
299
// r, g, b, dst_1, dst_2
300
YUV2RGBFUNC(yuv2rgb_c_4, uint8_t)
302
#define PUTRGB4(dst,src,i) \
304
acc = r[Y] + g[Y] + b[Y]; \
306
acc |= (r[Y] + g[Y] + b[Y])<<4; \
310
PUTRGB4(dst_1,py_1,0);
311
PUTRGB4(dst_2,py_2,0);
314
PUTRGB4(dst_2,py_2,1);
315
PUTRGB4(dst_1,py_1,1);
318
PUTRGB4(dst_1,py_1,2);
319
PUTRGB4(dst_2,py_2,2);
322
PUTRGB4(dst_2,py_2,3);
323
PUTRGB4(dst_1,py_1,3);
326
YUV2RGBFUNC(yuv2rgb_c_4_ordered_dither, uint8_t)
327
const uint8_t *d64 = dither_8x8_73[y&7];
328
const uint8_t *d128 = dither_8x8_220[y&7];
331
#define PUTRGB4D(dst,src,i,o) \
479
333
acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]]; \
481
335
acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4; \
484
#define DST2bpp4(i,o) \
486
acc = r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]]; \
488
acc |= (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4; \
339
PUTRGB4D(dst_1,py_1,0,0);
340
PUTRGB4D(dst_2,py_2,0,0+8);
343
PUTRGB4D(dst_2,py_2,1,2+8);
344
PUTRGB4D(dst_1,py_1,1,2);
347
PUTRGB4D(dst_1,py_1,2,4);
348
PUTRGB4D(dst_2,py_2,2,4+8);
351
PUTRGB4D(dst_2,py_2,3,6+8);
352
PUTRGB4D(dst_1,py_1,3,6);
509
355
// This is exactly the same code as yuv2rgb_c_32 except for the types of
510
356
// r, g, b, dst_1, dst_2
511
PROLOG(yuv2rgb_c_4b, uint8_t)
529
PROLOG(yuv2rgb_c_4b_ordered_dither, uint8_t)
530
const uint8_t *d64= dither_8x8_73[y&7];
531
const uint8_t *d128=dither_8x8_220[y&7];
533
#define DST1bpp4b(i,o) \
535
dst_1[2*i] = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]]; \
537
dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
539
#define DST2bpp4b(i,o) \
541
dst_2[2*i] = r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]]; \
543
dst_2[2*i+1] = r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
563
PROLOG(yuv2rgb_c_1_ordered_dither, uint8_t)
564
const uint8_t *d128=dither_8x8_220[y&7];
565
char out_1=0, out_2=0;
357
YUV2RGBFUNC(yuv2rgb_c_4b, uint8_t)
359
PUTRGB(dst_1,py_1,0,0);
360
PUTRGB(dst_2,py_2,0,1);
363
PUTRGB(dst_2,py_2,1,1);
364
PUTRGB(dst_1,py_1,1,0);
367
PUTRGB(dst_1,py_1,2,0);
368
PUTRGB(dst_2,py_2,2,1);
371
PUTRGB(dst_2,py_2,3,1);
372
PUTRGB(dst_1,py_1,3,0);
375
YUV2RGBFUNC(yuv2rgb_c_4b_ordered_dither, uint8_t)
376
const uint8_t *d64 = dither_8x8_73[y&7];
377
const uint8_t *d128 = dither_8x8_220[y&7];
379
#define PUTRGB4DB(dst,src,i,o) \
381
dst[2*i] = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]]; \
383
dst[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
386
PUTRGB4DB(dst_1,py_1,0,0);
387
PUTRGB4DB(dst_2,py_2,0,0+8);
390
PUTRGB4DB(dst_2,py_2,1,2+8);
391
PUTRGB4DB(dst_1,py_1,1,2);
394
PUTRGB4DB(dst_1,py_1,2,4);
395
PUTRGB4DB(dst_2,py_2,2,4+8);
398
PUTRGB4DB(dst_2,py_2,3,6+8);
399
PUTRGB4DB(dst_1,py_1,3,6);
402
YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t)
403
const uint8_t *d128 = dither_8x8_220[y&7];
404
char out_1 = 0, out_2 = 0;
566
405
g= c->table_gU[128] + c->table_gV[128];
568
#define DST1bpp1(i,o) \
570
out_1+= out_1 + g[Y+d128[0+o]]; \
572
out_1+= out_1 + g[Y+d128[1+o]];
574
#define DST2bpp1(i,o) \
576
out_2+= out_2 + g[Y+d128[8+o]]; \
578
out_2+= out_2 + g[Y+d128[9+o]];
407
#define PUTRGB1(out,src,i,o) \
409
out+= out + g[Y+d128[0+o]]; \
411
out+= out + g[Y+d128[1+o]];
413
PUTRGB1(out_1,py_1,0,0);
414
PUTRGB1(out_2,py_2,0,0+8);
416
PUTRGB1(out_2,py_2,1,2+8);
417
PUTRGB1(out_1,py_1,1,2);
419
PUTRGB1(out_1,py_1,2,4);
420
PUTRGB1(out_2,py_2,2,4+8);
422
PUTRGB1(out_2,py_2,3,6+8);
423
PUTRGB1(out_1,py_1,3,6);
596
SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
429
SwsFunc sws_yuv2rgb_get_func_ptr(SwsContext *c)
598
#if defined(HAVE_MMX2) || defined(HAVE_MMX)
599
if (c->flags & SWS_CPU_CAPS_MMX2){
600
switch(c->dstFormat){
432
#if (HAVE_MMX2 || HAVE_MMX) && CONFIG_GPL
433
if (c->flags & SWS_CPU_CAPS_MMX2) {
434
switch (c->dstFormat) {
601
435
case PIX_FMT_RGB32: return yuv420_rgb32_MMX2;
602
436
case PIX_FMT_BGR24: return yuv420_rgb24_MMX2;
603
case PIX_FMT_BGR565: return yuv420_rgb16_MMX2;
604
case PIX_FMT_BGR555: return yuv420_rgb15_MMX2;
437
case PIX_FMT_RGB565: return yuv420_rgb16_MMX2;
438
case PIX_FMT_RGB555: return yuv420_rgb15_MMX2;
607
if (c->flags & SWS_CPU_CAPS_MMX){
608
switch(c->dstFormat){
441
if (c->flags & SWS_CPU_CAPS_MMX) {
442
switch (c->dstFormat) {
609
443
case PIX_FMT_RGB32: return yuv420_rgb32_MMX;
610
444
case PIX_FMT_BGR24: return yuv420_rgb24_MMX;
611
case PIX_FMT_BGR565: return yuv420_rgb16_MMX;
612
case PIX_FMT_BGR555: return yuv420_rgb15_MMX;
445
case PIX_FMT_RGB565: return yuv420_rgb16_MMX;
446
case PIX_FMT_RGB555: return yuv420_rgb15_MMX;
618
SwsFunc t= yuv2rgb_init_vis(c);
624
SwsFunc t= yuv2rgb_init_mlib(c);
451
t = sws_yuv2rgb_init_vis(c);
454
t = sws_yuv2rgb_init_mlib(c);
629
457
if (c->flags & SWS_CPU_CAPS_ALTIVEC)
631
SwsFunc t = yuv2rgb_init_altivec(c);
458
t = sws_yuv2rgb_init_altivec(c);
637
462
if (c->flags & SWS_CPU_CAPS_BFIN)
639
SwsFunc t = ff_bfin_yuv2rgb_get_func_ptr (c);
463
t = sws_ff_bfin_yuv2rgb_get_func_ptr(c);
644
av_log(c, AV_LOG_WARNING, "No accelerated colorspace conversion found\n");
646
switch(c->dstFormat){
469
av_log(c, AV_LOG_WARNING, "No accelerated colorspace conversion found.\n");
471
switch (c->dstFormat) {
472
case PIX_FMT_BGR32_1:
473
case PIX_FMT_RGB32_1:
647
474
case PIX_FMT_BGR32:
648
case PIX_FMT_RGB32: return yuv2rgb_c_32;
649
case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
650
case PIX_FMT_BGR24: return yuv2rgb_c_24_bgr;
475
case PIX_FMT_RGB32: return yuv2rgb_c_32;
476
case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
477
case PIX_FMT_BGR24: return yuv2rgb_c_24_bgr;
651
478
case PIX_FMT_RGB565:
652
479
case PIX_FMT_BGR565:
653
480
case PIX_FMT_RGB555:
654
case PIX_FMT_BGR555: return yuv2rgb_c_16;
481
case PIX_FMT_BGR555: return yuv2rgb_c_16;
655
482
case PIX_FMT_RGB8:
656
case PIX_FMT_BGR8: return yuv2rgb_c_8_ordered_dither;
483
case PIX_FMT_BGR8: return yuv2rgb_c_8_ordered_dither;
657
484
case PIX_FMT_RGB4:
658
case PIX_FMT_BGR4: return yuv2rgb_c_4_ordered_dither;
485
case PIX_FMT_BGR4: return yuv2rgb_c_4_ordered_dither;
659
486
case PIX_FMT_RGB4_BYTE:
660
487
case PIX_FMT_BGR4_BYTE: return yuv2rgb_c_4b_ordered_dither;
661
488
case PIX_FMT_MONOBLACK: return yuv2rgb_c_1_ordered_dither;
696
543
int64_t cy = 1<<16;
699
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
704
crv= (crv*224) / 255;
705
cbu= (cbu*224) / 255;
706
cgu= (cgu*224) / 255;
707
cgv= (cgv*224) / 255;
552
crv = (crv*224) / 255;
553
cbu = (cbu*224) / 255;
554
cgu = (cgu*224) / 255;
555
cgv = (cgv*224) / 255;
710
cy = (cy *contrast )>>16;
711
crv= (crv*contrast * saturation)>>32;
712
cbu= (cbu*contrast * saturation)>>32;
713
cgu= (cgu*contrast * saturation)>>32;
714
cgv= (cgv*contrast * saturation)>>32;
715
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
558
cy = (cy *contrast ) >> 16;
559
crv = (crv*contrast * saturation) >> 32;
560
cbu = (cbu*contrast * saturation) >> 32;
561
cgu = (cgu*contrast * saturation) >> 32;
562
cgv = (cgv*contrast * saturation) >> 32;
716
563
oy -= 256*brightness;
718
for (i = 0; i < 1024; i++) {
565
//scale coefficients by cy
566
crv = ((crv << 16) + 0x8000) / cy;
567
cbu = ((cbu << 16) + 0x8000) / cy;
568
cgu = ((cgu << 16) + 0x8000) / cy;
569
cgv = ((cgv << 16) + 0x8000) / cy;
721
j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
722
j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
571
av_free(c->yuvTable);
728
table_start= table_32 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
730
entry_size = sizeof (uint32_t);
731
table_r = table_32 + 197;
732
table_b = table_32 + 197 + 685;
733
table_g = table_32 + 197 + 2*682;
735
for (i = -197; i < 256+197; i++)
736
((uint32_t *)table_r)[i] = table_Y[i+384] << (isRgb ? 16 : 0);
737
for (i = -132; i < 256+132; i++)
738
((uint32_t *)table_g)[i] = table_Y[i+384] << 8;
739
for (i = -232; i < 256+232; i++)
740
((uint32_t *)table_b)[i] = table_Y[i+384] << (isRgb ? 0 : 16);
744
table_start= table_8 = av_malloc ((256 + 2*232) * sizeof (uint8_t));
746
entry_size = sizeof (uint8_t);
747
table_r = table_g = table_b = table_8 + 232;
749
for (i = -232; i < 256+232; i++)
750
((uint8_t * )table_b)[i] = table_Y[i+384];
575
c->yuvTable = av_malloc(1024);
576
y_table = c->yuvTable;
577
yb = -(384<<16) - oy;
578
for (i = 0; i < 1024-110; i++) {
579
y_table[i+110] = av_clip_uint8((yb + 0x8000) >> 16) >> 7;
582
fill_table(c->table_gU, 1, cgu, y_table + yoffs);
583
fill_gv_table(c->table_gV, 1, cgv);
587
rbase = isRgb ? 3 : 0;
589
bbase = isRgb ? 0 : 3;
590
c->yuvTable = av_malloc(1024*3);
591
y_table = c->yuvTable;
592
yb = -(384<<16) - oy;
593
for (i = 0; i < 1024-110; i++) {
594
int yval = av_clip_uint8((yb + 0x8000) >> 16);
595
y_table[i+110 ] = (yval >> 7) << rbase;
596
y_table[i+ 37+1024] = ((yval + 43) / 85) << gbase;
597
y_table[i+110+2048] = (yval >> 7) << bbase;
600
fill_table(c->table_rV, 1, crv, y_table + yoffs);
601
fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024);
602
fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048);
603
fill_gv_table(c->table_gV, 1, cgv);
606
rbase = isRgb ? 5 : 0;
607
gbase = isRgb ? 2 : 3;
608
bbase = isRgb ? 0 : 6;
609
c->yuvTable = av_malloc(1024*3);
610
y_table = c->yuvTable;
611
yb = -(384<<16) - oy;
612
for (i = 0; i < 1024-38; i++) {
613
int yval = av_clip_uint8((yb + 0x8000) >> 16);
614
y_table[i+16 ] = ((yval + 18) / 36) << rbase;
615
y_table[i+16+1024] = ((yval + 18) / 36) << gbase;
616
y_table[i+37+2048] = ((yval + 43) / 85) << bbase;
619
fill_table(c->table_rV, 1, crv, y_table + yoffs);
620
fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024);
621
fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048);
622
fill_gv_table(c->table_gV, 1, cgv);
755
table_start= table_16 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
757
entry_size = sizeof (uint16_t);
758
table_r = table_16 + 197;
759
table_b = table_16 + 197 + 685;
760
table_g = table_16 + 197 + 2*682;
762
for (i = -197; i < 256+197; i++) {
763
int j = table_Y[i+384] >> 3;
766
j <<= ((bpp==16) ? 11 : 10);
768
((uint16_t *)table_r)[i] = j;
770
for (i = -132; i < 256+132; i++) {
771
int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
773
((uint16_t *)table_g)[i] = j << 5;
775
for (i = -232; i < 256+232; i++) {
776
int j = table_Y[i+384] >> 3;
779
j <<= ((bpp==16) ? 11 : 10);
781
((uint16_t *)table_b)[i] = j;
786
table_start= table_332 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
788
entry_size = sizeof (uint8_t);
789
table_r = table_332 + 197;
790
table_b = table_332 + 197 + 685;
791
table_g = table_332 + 197 + 2*682;
793
for (i = -197; i < 256+197; i++) {
794
int j = (table_Y[i+384 - 16] + 18)/36;
799
((uint8_t *)table_r)[i] = j;
801
for (i = -132; i < 256+132; i++) {
802
int j = (table_Y[i+384 - 16] + 18)/36;
807
((uint8_t *)table_g)[i] = j << 2;
809
for (i = -232; i < 256+232; i++) {
810
int j = (table_Y[i+384 - 37] + 43)/85;
815
((uint8_t *)table_b)[i] = j;
820
table_start= table_121 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
822
entry_size = sizeof (uint8_t);
823
table_r = table_121 + 197;
824
table_b = table_121 + 197 + 685;
825
table_g = table_121 + 197 + 2*682;
827
for (i = -197; i < 256+197; i++) {
828
int j = table_Y[i+384 - 110] >> 7;
833
((uint8_t *)table_r)[i] = j;
835
for (i = -132; i < 256+132; i++) {
836
int j = (table_Y[i+384 - 37]+ 43)/85;
838
((uint8_t *)table_g)[i] = j << 1;
840
for (i = -232; i < 256+232; i++) {
841
int j =table_Y[i+384 - 110] >> 7;
846
((uint8_t *)table_b)[i] = j;
851
table_start= table_1 = av_malloc (256*2 * sizeof (uint8_t));
853
entry_size = sizeof (uint8_t);
855
table_r = table_b = NULL;
857
for (i = 0; i < 256+256; i++) {
858
int j = table_Y[i + 384 - 110]>>7;
860
((uint8_t *)table_g)[i] = j;
626
rbase = isRgb ? bpp - 5 : 0;
628
bbase = isRgb ? 0 : (bpp - 5);
629
c->yuvTable = av_malloc(1024*3*2);
630
y_table16 = c->yuvTable;
631
yb = -(384<<16) - oy;
632
for (i = 0; i < 1024; i++) {
633
uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
634
y_table16[i ] = (yval >> 3) << rbase;
635
y_table16[i+1024] = (yval >> (18 - bpp)) << gbase;
636
y_table16[i+2048] = (yval >> 3) << bbase;
639
fill_table(c->table_rV, 2, crv, y_table16 + yoffs);
640
fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + 1024);
641
fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2048);
642
fill_gv_table(c->table_gV, 2, cgv);
645
c->yuvTable = av_malloc(1024);
646
y_table = c->yuvTable;
647
yb = -(384<<16) - oy;
648
for (i = 0; i < 1024; i++) {
649
y_table[i] = av_clip_uint8((yb + 0x8000) >> 16);
652
fill_table(c->table_rV, 1, crv, y_table + yoffs);
653
fill_table(c->table_gU, 1, cgu, y_table + yoffs);
654
fill_table(c->table_bU, 1, cbu, y_table + yoffs);
655
fill_gv_table(c->table_gV, 1, cgv);
658
base = (c->dstFormat == PIX_FMT_RGB32_1 || c->dstFormat == PIX_FMT_BGR32_1) ? 8 : 0;
659
rbase = base + (isRgb ? 16 : 0);
661
bbase = base + (isRgb ? 0 : 16);
662
abase = (base + 24) & 31;
663
c->yuvTable = av_malloc(1024*3*4);
664
y_table32 = c->yuvTable;
665
yb = -(384<<16) - oy;
666
for (i = 0; i < 1024; i++) {
667
uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
668
y_table32[i ] = (yval << rbase) + (255 << abase);
669
y_table32[i+1024] = yval << gbase;
670
y_table32[i+2048] = yval << bbase;
673
fill_table(c->table_rV, 4, crv, y_table32 + yoffs);
674
fill_table(c->table_gU, 4, cgu, y_table32 + yoffs + 1024);
675
fill_table(c->table_bU, 4, cbu, y_table32 + yoffs + 2048);
676
fill_gv_table(c->table_gV, 4, cgv);
866
680
av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);
871
for (i = 0; i < 256; i++) {
872
c->table_rV[i] = (uint8_t *)table_r + entry_size * div_round (crv * (i-128), 76309);
873
c->table_gU[i] = (uint8_t *)table_g + entry_size * div_round (cgu * (i-128), 76309);
874
c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
875
c->table_bU[i] = (uint8_t *)table_b + entry_size * div_round (cbu * (i-128), 76309);
878
av_free(c->yuvTable);
879
c->yuvTable= table_start;