Statistics
| Branch: | Revision:

ffmpeg / postproc / yuv2rgb.c @ 6d606c4f

History | View | Annotate | Download (20.8 KB)

1
/*
2
 * yuv2rgb.c, Software YUV to RGB coverter
3
 *
4
 *  Copyright (C) 1999, Aaron Holtzman <aholtzma@ess.engr.uvic.ca>
5
 *  All Rights Reserved.
6
 *
7
 *  Functions broken out from display_x11.c and several new modes
8
 *  added by H?kan Hjort <d95hjort@dtek.chalmers.se>
9
 *
10
 *  15 & 16 bpp support by Franck Sicard <Franck.Sicard@solsoft.fr>
11
 *
12
 *  This file is part of mpeg2dec, a free MPEG-2 video decoder
13
 *
14
 *  mpeg2dec is free software; you can redistribute it and/or modify
15
 *  it under the terms of the GNU General Public License as published by
16
 *  the Free Software Foundation; either version 2, or (at your option)
17
 *  any later version.
18
 *
19
 *  mpeg2dec is distributed in the hope that it will be useful,
20
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
21
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
22
 *  GNU General Public License for more details.
23
 *
24
 *  You should have received a copy of the GNU General Public License
25
 *  along with GNU Make; see the file COPYING.  If not, write to
26
 *  the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
27
 *
28
 * MMX/MMX2 Template stuff from Michael Niedermayer (michaelni@gmx.at) (needed for fast movntq support)
29
 * 1,4,8bpp support by Michael Niedermayer (michaelni@gmx.at)
30
 * context / deglobalize stuff by Michael Niedermayer
31
 */
32

    
33
#include <stdio.h>
34
#include <stdlib.h>
35
#include <inttypes.h>
36
#include <assert.h>
37

    
38
#include "config.h"
39
//#include "video_out.h"
40
#include "rgb2rgb.h"
41
#include "swscale.h"
42
#include "swscale_internal.h"
43
#include "../mangle.h"
44
#include "../libvo/img_format.h" //FIXME try to reduce dependency of such stuff
45

    
46
#ifdef HAVE_MLIB
47
#include "yuv2rgb_mlib.c"
48
#endif
49

    
50
#define DITHER1XBPP // only for mmx
51

    
52
const uint8_t  __attribute__((aligned(8))) dither_2x2_4[2][8]={
53
{  1,   3,   1,   3,   1,   3,   1,   3, },
54
{  2,   0,   2,   0,   2,   0,   2,   0, },
55
};
56

    
57
const uint8_t  __attribute__((aligned(8))) dither_2x2_8[2][8]={
58
{  6,   2,   6,   2,   6,   2,   6,   2, },
59
{  0,   4,   0,   4,   0,   4,   0,   4, },
60
};
61

    
62
const uint8_t  __attribute__((aligned(8))) dither_8x8_32[8][8]={
63
{ 17,   9,  23,  15,  16,   8,  22,  14, },
64
{  5,  29,   3,  27,   4,  28,   2,  26, },
65
{ 21,  13,  19,  11,  20,  12,  18,  10, },
66
{  0,  24,   6,  30,   1,  25,   7,  31, },
67
{ 16,   8,  22,  14,  17,   9,  23,  15, },
68
{  4,  28,   2,  26,   5,  29,   3,  27, },
69
{ 20,  12,  18,  10,  21,  13,  19,  11, },
70
{  1,  25,   7,  31,   0,  24,   6,  30, },
71
};
72

    
73
#if 0
74
const uint8_t  __attribute__((aligned(8))) dither_8x8_64[8][8]={
75
{  0,  48,  12,  60,   3,  51,  15,  63, },
76
{ 32,  16,  44,  28,  35,  19,  47,  31, },
77
{  8,  56,   4,  52,  11,  59,   7,  55, },
78
{ 40,  24,  36,  20,  43,  27,  39,  23, },
79
{  2,  50,  14,  62,   1,  49,  13,  61, },
80
{ 34,  18,  46,  30,  33,  17,  45,  29, },
81
{ 10,  58,   6,  54,   9,  57,   5,  53, },
82
{ 42,  26,  38,  22,  41,  25,  37,  21, },
83
};
84
#endif
85

    
86
const uint8_t  __attribute__((aligned(8))) dither_8x8_73[8][8]={
87
{  0,  55,  14,  68,   3,  58,  17,  72, },
88
{ 37,  18,  50,  32,  40,  22,  54,  35, },
89
{  9,  64,   5,  59,  13,  67,   8,  63, },
90
{ 46,  27,  41,  23,  49,  31,  44,  26, },
91
{  2,  57,  16,  71,   1,  56,  15,  70, },
92
{ 39,  21,  52,  34,  38,  19,  51,  33, },
93
{ 11,  66,   7,  62,  10,  65,   6,  60, },
94
{ 48,  30,  43,  25,  47,  29,  42,  24, },
95
};
96

    
97
#if 0
98
const uint8_t  __attribute__((aligned(8))) dither_8x8_128[8][8]={
99
{ 68,  36,  92,  60,  66,  34,  90,  58, },
100
{ 20, 116,  12, 108,  18, 114,  10, 106, },
101
{ 84,  52,  76,  44,  82,  50,  74,  42, },
102
{  0,  96,  24, 120,   6, 102,  30, 126, },
103
{ 64,  32,  88,  56,  70,  38,  94,  62, },
104
{ 16, 112,   8, 104,  22, 118,  14, 110, },
105
{ 80,  48,  72,  40,  86,  54,  78,  46, },
106
{  4, 100,  28, 124,   2,  98,  26, 122, },
107
};
108
#endif
109

    
110
#if 1
111
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
112
{117,  62, 158, 103, 113,  58, 155, 100, },
113
{ 34, 199,  21, 186,  31, 196,  17, 182, },
114
{144,  89, 131,  76, 141,  86, 127,  72, },
115
{  0, 165,  41, 206,  10, 175,  52, 217, },
116
{110,  55, 151,  96, 120,  65, 162, 107, },
117
{ 28, 193,  14, 179,  38, 203,  24, 189, },
118
{138,  83, 124,  69, 148,  93, 134,  79, },
119
{  7, 172,  48, 213,   3, 168,  45, 210, },
120
};
121
#elif 1
122
// tries to correct a gamma of 1.5
123
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
124
{  0, 143,  18, 200,   2, 156,  25, 215, },
125
{ 78,  28, 125,  64,  89,  36, 138,  74, },
126
{ 10, 180,   3, 161,  16, 195,   8, 175, },
127
{109,  51,  93,  38, 121,  60, 105,  47, },
128
{  1, 152,  23, 210,   0, 147,  20, 205, },
129
{ 85,  33, 134,  71,  81,  30, 130,  67, },
130
{ 14, 190,   6, 171,  12, 185,   5, 166, },
131
{117,  57, 101,  44, 113,  54,  97,  41, },
132
};
133
#elif 1
134
// tries to correct a gamma of 2.0
135
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
136
{  0, 124,   8, 193,   0, 140,  12, 213, },
137
{ 55,  14, 104,  42,  66,  19, 119,  52, },
138
{  3, 168,   1, 145,   6, 187,   3, 162, },
139
{ 86,  31,  70,  21,  99,  39,  82,  28, },
140
{  0, 134,  11, 206,   0, 129,   9, 200, },
141
{ 62,  17, 114,  48,  58,  16, 109,  45, },
142
{  5, 181,   2, 157,   4, 175,   1, 151, },
143
{ 95,  36,  78,  26,  90,  34,  74,  24, },
144
};
145
#else
146
// tries to correct a gamma of 2.5
147
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
148
{  0, 107,   3, 187,   0, 125,   6, 212, },
149
{ 39,   7,  86,  28,  49,  11, 102,  36, },
150
{  1, 158,   0, 131,   3, 180,   1, 151, },
151
{ 68,  19,  52,  12,  81,  25,  64,  17, },
152
{  0, 119,   5, 203,   0, 113,   4, 195, },
153
{ 45,   9,  96,  33,  42,   8,  91,  30, },
154
{  2, 172,   1, 144,   2, 165,   0, 137, },
155
{ 77,  23,  60,  15,  72,  21,  56,  14, },
156
};
157
#endif
158

    
159
#if defined(ARCH_X86) || defined(ARCH_X86_64)
160

    
161
/* hope these constant values are cache line aligned */
162
uint64_t attribute_used __attribute__((aligned(8))) mmx_00ffw = 0x00ff00ff00ff00ffULL;
163
uint64_t attribute_used __attribute__((aligned(8))) mmx_redmask = 0xf8f8f8f8f8f8f8f8ULL;
164
uint64_t attribute_used __attribute__((aligned(8))) mmx_grnmask = 0xfcfcfcfcfcfcfcfcULL;
165

    
166
uint64_t attribute_used __attribute__((aligned(8))) M24A=   0x00FF0000FF0000FFULL;
167
uint64_t attribute_used __attribute__((aligned(8))) M24B=   0xFF0000FF0000FF00ULL;
168
uint64_t attribute_used __attribute__((aligned(8))) M24C=   0x0000FF0000FF0000ULL;
169

    
170
// the volatile is required because gcc otherwise optimizes some writes away not knowing that these
171
// are read in the asm block
172
volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither;
173
volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither;
174
volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither;
175
volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither;
176

    
177
uint64_t __attribute__((aligned(8))) dither4[2]={
178
        0x0103010301030103LL,
179
        0x0200020002000200LL,};
180

    
181
uint64_t __attribute__((aligned(8))) dither8[2]={
182
        0x0602060206020602LL,
183
        0x0004000400040004LL,};
184

    
185
#undef HAVE_MMX
186

    
187
//MMX versions
188
#undef RENAME
189
#define HAVE_MMX
190
#undef HAVE_MMX2
191
#undef HAVE_3DNOW
192
#define RENAME(a) a ## _MMX
193
#include "yuv2rgb_template.c"
194

    
195
//MMX2 versions
196
#undef RENAME
197
#define HAVE_MMX
198
#define HAVE_MMX2
199
#undef HAVE_3DNOW
200
#define RENAME(a) a ## _MMX2
201
#include "yuv2rgb_template.c"
202

    
203
#endif // CAN_COMPILE_X86_ASM
204

    
205
const int32_t Inverse_Table_6_9[8][4] = {
206
    {117504, 138453, 13954, 34903}, /* no sequence_display_extension */
207
    {117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */
208
    {104597, 132201, 25675, 53279}, /* unspecified */
209
    {104597, 132201, 25675, 53279}, /* reserved */
210
    {104448, 132798, 24759, 53109}, /* FCC */
211
    {104597, 132201, 25675, 53279}, /* ITU-R Rec. 624-4 System B, G */
212
    {104597, 132201, 25675, 53279}, /* SMPTE 170M */
213
    {117579, 136230, 16907, 35559}  /* SMPTE 240M (1987) */
214
};
215

    
216
#define RGB(i)                                        \
217
        U = pu[i];                                \
218
        V = pv[i];                                \
219
        r = c->table_rV[V];                        \
220
        g = c->table_gU[U] + c->table_gV[V];                \
221
        b = c->table_bU[U];
222

    
223
#define DST1(i)                                        \
224
        Y = py_1[2*i];                                \
225
        dst_1[2*i] = r[Y] + g[Y] + b[Y];        \
226
        Y = py_1[2*i+1];                        \
227
        dst_1[2*i+1] = r[Y] + g[Y] + b[Y];
228

    
229
#define DST2(i)                                        \
230
        Y = py_2[2*i];                                \
231
        dst_2[2*i] = r[Y] + g[Y] + b[Y];        \
232
        Y = py_2[2*i+1];                        \
233
        dst_2[2*i+1] = r[Y] + g[Y] + b[Y];
234

    
235
#define DST1RGB(i)                                                        \
236
        Y = py_1[2*i];                                                        \
237
        dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y];        \
238
        Y = py_1[2*i+1];                                                \
239
        dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y];
240

    
241
#define DST2RGB(i)                                                        \
242
        Y = py_2[2*i];                                                        \
243
        dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y];        \
244
        Y = py_2[2*i+1];                                                \
245
        dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y];
246

    
247
#define DST1BGR(i)                                                        \
248
        Y = py_1[2*i];                                                        \
249
        dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y];        \
250
        Y = py_1[2*i+1];                                                \
251
        dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y];
252

    
253
#define DST2BGR(i)                                                        \
254
        Y = py_2[2*i];                                                        \
255
        dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y];        \
256
        Y = py_2[2*i+1];                                                \
257
        dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y];
258

    
259
#define PROLOG(func_name, dst_type) \
260
static int func_name(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, \
261
             int srcSliceH, uint8_t* dst[], int dstStride[]){\
262
    int y;\
263
\
264
    if(c->srcFormat == IMGFMT_422P){\
265
        srcStride[1] *= 2;\
266
        srcStride[2] *= 2;\
267
    }\
268
    for(y=0; y<srcSliceH; y+=2){\
269
        dst_type *dst_1= (dst_type*)(dst[0] + (y+srcSliceY  )*dstStride[0]);\
270
        dst_type *dst_2= (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
271
        dst_type *r, *g, *b;\
272
        uint8_t *py_1= src[0] + y*srcStride[0];\
273
        uint8_t *py_2= py_1 + srcStride[0];\
274
        uint8_t *pu= src[1] + (y>>1)*srcStride[1];\
275
        uint8_t *pv= src[2] + (y>>1)*srcStride[2];\
276
        unsigned int h_size= c->dstW>>3;\
277
        while (h_size--) {\
278
            int U, V, Y;\
279

    
280
#define EPILOG(dst_delta)\
281
            pu += 4;\
282
            pv += 4;\
283
            py_1 += 8;\
284
            py_2 += 8;\
285
            dst_1 += dst_delta;\
286
            dst_2 += dst_delta;\
287
        }\
288
    }\
289
    return srcSliceH;\
290
}
291

    
292
PROLOG(yuv2rgb_c_32, uint32_t)
293
        RGB(0);
294
        DST1(0);
295
        DST2(0);
296

    
297
        RGB(1);
298
        DST2(1);
299
        DST1(1);
300

    
301
        RGB(2);
302
        DST1(2);
303
        DST2(2);
304

    
305
        RGB(3);
306
        DST2(3);
307
        DST1(3);
308
EPILOG(8)
309

    
310
PROLOG(yuv2rgb_c_24_rgb, uint8_t)
311
        RGB(0);
312
        DST1RGB(0);
313
        DST2RGB(0);
314

    
315
        RGB(1);
316
        DST2RGB(1);
317
        DST1RGB(1);
318

    
319
        RGB(2);
320
        DST1RGB(2);
321
        DST2RGB(2);
322

    
323
        RGB(3);
324
        DST2RGB(3);
325
        DST1RGB(3);
326
EPILOG(24)
327

    
328
// only trivial mods from yuv2rgb_c_24_rgb
329
PROLOG(yuv2rgb_c_24_bgr, uint8_t)
330
        RGB(0);
331
        DST1BGR(0);
332
        DST2BGR(0);
333

    
334
        RGB(1);
335
        DST2BGR(1);
336
        DST1BGR(1);
337

    
338
        RGB(2);
339
        DST1BGR(2);
340
        DST2BGR(2);
341

    
342
        RGB(3);
343
        DST2BGR(3);
344
        DST1BGR(3);
345
EPILOG(24)
346

    
347
// This is exactly the same code as yuv2rgb_c_32 except for the types of
348
// r, g, b, dst_1, dst_2
349
PROLOG(yuv2rgb_c_16, uint16_t)
350
        RGB(0);
351
        DST1(0);
352
        DST2(0);
353

    
354
        RGB(1);
355
        DST2(1);
356
        DST1(1);
357

    
358
        RGB(2);
359
        DST1(2);
360
        DST2(2);
361

    
362
        RGB(3);
363
        DST2(3);
364
        DST1(3);
365
EPILOG(8)
366

    
367
// This is exactly the same code as yuv2rgb_c_32 except for the types of
368
// r, g, b, dst_1, dst_2
369
PROLOG(yuv2rgb_c_8, uint8_t)
370
        RGB(0);
371
        DST1(0);
372
        DST2(0);
373

    
374
        RGB(1);
375
        DST2(1);
376
        DST1(1);
377

    
378
        RGB(2);
379
        DST1(2);
380
        DST2(2);
381

    
382
        RGB(3);
383
        DST2(3);
384
        DST1(3);
385
EPILOG(8)
386

    
387
// r, g, b, dst_1, dst_2
388
PROLOG(yuv2rgb_c_8_ordered_dither, uint8_t)
389
        const uint8_t *d32= dither_8x8_32[y&7];
390
        const uint8_t *d64= dither_8x8_73[y&7];
391
#define DST1bpp8(i,o)                                        \
392
        Y = py_1[2*i];                                \
393
        dst_1[2*i] = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]];        \
394
        Y = py_1[2*i+1];                        \
395
        dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
396

    
397
#define DST2bpp8(i,o)                                        \
398
        Y = py_2[2*i];                                \
399
        dst_2[2*i] =  r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]];        \
400
        Y = py_2[2*i+1];                        \
401
        dst_2[2*i+1] =  r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
402

    
403

    
404
        RGB(0);
405
        DST1bpp8(0,0);
406
        DST2bpp8(0,0);
407

    
408
        RGB(1);
409
        DST2bpp8(1,2);
410
        DST1bpp8(1,2);
411

    
412
        RGB(2);
413
        DST1bpp8(2,4);
414
        DST2bpp8(2,4);
415

    
416
        RGB(3);
417
        DST2bpp8(3,6);
418
        DST1bpp8(3,6);
419
EPILOG(8)
420

    
421

    
422
// This is exactly the same code as yuv2rgb_c_32 except for the types of
423
// r, g, b, dst_1, dst_2
424
PROLOG(yuv2rgb_c_4, uint8_t)
425
        int acc;
426
#define DST1_4(i)                                        \
427
        Y = py_1[2*i];                                \
428
        acc = r[Y] + g[Y] + b[Y];        \
429
        Y = py_1[2*i+1];                        \
430
        acc |= (r[Y] + g[Y] + b[Y])<<4;\
431
        dst_1[i] = acc; 
432

    
433
#define DST2_4(i)                                        \
434
        Y = py_2[2*i];                                \
435
        acc = r[Y] + g[Y] + b[Y];        \
436
        Y = py_2[2*i+1];                        \
437
        acc |= (r[Y] + g[Y] + b[Y])<<4;\
438
        dst_2[i] = acc; 
439
        
440
        RGB(0);
441
        DST1_4(0);
442
        DST2_4(0);
443

    
444
        RGB(1);
445
        DST2_4(1);
446
        DST1_4(1);
447

    
448
        RGB(2);
449
        DST1_4(2);
450
        DST2_4(2);
451

    
452
        RGB(3);
453
        DST2_4(3);
454
        DST1_4(3);
455
EPILOG(4)
456

    
457
PROLOG(yuv2rgb_c_4_ordered_dither, uint8_t)
458
        const uint8_t *d64= dither_8x8_73[y&7];
459
        const uint8_t *d128=dither_8x8_220[y&7];
460
        int acc;
461

    
462
#define DST1bpp4(i,o)                                        \
463
        Y = py_1[2*i];                                \
464
        acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];        \
465
        Y = py_1[2*i+1];                        \
466
        acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4;\
467
        dst_1[i]= acc;
468

    
469
#define DST2bpp4(i,o)                                        \
470
        Y = py_2[2*i];                                \
471
        acc =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];        \
472
        Y = py_2[2*i+1];                        \
473
        acc |=  (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4;\
474
        dst_2[i]= acc;
475

    
476

    
477
        RGB(0);
478
        DST1bpp4(0,0);
479
        DST2bpp4(0,0);
480

    
481
        RGB(1);
482
        DST2bpp4(1,2);
483
        DST1bpp4(1,2);
484

    
485
        RGB(2);
486
        DST1bpp4(2,4);
487
        DST2bpp4(2,4);
488

    
489
        RGB(3);
490
        DST2bpp4(3,6);
491
        DST1bpp4(3,6);
492
EPILOG(4)
493

    
494
// This is exactly the same code as yuv2rgb_c_32 except for the types of
495
// r, g, b, dst_1, dst_2
496
PROLOG(yuv2rgb_c_4b, uint8_t)
497
        RGB(0);
498
        DST1(0);
499
        DST2(0);
500

    
501
        RGB(1);
502
        DST2(1);
503
        DST1(1);
504

    
505
        RGB(2);
506
        DST1(2);
507
        DST2(2);
508

    
509
        RGB(3);
510
        DST2(3);
511
        DST1(3);
512
EPILOG(8)
513

    
514
PROLOG(yuv2rgb_c_4b_ordered_dither, uint8_t)
515
        const uint8_t *d64= dither_8x8_73[y&7];
516
        const uint8_t *d128=dither_8x8_220[y&7];
517

    
518
#define DST1bpp4b(i,o)                                        \
519
        Y = py_1[2*i];                                \
520
        dst_1[2*i] = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];        \
521
        Y = py_1[2*i+1];                        \
522
        dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
523

    
524
#define DST2bpp4b(i,o)                                        \
525
        Y = py_2[2*i];                                \
526
        dst_2[2*i] =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];        \
527
        Y = py_2[2*i+1];                        \
528
        dst_2[2*i+1] =  r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
529

    
530

    
531
        RGB(0);
532
        DST1bpp4b(0,0);
533
        DST2bpp4b(0,0);
534

    
535
        RGB(1);
536
        DST2bpp4b(1,2);
537
        DST1bpp4b(1,2);
538

    
539
        RGB(2);
540
        DST1bpp4b(2,4);
541
        DST2bpp4b(2,4);
542

    
543
        RGB(3);
544
        DST2bpp4b(3,6);
545
        DST1bpp4b(3,6);
546
EPILOG(8)
547

    
548
PROLOG(yuv2rgb_c_1_ordered_dither, uint8_t)
549
        const uint8_t *d128=dither_8x8_220[y&7];
550
        char out_1=0, out_2=0;
551
        g= c->table_gU[128] + c->table_gV[128];
552

    
553
#define DST1bpp1(i,o)                                        \
554
        Y = py_1[2*i];                                \
555
        out_1+= out_1 + g[Y+d128[0+o]];        \
556
        Y = py_1[2*i+1];                        \
557
        out_1+= out_1 + g[Y+d128[1+o]];
558

    
559
#define DST2bpp1(i,o)                                        \
560
        Y = py_2[2*i];                                \
561
        out_2+= out_2 + g[Y+d128[8+o]];        \
562
        Y = py_2[2*i+1];                        \
563
        out_2+= out_2 + g[Y+d128[9+o]];
564

    
565
        DST1bpp1(0,0);
566
        DST2bpp1(0,0);
567

    
568
        DST2bpp1(1,2);
569
        DST1bpp1(1,2);
570

    
571
        DST1bpp1(2,4);
572
        DST2bpp1(2,4);
573

    
574
        DST2bpp1(3,6);
575
        DST1bpp1(3,6);
576
        
577
        dst_1[0]= out_1;
578
        dst_2[0]= out_2;
579
EPILOG(1)
580

    
581
SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
582
{
583
#if defined(ARCH_X86) || defined(ARCH_X86_64)
584
    if(c->flags & SWS_CPU_CAPS_MMX2){
585
        switch(c->dstFormat){
586
        case IMGFMT_BGR32: return yuv420_rgb32_MMX2;
587
        case IMGFMT_BGR24: return yuv420_rgb24_MMX2;
588
        case IMGFMT_BGR16: return yuv420_rgb16_MMX2;
589
        case IMGFMT_BGR15: return yuv420_rgb15_MMX2;
590
        }
591
    }
592
    if(c->flags & SWS_CPU_CAPS_MMX){
593
        switch(c->dstFormat){
594
        case IMGFMT_BGR32: return yuv420_rgb32_MMX;
595
        case IMGFMT_BGR24: return yuv420_rgb24_MMX;
596
        case IMGFMT_BGR16: return yuv420_rgb16_MMX;
597
        case IMGFMT_BGR15: return yuv420_rgb15_MMX;
598
        }
599
    }
600
#endif
601
#ifdef HAVE_MLIB
602
    {
603
        SwsFunc t= yuv2rgb_init_mlib(c);
604
        if(t) return t;
605
    }
606
#endif
607
#ifdef HAVE_ALTIVEC
608
    if (c->flags & SWS_CPU_CAPS_ALTIVEC)
609
    {
610
        SwsFunc t = yuv2rgb_init_altivec(c);
611
        if(t) return t;
612
    }
613
#endif
614

    
615
    MSG_WARN("No accelerated colorspace conversion found\n");
616

    
617
    switch(c->dstFormat){
618
    case IMGFMT_RGB32:
619
    case IMGFMT_BGR32: return yuv2rgb_c_32;
620
    case IMGFMT_RGB24: return yuv2rgb_c_24_rgb;
621
    case IMGFMT_BGR24: return yuv2rgb_c_24_bgr;
622
    case IMGFMT_RGB16:
623
    case IMGFMT_BGR16:
624
    case IMGFMT_RGB15:
625
    case IMGFMT_BGR15: return yuv2rgb_c_16;
626
    case IMGFMT_RGB8:
627
    case IMGFMT_BGR8:  return yuv2rgb_c_8_ordered_dither;
628
    case IMGFMT_RGB4:
629
    case IMGFMT_BGR4:  return yuv2rgb_c_4_ordered_dither;
630
    case IMGFMT_RG4B:
631
    case IMGFMT_BG4B:  return yuv2rgb_c_4b_ordered_dither;
632
    case IMGFMT_RGB1:
633
    case IMGFMT_BGR1:  return yuv2rgb_c_1_ordered_dither;
634
    default:
635
            assert(0);
636
    }
637
    return NULL;
638
}
639

    
640
static int div_round (int dividend, int divisor)
641
{
642
    if (dividend > 0)
643
        return (dividend + (divisor>>1)) / divisor;
644
    else
645
        return -((-dividend + (divisor>>1)) / divisor);
646
}
647

    
648
int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange, int brightness, int contrast, int saturation)
649
{  
650
    const int isRgb = IMGFMT_IS_BGR(c->dstFormat);
651
    const int bpp = isRgb?IMGFMT_RGB_DEPTH(c->dstFormat):IMGFMT_BGR_DEPTH(c->dstFormat);
652
    int i;
653
    uint8_t table_Y[1024];
654
    uint32_t *table_32 = 0;
655
    uint16_t *table_16 = 0;
656
    uint8_t *table_8 = 0;
657
    uint8_t *table_332 = 0;
658
    uint8_t *table_121 = 0;
659
    uint8_t *table_1 = 0;
660
    int entry_size = 0;
661
    void *table_r = 0, *table_g = 0, *table_b = 0;
662
    void *table_start;
663

    
664
    int64_t crv =  inv_table[0];
665
    int64_t cbu =  inv_table[1];
666
    int64_t cgu = -inv_table[2];
667
    int64_t cgv = -inv_table[3];
668
    int64_t cy  = 1<<16;
669
    int64_t oy  = 0;
670

    
671
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
672
    if(!fullRange){
673
        cy= (cy*255) / 219;
674
        oy= 16<<16;
675
    }
676
        
677
    cy = (cy *contrast             )>>16;
678
    crv= (crv*contrast * saturation)>>32;
679
    cbu= (cbu*contrast * saturation)>>32;
680
    cgu= (cgu*contrast * saturation)>>32;
681
    cgv= (cgv*contrast * saturation)>>32;
682
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
683
    oy -= 256*brightness;
684

    
685
    for (i = 0; i < 1024; i++) {
686
        int j;
687

    
688
        j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
689
        j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
690
        table_Y[i] = j;
691
    }
692

    
693
    switch (bpp) {
694
    case 32:
695
        table_start= table_32 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
696

    
697
        entry_size = sizeof (uint32_t);
698
        table_r = table_32 + 197;
699
        table_b = table_32 + 197 + 685;
700
        table_g = table_32 + 197 + 2*682;
701

    
702
        for (i = -197; i < 256+197; i++)
703
            ((uint32_t *)table_r)[i] = table_Y[i+384] << (isRgb ? 16 : 0);
704
        for (i = -132; i < 256+132; i++)
705
            ((uint32_t *)table_g)[i] = table_Y[i+384] << 8;
706
        for (i = -232; i < 256+232; i++)
707
            ((uint32_t *)table_b)[i] = table_Y[i+384] << (isRgb ? 0 : 16);
708
        break;
709

    
710
    case 24:
711
        table_start= table_8 = malloc ((256 + 2*232) * sizeof (uint8_t));
712

    
713
        entry_size = sizeof (uint8_t);
714
        table_r = table_g = table_b = table_8 + 232;
715

    
716
        for (i = -232; i < 256+232; i++)
717
            ((uint8_t * )table_b)[i] = table_Y[i+384];
718
        break;
719

    
720
    case 15:
721
    case 16:
722
        table_start= table_16 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
723

    
724
        entry_size = sizeof (uint16_t);
725
        table_r = table_16 + 197;
726
        table_b = table_16 + 197 + 685;
727
        table_g = table_16 + 197 + 2*682;
728

    
729
        for (i = -197; i < 256+197; i++) {
730
            int j = table_Y[i+384] >> 3;
731

    
732
            if (isRgb)
733
                j <<= ((bpp==16) ? 11 : 10);
734

    
735
            ((uint16_t *)table_r)[i] = j;
736
        }
737
        for (i = -132; i < 256+132; i++) {
738
            int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
739

    
740
            ((uint16_t *)table_g)[i] = j << 5;
741
        }
742
        for (i = -232; i < 256+232; i++) {
743
            int j = table_Y[i+384] >> 3;
744

    
745
            if (!isRgb)
746
                j <<= ((bpp==16) ? 11 : 10);
747

    
748
            ((uint16_t *)table_b)[i] = j;
749
        }
750
        break;
751

    
752
    case 8:
753
        table_start= table_332 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
754

    
755
        entry_size = sizeof (uint8_t);
756
        table_r = table_332 + 197;
757
        table_b = table_332 + 197 + 685;
758
        table_g = table_332 + 197 + 2*682;
759

    
760
        for (i = -197; i < 256+197; i++) {
761
            int j = (table_Y[i+384 - 16] + 18)/36;
762

    
763
            if (isRgb)
764
                j <<= 5;
765

    
766
            ((uint8_t *)table_r)[i] = j;
767
        }
768
        for (i = -132; i < 256+132; i++) {
769
            int j = (table_Y[i+384 - 16] + 18)/36;
770

    
771
            if (!isRgb)
772
                j <<= 1;
773

    
774
            ((uint8_t *)table_g)[i] = j << 2;
775
        }
776
        for (i = -232; i < 256+232; i++) {
777
            int j = (table_Y[i+384 - 37] + 43)/85;
778

    
779
            if (!isRgb)
780
                j <<= 6;
781

    
782
            ((uint8_t *)table_b)[i] = j;
783
        }
784
        break;
785
    case 4:
786
    case 4|128:
787
        table_start= table_121 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
788

    
789
        entry_size = sizeof (uint8_t);
790
        table_r = table_121 + 197;
791
        table_b = table_121 + 197 + 685;
792
        table_g = table_121 + 197 + 2*682;
793

    
794
        for (i = -197; i < 256+197; i++) {
795
            int j = table_Y[i+384 - 110] >> 7;
796

    
797
            if (isRgb)
798
                j <<= 3;
799

    
800
            ((uint8_t *)table_r)[i] = j;
801
        }
802
        for (i = -132; i < 256+132; i++) {
803
            int j = (table_Y[i+384 - 37]+ 43)/85;
804

    
805
            ((uint8_t *)table_g)[i] = j << 1;
806
        }
807
        for (i = -232; i < 256+232; i++) {
808
            int j =table_Y[i+384 - 110] >> 7;
809

    
810
            if (!isRgb)
811
                j <<= 3;
812

    
813
            ((uint8_t *)table_b)[i] = j;
814
        }
815
        break;
816

    
817
    case 1:
818
        table_start= table_1 = malloc (256*2 * sizeof (uint8_t));
819

    
820
        entry_size = sizeof (uint8_t);
821
        table_g = table_1;
822
        table_r = table_b = NULL;
823

    
824
        for (i = 0; i < 256+256; i++) {
825
            int j = table_Y[i + 384 - 110]>>7;
826

    
827
            ((uint8_t *)table_g)[i] = j;
828
        }
829
        break;
830

    
831
    default:
832
        table_start= NULL;
833
        MSG_ERR("%ibpp not supported by yuv2rgb\n", bpp);
834
        //free mem?
835
        return -1;
836
    }
837

    
838
    for (i = 0; i < 256; i++) {
839
        c->table_rV[i] = table_r + entry_size * div_round (crv * (i-128), 76309);
840
        c->table_gU[i] = table_g + entry_size * div_round (cgu * (i-128), 76309);
841
        c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
842
        c->table_bU[i] = table_b + entry_size * div_round (cbu * (i-128), 76309);
843
    }
844

    
845
    if(c->yuvTable) free(c->yuvTable);
846
    c->yuvTable= table_start;
847
    return 0;
848
}