Statistics
| Branch: | Revision:

ffmpeg / libswscale / swscale.c @ 300e497b

History | View | Annotate | Download (81.7 KB)

1
/*
2
 * Copyright (C) 2001-2003 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
/*
22
  supported Input formats: YV12, I420/IYUV, YUY2, UYVY, BGR32, BGR32_1, BGR24, BGR16, BGR15, RGB32, RGB32_1, RGB24, Y8/Y800, YVU9/IF09, PAL8
23
  supported output formats: YV12, I420/IYUV, YUY2, UYVY, {BGR,RGB}{1,4,8,15,16,24,32}, Y8/Y800, YVU9/IF09
24
  {BGR,RGB}{1,4,8,15,16} support dithering
25

26
  unscaled special converters (YV12=I420=IYUV, Y800=Y8)
27
  YV12 -> {BGR,RGB}{1,4,8,12,15,16,24,32}
28
  x -> x
29
  YUV9 -> YV12
30
  YUV9/YV12 -> Y800
31
  Y800 -> YUV9/YV12
32
  BGR24 -> BGR32 & RGB24 -> RGB32
33
  BGR32 -> BGR24 & RGB32 -> RGB24
34
  BGR15 -> BGR16
35
*/
36

    
37
/*
38
tested special converters (most are tested actually, but I did not write it down ...)
39
 YV12 -> BGR12/BGR16
40
 YV12 -> YV12
41
 BGR15 -> BGR16
42
 BGR16 -> BGR16
43
 YVU9 -> YV12
44

45
untested special converters
46
  YV12/I420 -> BGR15/BGR24/BGR32 (it is the yuv2rgb stuff, so it should be OK)
47
  YV12/I420 -> YV12/I420
48
  YUY2/BGR15/BGR24/BGR32/RGB24/RGB32 -> same format
49
  BGR24 -> BGR32 & RGB24 -> RGB32
50
  BGR32 -> BGR24 & RGB32 -> RGB24
51
  BGR24 -> YV12
52
*/
53

    
54
#include <inttypes.h>
55
#include <string.h>
56
#include <math.h>
57
#include <stdio.h>
58
#include "config.h"
59
#include <assert.h>
60
#include "swscale.h"
61
#include "swscale_internal.h"
62
#include "rgb2rgb.h"
63
#include "libavutil/intreadwrite.h"
64
#include "libavutil/x86_cpu.h"
65
#include "libavutil/avutil.h"
66
#include "libavutil/mathematics.h"
67
#include "libavutil/bswap.h"
68
#include "libavutil/pixdesc.h"
69

    
70
#undef MOVNTQ
71
#undef PAVGB
72

    
73
//#undef HAVE_MMX2
74
//#define HAVE_AMD3DNOW
75
//#undef HAVE_MMX
76
//#undef ARCH_X86
77
#define DITHER1XBPP
78

    
79
#define isPacked(x)         (       \
80
           (x)==PIX_FMT_PAL8        \
81
        || (x)==PIX_FMT_YUYV422     \
82
        || (x)==PIX_FMT_UYVY422     \
83
        || (x)==PIX_FMT_GRAY8A       \
84
        || isAnyRGB(x)              \
85
    )
86

    
87
#define RGB2YUV_SHIFT 15
88
#define BY ( (int)(0.114*219/255*(1<<RGB2YUV_SHIFT)+0.5))
89
#define BV (-(int)(0.081*224/255*(1<<RGB2YUV_SHIFT)+0.5))
90
#define BU ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
91
#define GY ( (int)(0.587*219/255*(1<<RGB2YUV_SHIFT)+0.5))
92
#define GV (-(int)(0.419*224/255*(1<<RGB2YUV_SHIFT)+0.5))
93
#define GU (-(int)(0.331*224/255*(1<<RGB2YUV_SHIFT)+0.5))
94
#define RY ( (int)(0.299*219/255*(1<<RGB2YUV_SHIFT)+0.5))
95
#define RV ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
96
#define RU (-(int)(0.169*224/255*(1<<RGB2YUV_SHIFT)+0.5))
97

    
98
static const double rgb2yuv_table[8][9]={
99
    {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5}, //ITU709
100
    {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5}, //ITU709
101
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M
102
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M
103
    {0.59  , 0.11  , 0.30  , -0.331, 0.5, -0.169, -0.421, -0.079, 0.5}, //FCC
104
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M
105
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M
106
    {0.701 , 0.087 , 0.212 , -0.384, 0.5, -0.116, -0.445, -0.055, 0.5}, //SMPTE 240M
107
};
108

    
109
/*
110
NOTES
111
Special versions: fast Y 1:1 scaling (no interpolation in y direction)
112

113
TODO
114
more intelligent misalignment avoidance for the horizontal scaler
115
write special vertical cubic upscale version
116
optimize C code (YV12 / minmax)
117
add support for packed pixel YUV input & output
118
add support for Y8 output
119
optimize BGR24 & BGR32
120
add BGR4 output support
121
write special BGR->BGR scaler
122
*/
123

    
124
#if ARCH_X86
125
DECLARE_ASM_CONST(8, uint64_t, bF8)=       0xF8F8F8F8F8F8F8F8LL;
126
DECLARE_ASM_CONST(8, uint64_t, bFC)=       0xFCFCFCFCFCFCFCFCLL;
127
DECLARE_ASM_CONST(8, uint64_t, w10)=       0x0010001000100010LL;
128
DECLARE_ASM_CONST(8, uint64_t, w02)=       0x0002000200020002LL;
129
DECLARE_ASM_CONST(8, uint64_t, bm00001111)=0x00000000FFFFFFFFLL;
130
DECLARE_ASM_CONST(8, uint64_t, bm00000111)=0x0000000000FFFFFFLL;
131
DECLARE_ASM_CONST(8, uint64_t, bm11111000)=0xFFFFFFFFFF000000LL;
132
DECLARE_ASM_CONST(8, uint64_t, bm01010101)=0x00FF00FF00FF00FFLL;
133

    
134
const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
135
        0x0103010301030103LL,
136
        0x0200020002000200LL,};
137

    
138
const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
139
        0x0602060206020602LL,
140
        0x0004000400040004LL,};
141

    
142
DECLARE_ASM_CONST(8, uint64_t, b16Mask)=   0x001F001F001F001FLL;
143
DECLARE_ASM_CONST(8, uint64_t, g16Mask)=   0x07E007E007E007E0LL;
144
DECLARE_ASM_CONST(8, uint64_t, r16Mask)=   0xF800F800F800F800LL;
145
DECLARE_ASM_CONST(8, uint64_t, b15Mask)=   0x001F001F001F001FLL;
146
DECLARE_ASM_CONST(8, uint64_t, g15Mask)=   0x03E003E003E003E0LL;
147
DECLARE_ASM_CONST(8, uint64_t, r15Mask)=   0x7C007C007C007C00LL;
148

    
149
DECLARE_ALIGNED(8, const uint64_t, ff_M24A)         = 0x00FF0000FF0000FFLL;
150
DECLARE_ALIGNED(8, const uint64_t, ff_M24B)         = 0xFF0000FF0000FF00LL;
151
DECLARE_ALIGNED(8, const uint64_t, ff_M24C)         = 0x0000FF0000FF0000LL;
152

    
153
#ifdef FAST_BGR2YV12
154
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000000210041000DULL;
155
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000FFEEFFDC0038ULL;
156
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00000038FFD2FFF8ULL;
157
#else
158
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000020E540830C8BULL;
159
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000ED0FDAC23831ULL;
160
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00003831D0E6F6EAULL;
161
#endif /* FAST_BGR2YV12 */
162
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset)  = 0x1010101010101010ULL;
163
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
164
DECLARE_ALIGNED(8, const uint64_t, ff_w1111)        = 0x0001000100010001ULL;
165

    
166
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY1Coeff) = 0x0C88000040870C88ULL;
167
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY2Coeff) = 0x20DE4087000020DEULL;
168
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY1Coeff) = 0x20DE0000408720DEULL;
169
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY2Coeff) = 0x0C88408700000C88ULL;
170
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toYOffset) = 0x0008400000084000ULL;
171

    
172
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUV)[2][4] = {
173
    {0x38380000DAC83838ULL, 0xECFFDAC80000ECFFULL, 0xF6E40000D0E3F6E4ULL, 0x3838D0E300003838ULL},
174
    {0xECFF0000DAC8ECFFULL, 0x3838DAC800003838ULL, 0x38380000D0E33838ULL, 0xF6E4D0E30000F6E4ULL},
175
};
176

    
177
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUVOffset)= 0x0040400000404000ULL;
178

    
179
#endif /* ARCH_X86 */
180

    
181
DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_4)[2][8]={
182
{  1,   3,   1,   3,   1,   3,   1,   3, },
183
{  2,   0,   2,   0,   2,   0,   2,   0, },
184
};
185

    
186
DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_8)[2][8]={
187
{  6,   2,   6,   2,   6,   2,   6,   2, },
188
{  0,   4,   0,   4,   0,   4,   0,   4, },
189
};
190

    
191
DECLARE_ALIGNED(8, const uint8_t, dither_4x4_16)[4][8]={
192
{  8,   4,  11,   7,   8,   4,  11,   7, },
193
{  2,  14,   1,  13,   2,  14,   1,  13, },
194
{ 10,   6,   9,   5,  10,   6,   9,   5, },
195
{  0,  12,   3,  15,   0,  12,   3,  15, },
196
};
197

    
198
DECLARE_ALIGNED(8, const uint8_t, dither_8x8_32)[8][8]={
199
{ 17,   9,  23,  15,  16,   8,  22,  14, },
200
{  5,  29,   3,  27,   4,  28,   2,  26, },
201
{ 21,  13,  19,  11,  20,  12,  18,  10, },
202
{  0,  24,   6,  30,   1,  25,   7,  31, },
203
{ 16,   8,  22,  14,  17,   9,  23,  15, },
204
{  4,  28,   2,  26,   5,  29,   3,  27, },
205
{ 20,  12,  18,  10,  21,  13,  19,  11, },
206
{  1,  25,   7,  31,   0,  24,   6,  30, },
207
};
208

    
209
DECLARE_ALIGNED(8, const uint8_t, dither_8x8_73)[8][8]={
210
{  0,  55,  14,  68,   3,  58,  17,  72, },
211
{ 37,  18,  50,  32,  40,  22,  54,  35, },
212
{  9,  64,   5,  59,  13,  67,   8,  63, },
213
{ 46,  27,  41,  23,  49,  31,  44,  26, },
214
{  2,  57,  16,  71,   1,  56,  15,  70, },
215
{ 39,  21,  52,  34,  38,  19,  51,  33, },
216
{ 11,  66,   7,  62,  10,  65,   6,  60, },
217
{ 48,  30,  43,  25,  47,  29,  42,  24, },
218
};
219

    
220
#if 1
221
DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
222
{117,  62, 158, 103, 113,  58, 155, 100, },
223
{ 34, 199,  21, 186,  31, 196,  17, 182, },
224
{144,  89, 131,  76, 141,  86, 127,  72, },
225
{  0, 165,  41, 206,  10, 175,  52, 217, },
226
{110,  55, 151,  96, 120,  65, 162, 107, },
227
{ 28, 193,  14, 179,  38, 203,  24, 189, },
228
{138,  83, 124,  69, 148,  93, 134,  79, },
229
{  7, 172,  48, 213,   3, 168,  45, 210, },
230
};
231
#elif 1
232
// tries to correct a gamma of 1.5
233
DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
234
{  0, 143,  18, 200,   2, 156,  25, 215, },
235
{ 78,  28, 125,  64,  89,  36, 138,  74, },
236
{ 10, 180,   3, 161,  16, 195,   8, 175, },
237
{109,  51,  93,  38, 121,  60, 105,  47, },
238
{  1, 152,  23, 210,   0, 147,  20, 205, },
239
{ 85,  33, 134,  71,  81,  30, 130,  67, },
240
{ 14, 190,   6, 171,  12, 185,   5, 166, },
241
{117,  57, 101,  44, 113,  54,  97,  41, },
242
};
243
#elif 1
244
// tries to correct a gamma of 2.0
245
DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
246
{  0, 124,   8, 193,   0, 140,  12, 213, },
247
{ 55,  14, 104,  42,  66,  19, 119,  52, },
248
{  3, 168,   1, 145,   6, 187,   3, 162, },
249
{ 86,  31,  70,  21,  99,  39,  82,  28, },
250
{  0, 134,  11, 206,   0, 129,   9, 200, },
251
{ 62,  17, 114,  48,  58,  16, 109,  45, },
252
{  5, 181,   2, 157,   4, 175,   1, 151, },
253
{ 95,  36,  78,  26,  90,  34,  74,  24, },
254
};
255
#else
256
// tries to correct a gamma of 2.5
257
DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
258
{  0, 107,   3, 187,   0, 125,   6, 212, },
259
{ 39,   7,  86,  28,  49,  11, 102,  36, },
260
{  1, 158,   0, 131,   3, 180,   1, 151, },
261
{ 68,  19,  52,  12,  81,  25,  64,  17, },
262
{  0, 119,   5, 203,   0, 113,   4, 195, },
263
{ 45,   9,  96,  33,  42,   8,  91,  30, },
264
{  2, 172,   1, 144,   2, 165,   0, 137, },
265
{ 77,  23,  60,  15,  72,  21,  56,  14, },
266
};
267
#endif
268

    
269
DECLARE_ALIGNED(8, const uint8_t, dithers)[8][8][8]={
270
{
271
  {   0,  1,  0,  1,  0,  1,  0,  1,},
272
  {   1,  0,  1,  0,  1,  0,  1,  0,},
273
  {   0,  1,  0,  1,  0,  1,  0,  1,},
274
  {   1,  0,  1,  0,  1,  0,  1,  0,},
275
  {   0,  1,  0,  1,  0,  1,  0,  1,},
276
  {   1,  0,  1,  0,  1,  0,  1,  0,},
277
  {   0,  1,  0,  1,  0,  1,  0,  1,},
278
  {   1,  0,  1,  0,  1,  0,  1,  0,},
279
},{
280
  {   1,  2,  1,  2,  1,  2,  1,  2,},
281
  {   3,  0,  3,  0,  3,  0,  3,  0,},
282
  {   1,  2,  1,  2,  1,  2,  1,  2,},
283
  {   3,  0,  3,  0,  3,  0,  3,  0,},
284
  {   1,  2,  1,  2,  1,  2,  1,  2,},
285
  {   3,  0,  3,  0,  3,  0,  3,  0,},
286
  {   1,  2,  1,  2,  1,  2,  1,  2,},
287
  {   3,  0,  3,  0,  3,  0,  3,  0,},
288
},{
289
  {   2,  4,  3,  5,  2,  4,  3,  5,},
290
  {   6,  0,  7,  1,  6,  0,  7,  1,},
291
  {   3,  5,  2,  4,  3,  5,  2,  4,},
292
  {   7,  1,  6,  0,  7,  1,  6,  0,},
293
  {   2,  4,  3,  5,  2,  4,  3,  5,},
294
  {   6,  0,  7,  1,  6,  0,  7,  1,},
295
  {   3,  5,  2,  4,  3,  5,  2,  4,},
296
  {   7,  1,  6,  0,  7,  1,  6,  0,},
297
},{
298
  {   4,  8,  7, 11,  4,  8,  7, 11,},
299
  {  12,  0, 15,  3, 12,  0, 15,  3,},
300
  {   6, 10,  5,  9,  6, 10,  5,  9,},
301
  {  14,  2, 13,  1, 14,  2, 13,  1,},
302
  {   4,  8,  7, 11,  4,  8,  7, 11,},
303
  {  12,  0, 15,  3, 12,  0, 15,  3,},
304
  {   6, 10,  5,  9,  6, 10,  5,  9,},
305
  {  14,  2, 13,  1, 14,  2, 13,  1,},
306
},{
307
  {   9, 17, 15, 23,  8, 16, 14, 22,},
308
  {  25,  1, 31,  7, 24,  0, 30,  6,},
309
  {  13, 21, 11, 19, 12, 20, 10, 18,},
310
  {  29,  5, 27,  3, 28,  4, 26,  2,},
311
  {   8, 16, 14, 22,  9, 17, 15, 23,},
312
  {  24,  0, 30,  6, 25,  1, 31,  7,},
313
  {  12, 20, 10, 18, 13, 21, 11, 19,},
314
  {  28,  4, 26,  2, 29,  5, 27,  3,},
315
},{
316
  {  18, 34, 30, 46, 17, 33, 29, 45,},
317
  {  50,  2, 62, 14, 49,  1, 61, 13,},
318
  {  26, 42, 22, 38, 25, 41, 21, 37,},
319
  {  58, 10, 54,  6, 57,  9, 53,  5,},
320
  {  16, 32, 28, 44, 19, 35, 31, 47,},
321
  {  48,  0, 60, 12, 51,  3, 63, 15,},
322
  {  24, 40, 20, 36, 27, 43, 23, 39,},
323
  {  56,  8, 52,  4, 59, 11, 55,  7,},
324
},{
325
  {  18, 34, 30, 46, 17, 33, 29, 45,},
326
  {  50,  2, 62, 14, 49,  1, 61, 13,},
327
  {  26, 42, 22, 38, 25, 41, 21, 37,},
328
  {  58, 10, 54,  6, 57,  9, 53,  5,},
329
  {  16, 32, 28, 44, 19, 35, 31, 47,},
330
  {  48,  0, 60, 12, 51,  3, 63, 15,},
331
  {  24, 40, 20, 36, 27, 43, 23, 39,},
332
  {  56,  8, 52,  4, 59, 11, 55,  7,},
333
},{
334
  {  36, 68, 60, 92, 34, 66, 58, 90,},
335
  { 100,  4,124, 28, 98,  2,122, 26,},
336
  {  52, 84, 44, 76, 50, 82, 42, 74,},
337
  { 116, 20,108, 12,114, 18,106, 10,},
338
  {  32, 64, 56, 88, 38, 70, 62, 94,},
339
  {  96,  0,120, 24,102,  6,126, 30,},
340
  {  48, 80, 40, 72, 54, 86, 46, 78,},
341
  { 112, 16,104,  8,118, 22,110, 14,},
342
}};
343

    
344
static av_always_inline void yuv2yuvX16inC_template(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
345
                                                    const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
346
                                                    const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest,
347
                                                    int dstW, int chrDstW, int big_endian)
348
{
349
    //FIXME Optimize (just quickly written not optimized..)
350
    int i;
351

    
352
    for (i = 0; i < dstW; i++) {
353
        int val = 1 << 10;
354
        int j;
355

    
356
        for (j = 0; j < lumFilterSize; j++)
357
            val += lumSrc[j][i] * lumFilter[j];
358

    
359
        if (big_endian) {
360
            AV_WB16(&dest[i], av_clip_uint16(val >> 11));
361
        } else {
362
            AV_WL16(&dest[i], av_clip_uint16(val >> 11));
363
        }
364
    }
365

    
366
    if (uDest) {
367
        for (i = 0; i < chrDstW; i++) {
368
            int u = 1 << 10;
369
            int v = 1 << 10;
370
            int j;
371

    
372
            for (j = 0; j < chrFilterSize; j++) {
373
                u += chrSrc[j][i       ] * chrFilter[j];
374
                v += chrSrc[j][i + VOFW] * chrFilter[j];
375
            }
376

    
377
            if (big_endian) {
378
                AV_WB16(&uDest[i], av_clip_uint16(u >> 11));
379
                AV_WB16(&vDest[i], av_clip_uint16(v >> 11));
380
            } else {
381
                AV_WL16(&uDest[i], av_clip_uint16(u >> 11));
382
                AV_WL16(&vDest[i], av_clip_uint16(v >> 11));
383
            }
384
        }
385
    }
386

    
387
    if (CONFIG_SWSCALE_ALPHA && aDest) {
388
        for (i = 0; i < dstW; i++) {
389
            int val = 1 << 10;
390
            int j;
391

    
392
            for (j = 0; j < lumFilterSize; j++)
393
                val += alpSrc[j][i] * lumFilter[j];
394

    
395
            if (big_endian) {
396
                AV_WB16(&aDest[i], av_clip_uint16(val >> 11));
397
            } else {
398
                AV_WL16(&aDest[i], av_clip_uint16(val >> 11));
399
            }
400
        }
401
    }
402
}
403

    
404
static av_always_inline void yuv2yuvXNinC_template(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
405
                                                   const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
406
                                                   const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest,
407
                                                   int dstW, int chrDstW, int big_endian, int depth)
408
{
409
    //FIXME Optimize (just quickly written not optimized..)
410
    int i;
411

    
412
    for (i = 0; i < dstW; i++) {
413
        int val = 1 << (26-depth);
414
        int j;
415

    
416
        for (j = 0; j < lumFilterSize; j++)
417
            val += lumSrc[j][i] * lumFilter[j];
418

    
419
        if (big_endian) {
420
            AV_WB16(&dest[i], av_clip(val >> (27-depth), 0, (1<<depth)-1));
421
        } else {
422
            AV_WL16(&dest[i], av_clip(val >> (27-depth), 0, (1<<depth)-1));
423
        }
424
    }
425

    
426
    if (uDest) {
427
        for (i = 0; i < chrDstW; i++) {
428
            int u = 1 << (26-depth);
429
            int v = 1 << (26-depth);
430
            int j;
431

    
432
            for (j = 0; j < chrFilterSize; j++) {
433
                u += chrSrc[j][i       ] * chrFilter[j];
434
                v += chrSrc[j][i + VOFW] * chrFilter[j];
435
            }
436

    
437
            if (big_endian) {
438
                AV_WB16(&uDest[i], av_clip(u >> (27-depth), 0, (1<<depth)-1));
439
                AV_WB16(&vDest[i], av_clip(v >> (27-depth), 0, (1<<depth)-1));
440
            } else {
441
                AV_WL16(&uDest[i], av_clip(u >> (27-depth), 0, (1<<depth)-1));
442
                AV_WL16(&vDest[i], av_clip(v >> (27-depth), 0, (1<<depth)-1));
443
            }
444
        }
445
    }
446
}
447

    
448
static inline void yuv2yuvX16inC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
449
                                 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
450
                                 const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest, int dstW, int chrDstW,
451
                                 enum PixelFormat dstFormat)
452
{
453
    if (isNBPS(dstFormat)) {
454
        const int depth = av_pix_fmt_descriptors[dstFormat].comp[0].depth_minus1+1;
455
        yuv2yuvXNinC_template(lumFilter, lumSrc, lumFilterSize,
456
                              chrFilter, chrSrc, chrFilterSize,
457
                              alpSrc,
458
                              dest, uDest, vDest, aDest,
459
                              dstW, chrDstW, isBE(dstFormat), depth);
460
    } else {
461
        if (isBE(dstFormat)) {
462
            yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
463
                                   chrFilter, chrSrc, chrFilterSize,
464
                                   alpSrc,
465
                                   dest, uDest, vDest, aDest,
466
                                   dstW, chrDstW, 1);
467
        } else {
468
            yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
469
                                   chrFilter, chrSrc, chrFilterSize,
470
                                   alpSrc,
471
                                   dest, uDest, vDest, aDest,
472
                                   dstW, chrDstW, 0);
473
        }
474
    }
475
}
476

    
477
static inline void yuv2yuvXinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
478
                               const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
479
                               const int16_t **alpSrc, uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, int dstW, int chrDstW)
480
{
481
    //FIXME Optimize (just quickly written not optimized..)
482
    int i;
483
    for (i=0; i<dstW; i++) {
484
        int val=1<<18;
485
        int j;
486
        for (j=0; j<lumFilterSize; j++)
487
            val += lumSrc[j][i] * lumFilter[j];
488

    
489
        dest[i]= av_clip_uint8(val>>19);
490
    }
491

    
492
    if (uDest)
493
        for (i=0; i<chrDstW; i++) {
494
            int u=1<<18;
495
            int v=1<<18;
496
            int j;
497
            for (j=0; j<chrFilterSize; j++) {
498
                u += chrSrc[j][i] * chrFilter[j];
499
                v += chrSrc[j][i + VOFW] * chrFilter[j];
500
            }
501

    
502
            uDest[i]= av_clip_uint8(u>>19);
503
            vDest[i]= av_clip_uint8(v>>19);
504
        }
505

    
506
    if (CONFIG_SWSCALE_ALPHA && aDest)
507
        for (i=0; i<dstW; i++) {
508
            int val=1<<18;
509
            int j;
510
            for (j=0; j<lumFilterSize; j++)
511
                val += alpSrc[j][i] * lumFilter[j];
512

    
513
            aDest[i]= av_clip_uint8(val>>19);
514
        }
515

    
516
}
517

    
518
static inline void yuv2nv12XinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
519
                                const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
520
                                uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat)
521
{
522
    //FIXME Optimize (just quickly written not optimized..)
523
    int i;
524
    for (i=0; i<dstW; i++) {
525
        int val=1<<18;
526
        int j;
527
        for (j=0; j<lumFilterSize; j++)
528
            val += lumSrc[j][i] * lumFilter[j];
529

    
530
        dest[i]= av_clip_uint8(val>>19);
531
    }
532

    
533
    if (!uDest)
534
        return;
535

    
536
    if (dstFormat == PIX_FMT_NV12)
537
        for (i=0; i<chrDstW; i++) {
538
            int u=1<<18;
539
            int v=1<<18;
540
            int j;
541
            for (j=0; j<chrFilterSize; j++) {
542
                u += chrSrc[j][i] * chrFilter[j];
543
                v += chrSrc[j][i + VOFW] * chrFilter[j];
544
            }
545

    
546
            uDest[2*i]= av_clip_uint8(u>>19);
547
            uDest[2*i+1]= av_clip_uint8(v>>19);
548
        }
549
    else
550
        for (i=0; i<chrDstW; i++) {
551
            int u=1<<18;
552
            int v=1<<18;
553
            int j;
554
            for (j=0; j<chrFilterSize; j++) {
555
                u += chrSrc[j][i] * chrFilter[j];
556
                v += chrSrc[j][i + VOFW] * chrFilter[j];
557
            }
558

    
559
            uDest[2*i]= av_clip_uint8(v>>19);
560
            uDest[2*i+1]= av_clip_uint8(u>>19);
561
        }
562
}
563

    
564
#define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha) \
565
    for (i=0; i<(dstW>>1); i++) {\
566
        int j;\
567
        int Y1 = 1<<18;\
568
        int Y2 = 1<<18;\
569
        int U  = 1<<18;\
570
        int V  = 1<<18;\
571
        int av_unused A1, A2;\
572
        type av_unused *r, *b, *g;\
573
        const int i2= 2*i;\
574
        \
575
        for (j=0; j<lumFilterSize; j++) {\
576
            Y1 += lumSrc[j][i2] * lumFilter[j];\
577
            Y2 += lumSrc[j][i2+1] * lumFilter[j];\
578
        }\
579
        for (j=0; j<chrFilterSize; j++) {\
580
            U += chrSrc[j][i] * chrFilter[j];\
581
            V += chrSrc[j][i+VOFW] * chrFilter[j];\
582
        }\
583
        Y1>>=19;\
584
        Y2>>=19;\
585
        U >>=19;\
586
        V >>=19;\
587
        if (alpha) {\
588
            A1 = 1<<18;\
589
            A2 = 1<<18;\
590
            for (j=0; j<lumFilterSize; j++) {\
591
                A1 += alpSrc[j][i2  ] * lumFilter[j];\
592
                A2 += alpSrc[j][i2+1] * lumFilter[j];\
593
            }\
594
            A1>>=19;\
595
            A2>>=19;\
596
        }
597

    
598
#define YSCALE_YUV_2_PACKEDX_C(type,alpha) \
599
        YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha)\
600
        if ((Y1|Y2|U|V)&256) {\
601
            if (Y1>255)   Y1=255; \
602
            else if (Y1<0)Y1=0;   \
603
            if (Y2>255)   Y2=255; \
604
            else if (Y2<0)Y2=0;   \
605
            if (U>255)    U=255;  \
606
            else if (U<0) U=0;    \
607
            if (V>255)    V=255;  \
608
            else if (V<0) V=0;    \
609
        }\
610
        if (alpha && ((A1|A2)&256)) {\
611
            A1=av_clip_uint8(A1);\
612
            A2=av_clip_uint8(A2);\
613
        }
614

    
615
#define YSCALE_YUV_2_PACKEDX_FULL_C(rnd,alpha) \
616
    for (i=0; i<dstW; i++) {\
617
        int j;\
618
        int Y = 0;\
619
        int U = -128<<19;\
620
        int V = -128<<19;\
621
        int av_unused A;\
622
        int R,G,B;\
623
        \
624
        for (j=0; j<lumFilterSize; j++) {\
625
            Y += lumSrc[j][i     ] * lumFilter[j];\
626
        }\
627
        for (j=0; j<chrFilterSize; j++) {\
628
            U += chrSrc[j][i     ] * chrFilter[j];\
629
            V += chrSrc[j][i+VOFW] * chrFilter[j];\
630
        }\
631
        Y >>=10;\
632
        U >>=10;\
633
        V >>=10;\
634
        if (alpha) {\
635
            A = rnd;\
636
            for (j=0; j<lumFilterSize; j++)\
637
                A += alpSrc[j][i     ] * lumFilter[j];\
638
            A >>=19;\
639
            if (A&256)\
640
                A = av_clip_uint8(A);\
641
        }
642

    
643
#define YSCALE_YUV_2_RGBX_FULL_C(rnd,alpha) \
644
    YSCALE_YUV_2_PACKEDX_FULL_C(rnd>>3,alpha)\
645
        Y-= c->yuv2rgb_y_offset;\
646
        Y*= c->yuv2rgb_y_coeff;\
647
        Y+= rnd;\
648
        R= Y + V*c->yuv2rgb_v2r_coeff;\
649
        G= Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;\
650
        B= Y +                          U*c->yuv2rgb_u2b_coeff;\
651
        if ((R|G|B)&(0xC0000000)) {\
652
            if (R>=(256<<22))   R=(256<<22)-1; \
653
            else if (R<0)R=0;   \
654
            if (G>=(256<<22))   G=(256<<22)-1; \
655
            else if (G<0)G=0;   \
656
            if (B>=(256<<22))   B=(256<<22)-1; \
657
            else if (B<0)B=0;   \
658
        }
659

    
660
#define YSCALE_YUV_2_GRAY16_C \
661
    for (i=0; i<(dstW>>1); i++) {\
662
        int j;\
663
        int Y1 = 1<<18;\
664
        int Y2 = 1<<18;\
665
        int U  = 1<<18;\
666
        int V  = 1<<18;\
667
        \
668
        const int i2= 2*i;\
669
        \
670
        for (j=0; j<lumFilterSize; j++) {\
671
            Y1 += lumSrc[j][i2] * lumFilter[j];\
672
            Y2 += lumSrc[j][i2+1] * lumFilter[j];\
673
        }\
674
        Y1>>=11;\
675
        Y2>>=11;\
676
        if ((Y1|Y2|U|V)&65536) {\
677
            if (Y1>65535)   Y1=65535; \
678
            else if (Y1<0)Y1=0;   \
679
            if (Y2>65535)   Y2=65535; \
680
            else if (Y2<0)Y2=0;   \
681
        }
682

    
683
#define YSCALE_YUV_2_RGBX_C(type,alpha) \
684
    YSCALE_YUV_2_PACKEDX_C(type,alpha)  /* FIXME fix tables so that clipping is not needed and then use _NOCLIP*/\
685
    r = (type *)c->table_rV[V];   \
686
    g = (type *)(c->table_gU[U] + c->table_gV[V]); \
687
    b = (type *)c->table_bU[U];
688

    
689
#define YSCALE_YUV_2_PACKED2_C(type,alpha)   \
690
    for (i=0; i<(dstW>>1); i++) { \
691
        const int i2= 2*i;       \
692
        int Y1= (buf0[i2  ]*yalpha1+buf1[i2  ]*yalpha)>>19;           \
693
        int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19;           \
694
        int U= (uvbuf0[i     ]*uvalpha1+uvbuf1[i     ]*uvalpha)>>19;  \
695
        int V= (uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19;  \
696
        type av_unused *r, *b, *g;                                    \
697
        int av_unused A1, A2;                                         \
698
        if (alpha) {\
699
            A1= (abuf0[i2  ]*yalpha1+abuf1[i2  ]*yalpha)>>19;         \
700
            A2= (abuf0[i2+1]*yalpha1+abuf1[i2+1]*yalpha)>>19;         \
701
        }
702

    
703
#define YSCALE_YUV_2_GRAY16_2_C   \
704
    for (i=0; i<(dstW>>1); i++) { \
705
        const int i2= 2*i;       \
706
        int Y1= (buf0[i2  ]*yalpha1+buf1[i2  ]*yalpha)>>11;           \
707
        int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>11;
708

    
709
#define YSCALE_YUV_2_RGB2_C(type,alpha) \
710
    YSCALE_YUV_2_PACKED2_C(type,alpha)\
711
    r = (type *)c->table_rV[V];\
712
    g = (type *)(c->table_gU[U] + c->table_gV[V]);\
713
    b = (type *)c->table_bU[U];
714

    
715
#define YSCALE_YUV_2_PACKED1_C(type,alpha) \
716
    for (i=0; i<(dstW>>1); i++) {\
717
        const int i2= 2*i;\
718
        int Y1= buf0[i2  ]>>7;\
719
        int Y2= buf0[i2+1]>>7;\
720
        int U= (uvbuf1[i     ])>>7;\
721
        int V= (uvbuf1[i+VOFW])>>7;\
722
        type av_unused *r, *b, *g;\
723
        int av_unused A1, A2;\
724
        if (alpha) {\
725
            A1= abuf0[i2  ]>>7;\
726
            A2= abuf0[i2+1]>>7;\
727
        }
728

    
729
#define YSCALE_YUV_2_GRAY16_1_C \
730
    for (i=0; i<(dstW>>1); i++) {\
731
        const int i2= 2*i;\
732
        int Y1= buf0[i2  ]<<1;\
733
        int Y2= buf0[i2+1]<<1;
734

    
735
#define YSCALE_YUV_2_RGB1_C(type,alpha) \
736
    YSCALE_YUV_2_PACKED1_C(type,alpha)\
737
    r = (type *)c->table_rV[V];\
738
    g = (type *)(c->table_gU[U] + c->table_gV[V]);\
739
    b = (type *)c->table_bU[U];
740

    
741
#define YSCALE_YUV_2_PACKED1B_C(type,alpha) \
742
    for (i=0; i<(dstW>>1); i++) {\
743
        const int i2= 2*i;\
744
        int Y1= buf0[i2  ]>>7;\
745
        int Y2= buf0[i2+1]>>7;\
746
        int U= (uvbuf0[i     ] + uvbuf1[i     ])>>8;\
747
        int V= (uvbuf0[i+VOFW] + uvbuf1[i+VOFW])>>8;\
748
        type av_unused *r, *b, *g;\
749
        int av_unused A1, A2;\
750
        if (alpha) {\
751
            A1= abuf0[i2  ]>>7;\
752
            A2= abuf0[i2+1]>>7;\
753
        }
754

    
755
#define YSCALE_YUV_2_RGB1B_C(type,alpha) \
756
    YSCALE_YUV_2_PACKED1B_C(type,alpha)\
757
    r = (type *)c->table_rV[V];\
758
    g = (type *)(c->table_gU[U] + c->table_gV[V]);\
759
    b = (type *)c->table_bU[U];
760

    
761
#define YSCALE_YUV_2_MONO2_C \
762
    const uint8_t * const d128=dither_8x8_220[y&7];\
763
    uint8_t *g= c->table_gU[128] + c->table_gV[128];\
764
    for (i=0; i<dstW-7; i+=8) {\
765
        int acc;\
766
        acc =       g[((buf0[i  ]*yalpha1+buf1[i  ]*yalpha)>>19) + d128[0]];\
767
        acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\
768
        acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\
769
        acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\
770
        acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\
771
        acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\
772
        acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\
773
        acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\
774
        ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
775
        dest++;\
776
    }
777

    
778
#define YSCALE_YUV_2_MONOX_C \
779
    const uint8_t * const d128=dither_8x8_220[y&7];\
780
    uint8_t *g= c->table_gU[128] + c->table_gV[128];\
781
    int acc=0;\
782
    for (i=0; i<dstW-1; i+=2) {\
783
        int j;\
784
        int Y1=1<<18;\
785
        int Y2=1<<18;\
786
\
787
        for (j=0; j<lumFilterSize; j++) {\
788
            Y1 += lumSrc[j][i] * lumFilter[j];\
789
            Y2 += lumSrc[j][i+1] * lumFilter[j];\
790
        }\
791
        Y1>>=19;\
792
        Y2>>=19;\
793
        if ((Y1|Y2)&256) {\
794
            if (Y1>255)   Y1=255;\
795
            else if (Y1<0)Y1=0;\
796
            if (Y2>255)   Y2=255;\
797
            else if (Y2<0)Y2=0;\
798
        }\
799
        acc+= acc + g[Y1+d128[(i+0)&7]];\
800
        acc+= acc + g[Y2+d128[(i+1)&7]];\
801
        if ((i&7)==6) {\
802
            ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
803
            dest++;\
804
        }\
805
    }
806

    
807
#define YSCALE_YUV_2_ANYRGB_C(func, func2, func_g16, func_monoblack)\
808
    switch(c->dstFormat) {\
809
    case PIX_FMT_RGB48BE:\
810
    case PIX_FMT_RGB48LE:\
811
        func(uint8_t,0)\
812
            ((uint8_t*)dest)[ 0]= r[Y1];\
813
            ((uint8_t*)dest)[ 1]= r[Y1];\
814
            ((uint8_t*)dest)[ 2]= g[Y1];\
815
            ((uint8_t*)dest)[ 3]= g[Y1];\
816
            ((uint8_t*)dest)[ 4]= b[Y1];\
817
            ((uint8_t*)dest)[ 5]= b[Y1];\
818
            ((uint8_t*)dest)[ 6]= r[Y2];\
819
            ((uint8_t*)dest)[ 7]= r[Y2];\
820
            ((uint8_t*)dest)[ 8]= g[Y2];\
821
            ((uint8_t*)dest)[ 9]= g[Y2];\
822
            ((uint8_t*)dest)[10]= b[Y2];\
823
            ((uint8_t*)dest)[11]= b[Y2];\
824
            dest+=12;\
825
        }\
826
        break;\
827
    case PIX_FMT_BGR48BE:\
828
    case PIX_FMT_BGR48LE:\
829
        func(uint8_t,0)\
830
            ((uint8_t*)dest)[ 0] = ((uint8_t*)dest)[ 1] = b[Y1];\
831
            ((uint8_t*)dest)[ 2] = ((uint8_t*)dest)[ 3] = g[Y1];\
832
            ((uint8_t*)dest)[ 4] = ((uint8_t*)dest)[ 5] = r[Y1];\
833
            ((uint8_t*)dest)[ 6] = ((uint8_t*)dest)[ 7] = b[Y2];\
834
            ((uint8_t*)dest)[ 8] = ((uint8_t*)dest)[ 9] = g[Y2];\
835
            ((uint8_t*)dest)[10] = ((uint8_t*)dest)[11] = r[Y2];\
836
            dest+=12;\
837
        }\
838
        break;\
839
    case PIX_FMT_RGBA:\
840
    case PIX_FMT_BGRA:\
841
        if (CONFIG_SMALL) {\
842
            int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
843
            func(uint32_t,needAlpha)\
844
                ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? (A1<<24) : 0);\
845
                ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? (A2<<24) : 0);\
846
            }\
847
        } else {\
848
            if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
849
                func(uint32_t,1)\
850
                    ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (A1<<24);\
851
                    ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (A2<<24);\
852
                }\
853
            } else {\
854
                func(uint32_t,0)\
855
                    ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
856
                    ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
857
                }\
858
            }\
859
        }\
860
        break;\
861
    case PIX_FMT_ARGB:\
862
    case PIX_FMT_ABGR:\
863
        if (CONFIG_SMALL) {\
864
            int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
865
            func(uint32_t,needAlpha)\
866
                ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? A1 : 0);\
867
                ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? A2 : 0);\
868
            }\
869
        } else {\
870
            if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
871
                func(uint32_t,1)\
872
                    ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + A1;\
873
                    ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + A2;\
874
                }\
875
            } else {\
876
                func(uint32_t,0)\
877
                    ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
878
                    ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
879
                }\
880
            }\
881
        }                \
882
        break;\
883
    case PIX_FMT_RGB24:\
884
        func(uint8_t,0)\
885
            ((uint8_t*)dest)[0]= r[Y1];\
886
            ((uint8_t*)dest)[1]= g[Y1];\
887
            ((uint8_t*)dest)[2]= b[Y1];\
888
            ((uint8_t*)dest)[3]= r[Y2];\
889
            ((uint8_t*)dest)[4]= g[Y2];\
890
            ((uint8_t*)dest)[5]= b[Y2];\
891
            dest+=6;\
892
        }\
893
        break;\
894
    case PIX_FMT_BGR24:\
895
        func(uint8_t,0)\
896
            ((uint8_t*)dest)[0]= b[Y1];\
897
            ((uint8_t*)dest)[1]= g[Y1];\
898
            ((uint8_t*)dest)[2]= r[Y1];\
899
            ((uint8_t*)dest)[3]= b[Y2];\
900
            ((uint8_t*)dest)[4]= g[Y2];\
901
            ((uint8_t*)dest)[5]= r[Y2];\
902
            dest+=6;\
903
        }\
904
        break;\
905
    case PIX_FMT_RGB565BE:\
906
    case PIX_FMT_RGB565LE:\
907
    case PIX_FMT_BGR565BE:\
908
    case PIX_FMT_BGR565LE:\
909
        {\
910
            const int dr1= dither_2x2_8[y&1    ][0];\
911
            const int dg1= dither_2x2_4[y&1    ][0];\
912
            const int db1= dither_2x2_8[(y&1)^1][0];\
913
            const int dr2= dither_2x2_8[y&1    ][1];\
914
            const int dg2= dither_2x2_4[y&1    ][1];\
915
            const int db2= dither_2x2_8[(y&1)^1][1];\
916
            func(uint16_t,0)\
917
                ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
918
                ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
919
            }\
920
        }\
921
        break;\
922
    case PIX_FMT_RGB555BE:\
923
    case PIX_FMT_RGB555LE:\
924
    case PIX_FMT_BGR555BE:\
925
    case PIX_FMT_BGR555LE:\
926
        {\
927
            const int dr1= dither_2x2_8[y&1    ][0];\
928
            const int dg1= dither_2x2_8[y&1    ][1];\
929
            const int db1= dither_2x2_8[(y&1)^1][0];\
930
            const int dr2= dither_2x2_8[y&1    ][1];\
931
            const int dg2= dither_2x2_8[y&1    ][0];\
932
            const int db2= dither_2x2_8[(y&1)^1][1];\
933
            func(uint16_t,0)\
934
                ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
935
                ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
936
            }\
937
        }\
938
        break;\
939
    case PIX_FMT_RGB444BE:\
940
    case PIX_FMT_RGB444LE:\
941
    case PIX_FMT_BGR444BE:\
942
    case PIX_FMT_BGR444LE:\
943
        {\
944
            const int dr1= dither_4x4_16[y&3    ][0];\
945
            const int dg1= dither_4x4_16[y&3    ][1];\
946
            const int db1= dither_4x4_16[(y&3)^3][0];\
947
            const int dr2= dither_4x4_16[y&3    ][1];\
948
            const int dg2= dither_4x4_16[y&3    ][0];\
949
            const int db2= dither_4x4_16[(y&3)^3][1];\
950
            func(uint16_t,0)\
951
                ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
952
                ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
953
            }\
954
        }\
955
        break;\
956
    case PIX_FMT_RGB8:\
957
    case PIX_FMT_BGR8:\
958
        {\
959
            const uint8_t * const d64= dither_8x8_73[y&7];\
960
            const uint8_t * const d32= dither_8x8_32[y&7];\
961
            func(uint8_t,0)\
962
                ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\
963
                ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\
964
            }\
965
        }\
966
        break;\
967
    case PIX_FMT_RGB4:\
968
    case PIX_FMT_BGR4:\
969
        {\
970
            const uint8_t * const d64= dither_8x8_73 [y&7];\
971
            const uint8_t * const d128=dither_8x8_220[y&7];\
972
            func(uint8_t,0)\
973
                ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\
974
                                 + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\
975
            }\
976
        }\
977
        break;\
978
    case PIX_FMT_RGB4_BYTE:\
979
    case PIX_FMT_BGR4_BYTE:\
980
        {\
981
            const uint8_t * const d64= dither_8x8_73 [y&7];\
982
            const uint8_t * const d128=dither_8x8_220[y&7];\
983
            func(uint8_t,0)\
984
                ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\
985
                ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\
986
            }\
987
        }\
988
        break;\
989
    case PIX_FMT_MONOBLACK:\
990
    case PIX_FMT_MONOWHITE:\
991
        {\
992
            func_monoblack\
993
        }\
994
        break;\
995
    case PIX_FMT_YUYV422:\
996
        func2\
997
            ((uint8_t*)dest)[2*i2+0]= Y1;\
998
            ((uint8_t*)dest)[2*i2+1]= U;\
999
            ((uint8_t*)dest)[2*i2+2]= Y2;\
1000
            ((uint8_t*)dest)[2*i2+3]= V;\
1001
        }                \
1002
        break;\
1003
    case PIX_FMT_UYVY422:\
1004
        func2\
1005
            ((uint8_t*)dest)[2*i2+0]= U;\
1006
            ((uint8_t*)dest)[2*i2+1]= Y1;\
1007
            ((uint8_t*)dest)[2*i2+2]= V;\
1008
            ((uint8_t*)dest)[2*i2+3]= Y2;\
1009
        }                \
1010
        break;\
1011
    case PIX_FMT_GRAY16BE:\
1012
        func_g16\
1013
            ((uint8_t*)dest)[2*i2+0]= Y1>>8;\
1014
            ((uint8_t*)dest)[2*i2+1]= Y1;\
1015
            ((uint8_t*)dest)[2*i2+2]= Y2>>8;\
1016
            ((uint8_t*)dest)[2*i2+3]= Y2;\
1017
        }                \
1018
        break;\
1019
    case PIX_FMT_GRAY16LE:\
1020
        func_g16\
1021
            ((uint8_t*)dest)[2*i2+0]= Y1;\
1022
            ((uint8_t*)dest)[2*i2+1]= Y1>>8;\
1023
            ((uint8_t*)dest)[2*i2+2]= Y2;\
1024
            ((uint8_t*)dest)[2*i2+3]= Y2>>8;\
1025
        }                \
1026
        break;\
1027
    }
1028

    
1029
static inline void yuv2packedXinC(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
1030
                                  const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
1031
                                  const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1032
{
1033
    int i;
1034
    YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGBX_C, YSCALE_YUV_2_PACKEDX_C(void,0), YSCALE_YUV_2_GRAY16_C, YSCALE_YUV_2_MONOX_C)
1035
}
1036

    
1037
static inline void yuv2rgbXinC_full(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
1038
                                    const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
1039
                                    const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1040
{
1041
    int i;
1042
    int step= c->dstFormatBpp/8;
1043
    int aidx= 3;
1044

    
1045
    switch(c->dstFormat) {
1046
    case PIX_FMT_ARGB:
1047
        dest++;
1048
        aidx= 0;
1049
    case PIX_FMT_RGB24:
1050
        aidx--;
1051
    case PIX_FMT_RGBA:
1052
        if (CONFIG_SMALL) {
1053
            int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
1054
            YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
1055
                dest[aidx]= needAlpha ? A : 255;
1056
                dest[0]= R>>22;
1057
                dest[1]= G>>22;
1058
                dest[2]= B>>22;
1059
                dest+= step;
1060
            }
1061
        } else {
1062
            if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1063
                YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
1064
                    dest[aidx]= A;
1065
                    dest[0]= R>>22;
1066
                    dest[1]= G>>22;
1067
                    dest[2]= B>>22;
1068
                    dest+= step;
1069
                }
1070
            } else {
1071
                YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
1072
                    dest[aidx]= 255;
1073
                    dest[0]= R>>22;
1074
                    dest[1]= G>>22;
1075
                    dest[2]= B>>22;
1076
                    dest+= step;
1077
                }
1078
            }
1079
        }
1080
        break;
1081
    case PIX_FMT_ABGR:
1082
        dest++;
1083
        aidx= 0;
1084
    case PIX_FMT_BGR24:
1085
        aidx--;
1086
    case PIX_FMT_BGRA:
1087
        if (CONFIG_SMALL) {
1088
            int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
1089
            YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
1090
                dest[aidx]= needAlpha ? A : 255;
1091
                dest[0]= B>>22;
1092
                dest[1]= G>>22;
1093
                dest[2]= R>>22;
1094
                dest+= step;
1095
            }
1096
        } else {
1097
            if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1098
                YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
1099
                    dest[aidx]= A;
1100
                    dest[0]= B>>22;
1101
                    dest[1]= G>>22;
1102
                    dest[2]= R>>22;
1103
                    dest+= step;
1104
                }
1105
            } else {
1106
                YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
1107
                    dest[aidx]= 255;
1108
                    dest[0]= B>>22;
1109
                    dest[1]= G>>22;
1110
                    dest[2]= R>>22;
1111
                    dest+= step;
1112
                }
1113
            }
1114
        }
1115
        break;
1116
    default:
1117
        assert(0);
1118
    }
1119
}
1120

    
1121
static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)
1122
{
1123
    int i;
1124
    uint8_t *ptr = plane + stride*y;
1125
    for (i=0; i<height; i++) {
1126
        memset(ptr, val, width);
1127
        ptr += stride;
1128
    }
1129
}
1130

    
1131
static inline void rgb48ToY(uint8_t *dst, const uint8_t *src, long width,
1132
                            uint32_t *unused)
1133
{
1134
    int i;
1135
    for (i = 0; i < width; i++) {
1136
        int r = src[i*6+0];
1137
        int g = src[i*6+2];
1138
        int b = src[i*6+4];
1139

    
1140
        dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1141
    }
1142
}
1143

    
1144
static inline void rgb48ToUV(uint8_t *dstU, uint8_t *dstV,
1145
                             const uint8_t *src1, const uint8_t *src2,
1146
                             long width, uint32_t *unused)
1147
{
1148
    int i;
1149
    assert(src1==src2);
1150
    for (i = 0; i < width; i++) {
1151
        int r = src1[6*i + 0];
1152
        int g = src1[6*i + 2];
1153
        int b = src1[6*i + 4];
1154

    
1155
        dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1156
        dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1157
    }
1158
}
1159

    
1160
static inline void rgb48ToUV_half(uint8_t *dstU, uint8_t *dstV,
1161
                                  const uint8_t *src1, const uint8_t *src2,
1162
                                  long width, uint32_t *unused)
1163
{
1164
    int i;
1165
    assert(src1==src2);
1166
    for (i = 0; i < width; i++) {
1167
        int r= src1[12*i + 0] + src1[12*i + 6];
1168
        int g= src1[12*i + 2] + src1[12*i + 8];
1169
        int b= src1[12*i + 4] + src1[12*i + 10];
1170

    
1171
        dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1172
        dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1173
    }
1174
}
1175

    
1176
static inline void bgr48ToY(uint8_t *dst, const uint8_t *src, long width,
1177
                            uint32_t *unused)
1178
{
1179
    int i;
1180
    for (i = 0; i < width; i++) {
1181
        int b = src[i*6+0];
1182
        int g = src[i*6+2];
1183
        int r = src[i*6+4];
1184

    
1185
        dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1186
    }
1187
}
1188

    
1189
static inline void bgr48ToUV(uint8_t *dstU, uint8_t *dstV,
1190
                             const uint8_t *src1, const uint8_t *src2,
1191
                             long width, uint32_t *unused)
1192
{
1193
    int i;
1194
    for (i = 0; i < width; i++) {
1195
        int b = src1[6*i + 0];
1196
        int g = src1[6*i + 2];
1197
        int r = src1[6*i + 4];
1198

    
1199
        dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1200
        dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1201
    }
1202
}
1203

    
1204
static inline void bgr48ToUV_half(uint8_t *dstU, uint8_t *dstV,
1205
                                  const uint8_t *src1, const uint8_t *src2,
1206
                                  long width, uint32_t *unused)
1207
{
1208
    int i;
1209
    for (i = 0; i < width; i++) {
1210
        int b= src1[12*i + 0] + src1[12*i + 6];
1211
        int g= src1[12*i + 2] + src1[12*i + 8];
1212
        int r= src1[12*i + 4] + src1[12*i + 10];
1213

    
1214
        dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1215
        dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1216
    }
1217
}
1218

    
1219
#define BGR2Y(type, name, shr, shg, shb, maskr, maskg, maskb, RY, GY, BY, S)\
1220
static inline void name(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)\
1221
{\
1222
    int i;\
1223
    for (i=0; i<width; i++) {\
1224
        int b= (((const type*)src)[i]>>shb)&maskb;\
1225
        int g= (((const type*)src)[i]>>shg)&maskg;\
1226
        int r= (((const type*)src)[i]>>shr)&maskr;\
1227
\
1228
        dst[i]= (((RY)*r + (GY)*g + (BY)*b + (33<<((S)-1)))>>(S));\
1229
    }\
1230
}
1231

    
1232
BGR2Y(uint32_t, bgr32ToY,16, 0, 0, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)
1233
BGR2Y(uint32_t,bgr321ToY,16,16, 0, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)
1234
BGR2Y(uint32_t, rgb32ToY, 0, 0,16, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)
1235
BGR2Y(uint32_t,rgb321ToY, 0,16,16, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)
1236
BGR2Y(uint16_t, bgr16ToY, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RY<<11, GY<<5, BY    , RGB2YUV_SHIFT+8)
1237
BGR2Y(uint16_t, bgr15ToY, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RY<<10, GY<<5, BY    , RGB2YUV_SHIFT+7)
1238
BGR2Y(uint16_t, rgb16ToY, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RY    , GY<<5, BY<<11, RGB2YUV_SHIFT+8)
1239
BGR2Y(uint16_t, rgb15ToY, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RY    , GY<<5, BY<<10, RGB2YUV_SHIFT+7)
1240

    
1241
static inline void abgrToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1242
{
1243
    int i;
1244
    for (i=0; i<width; i++) {
1245
        dst[i]= src[4*i];
1246
    }
1247
}
1248

    
1249
#define BGR2UV(type, name, shr, shg, shb, shp, maskr, maskg, maskb, RU, GU, BU, RV, GV, BV, S) \
1250
static inline void name(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\
1251
{\
1252
    int i;\
1253
    for (i=0; i<width; i++) {\
1254
        int b= ((((const type*)src)[i]>>shp)&maskb)>>shb;\
1255
        int g= ((((const type*)src)[i]>>shp)&maskg)>>shg;\
1256
        int r= ((((const type*)src)[i]>>shp)&maskr)>>shr;\
1257
\
1258
        dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<((S)-1)))>>(S);\
1259
        dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<((S)-1)))>>(S);\
1260
    }\
1261
}\
1262
static inline void name ## _half(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\
1263
{\
1264
    int i;\
1265
    for (i=0; i<width; i++) {\
1266
        int pix0= ((const type*)src)[2*i+0]>>shp;\
1267
        int pix1= ((const type*)src)[2*i+1]>>shp;\
1268
        int g= (pix0&~(maskr|maskb))+(pix1&~(maskr|maskb));\
1269
        int b= ((pix0+pix1-g)&(maskb|(2*maskb)))>>shb;\
1270
        int r= ((pix0+pix1-g)&(maskr|(2*maskr)))>>shr;\
1271
        g&= maskg|(2*maskg);\
1272
\
1273
        g>>=shg;\
1274
\
1275
        dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<(S)))>>((S)+1);\
1276
        dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<(S)))>>((S)+1);\
1277
    }\
1278
}
1279

    
1280
BGR2UV(uint32_t, bgr32ToUV,16, 0, 0, 0, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)
1281
BGR2UV(uint32_t,bgr321ToUV,16, 0, 0, 8, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)
1282
BGR2UV(uint32_t, rgb32ToUV, 0, 0,16, 0,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)
1283
BGR2UV(uint32_t,rgb321ToUV, 0, 0,16, 8,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)
1284
BGR2UV(uint16_t, bgr16ToUV, 0, 0, 0, 0,   0x001F, 0x07E0,   0xF800, RU<<11, GU<<5, BU    , RV<<11, GV<<5, BV    , RGB2YUV_SHIFT+8)
1285
BGR2UV(uint16_t, bgr15ToUV, 0, 0, 0, 0,   0x001F, 0x03E0,   0x7C00, RU<<10, GU<<5, BU    , RV<<10, GV<<5, BV    , RGB2YUV_SHIFT+7)
1286
BGR2UV(uint16_t, rgb16ToUV, 0, 0, 0, 0,   0xF800, 0x07E0,   0x001F, RU    , GU<<5, BU<<11, RV    , GV<<5, BV<<11, RGB2YUV_SHIFT+8)
1287
BGR2UV(uint16_t, rgb15ToUV, 0, 0, 0, 0,   0x7C00, 0x03E0,   0x001F, RU    , GU<<5, BU<<10, RV    , GV<<5, BV<<10, RGB2YUV_SHIFT+7)
1288

    
1289
static inline void palToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)
1290
{
1291
    int i;
1292
    for (i=0; i<width; i++) {
1293
        int d= src[i];
1294

    
1295
        dst[i]= pal[d] >> 24;
1296
    }
1297
}
1298

    
1299
static inline void palToY(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)
1300
{
1301
    int i;
1302
    for (i=0; i<width; i++) {
1303
        int d= src[i];
1304

    
1305
        dst[i]= pal[d] & 0xFF;
1306
    }
1307
}
1308

    
1309
static inline void palToUV(uint8_t *dstU, uint8_t *dstV,
1310
                           const uint8_t *src1, const uint8_t *src2,
1311
                           long width, uint32_t *pal)
1312
{
1313
    int i;
1314
    assert(src1 == src2);
1315
    for (i=0; i<width; i++) {
1316
        int p= pal[src1[i]];
1317

    
1318
        dstU[i]= p>>8;
1319
        dstV[i]= p>>16;
1320
    }
1321
}
1322

    
1323
static inline void monowhite2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1324
{
1325
    int i, j;
1326
    for (i=0; i<width/8; i++) {
1327
        int d= ~src[i];
1328
        for(j=0; j<8; j++)
1329
            dst[8*i+j]= ((d>>(7-j))&1)*255;
1330
    }
1331
}
1332

    
1333
static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1334
{
1335
    int i, j;
1336
    for (i=0; i<width/8; i++) {
1337
        int d= src[i];
1338
        for(j=0; j<8; j++)
1339
            dst[8*i+j]= ((d>>(7-j))&1)*255;
1340
    }
1341
}
1342

    
1343
//Note: we have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW+MMX2 one
1344
//Plain C versions
1345
#if CONFIG_RUNTIME_CPUDETECT
1346
#  define COMPILE_C 1
1347
#  if   ARCH_X86
1348
#    define COMPILE_MMX     1
1349
#    define COMPILE_MMX2    1
1350
#    define COMPILE_3DNOW   1
1351
#  elif ARCH_PPC
1352
#    define COMPILE_ALTIVEC HAVE_ALTIVEC
1353
#  endif
1354
#else /* CONFIG_RUNTIME_CPUDETECT */
1355
#  if   ARCH_X86
1356
#    if   HAVE_MMX2
1357
#      define COMPILE_MMX2  1
1358
#    elif HAVE_AMD3DNOW
1359
#      define COMPILE_3DNOW 1
1360
#    elif HAVE_MMX
1361
#      define COMPILE_MMX   1
1362
#    else
1363
#      define COMPILE_C     1
1364
#    endif
1365
#  elif ARCH_PPC && HAVE_ALTIVEC
1366
#    define COMPILE_ALTIVEC 1
1367
#  else
1368
#    define COMPILE_C       1
1369
#  endif
1370
#endif
1371

    
1372
#ifndef COMPILE_C
1373
#  define COMPILE_C 0
1374
#endif
1375
#ifndef COMPILE_MMX
1376
#  define COMPILE_MMX 0
1377
#endif
1378
#ifndef COMPILE_MMX2
1379
#  define COMPILE_MMX2 0
1380
#endif
1381
#ifndef COMPILE_3DNOW
1382
#  define COMPILE_3DNOW 0
1383
#endif
1384
#ifndef COMPILE_ALTIVEC
1385
#  define COMPILE_ALTIVEC 0
1386
#endif
1387

    
1388
#define COMPILE_TEMPLATE_MMX 0
1389
#define COMPILE_TEMPLATE_MMX2 0
1390
#define COMPILE_TEMPLATE_AMD3DNOW 0
1391
#define COMPILE_TEMPLATE_ALTIVEC 0
1392

    
1393
#if COMPILE_C
1394
#define RENAME(a) a ## _C
1395
#include "swscale_template.c"
1396
#endif
1397

    
1398
#if COMPILE_ALTIVEC
1399
#undef RENAME
1400
#undef COMPILE_TEMPLATE_ALTIVEC
1401
#define COMPILE_TEMPLATE_ALTIVEC 1
1402
#define RENAME(a) a ## _altivec
1403
#include "swscale_template.c"
1404
#endif
1405

    
1406
#if ARCH_X86
1407

    
1408
//MMX versions
1409
#if COMPILE_MMX
1410
#undef RENAME
1411
#undef COMPILE_TEMPLATE_MMX
1412
#undef COMPILE_TEMPLATE_MMX2
1413
#undef COMPILE_TEMPLATE_AMD3DNOW
1414
#define COMPILE_TEMPLATE_MMX 1
1415
#define COMPILE_TEMPLATE_MMX2 0
1416
#define COMPILE_TEMPLATE_AMD3DNOW 0
1417
#define RENAME(a) a ## _MMX
1418
#include "swscale_template.c"
1419
#endif
1420

    
1421
//MMX2 versions
1422
#if COMPILE_MMX2
1423
#undef RENAME
1424
#undef COMPILE_TEMPLATE_MMX
1425
#undef COMPILE_TEMPLATE_MMX2
1426
#undef COMPILE_TEMPLATE_AMD3DNOW
1427
#define COMPILE_TEMPLATE_MMX 1
1428
#define COMPILE_TEMPLATE_MMX2 1
1429
#define COMPILE_TEMPLATE_AMD3DNOW 0
1430
#define RENAME(a) a ## _MMX2
1431
#include "swscale_template.c"
1432
#endif
1433

    
1434
//3DNOW versions
1435
#if COMPILE_3DNOW
1436
#undef RENAME
1437
#undef COMPILE_TEMPLATE_MMX
1438
#undef COMPILE_TEMPLATE_MMX2
1439
#undef COMPILE_TEMPLATE_AMD3DNOW
1440
#define COMPILE_TEMPLATE_MMX 1
1441
#define COMPILE_TEMPLATE_MMX2 0
1442
#define COMPILE_TEMPLATE_AMD3DNOW 1
1443
#define RENAME(a) a ## _3DNow
1444
#include "swscale_template.c"
1445
#endif
1446

    
1447
#endif //ARCH_X86
1448

    
1449
SwsFunc ff_getSwsFunc(SwsContext *c)
1450
{
1451
#if CONFIG_RUNTIME_CPUDETECT
1452
    int flags = c->flags;
1453

    
1454
#if ARCH_X86
1455
    // ordered per speed fastest first
1456
    if (flags & SWS_CPU_CAPS_MMX2) {
1457
        sws_init_swScale_MMX2(c);
1458
        return swScale_MMX2;
1459
    } else if (flags & SWS_CPU_CAPS_3DNOW) {
1460
        sws_init_swScale_3DNow(c);
1461
        return swScale_3DNow;
1462
    } else if (flags & SWS_CPU_CAPS_MMX) {
1463
        sws_init_swScale_MMX(c);
1464
        return swScale_MMX;
1465
    } else {
1466
        sws_init_swScale_C(c);
1467
        return swScale_C;
1468
    }
1469

    
1470
#else
1471
#if COMPILE_ALTIVEC
1472
    if (flags & SWS_CPU_CAPS_ALTIVEC) {
1473
        sws_init_swScale_altivec(c);
1474
        return swScale_altivec;
1475
    } else {
1476
        sws_init_swScale_C(c);
1477
        return swScale_C;
1478
    }
1479
#endif
1480
    sws_init_swScale_C(c);
1481
    return swScale_C;
1482
#endif /* ARCH_X86 */
1483
#else //CONFIG_RUNTIME_CPUDETECT
1484
#if   COMPILE_TEMPLATE_MMX2
1485
    sws_init_swScale_MMX2(c);
1486
    return swScale_MMX2;
1487
#elif COMPILE_TEMPLATE_AMD3DNOW
1488
    sws_init_swScale_3DNow(c);
1489
    return swScale_3DNow;
1490
#elif COMPILE_TEMPLATE_MMX
1491
    sws_init_swScale_MMX(c);
1492
    return swScale_MMX;
1493
#elif COMPILE_TEMPLATE_ALTIVEC
1494
    sws_init_swScale_altivec(c);
1495
    return swScale_altivec;
1496
#else
1497
    sws_init_swScale_C(c);
1498
    return swScale_C;
1499
#endif
1500
#endif //!CONFIG_RUNTIME_CPUDETECT
1501
}
1502

    
1503
static void copyPlane(const uint8_t *src, int srcStride,
1504
                      int srcSliceY, int srcSliceH, int width,
1505
                      uint8_t *dst, int dstStride)
1506
{
1507
    dst += dstStride * srcSliceY;
1508
    if (dstStride == srcStride && srcStride > 0) {
1509
        memcpy(dst, src, srcSliceH * dstStride);
1510
    } else {
1511
        int i;
1512
        for (i=0; i<srcSliceH; i++) {
1513
            memcpy(dst, src, width);
1514
            src += srcStride;
1515
            dst += dstStride;
1516
        }
1517
    }
1518
}
1519

    
1520
static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1521
                               int srcSliceH, uint8_t* dstParam[], int dstStride[])
1522
{
1523
    uint8_t *dst = dstParam[1] + dstStride[1]*srcSliceY/2;
1524

    
1525
    copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
1526
              dstParam[0], dstStride[0]);
1527

    
1528
    if (c->dstFormat == PIX_FMT_NV12)
1529
        interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);
1530
    else
1531
        interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);
1532

    
1533
    return srcSliceH;
1534
}
1535

    
1536
static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1537
                               int srcSliceH, uint8_t* dstParam[], int dstStride[])
1538
{
1539
    uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1540

    
1541
    yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
1542

    
1543
    return srcSliceH;
1544
}
1545

    
1546
static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1547
                               int srcSliceH, uint8_t* dstParam[], int dstStride[])
1548
{
1549
    uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1550

    
1551
    yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
1552

    
1553
    return srcSliceH;
1554
}
1555

    
1556
static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1557
                                int srcSliceH, uint8_t* dstParam[], int dstStride[])
1558
{
1559
    uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1560

    
1561
    yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1562

    
1563
    return srcSliceH;
1564
}
1565

    
1566
static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1567
                                int srcSliceH, uint8_t* dstParam[], int dstStride[])
1568
{
1569
    uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1570

    
1571
    yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1572

    
1573
    return srcSliceH;
1574
}
1575

    
1576
static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1577
                               int srcSliceH, uint8_t* dstParam[], int dstStride[])
1578
{
1579
    uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1580
    uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1581
    uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1582

    
1583
    yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1584

    
1585
    if (dstParam[3])
1586
        fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1587

    
1588
    return srcSliceH;
1589
}
1590

    
1591
static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1592
                               int srcSliceH, uint8_t* dstParam[], int dstStride[])
1593
{
1594
    uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1595
    uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
1596
    uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
1597

    
1598
    yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1599

    
1600
    return srcSliceH;
1601
}
1602

    
1603
static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1604
                               int srcSliceH, uint8_t* dstParam[], int dstStride[])
1605
{
1606
    uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1607
    uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1608
    uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1609

    
1610
    uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1611

    
1612
    if (dstParam[3])
1613
        fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1614

    
1615
    return srcSliceH;
1616
}
1617

    
1618
static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1619
                               int srcSliceH, uint8_t* dstParam[], int dstStride[])
1620
{
1621
    uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1622
    uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
1623
    uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
1624

    
1625
    uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1626

    
1627
    return srcSliceH;
1628
}
1629

    
1630
static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1631
{
1632
    long i;
1633
    for (i=0; i<num_pixels; i++)
1634
        ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | (src[(i<<1)+1] << 24);
1635
}
1636

    
1637
static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1638
{
1639
    long i;
1640

    
1641
    for (i=0; i<num_pixels; i++)
1642
        ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | src[(i<<1)+1];
1643
}
1644

    
1645
static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1646
{
1647
    long i;
1648

    
1649
    for (i=0; i<num_pixels; i++) {
1650
        //FIXME slow?
1651
        dst[0]= palette[src[i<<1]*4+0];
1652
        dst[1]= palette[src[i<<1]*4+1];
1653
        dst[2]= palette[src[i<<1]*4+2];
1654
        dst+= 3;
1655
    }
1656
}
1657

    
1658
static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1659
                           int srcSliceH, uint8_t* dst[], int dstStride[])
1660
{
1661
    const enum PixelFormat srcFormat= c->srcFormat;
1662
    const enum PixelFormat dstFormat= c->dstFormat;
1663
    void (*conv)(const uint8_t *src, uint8_t *dst, long num_pixels,
1664
                 const uint8_t *palette)=NULL;
1665
    int i;
1666
    uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1667
    const uint8_t *srcPtr= src[0];
1668

    
1669
    if (srcFormat == PIX_FMT_GRAY8A) {
1670
        switch (dstFormat) {
1671
        case PIX_FMT_RGB32  : conv = gray8aToPacked32; break;
1672
        case PIX_FMT_BGR32  : conv = gray8aToPacked32; break;
1673
        case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;
1674
        case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;
1675
        case PIX_FMT_RGB24  : conv = gray8aToPacked24; break;
1676
        case PIX_FMT_BGR24  : conv = gray8aToPacked24; break;
1677
        }
1678
    } else if (usePal(srcFormat)) {
1679
        switch (dstFormat) {
1680
        case PIX_FMT_RGB32  : conv = sws_convertPalette8ToPacked32; break;
1681
        case PIX_FMT_BGR32  : conv = sws_convertPalette8ToPacked32; break;
1682
        case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;
1683
        case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;
1684
        case PIX_FMT_RGB24  : conv = sws_convertPalette8ToPacked24; break;
1685
        case PIX_FMT_BGR24  : conv = sws_convertPalette8ToPacked24; break;
1686
        }
1687
    }
1688

    
1689
    if (!conv)
1690
        av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1691
               sws_format_name(srcFormat), sws_format_name(dstFormat));
1692
    else {
1693
        for (i=0; i<srcSliceH; i++) {
1694
            conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);
1695
            srcPtr+= srcStride[0];
1696
            dstPtr+= dstStride[0];
1697
        }
1698
    }
1699

    
1700
    return srcSliceH;
1701
}
1702

    
1703
#define isRGBA32(x) (            \
1704
           (x) == PIX_FMT_ARGB   \
1705
        || (x) == PIX_FMT_RGBA   \
1706
        || (x) == PIX_FMT_BGRA   \
1707
        || (x) == PIX_FMT_ABGR   \
1708
        )
1709

    
1710
/* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */
1711
static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1712
                           int srcSliceH, uint8_t* dst[], int dstStride[])
1713
{
1714
    const enum PixelFormat srcFormat= c->srcFormat;
1715
    const enum PixelFormat dstFormat= c->dstFormat;
1716
    const int srcBpp= (c->srcFormatBpp + 7) >> 3;
1717
    const int dstBpp= (c->dstFormatBpp + 7) >> 3;
1718
    const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */
1719
    const int dstId= c->dstFormatBpp >> 2;
1720
    void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL;
1721

    
1722
#define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)
1723

    
1724
    if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {
1725
        if (     CONV_IS(ABGR, RGBA)
1726
              || CONV_IS(ARGB, BGRA)
1727
              || CONV_IS(BGRA, ARGB)
1728
              || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;
1729
        else if (CONV_IS(ABGR, ARGB)
1730
              || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;
1731
        else if (CONV_IS(ABGR, BGRA)
1732
              || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;
1733
        else if (CONV_IS(BGRA, RGBA)
1734
              || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;
1735
        else if (CONV_IS(BGRA, ABGR)
1736
              || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;
1737
    } else
1738
    /* BGR -> BGR */
1739
    if (  (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))
1740
       || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {
1741
        switch(srcId | (dstId<<4)) {
1742
        case 0x34: conv= rgb16to15; break;
1743
        case 0x36: conv= rgb24to15; break;
1744
        case 0x38: conv= rgb32to15; break;
1745
        case 0x43: conv= rgb15to16; break;
1746
        case 0x46: conv= rgb24to16; break;
1747
        case 0x48: conv= rgb32to16; break;
1748
        case 0x63: conv= rgb15to24; break;
1749
        case 0x64: conv= rgb16to24; break;
1750
        case 0x68: conv= rgb32to24; break;
1751
        case 0x83: conv= rgb15to32; break;
1752
        case 0x84: conv= rgb16to32; break;
1753
        case 0x86: conv= rgb24to32; break;
1754
        }
1755
    } else if (  (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))
1756
             || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {
1757
        switch(srcId | (dstId<<4)) {
1758
        case 0x33: conv= rgb15tobgr15; break;
1759
        case 0x34: conv= rgb16tobgr15; break;
1760
        case 0x36: conv= rgb24tobgr15; break;
1761
        case 0x38: conv= rgb32tobgr15; break;
1762
        case 0x43: conv= rgb15tobgr16; break;
1763
        case 0x44: conv= rgb16tobgr16; break;
1764
        case 0x46: conv= rgb24tobgr16; break;
1765
        case 0x48: conv= rgb32tobgr16; break;
1766
        case 0x63: conv= rgb15tobgr24; break;
1767
        case 0x64: conv= rgb16tobgr24; break;
1768
        case 0x66: conv= rgb24tobgr24; break;
1769
        case 0x68: conv= rgb32tobgr24; break;
1770
        case 0x83: conv= rgb15tobgr32; break;
1771
        case 0x84: conv= rgb16tobgr32; break;
1772
        case 0x86: conv= rgb24tobgr32; break;
1773
        }
1774
    }
1775

    
1776
    if (!conv) {
1777
        av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1778
               sws_format_name(srcFormat), sws_format_name(dstFormat));
1779
    } else {
1780
        const uint8_t *srcPtr= src[0];
1781
              uint8_t *dstPtr= dst[0];
1782
        if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))
1783
            srcPtr += ALT32_CORR;
1784

    
1785
        if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))
1786
            dstPtr += ALT32_CORR;
1787

    
1788
        if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0 && !(srcStride[0]%srcBpp))
1789
            conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);
1790
        else {
1791
            int i;
1792
            dstPtr += dstStride[0]*srcSliceY;
1793

    
1794
            for (i=0; i<srcSliceH; i++) {
1795
                conv(srcPtr, dstPtr, c->srcW*srcBpp);
1796
                srcPtr+= srcStride[0];
1797
                dstPtr+= dstStride[0];
1798
            }
1799
        }
1800
    }
1801
    return srcSliceH;
1802
}
1803

    
1804
static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1805
                              int srcSliceH, uint8_t* dst[], int dstStride[])
1806
{
1807
    rgb24toyv12(
1808
        src[0],
1809
        dst[0]+ srcSliceY    *dstStride[0],
1810
        dst[1]+(srcSliceY>>1)*dstStride[1],
1811
        dst[2]+(srcSliceY>>1)*dstStride[2],
1812
        c->srcW, srcSliceH,
1813
        dstStride[0], dstStride[1], srcStride[0]);
1814
    if (dst[3])
1815
        fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1816
    return srcSliceH;
1817
}
1818

    
1819
static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1820
                             int srcSliceH, uint8_t* dst[], int dstStride[])
1821
{
1822
    copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
1823
              dst[0], dstStride[0]);
1824

    
1825
    planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,
1826
             srcSliceH >> 2, srcStride[1], dstStride[1]);
1827
    planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,
1828
             srcSliceH >> 2, srcStride[2], dstStride[2]);
1829
    if (dst[3])
1830
        fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1831
    return srcSliceH;
1832
}
1833

    
1834
/* unscaled copy like stuff (assumes nearly identical formats) */
1835
static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1836
                             int srcSliceH, uint8_t* dst[], int dstStride[])
1837
{
1838
    if (dstStride[0]==srcStride[0] && srcStride[0] > 0)
1839
        memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);
1840
    else {
1841
        int i;
1842
        const uint8_t *srcPtr= src[0];
1843
        uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1844
        int length=0;
1845

    
1846
        /* universal length finder */
1847
        while(length+c->srcW <= FFABS(dstStride[0])
1848
           && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;
1849
        assert(length!=0);
1850

    
1851
        for (i=0; i<srcSliceH; i++) {
1852
            memcpy(dstPtr, srcPtr, length);
1853
            srcPtr+= srcStride[0];
1854
            dstPtr+= dstStride[0];
1855
        }
1856
    }
1857
    return srcSliceH;
1858
}
1859

    
1860
#define DITHER_COPY(dst, dstStride, src, srcStride)\
1861
    for (i = 0; i < height; i++) {\
1862
        uint8_t *dither= dithers[src_depth-9][i&7];\
1863
        for (j = 0; j < length-7; j+=8){\
1864
            dst[j+0] = (src[j+0] + dither[0])>>(src_depth-dst_depth);\
1865
            dst[j+1] = (src[j+1] + dither[1])>>(src_depth-dst_depth);\
1866
            dst[j+2] = (src[j+2] + dither[2])>>(src_depth-dst_depth);\
1867
            dst[j+3] = (src[j+3] + dither[3])>>(src_depth-dst_depth);\
1868
            dst[j+4] = (src[j+4] + dither[4])>>(src_depth-dst_depth);\
1869
            dst[j+5] = (src[j+5] + dither[5])>>(src_depth-dst_depth);\
1870
            dst[j+6] = (src[j+6] + dither[6])>>(src_depth-dst_depth);\
1871
            dst[j+7] = (src[j+7] + dither[7])>>(src_depth-dst_depth);\
1872
        }\
1873
        for (; j < length; j++)\
1874
            dst[j] = (src[j] + dither[j&7])>>(src_depth-dst_depth);\
1875
        dst += dstStride;\
1876
        src += srcStride;\
1877
    }
1878

    
1879

    
1880
static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1881
                             int srcSliceH, uint8_t* dst[], int dstStride[])
1882
{
1883
    int plane, i, j;
1884
    for (plane=0; plane<4; plane++) {
1885
        int length= (plane==0 || plane==3) ? c->srcW  : -((-c->srcW  )>>c->chrDstHSubSample);
1886
        int y=      (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);
1887
        int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);
1888
        const uint8_t *srcPtr= src[plane];
1889
        uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;
1890

    
1891
        if (!dst[plane]) continue;
1892
        // ignore palette for GRAY8
1893
        if (plane == 1 && !dst[2]) continue;
1894
        if (!src[plane] || (plane == 1 && !src[2])) {
1895
            if(is16BPS(c->dstFormat))
1896
                length*=2;
1897
            fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);
1898
        } else {
1899
            if(isNBPS(c->srcFormat) || isNBPS(c->dstFormat)) {
1900
                const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1+1;
1901
                const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;
1902
                uint16_t *srcPtr2 = (uint16_t*)srcPtr;
1903
                uint16_t *dstPtr2 = (uint16_t*)dstPtr;
1904

    
1905
                if (dst_depth == 8) {
1906
                    DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2)
1907
                } else if (src_depth == 8) {
1908
                    for (i = 0; i < height; i++) {
1909
                        for (j = 0; j < length; j++)
1910
                            dstPtr2[j] = (srcPtr[j]<<(dst_depth-8)) |
1911
                                (srcPtr[j]>>(2*8-dst_depth));
1912
                        dstPtr2 += dstStride[plane]/2;
1913
                        srcPtr  += srcStride[plane];
1914
                    }
1915
                } else if (src_depth < dst_depth) {
1916
                    for (i = 0; i < height; i++) {
1917
                        if(isBE(c->dstFormat)){
1918
                            for (j = 0; j < length; j++)
1919
                                AV_WB16(&dstPtr2[j], (srcPtr2[j]<<(dst_depth-src_depth)) |
1920
                                                     (srcPtr2[j]>>(2*src_depth-dst_depth)));
1921
                        }else{
1922
                            for (j = 0; j < length; j++)
1923
                                AV_WL16(&dstPtr2[j], (srcPtr2[j]<<(dst_depth-src_depth)) |
1924
                                                     (srcPtr2[j]>>(2*src_depth-dst_depth)));
1925
                        }
1926
                        dstPtr2 += dstStride[plane]/2;
1927
                        srcPtr2 += srcStride[plane]/2;
1928
                    }
1929
                } else {
1930
                    DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2)
1931
                }
1932
            } else if(is16BPS(c->srcFormat) && !is16BPS(c->dstFormat)) {
1933
                //FIXME add dither
1934
                if (!isBE(c->srcFormat)) srcPtr++;
1935
                for (i=0; i<height; i++) {
1936
                    for (j=0; j<length; j++) dstPtr[j] = srcPtr[j<<1];
1937
                    srcPtr+= srcStride[plane];
1938
                    dstPtr+= dstStride[plane];
1939
                }
1940
            } else if(!is16BPS(c->srcFormat) && is16BPS(c->dstFormat)) {
1941
                for (i=0; i<height; i++) {
1942
                    for (j=0; j<length; j++) {
1943
                        dstPtr[ j<<1   ] = srcPtr[j];
1944
                        dstPtr[(j<<1)+1] = srcPtr[j];
1945
                    }
1946
                    srcPtr+= srcStride[plane];
1947
                    dstPtr+= dstStride[plane];
1948
                }
1949
            } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)
1950
                  && isBE(c->srcFormat) != isBE(c->dstFormat)) {
1951

    
1952
                for (i=0; i<height; i++) {
1953
                    for (j=0; j<length; j++)
1954
                        ((uint16_t*)dstPtr)[j] = av_bswap16(((const uint16_t*)srcPtr)[j]);
1955
                    srcPtr+= srcStride[plane];
1956
                    dstPtr+= dstStride[plane];
1957
                }
1958
            } else if (dstStride[plane] == srcStride[plane] &&
1959
                       srcStride[plane] > 0 && srcStride[plane] == length) {
1960
                memcpy(dst[plane] + dstStride[plane]*y, src[plane],
1961
                       height*dstStride[plane]);
1962
            } else {
1963
                if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))
1964
                    length*=2;
1965
                for (i=0; i<height; i++) {
1966
                    memcpy(dstPtr, srcPtr, length);
1967
                    srcPtr+= srcStride[plane];
1968
                    dstPtr+= dstStride[plane];
1969
                }
1970
            }
1971
        }
1972
    }
1973
    return srcSliceH;
1974
}
1975

    
1976
int ff_hardcodedcpuflags(void)
1977
{
1978
    int flags = 0;
1979
#if   COMPILE_TEMPLATE_MMX2
1980
    flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2;
1981
#elif COMPILE_TEMPLATE_AMD3DNOW
1982
    flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW;
1983
#elif COMPILE_TEMPLATE_MMX
1984
    flags |= SWS_CPU_CAPS_MMX;
1985
#elif COMPILE_TEMPLATE_ALTIVEC
1986
    flags |= SWS_CPU_CAPS_ALTIVEC;
1987
#elif ARCH_BFIN
1988
    flags |= SWS_CPU_CAPS_BFIN;
1989
#endif
1990
    return flags;
1991
}
1992

    
1993
void ff_get_unscaled_swscale(SwsContext *c)
1994
{
1995
    const enum PixelFormat srcFormat = c->srcFormat;
1996
    const enum PixelFormat dstFormat = c->dstFormat;
1997
    const int flags = c->flags;
1998
    const int dstH = c->dstH;
1999
    int needsDither;
2000

    
2001
    needsDither= isAnyRGB(dstFormat)
2002
        &&  c->dstFormatBpp < 24
2003
        && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));
2004

    
2005
    /* yv12_to_nv12 */
2006
    if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {
2007
        c->swScale= planarToNv12Wrapper;
2008
    }
2009
    /* yuv2bgr */
2010
    if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)
2011
        && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {
2012
        c->swScale= ff_yuv2rgb_get_func_ptr(c);
2013
    }
2014

    
2015
    if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {
2016
        c->swScale= yvu9ToYv12Wrapper;
2017
    }
2018

    
2019
    /* bgr24toYV12 */
2020
    if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))
2021
        c->swScale= bgr24ToYv12Wrapper;
2022

    
2023
    /* RGB/BGR -> RGB/BGR (no dither needed forms) */
2024
    if (   isAnyRGB(srcFormat)
2025
        && isAnyRGB(dstFormat)
2026
        && srcFormat != PIX_FMT_BGR8      && dstFormat != PIX_FMT_BGR8
2027
        && srcFormat != PIX_FMT_RGB8      && dstFormat != PIX_FMT_RGB8
2028
        && srcFormat != PIX_FMT_BGR4      && dstFormat != PIX_FMT_BGR4
2029
        && srcFormat != PIX_FMT_RGB4      && dstFormat != PIX_FMT_RGB4
2030
        && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE
2031
        && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE
2032
        && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK
2033
        && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE
2034
        && srcFormat != PIX_FMT_RGB48LE   && dstFormat != PIX_FMT_RGB48LE
2035
        && srcFormat != PIX_FMT_RGB48BE   && dstFormat != PIX_FMT_RGB48BE
2036
        && srcFormat != PIX_FMT_BGR48LE   && dstFormat != PIX_FMT_BGR48LE
2037
        && srcFormat != PIX_FMT_BGR48BE   && dstFormat != PIX_FMT_BGR48BE
2038
        && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))
2039
        c->swScale= rgbToRgbWrapper;
2040

    
2041
    if ((usePal(srcFormat) && (
2042
        dstFormat == PIX_FMT_RGB32   ||
2043
        dstFormat == PIX_FMT_RGB32_1 ||
2044
        dstFormat == PIX_FMT_RGB24   ||
2045
        dstFormat == PIX_FMT_BGR32   ||
2046
        dstFormat == PIX_FMT_BGR32_1 ||
2047
        dstFormat == PIX_FMT_BGR24)))
2048
        c->swScale= palToRgbWrapper;
2049

    
2050
    if (srcFormat == PIX_FMT_YUV422P) {
2051
        if (dstFormat == PIX_FMT_YUYV422)
2052
            c->swScale= yuv422pToYuy2Wrapper;
2053
        else if (dstFormat == PIX_FMT_UYVY422)
2054
            c->swScale= yuv422pToUyvyWrapper;
2055
    }
2056

    
2057
    /* LQ converters if -sws 0 or -sws 4*/
2058
    if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {
2059
        /* yv12_to_yuy2 */
2060
        if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {
2061
            if (dstFormat == PIX_FMT_YUYV422)
2062
                c->swScale= planarToYuy2Wrapper;
2063
            else if (dstFormat == PIX_FMT_UYVY422)
2064
                c->swScale= planarToUyvyWrapper;
2065
        }
2066
    }
2067
    if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
2068
        c->swScale= yuyvToYuv420Wrapper;
2069
    if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
2070
        c->swScale= uyvyToYuv420Wrapper;
2071
    if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)
2072
        c->swScale= yuyvToYuv422Wrapper;
2073
    if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)
2074
        c->swScale= uyvyToYuv422Wrapper;
2075

    
2076
#if COMPILE_ALTIVEC
2077
    if ((c->flags & SWS_CPU_CAPS_ALTIVEC) &&
2078
        !(c->flags & SWS_BITEXACT) &&
2079
        srcFormat == PIX_FMT_YUV420P) {
2080
        // unscaled YV12 -> packed YUV, we want speed
2081
        if (dstFormat == PIX_FMT_YUYV422)
2082
            c->swScale= yv12toyuy2_unscaled_altivec;
2083
        else if (dstFormat == PIX_FMT_UYVY422)
2084
            c->swScale= yv12touyvy_unscaled_altivec;
2085
    }
2086
#endif
2087

    
2088
    /* simple copy */
2089
    if (  srcFormat == dstFormat
2090
        || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)
2091
        || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)
2092
        || (isPlanarYUV(srcFormat) && isGray(dstFormat))
2093
        || (isPlanarYUV(dstFormat) && isGray(srcFormat))
2094
        || (isGray(dstFormat) && isGray(srcFormat))
2095
        || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)
2096
            && c->chrDstHSubSample == c->chrSrcHSubSample
2097
            && c->chrDstVSubSample == c->chrSrcVSubSample
2098
            && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21
2099
            && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))
2100
    {
2101
        if (isPacked(c->srcFormat))
2102
            c->swScale= packedCopyWrapper;
2103
        else /* Planar YUV or gray */
2104
            c->swScale= planarCopyWrapper;
2105
    }
2106
#if ARCH_BFIN
2107
    if (flags & SWS_CPU_CAPS_BFIN)
2108
        ff_bfin_get_unscaled_swscale (c);
2109
#endif
2110
}
2111

    
2112
static void reset_ptr(const uint8_t* src[], int format)
2113
{
2114
    if(!isALPHA(format))
2115
        src[3]=NULL;
2116
    if(!isPlanarYUV(format)) {
2117
        src[3]=src[2]=NULL;
2118

    
2119
        if (!usePal(format))
2120
            src[1]= NULL;
2121
    }
2122
}
2123

    
2124
static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,
2125
                                const int linesizes[4])
2126
{
2127
    const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];
2128
    int i;
2129

    
2130
    for (i = 0; i < 4; i++) {
2131
        int plane = desc->comp[i].plane;
2132
        if (!data[plane] || !linesizes[plane])
2133
            return 0;
2134
    }
2135

    
2136
    return 1;
2137
}
2138

    
2139
/**
2140
 * swscale wrapper, so we don't need to export the SwsContext.
2141
 * Assumes planar YUV to be in YUV order instead of YVU.
2142
 */
2143
int sws_scale(SwsContext *c, const uint8_t* const src[], const int srcStride[], int srcSliceY,
2144
              int srcSliceH, uint8_t* const dst[], const int dstStride[])
2145
{
2146
    int i;
2147
    const uint8_t* src2[4]= {src[0], src[1], src[2], src[3]};
2148
    uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};
2149

    
2150
    // do not mess up sliceDir if we have a "trailing" 0-size slice
2151
    if (srcSliceH == 0)
2152
        return 0;
2153

    
2154
    if (!check_image_pointers(src, c->srcFormat, srcStride)) {
2155
        av_log(c, AV_LOG_ERROR, "bad src image pointers\n");
2156
        return 0;
2157
    }
2158
    if (!check_image_pointers(dst, c->dstFormat, dstStride)) {
2159
        av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");
2160
        return 0;
2161
    }
2162

    
2163
    if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {
2164
        av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");
2165
        return 0;
2166
    }
2167
    if (c->sliceDir == 0) {
2168
        if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;
2169
    }
2170

    
2171
    if (usePal(c->srcFormat)) {
2172
        for (i=0; i<256; i++) {
2173
            int p, r, g, b, y, u, v, a = 0xff;
2174
            if(c->srcFormat == PIX_FMT_PAL8) {
2175
                p=((const uint32_t*)(src[1]))[i];
2176
                a= (p>>24)&0xFF;
2177
                r= (p>>16)&0xFF;
2178
                g= (p>> 8)&0xFF;
2179
                b=  p     &0xFF;
2180
            } else if(c->srcFormat == PIX_FMT_RGB8) {
2181
                r= (i>>5    )*36;
2182
                g= ((i>>2)&7)*36;
2183
                b= (i&3     )*85;
2184
            } else if(c->srcFormat == PIX_FMT_BGR8) {
2185
                b= (i>>6    )*85;
2186
                g= ((i>>3)&7)*36;
2187
                r= (i&7     )*36;
2188
            } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {
2189
                r= (i>>3    )*255;
2190
                g= ((i>>1)&3)*85;
2191
                b= (i&1     )*255;
2192
            } else if(c->srcFormat == PIX_FMT_GRAY8 || c->srcFormat == PIX_FMT_GRAY8A) {
2193
                r = g = b = i;
2194
            } else {
2195
                assert(c->srcFormat == PIX_FMT_BGR4_BYTE);
2196
                b= (i>>3    )*255;
2197
                g= ((i>>1)&3)*85;
2198
                r= (i&1     )*255;
2199
            }
2200
            y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2201
            u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2202
            v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2203
            c->pal_yuv[i]= y + (u<<8) + (v<<16) + (a<<24);
2204

    
2205
            switch(c->dstFormat) {
2206
            case PIX_FMT_BGR32:
2207
#if !HAVE_BIGENDIAN
2208
            case PIX_FMT_RGB24:
2209
#endif
2210
                c->pal_rgb[i]=  r + (g<<8) + (b<<16) + (a<<24);
2211
                break;
2212
            case PIX_FMT_BGR32_1:
2213
#if HAVE_BIGENDIAN
2214
            case PIX_FMT_BGR24:
2215
#endif
2216
                c->pal_rgb[i]= a + (r<<8) + (g<<16) + (b<<24);
2217
                break;
2218
            case PIX_FMT_RGB32_1:
2219
#if HAVE_BIGENDIAN
2220
            case PIX_FMT_RGB24:
2221
#endif
2222
                c->pal_rgb[i]= a + (b<<8) + (g<<16) + (r<<24);
2223
                break;
2224
            case PIX_FMT_RGB32:
2225
#if !HAVE_BIGENDIAN
2226
            case PIX_FMT_BGR24:
2227
#endif
2228
            default:
2229
                c->pal_rgb[i]=  b + (g<<8) + (r<<16) + (a<<24);
2230
            }
2231
        }
2232
    }
2233

    
2234
    // copy strides, so they can safely be modified
2235
    if (c->sliceDir == 1) {
2236
        // slices go from top to bottom
2237
        int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};
2238
        int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};
2239

    
2240
        reset_ptr(src2, c->srcFormat);
2241
        reset_ptr((const uint8_t**)dst2, c->dstFormat);
2242

    
2243
        /* reset slice direction at end of frame */
2244
        if (srcSliceY + srcSliceH == c->srcH)
2245
            c->sliceDir = 0;
2246

    
2247
        return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);
2248
    } else {
2249
        // slices go from bottom to top => we flip the image internally
2250
        int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};
2251
        int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};
2252

    
2253
        src2[0] += (srcSliceH-1)*srcStride[0];
2254
        if (!usePal(c->srcFormat))
2255
            src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];
2256
        src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];
2257
        src2[3] += (srcSliceH-1)*srcStride[3];
2258
        dst2[0] += ( c->dstH                      -1)*dstStride[0];
2259
        dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];
2260
        dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];
2261
        dst2[3] += ( c->dstH                      -1)*dstStride[3];
2262

    
2263
        reset_ptr(src2, c->srcFormat);
2264
        reset_ptr((const uint8_t**)dst2, c->dstFormat);
2265

    
2266
        /* reset slice direction at end of frame */
2267
        if (!srcSliceY)
2268
            c->sliceDir = 0;
2269

    
2270
        return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);
2271
    }
2272
}
2273

    
2274
#if LIBSWSCALE_VERSION_MAJOR < 1
2275
int sws_scale_ordered(SwsContext *c, const uint8_t* const src[], int srcStride[], int srcSliceY,
2276
                      int srcSliceH, uint8_t* dst[], int dstStride[])
2277
{
2278
    return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride);
2279
}
2280
#endif
2281

    
2282
/* Convert the palette to the same packed 32-bit format as the palette */
2283
void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
2284
{
2285
    long i;
2286

    
2287
    for (i=0; i<num_pixels; i++)
2288
        ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];
2289
}
2290

    
2291
/* Palette format: ABCD -> dst format: ABC */
2292
void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
2293
{
2294
    long i;
2295

    
2296
    for (i=0; i<num_pixels; i++) {
2297
        //FIXME slow?
2298
        dst[0]= palette[src[i]*4+0];
2299
        dst[1]= palette[src[i]*4+1];
2300
        dst[2]= palette[src[i]*4+2];
2301
        dst+= 3;
2302
    }
2303
}