swscale.c 107 KB
Newer Older
1
/*
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public 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
 * the C code (not assembly, mmx, ...) of this file can be used
 * under the LGPL license too
22
 */
23

24
/*
25
  supported Input formats: YV12, I420/IYUV, YUY2, UYVY, BGR32, BGR32_1, BGR24, BGR16, BGR15, RGB32, RGB32_1, RGB24, Y8/Y800, YVU9/IF09, PAL8
26
  supported output formats: YV12, I420/IYUV, YUY2, UYVY, {BGR,RGB}{1,4,8,15,16,24,32}, Y8/Y800, YVU9/IF09
27
  {BGR,RGB}{1,4,8,15,16} support dithering
28

29 30 31 32 33 34
  unscaled special converters (YV12=I420=IYUV, Y800=Y8)
  YV12 -> {BGR,RGB}{1,4,8,15,16,24,32}
  x -> x
  YUV9 -> YV12
  YUV9/YV12 -> Y800
  Y800 -> YUV9/YV12
Michael Niedermayer's avatar
Michael Niedermayer committed
35 36
  BGR24 -> BGR32 & RGB24 -> RGB32
  BGR32 -> BGR24 & RGB32 -> RGB24
Michael Niedermayer's avatar
Michael Niedermayer committed
37
  BGR15 -> BGR16
Michael Niedermayer's avatar
Michael Niedermayer committed
38 39
*/

40
/*
Diego Biurrun's avatar
typos  
Diego Biurrun committed
41
tested special converters (most are tested actually, but I did not write it down ...)
42
 YV12 -> BGR16
Michael Niedermayer's avatar
Michael Niedermayer committed
43
 YV12 -> YV12
Michael Niedermayer's avatar
Michael Niedermayer committed
44
 BGR15 -> BGR16
45
 BGR16 -> BGR16
46
 YVU9 -> YV12
Michael Niedermayer's avatar
Michael Niedermayer committed
47 48

untested special converters
Diego Biurrun's avatar
Diego Biurrun committed
49
  YV12/I420 -> BGR15/BGR24/BGR32 (it is the yuv2rgb stuff, so it should be ok)
50 51
  YV12/I420 -> YV12/I420
  YUY2/BGR15/BGR24/BGR32/RGB24/RGB32 -> same format
Michael Niedermayer's avatar
Michael Niedermayer committed
52 53
  BGR24 -> BGR32 & RGB24 -> RGB32
  BGR32 -> BGR24 & RGB32 -> RGB24
Michael Niedermayer's avatar
Michael Niedermayer committed
54
  BGR24 -> YV12
55 56
*/

57
#define _SVID_SOURCE //needed for MAP_ANONYMOUS
58
#include <inttypes.h>
59
#include <string.h>
60
#include <math.h>
61
#include <stdio.h>
Bohdan Horst's avatar
Bohdan Horst committed
62
#include <unistd.h>
63
#include "config.h"
Michael Niedermayer's avatar
Michael Niedermayer committed
64
#include <assert.h>
65
#if HAVE_SYS_MMAN_H
66
#include <sys/mman.h>
67 68 69
#if defined(MAP_ANON) && !defined(MAP_ANONYMOUS)
#define MAP_ANONYMOUS MAP_ANON
#endif
70
#endif
71
#include "swscale.h"
72
#include "swscale_internal.h"
73
#include "rgb2rgb.h"
74 75
#include "libavutil/x86_cpu.h"
#include "libavutil/bswap.h"
Arpi's avatar
Arpi committed
76

77 78 79 80 81
unsigned swscale_version(void)
{
    return LIBSWSCALE_VERSION_INT;
}

Michael Niedermayer's avatar
Michael Niedermayer committed
82
#undef MOVNTQ
83
#undef PAVGB
84

85
//#undef HAVE_MMX2
Michael Niedermayer's avatar
101  
Michael Niedermayer committed
86
//#define HAVE_3DNOW
87
//#undef HAVE_MMX
88
//#undef ARCH_X86
Michael Niedermayer's avatar
101++  
Michael Niedermayer committed
89
//#define WORDS_BIGENDIAN
90
#define DITHER1XBPP
91

Michael Niedermayer's avatar
Michael Niedermayer committed
92 93
#define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit

94
#define RET 0xC3 //near return opcode for X86
95

96 97 98 99 100
#ifdef M_PI
#define PI M_PI
#else
#define PI 3.14159265358979323846
#endif
101

102 103
#define isSupportedIn(x)    (       \
           (x)==PIX_FMT_YUV420P     \
104
        || (x)==PIX_FMT_YUVA420P    \
105 106 107
        || (x)==PIX_FMT_YUYV422     \
        || (x)==PIX_FMT_UYVY422     \
        || (x)==PIX_FMT_RGB32       \
108
        || (x)==PIX_FMT_RGB32_1     \
109 110 111 112
        || (x)==PIX_FMT_BGR24       \
        || (x)==PIX_FMT_BGR565      \
        || (x)==PIX_FMT_BGR555      \
        || (x)==PIX_FMT_BGR32       \
113
        || (x)==PIX_FMT_BGR32_1     \
114 115 116 117 118
        || (x)==PIX_FMT_RGB24       \
        || (x)==PIX_FMT_RGB565      \
        || (x)==PIX_FMT_RGB555      \
        || (x)==PIX_FMT_GRAY8       \
        || (x)==PIX_FMT_YUV410P     \
119
        || (x)==PIX_FMT_YUV440P     \
120 121 122 123 124 125 126 127 128 129
        || (x)==PIX_FMT_GRAY16BE    \
        || (x)==PIX_FMT_GRAY16LE    \
        || (x)==PIX_FMT_YUV444P     \
        || (x)==PIX_FMT_YUV422P     \
        || (x)==PIX_FMT_YUV411P     \
        || (x)==PIX_FMT_PAL8        \
        || (x)==PIX_FMT_BGR8        \
        || (x)==PIX_FMT_RGB8        \
        || (x)==PIX_FMT_BGR4_BYTE   \
        || (x)==PIX_FMT_RGB4_BYTE   \
Andreas Öman's avatar
Andreas Öman committed
130
        || (x)==PIX_FMT_YUV440P     \
131 132
        || (x)==PIX_FMT_MONOWHITE   \
        || (x)==PIX_FMT_MONOBLACK   \
133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
    )
#define isSupportedOut(x)   (       \
           (x)==PIX_FMT_YUV420P     \
        || (x)==PIX_FMT_YUYV422     \
        || (x)==PIX_FMT_UYVY422     \
        || (x)==PIX_FMT_YUV444P     \
        || (x)==PIX_FMT_YUV422P     \
        || (x)==PIX_FMT_YUV411P     \
        || isRGB(x)                 \
        || isBGR(x)                 \
        || (x)==PIX_FMT_NV12        \
        || (x)==PIX_FMT_NV21        \
        || (x)==PIX_FMT_GRAY16BE    \
        || (x)==PIX_FMT_GRAY16LE    \
        || (x)==PIX_FMT_GRAY8       \
        || (x)==PIX_FMT_YUV410P     \
149
        || (x)==PIX_FMT_YUV440P     \
150 151 152 153 154 155 156 157
    )
#define isPacked(x)         (       \
           (x)==PIX_FMT_PAL8        \
        || (x)==PIX_FMT_YUYV422     \
        || (x)==PIX_FMT_UYVY422     \
        || isRGB(x)                 \
        || isBGR(x)                 \
    )
158 159 160 161 162 163 164
#define usePal(x)           (       \
           (x)==PIX_FMT_PAL8        \
        || (x)==PIX_FMT_BGR4_BYTE   \
        || (x)==PIX_FMT_RGB4_BYTE   \
        || (x)==PIX_FMT_BGR8        \
        || (x)==PIX_FMT_RGB8        \
    )
165

166
#define RGB2YUV_SHIFT 15
167 168 169 170 171 172 173 174 175
#define BY ( (int)(0.114*219/255*(1<<RGB2YUV_SHIFT)+0.5))
#define BV (-(int)(0.081*224/255*(1<<RGB2YUV_SHIFT)+0.5))
#define BU ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
#define GY ( (int)(0.587*219/255*(1<<RGB2YUV_SHIFT)+0.5))
#define GV (-(int)(0.419*224/255*(1<<RGB2YUV_SHIFT)+0.5))
#define GU (-(int)(0.331*224/255*(1<<RGB2YUV_SHIFT)+0.5))
#define RY ( (int)(0.299*219/255*(1<<RGB2YUV_SHIFT)+0.5))
#define RV ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
#define RU (-(int)(0.169*224/255*(1<<RGB2YUV_SHIFT)+0.5))
176

177 178
extern const int32_t Inverse_Table_6_9[8][4];

179 180 181 182 183 184 185 186 187 188 189
static const double rgb2yuv_table[8][9]={
    {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5},
    {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5},
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5},
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5},
    {0.59  , 0.11  , 0.30  , -0.331, 0.5, -0.169, -0.421, -0.079, 0.5}, //FCC
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5},
    {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //SMPTE 170M
    {0.701 , 0.087 , 0.212 , -0.384, 0.5  -0.116, -0.445, -0.055, 0.5}, //SMPTE 240M
};

190 191
/*
NOTES
192
Special versions: fast Y 1:1 scaling (no interpolation in y direction)
193

194
TODO
Diego Biurrun's avatar
typo  
Diego Biurrun committed
195
more intelligent misalignment avoidance for the horizontal scaler
196 197
write special vertical cubic upscale version
Optimize C code (yv12 / minmax)
198
add support for packed pixel yuv input & output
199 200
add support for Y8 output
optimize bgr24 & bgr32
201
add BGR4 output support
202
write special BGR->BGR scaler
203
*/
204

205
#if ARCH_X86 && CONFIG_GPL
206 207 208 209 210 211 212 213
DECLARE_ASM_CONST(8, uint64_t, bF8)=       0xF8F8F8F8F8F8F8F8LL;
DECLARE_ASM_CONST(8, uint64_t, bFC)=       0xFCFCFCFCFCFCFCFCLL;
DECLARE_ASM_CONST(8, uint64_t, w10)=       0x0010001000100010LL;
DECLARE_ASM_CONST(8, uint64_t, w02)=       0x0002000200020002LL;
DECLARE_ASM_CONST(8, uint64_t, bm00001111)=0x00000000FFFFFFFFLL;
DECLARE_ASM_CONST(8, uint64_t, bm00000111)=0x0000000000FFFFFFLL;
DECLARE_ASM_CONST(8, uint64_t, bm11111000)=0xFFFFFFFFFF000000LL;
DECLARE_ASM_CONST(8, uint64_t, bm01010101)=0x00FF00FF00FF00FFLL;
214

215
const DECLARE_ALIGNED(8, uint64_t, ff_dither4[2]) = {
216 217
        0x0103010301030103LL,
        0x0200020002000200LL,};
Michael Niedermayer's avatar
Michael Niedermayer committed
218

219
const DECLARE_ALIGNED(8, uint64_t, ff_dither8[2]) = {
220 221
        0x0602060206020602LL,
        0x0004000400040004LL,};
222

223 224 225 226 227 228
DECLARE_ASM_CONST(8, uint64_t, b16Mask)=   0x001F001F001F001FLL;
DECLARE_ASM_CONST(8, uint64_t, g16Mask)=   0x07E007E007E007E0LL;
DECLARE_ASM_CONST(8, uint64_t, r16Mask)=   0xF800F800F800F800LL;
DECLARE_ASM_CONST(8, uint64_t, b15Mask)=   0x001F001F001F001FLL;
DECLARE_ASM_CONST(8, uint64_t, g15Mask)=   0x03E003E003E003E0LL;
DECLARE_ASM_CONST(8, uint64_t, r15Mask)=   0x7C007C007C007C00LL;
229

230 231 232
DECLARE_ALIGNED(8, const uint64_t, ff_M24A)         = 0x00FF0000FF0000FFLL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24B)         = 0xFF0000FF0000FF00LL;
DECLARE_ALIGNED(8, const uint64_t, ff_M24C)         = 0x0000FF0000FF0000LL;
Michael Niedermayer's avatar
Michael Niedermayer committed
233

Michael Niedermayer's avatar
Michael Niedermayer committed
234
#ifdef FAST_BGR2YV12
235 236 237
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000000210041000DULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000FFEEFFDC0038ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00000038FFD2FFF8ULL;
Michael Niedermayer's avatar
Michael Niedermayer committed
238
#else
239 240 241
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000020E540830C8BULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000ED0FDAC23831ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00003831D0E6F6EAULL;
Diego Biurrun's avatar
Diego Biurrun committed
242
#endif /* FAST_BGR2YV12 */
243 244 245
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset)  = 0x1010101010101010ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
DECLARE_ALIGNED(8, const uint64_t, ff_w1111)        = 0x0001000100010001ULL;
246

247 248 249 250 251
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY1Coeff) = 0x0C88000040870C88ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY2Coeff) = 0x20DE4087000020DEULL;
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY1Coeff) = 0x20DE0000408720DEULL;
DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY2Coeff) = 0x0C88408700000C88ULL;
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toYOffset) = 0x0008400000084000ULL;
252

253
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUV[2][4]) = {
254 255 256 257
    {0x38380000DAC83838ULL, 0xECFFDAC80000ECFFULL, 0xF6E40000D0E3F6E4ULL, 0x3838D0E300003838ULL},
    {0xECFF0000DAC8ECFFULL, 0x3838DAC800003838ULL, 0x38380000D0E33838ULL, 0xF6E4D0E30000F6E4ULL},
};

258
DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUVOffset)= 0x0040400000404000ULL;
259

260
#endif /* ARCH_X86 */
261 262 263 264

// clipping helper table for C implementations:
static unsigned char clip_table[768];

265
static SwsVector *sws_getConvVec(SwsVector *a, SwsVector *b);
266

267
static const uint8_t  __attribute__((aligned(8))) dither_2x2_4[2][8]={
268 269 270 271
{  1,   3,   1,   3,   1,   3,   1,   3, },
{  2,   0,   2,   0,   2,   0,   2,   0, },
};

272
static const uint8_t  __attribute__((aligned(8))) dither_2x2_8[2][8]={
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372
{  6,   2,   6,   2,   6,   2,   6,   2, },
{  0,   4,   0,   4,   0,   4,   0,   4, },
};

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

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

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

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

#if 1
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
{117,  62, 158, 103, 113,  58, 155, 100, },
{ 34, 199,  21, 186,  31, 196,  17, 182, },
{144,  89, 131,  76, 141,  86, 127,  72, },
{  0, 165,  41, 206,  10, 175,  52, 217, },
{110,  55, 151,  96, 120,  65, 162, 107, },
{ 28, 193,  14, 179,  38, 203,  24, 189, },
{138,  83, 124,  69, 148,  93, 134,  79, },
{  7, 172,  48, 213,   3, 168,  45, 210, },
};
#elif 1
// tries to correct a gamma of 1.5
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
{  0, 143,  18, 200,   2, 156,  25, 215, },
{ 78,  28, 125,  64,  89,  36, 138,  74, },
{ 10, 180,   3, 161,  16, 195,   8, 175, },
{109,  51,  93,  38, 121,  60, 105,  47, },
{  1, 152,  23, 210,   0, 147,  20, 205, },
{ 85,  33, 134,  71,  81,  30, 130,  67, },
{ 14, 190,   6, 171,  12, 185,   5, 166, },
{117,  57, 101,  44, 113,  54,  97,  41, },
};
#elif 1
// tries to correct a gamma of 2.0
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
{  0, 124,   8, 193,   0, 140,  12, 213, },
{ 55,  14, 104,  42,  66,  19, 119,  52, },
{  3, 168,   1, 145,   6, 187,   3, 162, },
{ 86,  31,  70,  21,  99,  39,  82,  28, },
{  0, 134,  11, 206,   0, 129,   9, 200, },
{ 62,  17, 114,  48,  58,  16, 109,  45, },
{  5, 181,   2, 157,   4, 175,   1, 151, },
{ 95,  36,  78,  26,  90,  34,  74,  24, },
};
#else
// tries to correct a gamma of 2.5
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
{  0, 107,   3, 187,   0, 125,   6, 212, },
{ 39,   7,  86,  28,  49,  11, 102,  36, },
{  1, 158,   0, 131,   3, 180,   1, 151, },
{ 68,  19,  52,  12,  81,  25,  64,  17, },
{  0, 119,   5, 203,   0, 113,   4, 195, },
{ 45,   9,  96,  33,  42,   8,  91,  30, },
{  2, 172,   1, 144,   2, 165,   0, 137, },
{ 77,  23,  60,  15,  72,  21,  56,  14, },
};
#endif
373

374
const char *sws_format_name(enum PixelFormat format)
375
{
376 377 378
    switch (format) {
        case PIX_FMT_YUV420P:
            return "yuv420p";
379 380
        case PIX_FMT_YUVA420P:
            return "yuva420p";
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
        case PIX_FMT_YUYV422:
            return "yuyv422";
        case PIX_FMT_RGB24:
            return "rgb24";
        case PIX_FMT_BGR24:
            return "bgr24";
        case PIX_FMT_YUV422P:
            return "yuv422p";
        case PIX_FMT_YUV444P:
            return "yuv444p";
        case PIX_FMT_RGB32:
            return "rgb32";
        case PIX_FMT_YUV410P:
            return "yuv410p";
        case PIX_FMT_YUV411P:
            return "yuv411p";
        case PIX_FMT_RGB565:
            return "rgb565";
        case PIX_FMT_RGB555:
            return "rgb555";
Kostya Shishkov's avatar
Kostya Shishkov committed
401 402 403 404
        case PIX_FMT_GRAY16BE:
            return "gray16be";
        case PIX_FMT_GRAY16LE:
            return "gray16le";
405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452
        case PIX_FMT_GRAY8:
            return "gray8";
        case PIX_FMT_MONOWHITE:
            return "mono white";
        case PIX_FMT_MONOBLACK:
            return "mono black";
        case PIX_FMT_PAL8:
            return "Palette";
        case PIX_FMT_YUVJ420P:
            return "yuvj420p";
        case PIX_FMT_YUVJ422P:
            return "yuvj422p";
        case PIX_FMT_YUVJ444P:
            return "yuvj444p";
        case PIX_FMT_XVMC_MPEG2_MC:
            return "xvmc_mpeg2_mc";
        case PIX_FMT_XVMC_MPEG2_IDCT:
            return "xvmc_mpeg2_idct";
        case PIX_FMT_UYVY422:
            return "uyvy422";
        case PIX_FMT_UYYVYY411:
            return "uyyvyy411";
        case PIX_FMT_RGB32_1:
            return "rgb32x";
        case PIX_FMT_BGR32_1:
            return "bgr32x";
        case PIX_FMT_BGR32:
            return "bgr32";
        case PIX_FMT_BGR565:
            return "bgr565";
        case PIX_FMT_BGR555:
            return "bgr555";
        case PIX_FMT_BGR8:
            return "bgr8";
        case PIX_FMT_BGR4:
            return "bgr4";
        case PIX_FMT_BGR4_BYTE:
            return "bgr4 byte";
        case PIX_FMT_RGB8:
            return "rgb8";
        case PIX_FMT_RGB4:
            return "rgb4";
        case PIX_FMT_RGB4_BYTE:
            return "rgb4 byte";
        case PIX_FMT_NV12:
            return "nv12";
        case PIX_FMT_NV21:
            return "nv21";
Andreas Öman's avatar
Andreas Öman committed
453 454
        case PIX_FMT_YUV440P:
            return "yuv440p";
Carl Eugen Hoyos's avatar
Carl Eugen Hoyos committed
455 456
        case PIX_FMT_VDPAU_H264:
            return "vdpau_h264";
457 458 459 460
        case PIX_FMT_VDPAU_MPEG1:
            return "vdpau_mpeg1";
        case PIX_FMT_VDPAU_MPEG2:
            return "vdpau_mpeg2";
461 462 463
        default:
            return "Unknown format";
    }
464 465
}

466
static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
467 468
                               int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
                               uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW)
Michael Niedermayer's avatar
Michael Niedermayer committed
469
{
470 471 472 473 474 475 476 477 478 479 480 481
    //FIXME Optimize (just quickly writen not opti..)
    int i;
    for (i=0; i<dstW; i++)
    {
        int val=1<<18;
        int j;
        for (j=0; j<lumFilterSize; j++)
            val += lumSrc[j][i] * lumFilter[j];

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

482
    if (uDest)
483 484 485 486 487 488 489 490
        for (i=0; i<chrDstW; i++)
        {
            int u=1<<18;
            int v=1<<18;
            int j;
            for (j=0; j<chrFilterSize; j++)
            {
                u += chrSrc[j][i] * chrFilter[j];
491
                v += chrSrc[j][i + VOFW] * chrFilter[j];
492 493 494 495 496
            }

            uDest[i]= av_clip_uint8(u>>19);
            vDest[i]= av_clip_uint8(v>>19);
        }
Michael Niedermayer's avatar
Michael Niedermayer committed
497 498
}

Ville Syrjälä's avatar
Ville Syrjälä committed
499
static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
500 501
                                int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
                                uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat)
Ville Syrjälä's avatar
Ville Syrjälä committed
502
{
503 504 505 506 507 508 509 510 511 512 513 514
    //FIXME Optimize (just quickly writen not opti..)
    int i;
    for (i=0; i<dstW; i++)
    {
        int val=1<<18;
        int j;
        for (j=0; j<lumFilterSize; j++)
            val += lumSrc[j][i] * lumFilter[j];

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

515
    if (!uDest)
516 517 518 519 520 521 522 523 524 525 526
        return;

    if (dstFormat == PIX_FMT_NV12)
        for (i=0; i<chrDstW; i++)
        {
            int u=1<<18;
            int v=1<<18;
            int j;
            for (j=0; j<chrFilterSize; j++)
            {
                u += chrSrc[j][i] * chrFilter[j];
527
                v += chrSrc[j][i + VOFW] * chrFilter[j];
528 529 530 531 532 533 534 535 536 537 538 539 540 541
            }

            uDest[2*i]= av_clip_uint8(u>>19);
            uDest[2*i+1]= av_clip_uint8(v>>19);
        }
    else
        for (i=0; i<chrDstW; i++)
        {
            int u=1<<18;
            int v=1<<18;
            int j;
            for (j=0; j<chrFilterSize; j++)
            {
                u += chrSrc[j][i] * chrFilter[j];
542
                v += chrSrc[j][i + VOFW] * chrFilter[j];
543 544 545 546 547
            }

            uDest[2*i]= av_clip_uint8(v>>19);
            uDest[2*i+1]= av_clip_uint8(u>>19);
        }
Ville Syrjälä's avatar
Ville Syrjälä committed
548
}
Michael Niedermayer's avatar
Michael Niedermayer committed
549

550
#define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type) \
551 552 553 554 555 556
    for (i=0; i<(dstW>>1); i++){\
        int j;\
        int Y1 = 1<<18;\
        int Y2 = 1<<18;\
        int U  = 1<<18;\
        int V  = 1<<18;\
557
        type av_unused *r, *b, *g;\
558 559 560 561 562 563 564 565 566 567
        const int i2= 2*i;\
        \
        for (j=0; j<lumFilterSize; j++)\
        {\
            Y1 += lumSrc[j][i2] * lumFilter[j];\
            Y2 += lumSrc[j][i2+1] * lumFilter[j];\
        }\
        for (j=0; j<chrFilterSize; j++)\
        {\
            U += chrSrc[j][i] * chrFilter[j];\
568
            V += chrSrc[j][i+VOFW] * chrFilter[j];\
569 570 571 572 573
        }\
        Y1>>=19;\
        Y2>>=19;\
        U >>=19;\
        V >>=19;\
574 575 576

#define YSCALE_YUV_2_PACKEDX_C(type) \
        YSCALE_YUV_2_PACKEDX_NOCLIP_C(type)\
577 578 579 580 581 582 583 584 585 586 587
        if ((Y1|Y2|U|V)&256)\
        {\
            if (Y1>255)   Y1=255; \
            else if (Y1<0)Y1=0;   \
            if (Y2>255)   Y2=255; \
            else if (Y2<0)Y2=0;   \
            if (U>255)    U=255;  \
            else if (U<0) U=0;    \
            if (V>255)    V=255;  \
            else if (V<0) V=0;    \
        }
588

589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609
#define YSCALE_YUV_2_PACKEDX_FULL_C \
    for (i=0; i<dstW; i++){\
        int j;\
        int Y = 0;\
        int U = -128<<19;\
        int V = -128<<19;\
        int R,G,B;\
        \
        for (j=0; j<lumFilterSize; j++){\
            Y += lumSrc[j][i     ] * lumFilter[j];\
        }\
        for (j=0; j<chrFilterSize; j++){\
            U += chrSrc[j][i     ] * chrFilter[j];\
            V += chrSrc[j][i+VOFW] * chrFilter[j];\
        }\
        Y >>=10;\
        U >>=10;\
        V >>=10;\

#define YSCALE_YUV_2_RGBX_FULL_C(rnd) \
    YSCALE_YUV_2_PACKEDX_FULL_C\
610 611
        Y-= c->yuv2rgb_y_offset;\
        Y*= c->yuv2rgb_y_coeff;\
612
        Y+= rnd;\
613 614 615
        R= Y + V*c->yuv2rgb_v2r_coeff;\
        G= Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;\
        B= Y +                          U*c->yuv2rgb_u2b_coeff;\
616 617 618 619 620 621 622 623 624 625
        if ((R|G|B)&(0xC0000000)){\
            if (R>=(256<<22))   R=(256<<22)-1; \
            else if (R<0)R=0;   \
            if (G>=(256<<22))   G=(256<<22)-1; \
            else if (G<0)G=0;   \
            if (B>=(256<<22))   B=(256<<22)-1; \
            else if (B<0)B=0;   \
        }\


626
#define YSCALE_YUV_2_GRAY16_C \
627 628 629 630 631 632
    for (i=0; i<(dstW>>1); i++){\
        int j;\
        int Y1 = 1<<18;\
        int Y2 = 1<<18;\
        int U  = 1<<18;\
        int V  = 1<<18;\
633
        \
634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650
        const int i2= 2*i;\
        \
        for (j=0; j<lumFilterSize; j++)\
        {\
            Y1 += lumSrc[j][i2] * lumFilter[j];\
            Y2 += lumSrc[j][i2+1] * lumFilter[j];\
        }\
        Y1>>=11;\
        Y2>>=11;\
        if ((Y1|Y2|U|V)&65536)\
        {\
            if (Y1>65535)   Y1=65535; \
            else if (Y1<0)Y1=0;   \
            if (Y2>65535)   Y2=65535; \
            else if (Y2<0)Y2=0;   \
        }

Michael Niedermayer's avatar
Michael Niedermayer committed
651
#define YSCALE_YUV_2_RGBX_C(type) \
652
    YSCALE_YUV_2_PACKEDX_C(type)  /* FIXME fix tables so that cliping is not needed and then use _NOCLIP*/\
653 654 655 656 657 658 659 660 661 662
    r = (type *)c->table_rV[V];   \
    g = (type *)(c->table_gU[U] + c->table_gV[V]); \
    b = (type *)c->table_bU[U];   \

#define YSCALE_YUV_2_PACKED2_C   \
    for (i=0; i<(dstW>>1); i++){ \
        const int i2= 2*i;       \
        int Y1= (buf0[i2  ]*yalpha1+buf1[i2  ]*yalpha)>>19;           \
        int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19;           \
        int U= (uvbuf0[i     ]*uvalpha1+uvbuf1[i     ]*uvalpha)>>19;  \
663
        int V= (uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19;  \
Michael Niedermayer's avatar
Michael Niedermayer committed
664

665 666 667 668 669 670
#define YSCALE_YUV_2_GRAY16_2_C   \
    for (i=0; i<(dstW>>1); i++){ \
        const int i2= 2*i;       \
        int Y1= (buf0[i2  ]*yalpha1+buf1[i2  ]*yalpha)>>11;           \
        int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>11;           \

Michael Niedermayer's avatar
Michael Niedermayer committed
671
#define YSCALE_YUV_2_RGB2_C(type) \
672 673 674 675 676
    YSCALE_YUV_2_PACKED2_C\
    type *r, *b, *g;\
    r = (type *)c->table_rV[V];\
    g = (type *)(c->table_gU[U] + c->table_gV[V]);\
    b = (type *)c->table_bU[U];\
Michael Niedermayer's avatar
Michael Niedermayer committed
677

Michael Niedermayer's avatar
Michael Niedermayer committed
678
#define YSCALE_YUV_2_PACKED1_C \
679 680 681 682 683
    for (i=0; i<(dstW>>1); i++){\
        const int i2= 2*i;\
        int Y1= buf0[i2  ]>>7;\
        int Y2= buf0[i2+1]>>7;\
        int U= (uvbuf1[i     ])>>7;\
684
        int V= (uvbuf1[i+VOFW])>>7;\
Michael Niedermayer's avatar
Michael Niedermayer committed
685

686 687 688 689 690 691
#define YSCALE_YUV_2_GRAY16_1_C \
    for (i=0; i<(dstW>>1); i++){\
        const int i2= 2*i;\
        int Y1= buf0[i2  ]<<1;\
        int Y2= buf0[i2+1]<<1;\

Michael Niedermayer's avatar
Michael Niedermayer committed
692
#define YSCALE_YUV_2_RGB1_C(type) \
693 694 695 696 697
    YSCALE_YUV_2_PACKED1_C\
    type *r, *b, *g;\
    r = (type *)c->table_rV[V];\
    g = (type *)(c->table_gU[U] + c->table_gV[V]);\
    b = (type *)c->table_bU[U];\
Michael Niedermayer's avatar
Michael Niedermayer committed
698

Michael Niedermayer's avatar
Michael Niedermayer committed
699
#define YSCALE_YUV_2_PACKED1B_C \
700 701 702 703 704
    for (i=0; i<(dstW>>1); i++){\
        const int i2= 2*i;\
        int Y1= buf0[i2  ]>>7;\
        int Y2= buf0[i2+1]>>7;\
        int U= (uvbuf0[i     ] + uvbuf1[i     ])>>8;\
705
        int V= (uvbuf0[i+VOFW] + uvbuf1[i+VOFW])>>8;\
Michael Niedermayer's avatar
Michael Niedermayer committed
706 707

#define YSCALE_YUV_2_RGB1B_C(type) \
708 709 710 711 712
    YSCALE_YUV_2_PACKED1B_C\
    type *r, *b, *g;\
    r = (type *)c->table_rV[V];\
    g = (type *)(c->table_gU[U] + c->table_gV[V]);\
    b = (type *)c->table_bU[U];\
Michael Niedermayer's avatar
Michael Niedermayer committed
713

714
#define YSCALE_YUV_2_MONO2_C \
715 716 717 718 719 720 721 722 723 724 725 726
    const uint8_t * const d128=dither_8x8_220[y&7];\
    uint8_t *g= c->table_gU[128] + c->table_gV[128];\
    for (i=0; i<dstW-7; i+=8){\
        int acc;\
        acc =       g[((buf0[i  ]*yalpha1+buf1[i  ]*yalpha)>>19) + d128[0]];\
        acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\
        acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\
        acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\
        acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\
        acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\
        acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\
        acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\
727
        ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
728 729 730 731
        dest++;\
    }\


732
#define YSCALE_YUV_2_MONOX_C \
733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757
    const uint8_t * const d128=dither_8x8_220[y&7];\
    uint8_t *g= c->table_gU[128] + c->table_gV[128];\
    int acc=0;\
    for (i=0; i<dstW-1; i+=2){\
        int j;\
        int Y1=1<<18;\
        int Y2=1<<18;\
\
        for (j=0; j<lumFilterSize; j++)\
        {\
            Y1 += lumSrc[j][i] * lumFilter[j];\
            Y2 += lumSrc[j][i+1] * lumFilter[j];\
        }\
        Y1>>=19;\
        Y2>>=19;\
        if ((Y1|Y2)&256)\
        {\
            if (Y1>255)   Y1=255;\
            else if (Y1<0)Y1=0;\
            if (Y2>255)   Y2=255;\
            else if (Y2<0)Y2=0;\
        }\
        acc+= acc + g[Y1+d128[(i+0)&7]];\
        acc+= acc + g[Y2+d128[(i+1)&7]];\
        if ((i&7)==6){\
758
            ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
759 760 761 762 763 764
            dest++;\
        }\
    }


#define YSCALE_YUV_2_ANYRGB_C(func, func2, func_g16, func_monoblack)\
765 766 767 768
    switch(c->dstFormat)\
    {\
    case PIX_FMT_RGB32:\
    case PIX_FMT_BGR32:\
769 770
    case PIX_FMT_RGB32_1:\
    case PIX_FMT_BGR32_1:\
771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861
        func(uint32_t)\
            ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
            ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
        }                \
        break;\
    case PIX_FMT_RGB24:\
        func(uint8_t)\
            ((uint8_t*)dest)[0]= r[Y1];\
            ((uint8_t*)dest)[1]= g[Y1];\
            ((uint8_t*)dest)[2]= b[Y1];\
            ((uint8_t*)dest)[3]= r[Y2];\
            ((uint8_t*)dest)[4]= g[Y2];\
            ((uint8_t*)dest)[5]= b[Y2];\
            dest+=6;\
        }\
        break;\
    case PIX_FMT_BGR24:\
        func(uint8_t)\
            ((uint8_t*)dest)[0]= b[Y1];\
            ((uint8_t*)dest)[1]= g[Y1];\
            ((uint8_t*)dest)[2]= r[Y1];\
            ((uint8_t*)dest)[3]= b[Y2];\
            ((uint8_t*)dest)[4]= g[Y2];\
            ((uint8_t*)dest)[5]= r[Y2];\
            dest+=6;\
        }\
        break;\
    case PIX_FMT_RGB565:\
    case PIX_FMT_BGR565:\
        {\
            const int dr1= dither_2x2_8[y&1    ][0];\
            const int dg1= dither_2x2_4[y&1    ][0];\
            const int db1= dither_2x2_8[(y&1)^1][0];\
            const int dr2= dither_2x2_8[y&1    ][1];\
            const int dg2= dither_2x2_4[y&1    ][1];\
            const int db2= dither_2x2_8[(y&1)^1][1];\
            func(uint16_t)\
                ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
                ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
            }\
        }\
        break;\
    case PIX_FMT_RGB555:\
    case PIX_FMT_BGR555:\
        {\
            const int dr1= dither_2x2_8[y&1    ][0];\
            const int dg1= dither_2x2_8[y&1    ][1];\
            const int db1= dither_2x2_8[(y&1)^1][0];\
            const int dr2= dither_2x2_8[y&1    ][1];\
            const int dg2= dither_2x2_8[y&1    ][0];\
            const int db2= dither_2x2_8[(y&1)^1][1];\
            func(uint16_t)\
                ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
                ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
            }\
        }\
        break;\
    case PIX_FMT_RGB8:\
    case PIX_FMT_BGR8:\
        {\
            const uint8_t * const d64= dither_8x8_73[y&7];\
            const uint8_t * const d32= dither_8x8_32[y&7];\
            func(uint8_t)\
                ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\
                ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\
            }\
        }\
        break;\
    case PIX_FMT_RGB4:\
    case PIX_FMT_BGR4:\
        {\
            const uint8_t * const d64= dither_8x8_73 [y&7];\
            const uint8_t * const d128=dither_8x8_220[y&7];\
            func(uint8_t)\
                ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\
                                 + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\
            }\
        }\
        break;\
    case PIX_FMT_RGB4_BYTE:\
    case PIX_FMT_BGR4_BYTE:\
        {\
            const uint8_t * const d64= dither_8x8_73 [y&7];\
            const uint8_t * const d128=dither_8x8_220[y&7];\
            func(uint8_t)\
                ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\
                ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\
            }\
        }\
        break;\
    case PIX_FMT_MONOBLACK:\
862
    case PIX_FMT_MONOWHITE:\
863
        {\
864
            func_monoblack\
865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882
        }\
        break;\
    case PIX_FMT_YUYV422:\
        func2\
            ((uint8_t*)dest)[2*i2+0]= Y1;\
            ((uint8_t*)dest)[2*i2+1]= U;\
            ((uint8_t*)dest)[2*i2+2]= Y2;\
            ((uint8_t*)dest)[2*i2+3]= V;\
        }                \
        break;\
    case PIX_FMT_UYVY422:\
        func2\
            ((uint8_t*)dest)[2*i2+0]= U;\
            ((uint8_t*)dest)[2*i2+1]= Y1;\
            ((uint8_t*)dest)[2*i2+2]= V;\
            ((uint8_t*)dest)[2*i2+3]= Y2;\
        }                \
        break;\
883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898
    case PIX_FMT_GRAY16BE:\
        func_g16\
            ((uint8_t*)dest)[2*i2+0]= Y1>>8;\
            ((uint8_t*)dest)[2*i2+1]= Y1;\
            ((uint8_t*)dest)[2*i2+2]= Y2>>8;\
            ((uint8_t*)dest)[2*i2+3]= Y2;\
        }                \
        break;\
    case PIX_FMT_GRAY16LE:\
        func_g16\
            ((uint8_t*)dest)[2*i2+0]= Y1;\
            ((uint8_t*)dest)[2*i2+1]= Y1>>8;\
            ((uint8_t*)dest)[2*i2+2]= Y2;\
            ((uint8_t*)dest)[2*i2+3]= Y2>>8;\
        }                \
        break;\
899
    }\
Michael Niedermayer's avatar
Michael Niedermayer committed
900 901


Michael Niedermayer's avatar
Michael Niedermayer committed
902
static inline void yuv2packedXinC(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
903 904
                                  int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
                                  uint8_t *dest, int dstW, int y)
Michael Niedermayer's avatar
Michael Niedermayer committed
905
{
906
    int i;
907
    YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGBX_C, YSCALE_YUV_2_PACKEDX_C(void), YSCALE_YUV_2_GRAY16_C, YSCALE_YUV_2_MONOX_C)
Michael Niedermayer's avatar
Michael Niedermayer committed
908 909
}

910 911 912 913 914 915
static inline void yuv2rgbXinC_full(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
                                    int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
                                    uint8_t *dest, int dstW, int y)
{
    int i;
    int step= fmt_depth(c->dstFormat)/8;
916
    int aidx= 3;
917 918 919 920

    switch(c->dstFormat){
    case PIX_FMT_ARGB:
        dest++;
921
        aidx= 0;
922
    case PIX_FMT_RGB24:
923
        aidx--;
924 925
    case PIX_FMT_RGBA:
        YSCALE_YUV_2_RGBX_FULL_C(1<<21)
926
            dest[aidx]= 0;
927 928 929 930 931 932 933 934
            dest[0]= R>>22;
            dest[1]= G>>22;
            dest[2]= B>>22;
            dest+= step;
        }
        break;
    case PIX_FMT_ABGR:
        dest++;
935
        aidx= 0;
936
    case PIX_FMT_BGR24:
937
        aidx--;
938 939
    case PIX_FMT_BGRA:
        YSCALE_YUV_2_RGBX_FULL_C(1<<21)
940
            dest[aidx]= 0;
941 942 943 944 945 946 947 948 949 950
            dest[0]= B>>22;
            dest[1]= G>>22;
            dest[2]= R>>22;
            dest+= step;
        }
        break;
    default:
        assert(0);
    }
}
Michael Niedermayer's avatar
Michael Niedermayer committed
951

Michael Niedermayer's avatar
Michael Niedermayer committed
952 953
//Note: we have C, X86, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
//Plain C versions
954
#if !HAVE_MMX || defined (RUNTIME_CPUDETECT) || !CONFIG_GPL
955 956 957
#define COMPILE_C
#endif

958 959
#if ARCH_PPC
#if (HAVE_ALTIVEC || defined (RUNTIME_CPUDETECT)) && CONFIG_GPL
960 961
#define COMPILE_ALTIVEC
#endif //HAVE_ALTIVEC
962
#endif //ARCH_PPC
963

964
#if ARCH_X86
965

966
#if ((HAVE_MMX && !HAVE_3DNOW && !HAVE_MMX2) || defined (RUNTIME_CPUDETECT)) && CONFIG_GPL
967 968 969
#define COMPILE_MMX
#endif

970
#if (HAVE_MMX2 || defined (RUNTIME_CPUDETECT)) && CONFIG_GPL
971 972 973
#define COMPILE_MMX2
#endif

974
#if ((HAVE_3DNOW && !HAVE_MMX2) || defined (RUNTIME_CPUDETECT)) && CONFIG_GPL
975 976
#define COMPILE_3DNOW
#endif
977
#endif //ARCH_X86 || ARCH_X86_64
978 979 980 981

#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
982 983 984 985 986
#undef HAVE_ALTIVEC
#define HAVE_MMX 0
#define HAVE_MMX2 0
#define HAVE_3DNOW 0
#define HAVE_ALTIVEC 0
987 988

#ifdef COMPILE_C
Michael Niedermayer's avatar
Michael Niedermayer committed
989 990
#define RENAME(a) a ## _C
#include "swscale_template.c"
991
#endif
Michael Niedermayer's avatar
Michael Niedermayer committed
992

993 994
#ifdef COMPILE_ALTIVEC
#undef RENAME
995 996
#undef HAVE_ALTIVEC
#define HAVE_ALTIVEC 1
997 998 999 1000
#define RENAME(a) a ## _altivec
#include "swscale_template.c"
#endif

1001
#if ARCH_X86
Michael Niedermayer's avatar
Michael Niedermayer committed
1002

Michael Niedermayer's avatar
Michael Niedermayer committed
1003 1004 1005 1006 1007 1008 1009 1010 1011
//X86 versions
/*
#undef RENAME
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
#define ARCH_X86
#define RENAME(a) a ## _X86
#include "swscale_template.c"
Michael Niedermayer's avatar
Michael Niedermayer committed
1012
*/
Michael Niedermayer's avatar
Michael Niedermayer committed
1013
//MMX versions
1014
#ifdef COMPILE_MMX
Michael Niedermayer's avatar
Michael Niedermayer committed
1015
#undef RENAME
1016
#undef HAVE_MMX
Michael Niedermayer's avatar
Michael Niedermayer committed
1017 1018
#undef HAVE_MMX2
#undef HAVE_3DNOW
1019 1020 1021
#define HAVE_MMX 1
#define HAVE_MMX2 0
#define HAVE_3DNOW 0
Michael Niedermayer's avatar
Michael Niedermayer committed
1022 1023
#define RENAME(a) a ## _MMX
#include "swscale_template.c"
1024
#endif
Michael Niedermayer's avatar
Michael Niedermayer committed
1025 1026

//MMX2 versions
1027
#ifdef COMPILE_MMX2
Michael Niedermayer's avatar
Michael Niedermayer committed
1028
#undef RENAME
1029 1030
#undef HAVE_MMX
#undef HAVE_MMX2
Michael Niedermayer's avatar
Michael Niedermayer committed
1031
#undef HAVE_3DNOW
1032 1033 1034
#define HAVE_MMX 1
#define HAVE_MMX2 1
#define HAVE_3DNOW 0
Michael Niedermayer's avatar
Michael Niedermayer committed
1035 1036
#define RENAME(a) a ## _MMX2
#include "swscale_template.c"
1037
#endif
Michael Niedermayer's avatar
Michael Niedermayer committed
1038 1039

//3DNOW versions
1040
#ifdef COMPILE_3DNOW
Michael Niedermayer's avatar
Michael Niedermayer committed
1041
#undef RENAME
1042
#undef HAVE_MMX
Michael Niedermayer's avatar
Michael Niedermayer committed
1043
#undef HAVE_MMX2
1044 1045 1046 1047
#undef HAVE_3DNOW
#define HAVE_MMX 1
#define HAVE_MMX2 0
#define HAVE_3DNOW 1
Michael Niedermayer's avatar
Michael Niedermayer committed
1048 1049
#define RENAME(a) a ## _3DNow
#include "swscale_template.c"
1050
#endif
Michael Niedermayer's avatar
Michael Niedermayer committed
1051

1052
#endif //ARCH_X86 || ARCH_X86_64
Michael Niedermayer's avatar
Michael Niedermayer committed
1053

Gabucino's avatar
Gabucino committed
1054
// minor note: the HAVE_xyz is messed up after that line so don't use it
1055

Michael Niedermayer's avatar
Michael Niedermayer committed
1056 1057
static double getSplineCoeff(double a, double b, double c, double d, double dist)
{
1058 1059 1060 1061 1062 1063 1064
//    printf("%f %f %f %f %f\n", a,b,c,d,dist);
    if (dist<=1.0)      return ((d*dist + c)*dist + b)*dist +a;
    else                return getSplineCoeff(        0.0,
                                             b+ 2.0*c + 3.0*d,
                                                    c + 3.0*d,
                                            -b- 3.0*c - 6.0*d,
                                            dist-1.0);
Michael Niedermayer's avatar
Michael Niedermayer committed
1065
}
1066

1067
static inline int initFilter(int16_t **outFilter, int16_t **filterPos, int *outFilterSize, int xInc,
1068 1069
                             int srcW, int dstW, int filterAlign, int one, int flags,
                             SwsVector *srcFilter, SwsVector *dstFilter, double param[2])
1070
{
1071 1072 1073 1074
    int i;
    int filterSize;
    int filter2Size;
    int minFilterSize;
1075 1076 1077
    int64_t *filter=NULL;
    int64_t *filter2=NULL;
    const int64_t fone= 1LL<<54;
Michael Niedermayer's avatar
Michael Niedermayer committed
1078
    int ret= -1;
1079
#if ARCH_X86
1080
    if (flags & SWS_CPU_CAPS_MMX)
1081
        __asm__ volatile("emms\n\t"::: "memory"); //FIXME this should not be required but it IS (even for non-MMX versions)
1082
#endif
1083

1084 1085 1086 1087 1088 1089 1090
    // Note the +1 is for the MMXscaler which reads over the end
    *filterPos = av_malloc((dstW+1)*sizeof(int16_t));

    if (FFABS(xInc - 0x10000) <10) // unscaled
    {
        int i;
        filterSize= 1;
1091
        filter= av_mallocz(dstW*sizeof(*filter)*filterSize);
1092 1093 1094

        for (i=0; i<dstW; i++)
        {
Michael Niedermayer's avatar