rgb2rgb.c 26.9 KB
Newer Older
1
/*
2 3
 *
 *  rgb2rgb.c, Software RGB to RGB convertor
4 5 6
 *  pluralize by Software PAL8 to RGB convertor
 *               Software YUV to YUV convertor
 *               Software YUV to RGB convertor
7
 *  Written by Nick Kurshev.
8
 *  palette & yuv & runtime cpu stuff by Michael (michaelni@gmx.at) (under GPL)
9
 */
Nick Kurshev's avatar
Nick Kurshev committed
10 11 12
#include <inttypes.h>
#include "../config.h"
#include "rgb2rgb.h"
13
#include "../cpudetect.h"
Felix Bünemann's avatar
Felix Bünemann committed
14
#include "../mangle.h"
15
#include "../bswap.h"
16

17 18 19 20
#ifdef ARCH_X86
#define CAN_COMPILE_X86_ASM
#endif

21 22
#define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit

23
#ifdef CAN_COMPILE_X86_ASM
Arpi's avatar
Arpi committed
24 25
static const uint64_t mmx_null  __attribute__((aligned(8))) = 0x0000000000000000ULL;
static const uint64_t mmx_one   __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
26 27 28
static const uint64_t mask32b  __attribute__((aligned(8))) = 0x000000FF000000FFULL;
static const uint64_t mask32g  __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
static const uint64_t mask32r  __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
29
static const uint64_t mask32   __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
Michael Niedermayer's avatar
Michael Niedermayer committed
30 31 32
static const uint64_t mask24b  __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
static const uint64_t mask24g  __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
static const uint64_t mask24r  __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
33 34
static const uint64_t mask24l  __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
static const uint64_t mask24h  __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
35 36 37
static const uint64_t mask24hh  __attribute__((aligned(8))) = 0xffff000000000000ULL;
static const uint64_t mask24hhh  __attribute__((aligned(8))) = 0xffffffff00000000ULL;
static const uint64_t mask24hhhh  __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
38 39
static const uint64_t mask15b  __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111  xxB */
static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000  RGx */
40
static const uint64_t mask15s  __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
Arpi's avatar
Arpi committed
41 42 43 44 45
static const uint64_t mask15g  __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
static const uint64_t mask15r  __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
#define mask16b mask15b
static const uint64_t mask16g  __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
static const uint64_t mask16r  __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
46 47 48 49 50 51
static const uint64_t red_16mask  __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
static const uint64_t green_16mask __attribute__((aligned(8)))= 0x000007e0000007e0ULL;
static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
static const uint64_t red_15mask  __attribute__((aligned(8))) = 0x00007c000000f800ULL;
static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000007e0ULL;
static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
52 53 54 55 56 57 58 59 60 61 62 63 64 65

#ifdef FAST_BGR2YV12
static const uint64_t bgr2YCoeff  __attribute__((aligned(8))) = 0x000000210041000DULL;
static const uint64_t bgr2UCoeff  __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
static const uint64_t bgr2VCoeff  __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
#else
static const uint64_t bgr2YCoeff  __attribute__((aligned(8))) = 0x000020E540830C8BULL;
static const uint64_t bgr2UCoeff  __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
static const uint64_t bgr2VCoeff  __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
#endif
static const uint64_t bgr2YOffset __attribute__((aligned(8))) = 0x1010101010101010ULL;
static const uint64_t bgr2UVOffset __attribute__((aligned(8)))= 0x8080808080808080ULL;
static const uint64_t w1111       __attribute__((aligned(8))) = 0x0001000100010001ULL;

66 67 68 69 70 71 72 73 74 75 76 77 78 79
#if 0
static volatile uint64_t __attribute__((aligned(8))) b5Dither;
static volatile uint64_t __attribute__((aligned(8))) g5Dither;
static volatile uint64_t __attribute__((aligned(8))) g6Dither;
static volatile uint64_t __attribute__((aligned(8))) r5Dither;

static uint64_t __attribute__((aligned(8))) dither4[2]={
	0x0103010301030103LL,
	0x0200020002000200LL,};

static uint64_t __attribute__((aligned(8))) dither8[2]={
	0x0602060206020602LL,
	0x0004000400040004LL,};
#endif
80
#endif
Nick Kurshev's avatar
Nick Kurshev committed
81

82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98
#define RGB2YUV_SHIFT 8
#define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
#define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
#define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
#define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
#define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
#define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
#define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
#define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
#define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))

//Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
//Plain C versions
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
#undef ARCH_X86
Michael Niedermayer's avatar
Michael Niedermayer committed
99
#undef HAVE_SSE2
100 101 102 103 104 105 106 107 108 109
#define RENAME(a) a ## _C
#include "rgb2rgb_template.c"

#ifdef CAN_COMPILE_X86_ASM

//MMX versions
#undef RENAME
#define HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
Michael Niedermayer's avatar
Michael Niedermayer committed
110
#undef HAVE_SSE2
111 112 113 114 115 116 117 118 119
#define ARCH_X86
#define RENAME(a) a ## _MMX
#include "rgb2rgb_template.c"

//MMX2 versions
#undef RENAME
#define HAVE_MMX
#define HAVE_MMX2
#undef HAVE_3DNOW
Michael Niedermayer's avatar
Michael Niedermayer committed
120
#undef HAVE_SSE2
121 122 123 124 125 126 127 128 129
#define ARCH_X86
#define RENAME(a) a ## _MMX2
#include "rgb2rgb_template.c"

//3DNOW versions
#undef RENAME
#define HAVE_MMX
#undef HAVE_MMX2
#define HAVE_3DNOW
Michael Niedermayer's avatar
Michael Niedermayer committed
130
#undef HAVE_SSE2
131 132 133 134 135 136
#define ARCH_X86
#define RENAME(a) a ## _3DNow
#include "rgb2rgb_template.c"

#endif //CAN_COMPILE_X86_ASM

137
void rgb24to32(const uint8_t *src,uint8_t *dst,unsigned src_size)
Nick Kurshev's avatar
Nick Kurshev committed
138
{
139 140 141 142 143 144 145 146 147
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb24to32_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb24to32_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb24to32_MMX(src, dst, src_size);
	else
Arpi's avatar
Arpi committed
148
#endif
149
		rgb24to32_C(src, dst, src_size);
Arpi's avatar
Arpi committed
150 151 152 153 154 155 156 157 158 159 160 161 162
}

void rgb15to24(const uint8_t *src,uint8_t *dst,unsigned src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb15to24_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb15to24_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb15to24_MMX(src, dst, src_size);
	else
Nick Kurshev's avatar
Nick Kurshev committed
163
#endif
Arpi's avatar
Arpi committed
164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209
		rgb15to24_C(src, dst, src_size);
}

void rgb16to24(const uint8_t *src,uint8_t *dst,unsigned src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb16to24_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb16to24_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb16to24_MMX(src, dst, src_size);
	else
#endif
		rgb16to24_C(src, dst, src_size);
}

void rgb15to32(const uint8_t *src,uint8_t *dst,unsigned src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb15to32_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb15to32_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb15to32_MMX(src, dst, src_size);
	else
#endif
		rgb15to32_C(src, dst, src_size);
}

void rgb16to32(const uint8_t *src,uint8_t *dst,unsigned src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb16to32_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb16to32_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb16to32_MMX(src, dst, src_size);
	else
#endif
		rgb16to32_C(src, dst, src_size);
Nick Kurshev's avatar
Nick Kurshev committed
210
}
211

212
void rgb32to24(const uint8_t *src,uint8_t *dst,unsigned src_size)
213
{
214 215 216 217 218 219 220 221 222
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb32to24_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb32to24_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb32to24_MMX(src, dst, src_size);
	else
223
#endif
Arpi's avatar
Arpi committed
224
		rgb32to24_C(src, dst, src_size);
225
}
226

227 228 229
/*
 Original by Strepto/Astral
 ported to gcc & bugfixed : A'rpi
Nick Kurshev's avatar
Nick Kurshev committed
230
 MMX2, 3DNOW optimization by Nick Kurshev
231
 32bit c version, and and&add trick by Michael Niedermayer
232
*/
233
void rgb15to16(const uint8_t *src,uint8_t *dst,unsigned src_size)
234
{
235 236 237 238 239 240 241 242 243
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb15to16_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb15to16_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb15to16_MMX(src, dst, src_size);
	else
244
#endif
Arpi's avatar
Arpi committed
245
		rgb15to16_C(src, dst, src_size);
246
}
247

248 249 250 251 252 253 254 255 256 257 258 259 260 261
void rgb16to15(const uint8_t *src,uint8_t *dst,unsigned src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb16to15_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb16to15_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb16to15_MMX(src, dst, src_size);
	else
#endif
		rgb16to15_C(src, dst, src_size);
}
262 263 264
/**
 * Pallete is assumed to contain bgr32
 */
265
void palette8torgb32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
266
{
267
	unsigned i;
268 269

/*
Michael Niedermayer's avatar
Michael Niedermayer committed
270
	for(i=0; i<num_pixels; i++)
271
		((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
272
*/
273

274 275 276
	for(i=0; i<num_pixels; i++)
	{
		//FIXME slow?
277 278 279 280
		dst[0]= palette[ src[i]*4+2 ];
		dst[1]= palette[ src[i]*4+1 ];
		dst[2]= palette[ src[i]*4+0 ];
//		dst[3]= 0; /* do we need this cleansing? */
281 282 283
		dst+= 4;
	}
}
284

285 286 287 288
void palette8tobgr32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
{
	unsigned i;
	for(i=0; i<num_pixels; i++)
289 290 291 292 293 294 295 296
	{
		//FIXME slow?
		dst[0]= palette[ src[i]*4+0 ];
		dst[1]= palette[ src[i]*4+1 ];
		dst[2]= palette[ src[i]*4+2 ];
//		dst[3]= 0; /* do we need this cleansing? */
		dst+= 4;
	}
297 298
}

Michael Niedermayer's avatar
Michael Niedermayer committed
299 300 301
/**
 * Pallete is assumed to contain bgr32
 */
302
void palette8torgb24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
Michael Niedermayer's avatar
Michael Niedermayer committed
303
{
304
	unsigned i;
Michael Niedermayer's avatar
Michael Niedermayer committed
305 306
/*
	writes 1 byte o much and might cause alignment issues on some architectures?
Michael Niedermayer's avatar
Michael Niedermayer committed
307
	for(i=0; i<num_pixels; i++)
308
		((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
Michael Niedermayer's avatar
Michael Niedermayer committed
309
*/
Michael Niedermayer's avatar
Michael Niedermayer committed
310
	for(i=0; i<num_pixels; i++)
Michael Niedermayer's avatar
Michael Niedermayer committed
311 312
	{
		//FIXME slow?
313
		dst[0]= palette[ src[i]*4+2 ];
Michael Niedermayer's avatar
Michael Niedermayer committed
314
		dst[1]= palette[ src[i]*4+1 ];
315
		dst[2]= palette[ src[i]*4+0 ];
Michael Niedermayer's avatar
Michael Niedermayer committed
316 317 318 319
		dst+= 3;
	}
}

320 321 322 323 324 325 326 327 328 329 330
void palette8tobgr24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
{
	unsigned i;
/*
	writes 1 byte o much and might cause alignment issues on some architectures?
	for(i=0; i<num_pixels; i++)
		((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
*/
	for(i=0; i<num_pixels; i++)
	{
		//FIXME slow?
331 332 333
		dst[0]= palette[ src[i]*4+0 ];
		dst[1]= palette[ src[i]*4+1 ];
		dst[2]= palette[ src[i]*4+2 ];
334 335 336 337
		dst+= 3;
	}
}

338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
void bgr24torgb24(const uint8_t *src, uint8_t *dst, unsigned src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		bgr24torgb24_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		bgr24torgb24_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		bgr24torgb24_MMX(src, dst, src_size);
	else
		bgr24torgb24_C(src, dst, src_size);
#else
		bgr24torgb24_C(src, dst, src_size);
#endif
}

355
void rgb32to16(const uint8_t *src, uint8_t *dst, unsigned src_size)
356
{
357 358 359 360 361 362 363 364 365
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb32to16_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb32to16_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb32to16_MMX(src, dst, src_size);
	else
366
#endif
Arpi's avatar
Arpi committed
367
		rgb32to16_C(src, dst, src_size);
368 369
}

370
void rgb32to15(const uint8_t *src, uint8_t *dst, unsigned src_size)
371
{
372 373 374 375 376 377 378 379 380
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb32to15_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb32to15_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb32to15_MMX(src, dst, src_size);
	else
381
#endif
Arpi's avatar
Arpi committed
382
		rgb32to15_C(src, dst, src_size);
383 384
}

385 386
void rgb24to16(const uint8_t *src, uint8_t *dst, unsigned src_size)
{
387 388 389 390 391 392 393 394 395
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb24to16_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb24to16_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb24to16_MMX(src, dst, src_size);
	else
Nick Kurshev's avatar
Nick Kurshev committed
396
#endif
Arpi's avatar
Arpi committed
397
		rgb24to16_C(src, dst, src_size);
398 399 400 401
}

void rgb24to15(const uint8_t *src, uint8_t *dst, unsigned src_size)
{
402 403 404 405 406 407 408 409 410
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb24to15_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb24to15_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb24to15_MMX(src, dst, src_size);
	else
411
#endif
Arpi's avatar
Arpi committed
412
		rgb24to15_C(src, dst, src_size);
413
}
414 415 416 417

/**
 * Palette is assumed to contain bgr16, see rgb32to16 to convert the palette
 */
418
void palette8torgb16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
419
{
420
	unsigned i;
Michael Niedermayer's avatar
Michael Niedermayer committed
421
	for(i=0; i<num_pixels; i++)
422 423
		((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
}
424 425 426 427 428 429
void palette8tobgr16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
{
	unsigned i;
	for(i=0; i<num_pixels; i++)
		((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
}
430 431 432 433

/**
 * Pallete is assumed to contain bgr15, see rgb32to15 to convert the palette
 */
434
void palette8torgb15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
435
{
436
	unsigned i;
Michael Niedermayer's avatar
Michael Niedermayer committed
437
	for(i=0; i<num_pixels; i++)
438
		((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
Michael Niedermayer's avatar
Michael Niedermayer committed
439
}
440 441 442 443 444 445
void palette8tobgr15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
{
	unsigned i;
	for(i=0; i<num_pixels; i++)
		((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
}
446 447 448

void rgb32tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
449 450 451 452 453 454 455 456 457
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb32tobgr32_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb32tobgr32_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb32tobgr32_MMX(src, dst, src_size);
	else
458
#endif
Arpi's avatar
Arpi committed
459
		rgb32tobgr32_C(src, dst, src_size);
460 461
}

462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506
void rgb32tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	unsigned i;
	unsigned num_pixels = src_size >> 2;
	for(i=0; i<num_pixels; i++)
	{
		dst[3*i + 0] = src[4*i + 2];
		dst[3*i + 1] = src[4*i + 1];
		dst[3*i + 2] = src[4*i + 0];
	}
}

void rgb32tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb32tobgr16_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb32tobgr16_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb32tobgr16_MMX(src, dst, src_size);
	else
#endif
		rgb32tobgr16_C(src, dst, src_size);
}

void rgb32tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb32tobgr15_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb32tobgr15_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb32tobgr15_MMX(src, dst, src_size);
	else
#endif
		rgb32tobgr15_C(src, dst, src_size);
}

void rgb24tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	unsigned i;
507
	for(i=0; 3*i<src_size; i++)
508 509 510 511 512 513 514 515
	{
		dst[4*i + 0] = src[3*i + 2];
		dst[4*i + 1] = src[3*i + 1];
		dst[4*i + 2] = src[3*i + 0];
		dst[4*i + 3] = 0;
	}
}

Michael Niedermayer's avatar
Michael Niedermayer committed
516 517 518 519 520 521 522 523 524 525 526 527
void rgb24tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb24tobgr24_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb24tobgr24_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb24tobgr24_MMX(src, dst, src_size);
	else
#endif
Arpi's avatar
Arpi committed
528
		rgb24tobgr24_C(src, dst, src_size);
Michael Niedermayer's avatar
Michael Niedermayer committed
529 530
}

531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710
void rgb24tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb24tobgr16_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb24tobgr16_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb24tobgr16_MMX(src, dst, src_size);
	else
#endif
		rgb24tobgr16_C(src, dst, src_size);
}

void rgb24tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb24tobgr15_MMX2(src, dst, src_size);
	else if(gCpuCaps.has3DNow)
		rgb24tobgr15_3DNow(src, dst, src_size);
	else if(gCpuCaps.hasMMX)
		rgb24tobgr15_MMX(src, dst, src_size);
	else
#endif
		rgb24tobgr15_C(src, dst, src_size);
}

void rgb16tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	const uint16_t *end;
	uint8_t *d = (uint8_t *)dst;
	const uint16_t *s = (uint16_t *)src;
	end = s + src_size/2;
	while(s < end)
	{
		register uint16_t bgr;
		bgr = *s++;
		*d++ = (bgr&0xF800)>>8;
		*d++ = (bgr&0x7E0)>>3;
		*d++ = (bgr&0x1F)<<3;
		*d++ = 0;
	}
}

void rgb16tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	const uint16_t *end;
	uint8_t *d = (uint8_t *)dst;
	const uint16_t *s = (const uint16_t *)src;
	end = s + src_size/2;
	while(s < end)
	{
		register uint16_t bgr;
		bgr = *s++;
		*d++ = (bgr&0xF800)>>8;
		*d++ = (bgr&0x7E0)>>3;
		*d++ = (bgr&0x1F)<<3;
	}
}

void rgb16tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	unsigned i;
	unsigned num_pixels = src_size >> 1;
	
	for(i=0; i<num_pixels; i++)
	{
	    unsigned b,g,r;
	    register uint16_t rgb;
	    rgb = src[2*i];
	    r = rgb&0x1F;
	    g = (rgb&0x7E0)>>5;
	    b = (rgb&0xF800)>>11;
	    dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
	}
}

void rgb16tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	unsigned i;
	unsigned num_pixels = src_size >> 1;
	
	for(i=0; i<num_pixels; i++)
	{
	    unsigned b,g,r;
	    register uint16_t rgb;
	    rgb = src[2*i];
	    r = rgb&0x1F;
	    g = (rgb&0x7E0)>>5;
	    b = (rgb&0xF800)>>11;
	    dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
	}
}

void rgb15tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	const uint16_t *end;
	uint8_t *d = (uint8_t *)dst;
	const uint16_t *s = (const uint16_t *)src;
	end = s + src_size/2;
	while(s < end)
	{
		register uint16_t bgr;
		bgr = *s++;
		*d++ = (bgr&0x7C00)>>7;
		*d++ = (bgr&0x3E0)>>2;
		*d++ = (bgr&0x1F)<<3;
		*d++ = 0;
	}
}

void rgb15tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	const uint16_t *end;
	uint8_t *d = (uint8_t *)dst;
	const uint16_t *s = (uint16_t *)src;
	end = s + src_size/2;
	while(s < end)
	{
		register uint16_t bgr;
		bgr = *s++;
		*d++ = (bgr&0x7C00)>>7;
		*d++ = (bgr&0x3E0)>>2;
		*d++ = (bgr&0x1F)<<3;
	}
}

void rgb15tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	unsigned i;
	unsigned num_pixels = src_size >> 1;
	
	for(i=0; i<num_pixels; i++)
	{
	    unsigned b,g,r;
	    register uint16_t rgb;
	    rgb = src[2*i];
	    r = rgb&0x1F;
	    g = (rgb&0x3E0)>>5;
	    b = (rgb&0x7C00)>>10;
	    dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
	}
}

void rgb15tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	unsigned i;
	unsigned num_pixels = src_size >> 1;
	
	for(i=0; i<num_pixels; i++)
	{
	    unsigned b,g,r;
	    register uint16_t rgb;
	    rgb = src[2*i];
	    r = rgb&0x1F;
	    g = (rgb&0x3E0)>>5;
	    b = (rgb&0x7C00)>>10;
	    dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
	}
}

void rgb8tobgr8(const uint8_t *src, uint8_t *dst, unsigned int src_size)
{
	unsigned i;
	unsigned num_pixels = src_size;
	for(i=0; i<num_pixels; i++)
	{
	    unsigned b,g,r;
	    register uint8_t rgb;
	    rgb = src[i];
	    r = (rgb&0x07);
	    g = (rgb&0x38)>>3;
	    b = (rgb&0xC0)>>6;
	    dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
	}
}

Michael Niedermayer's avatar
Michael Niedermayer committed
711 712
/**
 *
713 714
 * height should be a multiple of 2 and width should be a multiple of 16 (if this is a
 * problem for anyone then tell me, and ill fix it)
Michael Niedermayer's avatar
Michael Niedermayer committed
715
 */
Michael Niedermayer's avatar
Michael Niedermayer committed
716
void yv12toyuy2(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
Michael Niedermayer's avatar
Michael Niedermayer committed
717 718
	unsigned int width, unsigned int height,
	unsigned int lumStride, unsigned int chromStride, unsigned int dstStride)
Michael Niedermayer's avatar
Michael Niedermayer committed
719
{
720 721 722 723 724 725 726 727 728
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		yv12toyuy2_MMX2(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
	else if(gCpuCaps.has3DNow)
		yv12toyuy2_3DNow(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
	else if(gCpuCaps.hasMMX)
		yv12toyuy2_MMX(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
	else
Michael Niedermayer's avatar
Michael Niedermayer committed
729
#endif
Arpi's avatar
Arpi committed
730
		yv12toyuy2_C(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
Michael Niedermayer's avatar
Michael Niedermayer committed
731 732
}

733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
/**
 *
 * width should be a multiple of 16
 */
void yuv422ptoyuy2(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
	unsigned int width, unsigned int height,
	unsigned int lumStride, unsigned int chromStride, unsigned int dstStride)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		yuv422ptoyuy2_MMX2(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
	else if(gCpuCaps.has3DNow)
		yuv422ptoyuy2_3DNow(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
	else if(gCpuCaps.hasMMX)
		yuv422ptoyuy2_MMX(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
	else
#endif
Arpi's avatar
Arpi committed
751
		yuv422ptoyuy2_C(ysrc, usrc, vsrc, dst, width, height, lumStride, chromStride, dstStride);
752 753
}

754 755 756 757 758 759
/**
 *
 * height should be a multiple of 2 and width should be a multiple of 16 (if this is a
 * problem for anyone then tell me, and ill fix it)
 */
void yuy2toyv12(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
Michael Niedermayer's avatar
Michael Niedermayer committed
760 761
	unsigned int width, unsigned int height,
	unsigned int lumStride, unsigned int chromStride, unsigned int srcStride)
Michael Niedermayer's avatar
Michael Niedermayer committed
762
{
763 764 765 766 767 768 769 770 771
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		yuy2toyv12_MMX2(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else if(gCpuCaps.has3DNow)
		yuy2toyv12_3DNow(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else if(gCpuCaps.hasMMX)
		yuy2toyv12_MMX(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else
Michael Niedermayer's avatar
Michael Niedermayer committed
772
#endif
Arpi's avatar
Arpi committed
773
		yuy2toyv12_C(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
Michael Niedermayer's avatar
Michael Niedermayer committed
774
}
Arpi's avatar
Arpi committed
775 776 777 778 779

/**
 *
 * height should be a multiple of 2 and width should be a multiple of 16 (if this is a
 * problem for anyone then tell me, and ill fix it)
780
 * chrominance data is only taken from every secound line others are ignored FIXME write HQ version
Arpi's avatar
Arpi committed
781 782 783 784 785
 */
void uyvytoyv12(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
	unsigned int width, unsigned int height,
	unsigned int lumStride, unsigned int chromStride, unsigned int srcStride)
{
786 787 788 789 790 791 792 793 794 795
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		uyvytoyv12_MMX2(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else if(gCpuCaps.has3DNow)
		uyvytoyv12_3DNow(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else if(gCpuCaps.hasMMX)
		uyvytoyv12_MMX(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else
		uyvytoyv12_C(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
796
#else
797
		uyvytoyv12_C(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
798
#endif
Arpi's avatar
Arpi committed
799 800
}

801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
void yvu9toyv12(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc,
	uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
	unsigned int width, unsigned int height,
	unsigned int lumStride, unsigned int chromStride)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		yvu9toyv12_MMX2(ysrc, usrc, vsrc, ydst, udst, vdst, width, height, lumStride, chromStride);
	else if(gCpuCaps.has3DNow)
		yvu9toyv12_3DNow(ysrc, usrc, vsrc, ydst, udst, vdst, width, height, lumStride, chromStride);
	else if(gCpuCaps.hasMMX)
		yvu9toyv12_MMX(ysrc, usrc, vsrc, ydst, udst, vdst, width, height, lumStride, chromStride);
	else
		yvu9toyv12_C(ysrc, usrc, vsrc, ydst, udst, vdst, width, height, lumStride, chromStride);
#else
		yvu9toyv12_C(ysrc, usrc, vsrc, ydst, udst, vdst, width, height, lumStride, chromStride);
#endif
}

821 822 823 824 825 826 827 828 829 830 831 832 833
void planar2x(const uint8_t *src, uint8_t *dst, int width, int height, int srcStride, int dstStride)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		planar2x_MMX2(src, dst, width, height, srcStride, dstStride);
	else if(gCpuCaps.has3DNow)
		planar2x_3DNow(src, dst, width, height, srcStride, dstStride);
	else
#endif
		planar2x_C(src, dst, width, height, srcStride, dstStride);
}

834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853
/**
 *
 * height should be a multiple of 2 and width should be a multiple of 2 (if this is a
 * problem for anyone then tell me, and ill fix it)
 * chrominance data is only taken from every secound line others are ignored FIXME write HQ version
 */
void rgb24toyv12(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
	unsigned int width, unsigned int height,
	unsigned int lumStride, unsigned int chromStride, unsigned int srcStride)
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		rgb24toyv12_MMX2(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else if(gCpuCaps.has3DNow)
		rgb24toyv12_3DNow(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else if(gCpuCaps.hasMMX)
		rgb24toyv12_MMX(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
	else
#endif
Arpi's avatar
Arpi committed
854
		rgb24toyv12_C(src, ydst, udst, vdst, width,  height, lumStride, chromStride, srcStride);
855
}
856 857

void interleaveBytes(uint8_t *src1, uint8_t *src2, uint8_t *dst,
Arpi's avatar
Arpi committed
858 859
		     unsigned width, unsigned height, unsigned src1Stride,
		     unsigned src2Stride, unsigned dstStride)
860 861 862 863 864 865 866 867 868 869 870
{
#ifdef CAN_COMPILE_X86_ASM
	// ordered per speed fasterst first
	if(gCpuCaps.hasMMX2)
		interleaveBytes_MMX2(src1, src2, dst, width, height, src1Stride, src2Stride, dstStride);
	else if(gCpuCaps.has3DNow)
		interleaveBytes_3DNow(src1, src2, dst, width, height, src1Stride, src2Stride, dstStride);
	else if(gCpuCaps.hasMMX)
		interleaveBytes_MMX(src1, src2, dst, width, height, src1Stride, src2Stride, dstStride);
	else
#endif
Arpi's avatar
Arpi committed
871
		interleaveBytes_C(src1, src2, dst, width, height, src1Stride, src2Stride, dstStride);
872
}
873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908

void vu9_to_vu12(const uint8_t *src1, const uint8_t *src2,
		uint8_t *dst1, uint8_t *dst2,
		unsigned width, unsigned height,
		unsigned srcStride1, unsigned srcStride2,
		unsigned dstStride1, unsigned dstStride2)
{
#ifdef CAN_COMPILE_X86_ASM
	if(gCpuCaps.hasMMX2)
		vu9_to_vu12_MMX2(src1, src2, dst1, dst2, width, height, srcStride1, srcStride2, dstStride1, dstStride2);
	else if(gCpuCaps.has3DNow)
		vu9_to_vu12_3DNow(src1, src2, dst1, dst2, width, height, srcStride1, srcStride2, dstStride1, dstStride2);
	else if(gCpuCaps.hasMMX)
		vu9_to_vu12_MMX(src1, src2, dst1, dst2, width, height, srcStride1, srcStride2, dstStride1, dstStride2);
	else
#endif
		vu9_to_vu12_C(src1, src2, dst1, dst2, width, height, srcStride1, srcStride2, dstStride1, dstStride2);
}

void yvu9_to_yuy2(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
		uint8_t *dst,
		unsigned width, unsigned height,
		unsigned srcStride1, unsigned srcStride2,
		unsigned srcStride3, unsigned dstStride)
{
#ifdef CAN_COMPILE_X86_ASM
	if(gCpuCaps.hasMMX2)
		yvu9_to_yuy2_MMX2(src1, src2, src3, dst, width, height, srcStride1, srcStride2, srcStride3, dstStride);
	else if(gCpuCaps.has3DNow)
		yvu9_to_yuy2_3DNow(src1, src2, src3, dst, width, height, srcStride1, srcStride2, srcStride3, dstStride);
	else if(gCpuCaps.hasMMX)
		yvu9_to_yuy2_MMX(src1, src2, src3, dst, width, height, srcStride1, srcStride2, srcStride3, dstStride);
	else
#endif
		yvu9_to_yuy2_C(src1, src2, src3, dst, width, height, srcStride1, srcStride2, srcStride3, dstStride);
}