git.sesse.net Git - ffmpeg/blob - postproc/rgb2rgb.c

   1 /*
   2  *
   3  *  rgb2rgb.c, Software RGB to RGB convertor
   4  *  pluralize by Software PAL8 to RGB convertor
   5  *               Software YUV to YUV convertor
   6  *               Software YUV to RGB convertor
   7  *  Written by Nick Kurshev.
   8  *  palette & yuv & runtime cpu stuff by Michael (michaelni@gmx.at) (under GPL)
   9  */
  10 #include <inttypes.h>
  11 #include "../config.h"
  12 #include "rgb2rgb.h"
  13 #include "swscale.h"
  14 #include "../mangle.h"
  15 #include "../bswap.h"
  16 #include "../libvo/fastmemcpy.h"
  17
  18 #define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit
  19
  20 void (*rgb24to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  21 void (*rgb24to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  22 void (*rgb24to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  23 void (*rgb32to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  24 void (*rgb32to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  25 void (*rgb32to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  26 void (*rgb15to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  27 void (*rgb15to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  28 void (*rgb15to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  29 void (*rgb16to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  30 void (*rgb16to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  31 void (*rgb16to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  32 //void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  33 void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  34 void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  35 void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  36 void (*rgb32tobgr32)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  37 //void (*rgb32tobgr24)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  38 void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  39 void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  40
  41 void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
  42         unsigned int width, unsigned int height,
  43         int lumStride, int chromStride, int dstStride);
  44 void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
  45         unsigned int width, unsigned int height,
  46         int lumStride, int chromStride, int dstStride);
  47 void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
  48         unsigned int width, unsigned int height,
  49         int lumStride, int chromStride, int srcStride);
  50 void (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
  51         unsigned int width, unsigned int height,
  52         int lumStride, int chromStride, int srcStride);
  53 void (*planar2x)(const uint8_t *src, uint8_t *dst, int width, int height, int srcStride, int dstStride);
  54 void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
  55                             unsigned width, unsigned height, int src1Stride,
  56                             int src2Stride, int dstStride);
  57 void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
  58                         uint8_t *dst1, uint8_t *dst2,
  59                         unsigned width, unsigned height,
  60                         int srcStride1, int srcStride2,
  61                         int dstStride1, int dstStride2);
  62 void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
  63                         uint8_t *dst,
  64                         unsigned width, unsigned height,
  65                         int srcStride1, int srcStride2,
  66                         int srcStride3, int dstStride);
  67
  68 #ifdef ARCH_X86
  69 static const uint64_t mmx_null  __attribute__((aligned(8))) = 0x0000000000000000ULL;
  70 static const uint64_t mmx_one   __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
  71 static const uint64_t mask32b  __attribute__((aligned(8))) = 0x000000FF000000FFULL;
  72 static const uint64_t mask32g  __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
  73 static const uint64_t mask32r  __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
  74 static const uint64_t mask32   __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
  75 static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL;
  76 static const uint64_t mask3216g  __attribute__((aligned(8)))=0x0000FC000000FC00ULL;
  77 static const uint64_t mask3215g  __attribute__((aligned(8)))=0x0000F8000000F800ULL;
  78 static const uint64_t mul3216  __attribute__((aligned(8))) = 0x2000000420000004ULL;
  79 static const uint64_t mul3215  __attribute__((aligned(8))) = 0x2000000820000008ULL;
  80 static const uint64_t mask24b  __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
  81 static const uint64_t mask24g  __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
  82 static const uint64_t mask24r  __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
  83 static const uint64_t mask24l  __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
  84 static const uint64_t mask24h  __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
  85 static const uint64_t mask24hh  __attribute__((aligned(8))) = 0xffff000000000000ULL;
  86 static const uint64_t mask24hhh  __attribute__((aligned(8))) = 0xffffffff00000000ULL;
  87 static const uint64_t mask24hhhh  __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
  88 static const uint64_t mask15b  __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111  xxB */
  89 static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000  RGx */
  90 static const uint64_t mask15s  __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
  91 static const uint64_t mask15g  __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
  92 static const uint64_t mask15r  __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
  93 #define mask16b mask15b
  94 static const uint64_t mask16g  __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
  95 static const uint64_t mask16r  __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
  96 static const uint64_t red_16mask  __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
  97 static const uint64_t green_16mask __attribute__((aligned(8)))= 0x000007e0000007e0ULL;
  98 static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
  99 static const uint64_t red_15mask  __attribute__((aligned(8))) = 0x00007c000000f800ULL;
 100 static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000007e0ULL;
 101 static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
 102
 103 #ifdef FAST_BGR2YV12
 104 static const uint64_t bgr2YCoeff  __attribute__((aligned(8))) = 0x000000210041000DULL;
 105 static const uint64_t bgr2UCoeff  __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
 106 static const uint64_t bgr2VCoeff  __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
 107 #else
 108 static const uint64_t bgr2YCoeff  __attribute__((aligned(8))) = 0x000020E540830C8BULL;
 109 static const uint64_t bgr2UCoeff  __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
 110 static const uint64_t bgr2VCoeff  __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
 111 #endif
 112 static const uint64_t bgr2YOffset __attribute__((aligned(8))) = 0x1010101010101010ULL;
 113 static const uint64_t bgr2UVOffset __attribute__((aligned(8)))= 0x8080808080808080ULL;
 114 static const uint64_t w1111       __attribute__((aligned(8))) = 0x0001000100010001ULL;
 115
 116 #if 0
 117 static volatile uint64_t __attribute__((aligned(8))) b5Dither;
 118 static volatile uint64_t __attribute__((aligned(8))) g5Dither;
 119 static volatile uint64_t __attribute__((aligned(8))) g6Dither;
 120 static volatile uint64_t __attribute__((aligned(8))) r5Dither;
 121
 122 static uint64_t __attribute__((aligned(8))) dither4[2]={
 123         0x0103010301030103LL,
 124         0x0200020002000200LL,};
 125
 126 static uint64_t __attribute__((aligned(8))) dither8[2]={
 127         0x0602060206020602LL,
 128         0x0004000400040004LL,};
 129 #endif
 130 #endif
 131
 132 #define RGB2YUV_SHIFT 8
 133 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
 134 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
 135 #define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
 136 #define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
 137 #define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
 138 #define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
 139 #define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
 140 #define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
 141 #define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
 142
 143 //Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
 144 //Plain C versions
 145 #undef HAVE_MMX
 146 #undef HAVE_MMX2
 147 #undef HAVE_3DNOW
 148 #undef ARCH_X86
 149 #undef HAVE_SSE2
 150 #define RENAME(a) a ## _C
 151 #include "rgb2rgb_template.c"
 152
 153 #ifdef ARCH_X86
 154
 155 //MMX versions
 156 #undef RENAME
 157 #define HAVE_MMX
 158 #undef HAVE_MMX2
 159 #undef HAVE_3DNOW
 160 #undef HAVE_SSE2
 161 #define ARCH_X86
 162 #define RENAME(a) a ## _MMX
 163 #include "rgb2rgb_template.c"
 164
 165 //MMX2 versions
 166 #undef RENAME
 167 #define HAVE_MMX
 168 #define HAVE_MMX2
 169 #undef HAVE_3DNOW
 170 #undef HAVE_SSE2
 171 #define ARCH_X86
 172 #define RENAME(a) a ## _MMX2
 173 #include "rgb2rgb_template.c"
 174
 175 //3DNOW versions
 176 #undef RENAME
 177 #define HAVE_MMX
 178 #undef HAVE_MMX2
 179 #define HAVE_3DNOW
 180 #undef HAVE_SSE2
 181 #define ARCH_X86
 182 #define RENAME(a) a ## _3DNow
 183 #include "rgb2rgb_template.c"
 184
 185 #endif //ARCH_X86
 186
 187 /*
 188  rgb15->rgb16 Original by Strepto/Astral
 189  ported to gcc & bugfixed : A'rpi
 190  MMX2, 3DNOW optimization by Nick Kurshev
 191  32bit c version, and and&add trick by Michael Niedermayer
 192 */
 193
 194 void sws_rgb2rgb_init(int flags){
 195 #ifdef ARCH_X86
 196         if(flags & SWS_CPU_CAPS_MMX2){
 197                 rgb15to16= rgb15to16_MMX2;
 198                 rgb15to24= rgb15to24_MMX2;
 199                 rgb15to32= rgb15to32_MMX2;
 200                 rgb16to24= rgb16to24_MMX2;
 201                 rgb16to32= rgb16to32_MMX2;
 202                 rgb16to15= rgb16to15_MMX2;
 203                 rgb24to16= rgb24to16_MMX2;
 204                 rgb24to15= rgb24to15_MMX2;
 205                 rgb24to32= rgb24to32_MMX2;
 206                 rgb32to16= rgb32to16_MMX2;
 207                 rgb32to15= rgb32to15_MMX2;
 208                 rgb32to24= rgb32to24_MMX2;
 209                 rgb24tobgr15= rgb24tobgr15_MMX2;
 210                 rgb24tobgr16= rgb24tobgr16_MMX2;
 211                 rgb24tobgr24= rgb24tobgr24_MMX2;
 212                 rgb32tobgr32= rgb32tobgr32_MMX2;
 213                 rgb32tobgr16= rgb32tobgr16_MMX2;
 214                 rgb32tobgr15= rgb32tobgr15_MMX2;
 215                 yv12toyuy2= yv12toyuy2_MMX2;
 216                 yuv422ptoyuy2= yuv422ptoyuy2_MMX2;
 217                 yuy2toyv12= yuy2toyv12_MMX2;
 218                 uyvytoyv12= uyvytoyv12_MMX2;
 219                 yvu9toyv12= yvu9toyv12_MMX2;
 220                 planar2x= planar2x_MMX2;
 221                 rgb24toyv12= rgb24toyv12_MMX2;
 222                 interleaveBytes= interleaveBytes_MMX2;
 223                 vu9_to_vu12= vu9_to_vu12_MMX2;
 224                 yvu9_to_yuy2= yvu9_to_yuy2_MMX2;
 225         }else if(flags & SWS_CPU_CAPS_3DNOW){
 226                 rgb15to16= rgb15to16_3DNOW;
 227                 rgb15to24= rgb15to24_3DNOW;
 228                 rgb15to32= rgb15to32_3DNOW;
 229                 rgb16to24= rgb16to24_3DNOW;
 230                 rgb16to32= rgb16to32_3DNOW;
 231                 rgb16to15= rgb16to15_3DNOW;
 232                 rgb24to16= rgb24to16_3DNOW;
 233                 rgb24to15= rgb24to15_3DNOW;
 234                 rgb24to32= rgb24to32_3DNOW;
 235                 rgb32to16= rgb32to16_3DNOW;
 236                 rgb32to15= rgb32to15_3DNOW;
 237                 rgb32to24= rgb32to24_3DNOW;
 238                 rgb24tobgr15= rgb24tobgr15_3DNOW;
 239                 rgb24tobgr16= rgb24tobgr16_3DNOW;
 240                 rgb24tobgr24= rgb24tobgr24_3DNOW;
 241                 rgb32tobgr32= rgb32tobgr32_3DNOW;
 242                 rgb32tobgr16= rgb32tobgr16_3DNOW;
 243                 rgb32tobgr15= rgb32tobgr15_3DNOW;
 244                 yv12toyuy2= yv12toyuy2_3DNOW;
 245                 yuv422ptoyuy2= yuv422ptoyuy2_3DNOW;
 246                 yuy2toyv12= yuy2toyv12_3DNOW;
 247                 uyvytoyv12= uyvytoyv12_3DNOW;
 248                 yvu9toyv12= yvu9toyv12_3DNOW;
 249                 planar2x= planar2x_3DNOW;
 250                 rgb24toyv12= rgb24toyv12_3DNOW;
 251                 interleaveBytes= interleaveBytes_3DNOW;
 252                 vu9_to_vu12= vu9_to_vu12_3DNOW;
 253                 yvu9_to_yuy2= yvu9_to_yuy2_3DNOW;
 254         }else if(flags & SWS_CPU_CAPS_MMX){
 255                 rgb15to16= rgb15to16_MMX;
 256                 rgb15to24= rgb15to24_MMX;
 257                 rgb15to32= rgb15to32_MMX;
 258                 rgb16to24= rgb16to24_MMX;
 259                 rgb16to32= rgb16to32_MMX;
 260                 rgb16to15= rgb16to15_MMX;
 261                 rgb24to16= rgb24to16_MMX;
 262                 rgb24to15= rgb24to15_MMX;
 263                 rgb24to32= rgb24to32_MMX;
 264                 rgb32to16= rgb32to16_MMX;
 265                 rgb32to15= rgb32to15_MMX;
 266                 rgb32to24= rgb32to24_MMX;
 267                 rgb24tobgr15= rgb24tobgr15_MMX;
 268                 rgb24tobgr16= rgb24tobgr16_MMX;
 269                 rgb24tobgr24= rgb24tobgr24_MMX;
 270                 rgb32tobgr32= rgb32tobgr32_MMX;
 271                 rgb32tobgr16= rgb32tobgr16_MMX;
 272                 rgb32tobgr15= rgb32tobgr15_MMX;
 273                 yv12toyuy2= yv12toyuy2_MMX;
 274                 yuv422ptoyuy2= yuv422ptoyuy2_MMX;
 275                 yuy2toyv12= yuy2toyv12_MMX;
 276                 uyvytoyv12= uyvytoyv12_MMX;
 277                 yvu9toyv12= yvu9toyv12_MMX;
 278                 planar2x= planar2x_MMX;
 279                 rgb24toyv12= rgb24toyv12_MMX;
 280                 interleaveBytes= interleaveBytes_MMX;
 281                 vu9_to_vu12= vu9_to_vu12_MMX;
 282                 yvu9_to_yuy2= yvu9_to_yuy2_MMX;
 283         }else
 284 #endif
 285         {
 286                 rgb15to16= rgb15to16_C;
 287                 rgb15to24= rgb15to24_C;
 288                 rgb15to32= rgb15to32_C;
 289                 rgb16to24= rgb16to24_C;
 290                 rgb16to32= rgb16to32_C;
 291                 rgb16to15= rgb16to15_C;
 292                 rgb24to16= rgb24to16_C;
 293                 rgb24to15= rgb24to15_C;
 294                 rgb24to32= rgb24to32_C;
 295                 rgb32to16= rgb32to16_C;
 296                 rgb32to15= rgb32to15_C;
 297                 rgb32to24= rgb32to24_C;
 298                 rgb24tobgr15= rgb24tobgr15_C;
 299                 rgb24tobgr16= rgb24tobgr16_C;
 300                 rgb24tobgr24= rgb24tobgr24_C;
 301                 rgb32tobgr32= rgb32tobgr32_C;
 302                 rgb32tobgr16= rgb32tobgr16_C;
 303                 rgb32tobgr15= rgb32tobgr15_C;
 304                 yv12toyuy2= yv12toyuy2_C;
 305                 yuv422ptoyuy2= yuv422ptoyuy2_C;
 306                 yuy2toyv12= yuy2toyv12_C;
 307 //              uyvytoyv12= uyvytoyv12_C;
 308 //              yvu9toyv12= yvu9toyv12_C;
 309                 planar2x= planar2x_C;
 310                 rgb24toyv12= rgb24toyv12_C;
 311                 interleaveBytes= interleaveBytes_C;
 312                 vu9_to_vu12= vu9_to_vu12_C;
 313                 yvu9_to_yuy2= yvu9_to_yuy2_C;
 314         }
 315 }
 316
 317 /**
 318  * Pallete is assumed to contain bgr32
 319  */
 320 void palette8torgb32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 321 {
 322         unsigned i;
 323
 324 /*
 325         for(i=0; i<num_pixels; i++)
 326                 ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
 327 */
 328
 329         for(i=0; i<num_pixels; i++)
 330         {
 331                 //FIXME slow?
 332                 dst[0]= palette[ src[i]*4+2 ];
 333                 dst[1]= palette[ src[i]*4+1 ];
 334                 dst[2]= palette[ src[i]*4+0 ];
 335 //              dst[3]= 0; /* do we need this cleansing? */
 336                 dst+= 4;
 337         }
 338 }
 339
 340 void palette8tobgr32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 341 {
 342         unsigned i;
 343         for(i=0; i<num_pixels; i++)
 344         {
 345                 //FIXME slow?
 346                 dst[0]= palette[ src[i]*4+0 ];
 347                 dst[1]= palette[ src[i]*4+1 ];
 348                 dst[2]= palette[ src[i]*4+2 ];
 349 //              dst[3]= 0; /* do we need this cleansing? */
 350                 dst+= 4;
 351         }
 352 }
 353
 354 /**
 355  * Pallete is assumed to contain bgr32
 356  */
 357 void palette8torgb24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 358 {
 359         unsigned i;
 360 /*
 361         writes 1 byte o much and might cause alignment issues on some architectures?
 362         for(i=0; i<num_pixels; i++)
 363                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
 364 */
 365         for(i=0; i<num_pixels; i++)
 366         {
 367                 //FIXME slow?
 368                 dst[0]= palette[ src[i]*4+2 ];
 369                 dst[1]= palette[ src[i]*4+1 ];
 370                 dst[2]= palette[ src[i]*4+0 ];
 371                 dst+= 3;
 372         }
 373 }
 374
 375 void palette8tobgr24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 376 {
 377         unsigned i;
 378 /*
 379         writes 1 byte o much and might cause alignment issues on some architectures?
 380         for(i=0; i<num_pixels; i++)
 381                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
 382 */
 383         for(i=0; i<num_pixels; i++)
 384         {
 385                 //FIXME slow?
 386                 dst[0]= palette[ src[i]*4+0 ];
 387                 dst[1]= palette[ src[i]*4+1 ];
 388                 dst[2]= palette[ src[i]*4+2 ];
 389                 dst+= 3;
 390         }
 391 }
 392
 393 /**
 394  * Palette is assumed to contain bgr16, see rgb32to16 to convert the palette
 395  */
 396 void palette8torgb16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 397 {
 398         unsigned i;
 399         for(i=0; i<num_pixels; i++)
 400                 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
 401 }
 402 void palette8tobgr16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 403 {
 404         unsigned i;
 405         for(i=0; i<num_pixels; i++)
 406                 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
 407 }
 408
 409 /**
 410  * Pallete is assumed to contain bgr15, see rgb32to15 to convert the palette
 411  */
 412 void palette8torgb15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 413 {
 414         unsigned i;
 415         for(i=0; i<num_pixels; i++)
 416                 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
 417 }
 418 void palette8tobgr15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 419 {
 420         unsigned i;
 421         for(i=0; i<num_pixels; i++)
 422                 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
 423 }
 424
 425 void rgb32tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 426 {
 427         unsigned i;
 428         unsigned num_pixels = src_size >> 2;
 429         for(i=0; i<num_pixels; i++)
 430         {
 431                 dst[3*i + 0] = src[4*i + 2];
 432                 dst[3*i + 1] = src[4*i + 1];
 433                 dst[3*i + 2] = src[4*i + 0];
 434         }
 435 }
 436
 437 void rgb24tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 438 {
 439         unsigned i;
 440         for(i=0; 3*i<src_size; i++)
 441         {
 442                 dst[4*i + 0] = src[3*i + 2];
 443                 dst[4*i + 1] = src[3*i + 1];
 444                 dst[4*i + 2] = src[3*i + 0];
 445                 dst[4*i + 3] = 0;
 446         }
 447 }
 448
 449 void rgb16tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 450 {
 451         const uint16_t *end;
 452         uint8_t *d = (uint8_t *)dst;
 453         const uint16_t *s = (uint16_t *)src;
 454         end = s + src_size/2;
 455         while(s < end)
 456         {
 457                 register uint16_t bgr;
 458                 bgr = *s++;
 459                 *d++ = (bgr&0xF800)>>8;
 460                 *d++ = (bgr&0x7E0)>>3;
 461                 *d++ = (bgr&0x1F)<<3;
 462                 *d++ = 0;
 463         }
 464 }
 465
 466 void rgb16tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 467 {
 468         const uint16_t *end;
 469         uint8_t *d = (uint8_t *)dst;
 470         const uint16_t *s = (const uint16_t *)src;
 471         end = s + src_size/2;
 472         while(s < end)
 473         {
 474                 register uint16_t bgr;
 475                 bgr = *s++;
 476                 *d++ = (bgr&0xF800)>>8;
 477                 *d++ = (bgr&0x7E0)>>3;
 478                 *d++ = (bgr&0x1F)<<3;
 479         }
 480 }
 481
 482 void rgb16tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 483 {
 484         unsigned i;
 485         unsigned num_pixels = src_size >> 1;
 486
 487         for(i=0; i<num_pixels; i++)
 488         {
 489             unsigned b,g,r;
 490             register uint16_t rgb;
 491             rgb = src[2*i];
 492             r = rgb&0x1F;
 493             g = (rgb&0x7E0)>>5;
 494             b = (rgb&0xF800)>>11;
 495             dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
 496         }
 497 }
 498
 499 void rgb16tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 500 {
 501         unsigned i;
 502         unsigned num_pixels = src_size >> 1;
 503
 504         for(i=0; i<num_pixels; i++)
 505         {
 506             unsigned b,g,r;
 507             register uint16_t rgb;
 508             rgb = src[2*i];
 509             r = rgb&0x1F;
 510             g = (rgb&0x7E0)>>5;
 511             b = (rgb&0xF800)>>11;
 512             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
 513         }
 514 }
 515
 516 void rgb15tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 517 {
 518         const uint16_t *end;
 519         uint8_t *d = (uint8_t *)dst;
 520         const uint16_t *s = (const uint16_t *)src;
 521         end = s + src_size/2;
 522         while(s < end)
 523         {
 524                 register uint16_t bgr;
 525                 bgr = *s++;
 526                 *d++ = (bgr&0x7C00)>>7;
 527                 *d++ = (bgr&0x3E0)>>2;
 528                 *d++ = (bgr&0x1F)<<3;
 529                 *d++ = 0;
 530         }
 531 }
 532
 533 void rgb15tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 534 {
 535         const uint16_t *end;
 536         uint8_t *d = (uint8_t *)dst;
 537         const uint16_t *s = (uint16_t *)src;
 538         end = s + src_size/2;
 539         while(s < end)
 540         {
 541                 register uint16_t bgr;
 542                 bgr = *s++;
 543                 *d++ = (bgr&0x7C00)>>7;
 544                 *d++ = (bgr&0x3E0)>>2;
 545                 *d++ = (bgr&0x1F)<<3;
 546         }
 547 }
 548
 549 void rgb15tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 550 {
 551         unsigned i;
 552         unsigned num_pixels = src_size >> 1;
 553
 554         for(i=0; i<num_pixels; i++)
 555         {
 556             unsigned b,g,r;
 557             register uint16_t rgb;
 558             rgb = src[2*i];
 559             r = rgb&0x1F;
 560             g = (rgb&0x3E0)>>5;
 561             b = (rgb&0x7C00)>>10;
 562             dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
 563         }
 564 }
 565
 566 void rgb15tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 567 {
 568         unsigned i;
 569         unsigned num_pixels = src_size >> 1;
 570
 571         for(i=0; i<num_pixels; i++)
 572         {
 573             unsigned b,g,r;
 574             register uint16_t rgb;
 575             rgb = src[2*i];
 576             r = rgb&0x1F;
 577             g = (rgb&0x3E0)>>5;
 578             b = (rgb&0x7C00)>>10;
 579             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
 580         }
 581 }
 582
 583 void rgb8tobgr8(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 584 {
 585         unsigned i;
 586         unsigned num_pixels = src_size;
 587         for(i=0; i<num_pixels; i++)
 588         {
 589             unsigned b,g,r;
 590             register uint8_t rgb;
 591             rgb = src[i];
 592             r = (rgb&0x07);
 593             g = (rgb&0x38)>>3;
 594             b = (rgb&0xC0)>>6;
 595             dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
 596         }
 597 }