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 (*yv12touyvy)(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 (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
  48         unsigned int width, unsigned int height,
  49         int lumStride, int chromStride, int dstStride);
  50 void (*yuy2toyv12)(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 (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
  54         unsigned int width, unsigned int height,
  55         int lumStride, int chromStride, int srcStride);
  56 void (*planar2x)(const uint8_t *src, uint8_t *dst, int width, int height, int srcStride, int dstStride);
  57 void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
  58                             unsigned width, unsigned height, int src1Stride,
  59                             int src2Stride, int dstStride);
  60 void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
  61                         uint8_t *dst1, uint8_t *dst2,
  62                         unsigned width, unsigned height,
  63                         int srcStride1, int srcStride2,
  64                         int dstStride1, int dstStride2);
  65 void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
  66                         uint8_t *dst,
  67                         unsigned width, unsigned height,
  68                         int srcStride1, int srcStride2,
  69                         int srcStride3, int dstStride);
  70
  71 #ifdef ARCH_X86
  72 static const uint64_t mmx_null  __attribute__((aligned(8))) = 0x0000000000000000ULL;
  73 static const uint64_t mmx_one   __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
  74 static const uint64_t mask32b  __attribute__((aligned(8))) = 0x000000FF000000FFULL;
  75 static const uint64_t mask32g  __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
  76 static const uint64_t mask32r  __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
  77 static const uint64_t mask32   __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
  78 static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL;
  79 static const uint64_t mask3216g  __attribute__((aligned(8)))=0x0000FC000000FC00ULL;
  80 static const uint64_t mask3215g  __attribute__((aligned(8)))=0x0000F8000000F800ULL;
  81 static const uint64_t mul3216  __attribute__((aligned(8))) = 0x2000000420000004ULL;
  82 static const uint64_t mul3215  __attribute__((aligned(8))) = 0x2000000820000008ULL;
  83 static const uint64_t mask24b  __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
  84 static const uint64_t mask24g  __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
  85 static const uint64_t mask24r  __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
  86 static const uint64_t mask24l  __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
  87 static const uint64_t mask24h  __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
  88 static const uint64_t mask24hh  __attribute__((aligned(8))) = 0xffff000000000000ULL;
  89 static const uint64_t mask24hhh  __attribute__((aligned(8))) = 0xffffffff00000000ULL;
  90 static const uint64_t mask24hhhh  __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
  91 static const uint64_t mask15b  __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111  xxB */
  92 static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000  RGx */
  93 static const uint64_t mask15s  __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
  94 static const uint64_t mask15g  __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
  95 static const uint64_t mask15r  __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
  96 #define mask16b mask15b
  97 static const uint64_t mask16g  __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
  98 static const uint64_t mask16r  __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
  99 static const uint64_t red_16mask  __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
 100 static const uint64_t green_16mask __attribute__((aligned(8)))= 0x000007e0000007e0ULL;
 101 static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
 102 static const uint64_t red_15mask  __attribute__((aligned(8))) = 0x00007c000000f800ULL;
 103 static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000007e0ULL;
 104 static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
 105
 106 #ifdef FAST_BGR2YV12
 107 static const uint64_t bgr2YCoeff  __attribute__((aligned(8))) = 0x000000210041000DULL;
 108 static const uint64_t bgr2UCoeff  __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
 109 static const uint64_t bgr2VCoeff  __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
 110 #else
 111 static const uint64_t bgr2YCoeff  __attribute__((aligned(8))) = 0x000020E540830C8BULL;
 112 static const uint64_t bgr2UCoeff  __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
 113 static const uint64_t bgr2VCoeff  __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
 114 #endif
 115 static const uint64_t bgr2YOffset __attribute__((aligned(8))) = 0x1010101010101010ULL;
 116 static const uint64_t bgr2UVOffset __attribute__((aligned(8)))= 0x8080808080808080ULL;
 117 static const uint64_t w1111       __attribute__((aligned(8))) = 0x0001000100010001ULL;
 118
 119 #if 0
 120 static volatile uint64_t __attribute__((aligned(8))) b5Dither;
 121 static volatile uint64_t __attribute__((aligned(8))) g5Dither;
 122 static volatile uint64_t __attribute__((aligned(8))) g6Dither;
 123 static volatile uint64_t __attribute__((aligned(8))) r5Dither;
 124
 125 static uint64_t __attribute__((aligned(8))) dither4[2]={
 126         0x0103010301030103LL,
 127         0x0200020002000200LL,};
 128
 129 static uint64_t __attribute__((aligned(8))) dither8[2]={
 130         0x0602060206020602LL,
 131         0x0004000400040004LL,};
 132 #endif
 133 #endif
 134
 135 #define RGB2YUV_SHIFT 8
 136 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
 137 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
 138 #define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
 139 #define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
 140 #define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
 141 #define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
 142 #define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
 143 #define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
 144 #define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
 145
 146 //Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
 147 //Plain C versions
 148 #undef HAVE_MMX
 149 #undef HAVE_MMX2
 150 #undef HAVE_3DNOW
 151 #undef HAVE_SSE2
 152 #define RENAME(a) a ## _C
 153 #include "rgb2rgb_template.c"
 154
 155 #ifdef ARCH_X86
 156
 157 //MMX versions
 158 #undef RENAME
 159 #define HAVE_MMX
 160 #undef HAVE_MMX2
 161 #undef HAVE_3DNOW
 162 #undef HAVE_SSE2
 163 #define RENAME(a) a ## _MMX
 164 #include "rgb2rgb_template.c"
 165
 166 //MMX2 versions
 167 #undef RENAME
 168 #define HAVE_MMX
 169 #define HAVE_MMX2
 170 #undef HAVE_3DNOW
 171 #undef HAVE_SSE2
 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 RENAME(a) a ## _3DNOW
 182 #include "rgb2rgb_template.c"
 183
 184 #endif //ARCH_X86
 185
 186 /*
 187  rgb15->rgb16 Original by Strepto/Astral
 188  ported to gcc & bugfixed : A'rpi
 189  MMX2, 3DNOW optimization by Nick Kurshev
 190  32bit c version, and and&add trick by Michael Niedermayer
 191 */
 192
 193 void sws_rgb2rgb_init(int flags){
 194 #ifdef ARCH_X86
 195         if(flags & SWS_CPU_CAPS_MMX2){
 196                 rgb15to16= rgb15to16_MMX2;
 197                 rgb15to24= rgb15to24_MMX2;
 198                 rgb15to32= rgb15to32_MMX2;
 199                 rgb16to24= rgb16to24_MMX2;
 200                 rgb16to32= rgb16to32_MMX2;
 201                 rgb16to15= rgb16to15_MMX2;
 202                 rgb24to16= rgb24to16_MMX2;
 203                 rgb24to15= rgb24to15_MMX2;
 204                 rgb24to32= rgb24to32_MMX2;
 205                 rgb32to16= rgb32to16_MMX2;
 206                 rgb32to15= rgb32to15_MMX2;
 207                 rgb32to24= rgb32to24_MMX2;
 208                 rgb24tobgr15= rgb24tobgr15_MMX2;
 209                 rgb24tobgr16= rgb24tobgr16_MMX2;
 210                 rgb24tobgr24= rgb24tobgr24_MMX2;
 211                 rgb32tobgr32= rgb32tobgr32_MMX2;
 212                 rgb32tobgr16= rgb32tobgr16_MMX2;
 213                 rgb32tobgr15= rgb32tobgr15_MMX2;
 214                 yv12toyuy2= yv12toyuy2_MMX2;
 215                 yv12touyvy= yv12touyvy_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                 yv12touyvy= yv12touyvy_3DNOW;
 246                 yuv422ptoyuy2= yuv422ptoyuy2_3DNOW;
 247                 yuy2toyv12= yuy2toyv12_3DNOW;
 248 //              uyvytoyv12= uyvytoyv12_3DNOW;
 249 //              yvu9toyv12= yvu9toyv12_3DNOW;
 250                 planar2x= planar2x_3DNOW;
 251                 rgb24toyv12= rgb24toyv12_3DNOW;
 252                 interleaveBytes= interleaveBytes_3DNOW;
 253                 vu9_to_vu12= vu9_to_vu12_3DNOW;
 254                 yvu9_to_yuy2= yvu9_to_yuy2_3DNOW;
 255         }else if(flags & SWS_CPU_CAPS_MMX){
 256                 rgb15to16= rgb15to16_MMX;
 257                 rgb15to24= rgb15to24_MMX;
 258                 rgb15to32= rgb15to32_MMX;
 259                 rgb16to24= rgb16to24_MMX;
 260                 rgb16to32= rgb16to32_MMX;
 261                 rgb16to15= rgb16to15_MMX;
 262                 rgb24to16= rgb24to16_MMX;
 263                 rgb24to15= rgb24to15_MMX;
 264                 rgb24to32= rgb24to32_MMX;
 265                 rgb32to16= rgb32to16_MMX;
 266                 rgb32to15= rgb32to15_MMX;
 267                 rgb32to24= rgb32to24_MMX;
 268                 rgb24tobgr15= rgb24tobgr15_MMX;
 269                 rgb24tobgr16= rgb24tobgr16_MMX;
 270                 rgb24tobgr24= rgb24tobgr24_MMX;
 271                 rgb32tobgr32= rgb32tobgr32_MMX;
 272                 rgb32tobgr16= rgb32tobgr16_MMX;
 273                 rgb32tobgr15= rgb32tobgr15_MMX;
 274                 yv12toyuy2= yv12toyuy2_MMX;
 275                 yv12touyvy= yv12touyvy_MMX;
 276                 yuv422ptoyuy2= yuv422ptoyuy2_MMX;
 277                 yuy2toyv12= yuy2toyv12_MMX;
 278 //              uyvytoyv12= uyvytoyv12_MMX;
 279 //              yvu9toyv12= yvu9toyv12_MMX;
 280                 planar2x= planar2x_MMX;
 281                 rgb24toyv12= rgb24toyv12_MMX;
 282                 interleaveBytes= interleaveBytes_MMX;
 283                 vu9_to_vu12= vu9_to_vu12_MMX;
 284                 yvu9_to_yuy2= yvu9_to_yuy2_MMX;
 285         }else
 286 #endif
 287         {
 288                 rgb15to16= rgb15to16_C;
 289                 rgb15to24= rgb15to24_C;
 290                 rgb15to32= rgb15to32_C;
 291                 rgb16to24= rgb16to24_C;
 292                 rgb16to32= rgb16to32_C;
 293                 rgb16to15= rgb16to15_C;
 294                 rgb24to16= rgb24to16_C;
 295                 rgb24to15= rgb24to15_C;
 296                 rgb24to32= rgb24to32_C;
 297                 rgb32to16= rgb32to16_C;
 298                 rgb32to15= rgb32to15_C;
 299                 rgb32to24= rgb32to24_C;
 300                 rgb24tobgr15= rgb24tobgr15_C;
 301                 rgb24tobgr16= rgb24tobgr16_C;
 302                 rgb24tobgr24= rgb24tobgr24_C;
 303                 rgb32tobgr32= rgb32tobgr32_C;
 304                 rgb32tobgr16= rgb32tobgr16_C;
 305                 rgb32tobgr15= rgb32tobgr15_C;
 306                 yv12toyuy2= yv12toyuy2_C;
 307                 yv12touyvy= yv12touyvy_C;
 308                 yuv422ptoyuy2= yuv422ptoyuy2_C;
 309                 yuy2toyv12= yuy2toyv12_C;
 310 //              uyvytoyv12= uyvytoyv12_C;
 311 //              yvu9toyv12= yvu9toyv12_C;
 312                 planar2x= planar2x_C;
 313                 rgb24toyv12= rgb24toyv12_C;
 314                 interleaveBytes= interleaveBytes_C;
 315                 vu9_to_vu12= vu9_to_vu12_C;
 316                 yvu9_to_yuy2= yvu9_to_yuy2_C;
 317         }
 318 }
 319
 320 /**
 321  * Pallete is assumed to contain bgr32
 322  */
 323 void palette8torgb32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 324 {
 325         unsigned i;
 326
 327 /*
 328         for(i=0; i<num_pixels; i++)
 329                 ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
 330 */
 331
 332         for(i=0; i<num_pixels; i++)
 333         {
 334                 //FIXME slow?
 335                 dst[0]= palette[ src[i]*4+2 ];
 336                 dst[1]= palette[ src[i]*4+1 ];
 337                 dst[2]= palette[ src[i]*4+0 ];
 338 //              dst[3]= 0; /* do we need this cleansing? */
 339                 dst+= 4;
 340         }
 341 }
 342
 343 void palette8tobgr32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 344 {
 345         unsigned i;
 346         for(i=0; i<num_pixels; i++)
 347         {
 348                 //FIXME slow?
 349                 dst[0]= palette[ src[i]*4+0 ];
 350                 dst[1]= palette[ src[i]*4+1 ];
 351                 dst[2]= palette[ src[i]*4+2 ];
 352 //              dst[3]= 0; /* do we need this cleansing? */
 353                 dst+= 4;
 354         }
 355 }
 356
 357 /**
 358  * Pallete is assumed to contain bgr32
 359  */
 360 void palette8torgb24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 361 {
 362         unsigned i;
 363 /*
 364         writes 1 byte o much and might cause alignment issues on some architectures?
 365         for(i=0; i<num_pixels; i++)
 366                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
 367 */
 368         for(i=0; i<num_pixels; i++)
 369         {
 370                 //FIXME slow?
 371                 dst[0]= palette[ src[i]*4+2 ];
 372                 dst[1]= palette[ src[i]*4+1 ];
 373                 dst[2]= palette[ src[i]*4+0 ];
 374                 dst+= 3;
 375         }
 376 }
 377
 378 void palette8tobgr24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 379 {
 380         unsigned i;
 381 /*
 382         writes 1 byte o much and might cause alignment issues on some architectures?
 383         for(i=0; i<num_pixels; i++)
 384                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
 385 */
 386         for(i=0; i<num_pixels; i++)
 387         {
 388                 //FIXME slow?
 389                 dst[0]= palette[ src[i]*4+0 ];
 390                 dst[1]= palette[ src[i]*4+1 ];
 391                 dst[2]= palette[ src[i]*4+2 ];
 392                 dst+= 3;
 393         }
 394 }
 395
 396 /**
 397  * Palette is assumed to contain bgr16, see rgb32to16 to convert the palette
 398  */
 399 void palette8torgb16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 400 {
 401         unsigned i;
 402         for(i=0; i<num_pixels; i++)
 403                 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
 404 }
 405 void palette8tobgr16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 406 {
 407         unsigned i;
 408         for(i=0; i<num_pixels; i++)
 409                 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
 410 }
 411
 412 /**
 413  * Pallete is assumed to contain bgr15, see rgb32to15 to convert the palette
 414  */
 415 void palette8torgb15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 416 {
 417         unsigned i;
 418         for(i=0; i<num_pixels; i++)
 419                 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
 420 }
 421 void palette8tobgr15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 422 {
 423         unsigned i;
 424         for(i=0; i<num_pixels; i++)
 425                 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
 426 }
 427
 428 void rgb32tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 429 {
 430         unsigned i;
 431         unsigned num_pixels = src_size >> 2;
 432         for(i=0; i<num_pixels; i++)
 433         {
 434                 dst[3*i + 0] = src[4*i + 2];
 435                 dst[3*i + 1] = src[4*i + 1];
 436                 dst[3*i + 2] = src[4*i + 0];
 437         }
 438 }
 439
 440 void rgb24tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 441 {
 442         unsigned i;
 443         for(i=0; 3*i<src_size; i++)
 444         {
 445                 dst[4*i + 0] = src[3*i + 2];
 446                 dst[4*i + 1] = src[3*i + 1];
 447                 dst[4*i + 2] = src[3*i + 0];
 448                 dst[4*i + 3] = 0;
 449         }
 450 }
 451
 452 void rgb16tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 453 {
 454         const uint16_t *end;
 455         uint8_t *d = (uint8_t *)dst;
 456         const uint16_t *s = (uint16_t *)src;
 457         end = s + src_size/2;
 458         while(s < end)
 459         {
 460                 register uint16_t bgr;
 461                 bgr = *s++;
 462                 *d++ = (bgr&0xF800)>>8;
 463                 *d++ = (bgr&0x7E0)>>3;
 464                 *d++ = (bgr&0x1F)<<3;
 465                 *d++ = 0;
 466         }
 467 }
 468
 469 void rgb16tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 470 {
 471         const uint16_t *end;
 472         uint8_t *d = (uint8_t *)dst;
 473         const uint16_t *s = (const uint16_t *)src;
 474         end = s + src_size/2;
 475         while(s < end)
 476         {
 477                 register uint16_t bgr;
 478                 bgr = *s++;
 479                 *d++ = (bgr&0xF800)>>8;
 480                 *d++ = (bgr&0x7E0)>>3;
 481                 *d++ = (bgr&0x1F)<<3;
 482         }
 483 }
 484
 485 void rgb16tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 486 {
 487         unsigned i;
 488         unsigned num_pixels = src_size >> 1;
 489
 490         for(i=0; i<num_pixels; i++)
 491         {
 492             unsigned b,g,r;
 493             register uint16_t rgb;
 494             rgb = src[2*i];
 495             r = rgb&0x1F;
 496             g = (rgb&0x7E0)>>5;
 497             b = (rgb&0xF800)>>11;
 498             dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
 499         }
 500 }
 501
 502 void rgb16tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 503 {
 504         unsigned i;
 505         unsigned num_pixels = src_size >> 1;
 506
 507         for(i=0; i<num_pixels; i++)
 508         {
 509             unsigned b,g,r;
 510             register uint16_t rgb;
 511             rgb = src[2*i];
 512             r = rgb&0x1F;
 513             g = (rgb&0x7E0)>>5;
 514             b = (rgb&0xF800)>>11;
 515             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
 516         }
 517 }
 518
 519 void rgb15tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 520 {
 521         const uint16_t *end;
 522         uint8_t *d = (uint8_t *)dst;
 523         const uint16_t *s = (const uint16_t *)src;
 524         end = s + src_size/2;
 525         while(s < end)
 526         {
 527                 register uint16_t bgr;
 528                 bgr = *s++;
 529                 *d++ = (bgr&0x7C00)>>7;
 530                 *d++ = (bgr&0x3E0)>>2;
 531                 *d++ = (bgr&0x1F)<<3;
 532                 *d++ = 0;
 533         }
 534 }
 535
 536 void rgb15tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 537 {
 538         const uint16_t *end;
 539         uint8_t *d = (uint8_t *)dst;
 540         const uint16_t *s = (uint16_t *)src;
 541         end = s + src_size/2;
 542         while(s < end)
 543         {
 544                 register uint16_t bgr;
 545                 bgr = *s++;
 546                 *d++ = (bgr&0x7C00)>>7;
 547                 *d++ = (bgr&0x3E0)>>2;
 548                 *d++ = (bgr&0x1F)<<3;
 549         }
 550 }
 551
 552 void rgb15tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 553 {
 554         unsigned i;
 555         unsigned num_pixels = src_size >> 1;
 556
 557         for(i=0; i<num_pixels; i++)
 558         {
 559             unsigned b,g,r;
 560             register uint16_t rgb;
 561             rgb = src[2*i];
 562             r = rgb&0x1F;
 563             g = (rgb&0x3E0)>>5;
 564             b = (rgb&0x7C00)>>10;
 565             dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
 566         }
 567 }
 568
 569 void rgb15tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 570 {
 571         unsigned i;
 572         unsigned num_pixels = src_size >> 1;
 573
 574         for(i=0; i<num_pixels; i++)
 575         {
 576             unsigned b,g,r;
 577             register uint16_t rgb;
 578             rgb = src[2*i];
 579             r = rgb&0x1F;
 580             g = (rgb&0x3E0)>>5;
 581             b = (rgb&0x7C00)>>10;
 582             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
 583         }
 584 }
 585
 586 void rgb8tobgr8(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 587 {
 588         unsigned i;
 589         unsigned num_pixels = src_size;
 590         for(i=0; i<num_pixels; i++)
 591         {
 592             unsigned b,g,r;
 593             register uint8_t rgb;
 594             rgb = src[i];
 595             r = (rgb&0x07);
 596             g = (rgb&0x38)>>3;
 597             b = (rgb&0xC0)>>6;
 598             dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
 599         }
 600 }