git.sesse.net Git - vlc/blob - modules/video_filter/swscale/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 "common.h"
  15
  16 #define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit
  17
  18 void (*rgb24to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  19 void (*rgb24to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  20 void (*rgb24to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  21 void (*rgb32to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  22 void (*rgb32to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  23 void (*rgb32to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  24 void (*rgb15to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  25 void (*rgb15to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  26 void (*rgb15to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  27 void (*rgb16to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  28 void (*rgb16to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  29 void (*rgb16to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
  30 //void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  31 void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  32 void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  33 void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  34 void (*rgb32tobgr32)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  35 //void (*rgb32tobgr24)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  36 void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  37 void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, unsigned src_size);
  38
  39 void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
  40         unsigned int width, unsigned int height,
  41         int lumStride, int chromStride, int dstStride);
  42 void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
  43         unsigned int width, unsigned int height,
  44         int lumStride, int chromStride, int dstStride);
  45 void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
  46         unsigned int width, unsigned int height,
  47         int lumStride, int chromStride, int dstStride);
  48 void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
  49         unsigned int width, unsigned int height,
  50         int lumStride, int chromStride, int srcStride);
  51 void (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
  52         unsigned int width, unsigned int height,
  53         int lumStride, int chromStride, int srcStride);
  54 void (*planar2x)(const uint8_t *src, uint8_t *dst, int width, int height, int srcStride, int dstStride);
  55 void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
  56                             unsigned width, unsigned height, int src1Stride,
  57                             int src2Stride, int dstStride);
  58 void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
  59                         uint8_t *dst1, uint8_t *dst2,
  60                         unsigned width, unsigned height,
  61                         int srcStride1, int srcStride2,
  62                         int dstStride1, int dstStride2);
  63 void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
  64                         uint8_t *dst,
  65                         unsigned width, unsigned height,
  66                         int srcStride1, int srcStride2,
  67                         int srcStride3, int dstStride);
  68
  69 #ifdef ARCH_X86
  70 static const uint64_t mmx_null  __attribute__((aligned(8))) = 0x0000000000000000ULL;
  71 static const uint64_t mmx_one   __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
  72 static const uint64_t mask32b  attribute_used __attribute__((aligned(8))) = 0x000000FF000000FFULL;
  73 static const uint64_t mask32g  attribute_used __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
  74 static const uint64_t mask32r  attribute_used __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
  75 static const uint64_t mask32   __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
  76 static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL;
  77 static const uint64_t mask3216g  __attribute__((aligned(8)))=0x0000FC000000FC00ULL;
  78 static const uint64_t mask3215g  __attribute__((aligned(8)))=0x0000F8000000F800ULL;
  79 static const uint64_t mul3216  __attribute__((aligned(8))) = 0x2000000420000004ULL;
  80 static const uint64_t mul3215  __attribute__((aligned(8))) = 0x2000000820000008ULL;
  81 static const uint64_t mask24b  attribute_used __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
  82 static const uint64_t mask24g  attribute_used __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
  83 static const uint64_t mask24r  attribute_used __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
  84 static const uint64_t mask24l  __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
  85 static const uint64_t mask24h  __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
  86 static const uint64_t mask24hh  __attribute__((aligned(8))) = 0xffff000000000000ULL;
  87 static const uint64_t mask24hhh  __attribute__((aligned(8))) = 0xffffffff00000000ULL;
  88 static const uint64_t mask24hhhh  __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
  89 static const uint64_t mask15b  __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111  xxB */
  90 static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000  RGx */
  91 static const uint64_t mask15s  __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
  92 static const uint64_t mask15g  __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
  93 static const uint64_t mask15r  __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
  94 #define mask16b mask15b
  95 static const uint64_t mask16g  __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
  96 static const uint64_t mask16r  __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
  97 static const uint64_t red_16mask  __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
  98 static const uint64_t green_16mask __attribute__((aligned(8)))= 0x000007e0000007e0ULL;
  99 static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
 100 static const uint64_t red_15mask  __attribute__((aligned(8))) = 0x00007c000000f800ULL;
 101 static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000007e0ULL;
 102 static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
 103
 104 #ifdef FAST_BGR2YV12
 105 static const uint64_t bgr2YCoeff  attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL;
 106 static const uint64_t bgr2UCoeff  attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
 107 static const uint64_t bgr2VCoeff  attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
 108 #else
 109 static const uint64_t bgr2YCoeff  attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL;
 110 static const uint64_t bgr2UCoeff  attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
 111 static const uint64_t bgr2VCoeff  attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
 112 #endif
 113 static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL;
 114 static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8)))= 0x8080808080808080ULL;
 115 static const uint64_t w1111       attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL;
 116
 117 #if 0
 118 static volatile uint64_t __attribute__((aligned(8))) b5Dither;
 119 static volatile uint64_t __attribute__((aligned(8))) g5Dither;
 120 static volatile uint64_t __attribute__((aligned(8))) g6Dither;
 121 static volatile uint64_t __attribute__((aligned(8))) r5Dither;
 122
 123 static uint64_t __attribute__((aligned(8))) dither4[2]={
 124         0x0103010301030103LL,
 125         0x0200020002000200LL,};
 126
 127 static uint64_t __attribute__((aligned(8))) dither8[2]={
 128         0x0602060206020602LL,
 129         0x0004000400040004LL,};
 130 #endif
 131 #endif
 132
 133 #define RGB2YUV_SHIFT 8
 134 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
 135 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
 136 #define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
 137 #define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
 138 #define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
 139 #define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
 140 #define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
 141 #define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
 142 #define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
 143
 144 //Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
 145 //Plain C versions
 146 #undef HAVE_MMX
 147 #undef HAVE_MMX2
 148 #undef HAVE_3DNOW
 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 RENAME(a) a ## _MMX
 162 #include "rgb2rgb_template.c"
 163
 164 //MMX2 versions
 165 #undef RENAME
 166 #define HAVE_MMX
 167 #define HAVE_MMX2
 168 #undef HAVE_3DNOW
 169 #undef HAVE_SSE2
 170 #define RENAME(a) a ## _MMX2
 171 #include "rgb2rgb_template.c"
 172
 173 //3DNOW versions
 174 #undef RENAME
 175 #define HAVE_MMX
 176 #undef HAVE_MMX2
 177 #define HAVE_3DNOW
 178 #undef HAVE_SSE2
 179 #define RENAME(a) a ## _3DNOW
 180 #include "rgb2rgb_template.c"
 181
 182 #endif //ARCH_X86
 183
 184 /*
 185  rgb15->rgb16 Original by Strepto/Astral
 186  ported to gcc & bugfixed : A'rpi
 187  MMX2, 3DNOW optimization by Nick Kurshev
 188  32bit c version, and and&add trick by Michael Niedermayer
 189 */
 190
 191 void sws_rgb2rgb_init(int flags){
 192 #ifdef ARCH_X86
 193         if(flags & SWS_CPU_CAPS_MMX2){
 194                 rgb15to16= rgb15to16_MMX2;
 195                 rgb15to24= rgb15to24_MMX2;
 196                 rgb15to32= rgb15to32_MMX2;
 197                 rgb16to24= rgb16to24_MMX2;
 198                 rgb16to32= rgb16to32_MMX2;
 199                 rgb16to15= rgb16to15_MMX2;
 200                 rgb24to16= rgb24to16_MMX2;
 201                 rgb24to15= rgb24to15_MMX2;
 202                 rgb24to32= rgb24to32_MMX2;
 203                 rgb32to16= rgb32to16_MMX2;
 204                 rgb32to15= rgb32to15_MMX2;
 205                 rgb32to24= rgb32to24_MMX2;
 206                 rgb24tobgr15= rgb24tobgr15_MMX2;
 207                 rgb24tobgr16= rgb24tobgr16_MMX2;
 208                 rgb24tobgr24= rgb24tobgr24_MMX2;
 209                 rgb32tobgr32= rgb32tobgr32_MMX2;
 210                 rgb32tobgr16= rgb32tobgr16_MMX2;
 211                 rgb32tobgr15= rgb32tobgr15_MMX2;
 212                 yv12toyuy2= yv12toyuy2_MMX2;
 213                 yv12touyvy= yv12touyvy_MMX2;
 214                 yuv422ptoyuy2= yuv422ptoyuy2_MMX2;
 215                 yuy2toyv12= yuy2toyv12_MMX2;
 216 //              uyvytoyv12= uyvytoyv12_MMX2;
 217 //              yvu9toyv12= yvu9toyv12_MMX2;
 218                 planar2x= planar2x_MMX2;
 219                 rgb24toyv12= rgb24toyv12_MMX2;
 220                 interleaveBytes= interleaveBytes_MMX2;
 221                 vu9_to_vu12= vu9_to_vu12_MMX2;
 222                 yvu9_to_yuy2= yvu9_to_yuy2_MMX2;
 223         }else if(flags & SWS_CPU_CAPS_3DNOW){
 224                 rgb15to16= rgb15to16_3DNOW;
 225                 rgb15to24= rgb15to24_3DNOW;
 226                 rgb15to32= rgb15to32_3DNOW;
 227                 rgb16to24= rgb16to24_3DNOW;
 228                 rgb16to32= rgb16to32_3DNOW;
 229                 rgb16to15= rgb16to15_3DNOW;
 230                 rgb24to16= rgb24to16_3DNOW;
 231                 rgb24to15= rgb24to15_3DNOW;
 232                 rgb24to32= rgb24to32_3DNOW;
 233                 rgb32to16= rgb32to16_3DNOW;
 234                 rgb32to15= rgb32to15_3DNOW;
 235                 rgb32to24= rgb32to24_3DNOW;
 236                 rgb24tobgr15= rgb24tobgr15_3DNOW;
 237                 rgb24tobgr16= rgb24tobgr16_3DNOW;
 238                 rgb24tobgr24= rgb24tobgr24_3DNOW;
 239                 rgb32tobgr32= rgb32tobgr32_3DNOW;
 240                 rgb32tobgr16= rgb32tobgr16_3DNOW;
 241                 rgb32tobgr15= rgb32tobgr15_3DNOW;
 242                 yv12toyuy2= yv12toyuy2_3DNOW;
 243                 yv12touyvy= yv12touyvy_3DNOW;
 244                 yuv422ptoyuy2= yuv422ptoyuy2_3DNOW;
 245                 yuy2toyv12= yuy2toyv12_3DNOW;
 246 //              uyvytoyv12= uyvytoyv12_3DNOW;
 247 //              yvu9toyv12= yvu9toyv12_3DNOW;
 248                 planar2x= planar2x_3DNOW;
 249                 rgb24toyv12= rgb24toyv12_3DNOW;
 250                 interleaveBytes= interleaveBytes_3DNOW;
 251                 vu9_to_vu12= vu9_to_vu12_3DNOW;
 252                 yvu9_to_yuy2= yvu9_to_yuy2_3DNOW;
 253         }else if(flags & SWS_CPU_CAPS_MMX){
 254                 rgb15to16= rgb15to16_MMX;
 255                 rgb15to24= rgb15to24_MMX;
 256                 rgb15to32= rgb15to32_MMX;
 257                 rgb16to24= rgb16to24_MMX;
 258                 rgb16to32= rgb16to32_MMX;
 259                 rgb16to15= rgb16to15_MMX;
 260                 rgb24to16= rgb24to16_MMX;
 261                 rgb24to15= rgb24to15_MMX;
 262                 rgb24to32= rgb24to32_MMX;
 263                 rgb32to16= rgb32to16_MMX;
 264                 rgb32to15= rgb32to15_MMX;
 265                 rgb32to24= rgb32to24_MMX;
 266                 rgb24tobgr15= rgb24tobgr15_MMX;
 267                 rgb24tobgr16= rgb24tobgr16_MMX;
 268                 rgb24tobgr24= rgb24tobgr24_MMX;
 269                 rgb32tobgr32= rgb32tobgr32_MMX;
 270                 rgb32tobgr16= rgb32tobgr16_MMX;
 271                 rgb32tobgr15= rgb32tobgr15_MMX;
 272                 yv12toyuy2= yv12toyuy2_MMX;
 273                 yv12touyvy= yv12touyvy_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                 yv12touyvy= yv12touyvy_C;
 306                 yuv422ptoyuy2= yuv422ptoyuy2_C;
 307                 yuy2toyv12= yuy2toyv12_C;
 308 //              uyvytoyv12= uyvytoyv12_C;
 309 //              yvu9toyv12= yvu9toyv12_C;
 310                 planar2x= planar2x_C;
 311                 rgb24toyv12= rgb24toyv12_C;
 312                 interleaveBytes= interleaveBytes_C;
 313                 vu9_to_vu12= vu9_to_vu12_C;
 314                 yvu9_to_yuy2= yvu9_to_yuy2_C;
 315         }
 316 }
 317
 318 /**
 319  * Pallete is assumed to contain bgr32
 320  */
 321 void palette8torgb32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 322 {
 323         unsigned i;
 324
 325 /*
 326         for(i=0; i<num_pixels; i++)
 327                 ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
 328 */
 329
 330         for(i=0; i<num_pixels; i++)
 331         {
 332                 //FIXME slow?
 333                 dst[0]= palette[ src[i]*4+2 ];
 334                 dst[1]= palette[ src[i]*4+1 ];
 335                 dst[2]= palette[ src[i]*4+0 ];
 336 //              dst[3]= 0; /* do we need this cleansing? */
 337                 dst+= 4;
 338         }
 339 }
 340
 341 void palette8tobgr32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 342 {
 343         unsigned i;
 344         for(i=0; i<num_pixels; i++)
 345         {
 346                 //FIXME slow?
 347                 dst[0]= palette[ src[i]*4+0 ];
 348                 dst[1]= palette[ src[i]*4+1 ];
 349                 dst[2]= palette[ src[i]*4+2 ];
 350 //              dst[3]= 0; /* do we need this cleansing? */
 351                 dst+= 4;
 352         }
 353 }
 354
 355 /**
 356  * Pallete is assumed to contain bgr32
 357  */
 358 void palette8torgb24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 359 {
 360         unsigned i;
 361 /*
 362         writes 1 byte o much and might cause alignment issues on some architectures?
 363         for(i=0; i<num_pixels; i++)
 364                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
 365 */
 366         for(i=0; i<num_pixels; i++)
 367         {
 368                 //FIXME slow?
 369                 dst[0]= palette[ src[i]*4+2 ];
 370                 dst[1]= palette[ src[i]*4+1 ];
 371                 dst[2]= palette[ src[i]*4+0 ];
 372                 dst+= 3;
 373         }
 374 }
 375
 376 void palette8tobgr24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 377 {
 378         unsigned i;
 379 /*
 380         writes 1 byte o much and might cause alignment issues on some architectures?
 381         for(i=0; i<num_pixels; i++)
 382                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
 383 */
 384         for(i=0; i<num_pixels; i++)
 385         {
 386                 //FIXME slow?
 387                 dst[0]= palette[ src[i]*4+0 ];
 388                 dst[1]= palette[ src[i]*4+1 ];
 389                 dst[2]= palette[ src[i]*4+2 ];
 390                 dst+= 3;
 391         }
 392 }
 393
 394 /**
 395  * Palette is assumed to contain bgr16, see rgb32to16 to convert the palette
 396  */
 397 void palette8torgb16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 398 {
 399         unsigned i;
 400         for(i=0; i<num_pixels; i++)
 401                 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
 402 }
 403 void palette8tobgr16(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 404 {
 405         unsigned i;
 406         for(i=0; i<num_pixels; i++)
 407                 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
 408 }
 409
 410 /**
 411  * Pallete is assumed to contain bgr15, see rgb32to15 to convert the palette
 412  */
 413 void palette8torgb15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 414 {
 415         unsigned i;
 416         for(i=0; i<num_pixels; i++)
 417                 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
 418 }
 419 void palette8tobgr15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
 420 {
 421         unsigned i;
 422         for(i=0; i<num_pixels; i++)
 423                 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
 424 }
 425
 426 void rgb32tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 427 {
 428         unsigned i;
 429         unsigned num_pixels = src_size >> 2;
 430         for(i=0; i<num_pixels; i++)
 431         {
 432                 dst[3*i + 0] = src[4*i + 2];
 433                 dst[3*i + 1] = src[4*i + 1];
 434                 dst[3*i + 2] = src[4*i + 0];
 435         }
 436 }
 437
 438 void rgb24tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 439 {
 440         unsigned i;
 441         for(i=0; 3*i<src_size; i++)
 442         {
 443                 dst[4*i + 0] = src[3*i + 2];
 444                 dst[4*i + 1] = src[3*i + 1];
 445                 dst[4*i + 2] = src[3*i + 0];
 446                 dst[4*i + 3] = 0;
 447         }
 448 }
 449
 450 void rgb16tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 451 {
 452         const uint16_t *end;
 453         uint8_t *d = (uint8_t *)dst;
 454         const uint16_t *s = (uint16_t *)src;
 455         end = s + src_size/2;
 456         while(s < end)
 457         {
 458                 register uint16_t bgr;
 459                 bgr = *s++;
 460                 *d++ = (bgr&0xF800)>>8;
 461                 *d++ = (bgr&0x7E0)>>3;
 462                 *d++ = (bgr&0x1F)<<3;
 463                 *d++ = 0;
 464         }
 465 }
 466
 467 void rgb16tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 468 {
 469         const uint16_t *end;
 470         uint8_t *d = (uint8_t *)dst;
 471         const uint16_t *s = (const uint16_t *)src;
 472         end = s + src_size/2;
 473         while(s < end)
 474         {
 475                 register uint16_t bgr;
 476                 bgr = *s++;
 477                 *d++ = (bgr&0xF800)>>8;
 478                 *d++ = (bgr&0x7E0)>>3;
 479                 *d++ = (bgr&0x1F)<<3;
 480         }
 481 }
 482
 483 void rgb16tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 484 {
 485         unsigned i;
 486         unsigned num_pixels = src_size >> 1;
 487
 488         for(i=0; i<num_pixels; i++)
 489         {
 490             unsigned b,g,r;
 491             register uint16_t rgb;
 492             rgb = src[2*i];
 493             r = rgb&0x1F;
 494             g = (rgb&0x7E0)>>5;
 495             b = (rgb&0xF800)>>11;
 496             dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
 497         }
 498 }
 499
 500 void rgb16tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 501 {
 502         unsigned i;
 503         unsigned num_pixels = src_size >> 1;
 504
 505         for(i=0; i<num_pixels; i++)
 506         {
 507             unsigned b,g,r;
 508             register uint16_t rgb;
 509             rgb = src[2*i];
 510             r = rgb&0x1F;
 511             g = (rgb&0x7E0)>>5;
 512             b = (rgb&0xF800)>>11;
 513             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
 514         }
 515 }
 516
 517 void rgb15tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 518 {
 519         const uint16_t *end;
 520         uint8_t *d = (uint8_t *)dst;
 521         const uint16_t *s = (const uint16_t *)src;
 522         end = s + src_size/2;
 523         while(s < end)
 524         {
 525                 register uint16_t bgr;
 526                 bgr = *s++;
 527                 *d++ = (bgr&0x7C00)>>7;
 528                 *d++ = (bgr&0x3E0)>>2;
 529                 *d++ = (bgr&0x1F)<<3;
 530                 *d++ = 0;
 531         }
 532 }
 533
 534 void rgb15tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 535 {
 536         const uint16_t *end;
 537         uint8_t *d = (uint8_t *)dst;
 538         const uint16_t *s = (uint16_t *)src;
 539         end = s + src_size/2;
 540         while(s < end)
 541         {
 542                 register uint16_t bgr;
 543                 bgr = *s++;
 544                 *d++ = (bgr&0x7C00)>>7;
 545                 *d++ = (bgr&0x3E0)>>2;
 546                 *d++ = (bgr&0x1F)<<3;
 547         }
 548 }
 549
 550 void rgb15tobgr16(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 551 {
 552         unsigned i;
 553         unsigned num_pixels = src_size >> 1;
 554
 555         for(i=0; i<num_pixels; i++)
 556         {
 557             unsigned b,g,r;
 558             register uint16_t rgb;
 559             rgb = src[2*i];
 560             r = rgb&0x1F;
 561             g = (rgb&0x3E0)>>5;
 562             b = (rgb&0x7C00)>>10;
 563             dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
 564         }
 565 }
 566
 567 void rgb15tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 568 {
 569         unsigned i;
 570         unsigned num_pixels = src_size >> 1;
 571
 572         for(i=0; i<num_pixels; i++)
 573         {
 574             unsigned b,g,r;
 575             register uint16_t rgb;
 576             rgb = src[2*i];
 577             r = rgb&0x1F;
 578             g = (rgb&0x3E0)>>5;
 579             b = (rgb&0x7C00)>>10;
 580             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
 581         }
 582 }
 583
 584 void rgb8tobgr8(const uint8_t *src, uint8_t *dst, unsigned int src_size)
 585 {
 586         unsigned i;
 587         unsigned num_pixels = src_size;
 588         for(i=0; i<num_pixels; i++)
 589         {
 590             unsigned b,g,r;
 591             register uint8_t rgb;
 592             rgb = src[i];
 593             r = (rgb&0x07);
 594             g = (rgb&0x38)>>3;
 595             b = (rgb&0xC0)>>6;
 596             dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
 597         }
 598 }