00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023
00024
00025
00026
00027
00028 #include <inttypes.h>
00029 #include "config.h"
00030 #include "rgb2rgb.h"
00031 #include "swscale.h"
00032 #include "swscale_internal.h"
00033 #include "x86_cpu.h"
00034 #include "bswap.h"
00035
00036 #define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit
00037
00038 void (*rgb24to32)(const uint8_t *src, uint8_t *dst, long src_size);
00039 void (*rgb24to16)(const uint8_t *src, uint8_t *dst, long src_size);
00040 void (*rgb24to15)(const uint8_t *src, uint8_t *dst, long src_size);
00041 void (*rgb32to24)(const uint8_t *src, uint8_t *dst, long src_size);
00042 void (*rgb32to16)(const uint8_t *src, uint8_t *dst, long src_size);
00043 void (*rgb32to15)(const uint8_t *src, uint8_t *dst, long src_size);
00044 void (*rgb15to16)(const uint8_t *src, uint8_t *dst, long src_size);
00045 void (*rgb15to24)(const uint8_t *src, uint8_t *dst, long src_size);
00046 void (*rgb15to32)(const uint8_t *src, uint8_t *dst, long src_size);
00047 void (*rgb16to15)(const uint8_t *src, uint8_t *dst, long src_size);
00048 void (*rgb16to24)(const uint8_t *src, uint8_t *dst, long src_size);
00049 void (*rgb16to32)(const uint8_t *src, uint8_t *dst, long src_size);
00050
00051 void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
00052 void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, long src_size);
00053 void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, long src_size);
00054 void (*rgb32tobgr32)(const uint8_t *src, uint8_t *dst, long src_size);
00055
00056 void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, long src_size);
00057 void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, long src_size);
00058
00059 void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
00060 long width, long height,
00061 long lumStride, long chromStride, long dstStride);
00062 void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
00063 long width, long height,
00064 long lumStride, long chromStride, long dstStride);
00065 void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
00066 long width, long height,
00067 long lumStride, long chromStride, long dstStride);
00068 void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
00069 long width, long height,
00070 long lumStride, long chromStride, long srcStride);
00071 void (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
00072 long width, long height,
00073 long lumStride, long chromStride, long srcStride);
00074 void (*planar2x)(const uint8_t *src, uint8_t *dst, long width, long height,
00075 long srcStride, long dstStride);
00076 void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
00077 long width, long height, long src1Stride,
00078 long src2Stride, long dstStride);
00079 void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
00080 uint8_t *dst1, uint8_t *dst2,
00081 long width, long height,
00082 long srcStride1, long srcStride2,
00083 long dstStride1, long dstStride2);
00084 void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
00085 uint8_t *dst,
00086 long width, long height,
00087 long srcStride1, long srcStride2,
00088 long srcStride3, long dstStride);
00089
00090 #if defined(ARCH_X86) && defined(CONFIG_GPL)
00091 DECLARE_ASM_CONST(8, uint64_t, mmx_null) = 0x0000000000000000ULL;
00092 DECLARE_ASM_CONST(8, uint64_t, mmx_one) = 0xFFFFFFFFFFFFFFFFULL;
00093 DECLARE_ASM_CONST(8, uint64_t, mask32b) = 0x000000FF000000FFULL;
00094 DECLARE_ASM_CONST(8, uint64_t, mask32g) = 0x0000FF000000FF00ULL;
00095 DECLARE_ASM_CONST(8, uint64_t, mask32r) = 0x00FF000000FF0000ULL;
00096 DECLARE_ASM_CONST(8, uint64_t, mask32) = 0x00FFFFFF00FFFFFFULL;
00097 DECLARE_ASM_CONST(8, uint64_t, mask3216br) = 0x00F800F800F800F8ULL;
00098 DECLARE_ASM_CONST(8, uint64_t, mask3216g) = 0x0000FC000000FC00ULL;
00099 DECLARE_ASM_CONST(8, uint64_t, mask3215g) = 0x0000F8000000F800ULL;
00100 DECLARE_ASM_CONST(8, uint64_t, mul3216) = 0x2000000420000004ULL;
00101 DECLARE_ASM_CONST(8, uint64_t, mul3215) = 0x2000000820000008ULL;
00102 DECLARE_ASM_CONST(8, uint64_t, mask24b) = 0x00FF0000FF0000FFULL;
00103 DECLARE_ASM_CONST(8, uint64_t, mask24g) = 0xFF0000FF0000FF00ULL;
00104 DECLARE_ASM_CONST(8, uint64_t, mask24r) = 0x0000FF0000FF0000ULL;
00105 DECLARE_ASM_CONST(8, uint64_t, mask24l) = 0x0000000000FFFFFFULL;
00106 DECLARE_ASM_CONST(8, uint64_t, mask24h) = 0x0000FFFFFF000000ULL;
00107 DECLARE_ASM_CONST(8, uint64_t, mask24hh) = 0xffff000000000000ULL;
00108 DECLARE_ASM_CONST(8, uint64_t, mask24hhh) = 0xffffffff00000000ULL;
00109 DECLARE_ASM_CONST(8, uint64_t, mask24hhhh) = 0xffffffffffff0000ULL;
00110 DECLARE_ASM_CONST(8, uint64_t, mask15b) = 0x001F001F001F001FULL;
00111 DECLARE_ASM_CONST(8, uint64_t, mask15rg) = 0x7FE07FE07FE07FE0ULL;
00112 DECLARE_ASM_CONST(8, uint64_t, mask15s) = 0xFFE0FFE0FFE0FFE0ULL;
00113 DECLARE_ASM_CONST(8, uint64_t, mask15g) = 0x03E003E003E003E0ULL;
00114 DECLARE_ASM_CONST(8, uint64_t, mask15r) = 0x7C007C007C007C00ULL;
00115 #define mask16b mask15b
00116 DECLARE_ASM_CONST(8, uint64_t, mask16g) = 0x07E007E007E007E0ULL;
00117 DECLARE_ASM_CONST(8, uint64_t, mask16r) = 0xF800F800F800F800ULL;
00118 DECLARE_ASM_CONST(8, uint64_t, red_16mask) = 0x0000f8000000f800ULL;
00119 DECLARE_ASM_CONST(8, uint64_t, green_16mask) = 0x000007e0000007e0ULL;
00120 DECLARE_ASM_CONST(8, uint64_t, blue_16mask) = 0x0000001f0000001fULL;
00121 DECLARE_ASM_CONST(8, uint64_t, red_15mask) = 0x00007c0000007c00ULL;
00122 DECLARE_ASM_CONST(8, uint64_t, green_15mask) = 0x000003e0000003e0ULL;
00123 DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
00124
00125 #if 0
00126 static volatile uint64_t __attribute__((aligned(8))) b5Dither;
00127 static volatile uint64_t __attribute__((aligned(8))) g5Dither;
00128 static volatile uint64_t __attribute__((aligned(8))) g6Dither;
00129 static volatile uint64_t __attribute__((aligned(8))) r5Dither;
00130
00131 static uint64_t __attribute__((aligned(8))) dither4[2]={
00132 0x0103010301030103LL,
00133 0x0200020002000200LL,};
00134
00135 static uint64_t __attribute__((aligned(8))) dither8[2]={
00136 0x0602060206020602LL,
00137 0x0004000400040004LL,};
00138 #endif
00139 #endif
00140
00141 #define RGB2YUV_SHIFT 8
00142 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
00143 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
00144 #define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
00145 #define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
00146 #define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
00147 #define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
00148 #define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
00149 #define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
00150 #define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
00151
00152
00153
00154 #undef HAVE_MMX
00155 #undef HAVE_MMX2
00156 #undef HAVE_3DNOW
00157 #undef HAVE_SSE2
00158 #define RENAME(a) a ## _C
00159 #include "rgb2rgb_template.c"
00160
00161 #if defined(ARCH_X86) && defined(CONFIG_GPL)
00162
00163
00164 #undef RENAME
00165 #define HAVE_MMX
00166 #undef HAVE_MMX2
00167 #undef HAVE_3DNOW
00168 #undef HAVE_SSE2
00169 #define RENAME(a) a ## _MMX
00170 #include "rgb2rgb_template.c"
00171
00172
00173 #undef RENAME
00174 #define HAVE_MMX
00175 #define HAVE_MMX2
00176 #undef HAVE_3DNOW
00177 #undef HAVE_SSE2
00178 #define RENAME(a) a ## _MMX2
00179 #include "rgb2rgb_template.c"
00180
00181
00182 #undef RENAME
00183 #define HAVE_MMX
00184 #undef HAVE_MMX2
00185 #define HAVE_3DNOW
00186 #undef HAVE_SSE2
00187 #define RENAME(a) a ## _3DNOW
00188 #include "rgb2rgb_template.c"
00189
00190 #endif //ARCH_X86 || ARCH_X86_64
00191
00192
00193
00194
00195
00196
00197
00198
00199 void sws_rgb2rgb_init(int flags){
00200 #if (defined(HAVE_MMX2) || defined(HAVE_3DNOW) || defined(HAVE_MMX)) && defined(CONFIG_GPL)
00201 if (flags & SWS_CPU_CAPS_MMX2)
00202 rgb2rgb_init_MMX2();
00203 else if (flags & SWS_CPU_CAPS_3DNOW)
00204 rgb2rgb_init_3DNOW();
00205 else if (flags & SWS_CPU_CAPS_MMX)
00206 rgb2rgb_init_MMX();
00207 else
00208 #endif
00209 rgb2rgb_init_C();
00210 }
00211
00215 void palette8torgb32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00216 {
00217 long i;
00218
00219
00220
00221
00222
00223
00224 for (i=0; i<num_pixels; i++)
00225 {
00226 #ifdef WORDS_BIGENDIAN
00227 dst[3]= palette[src[i]*4+2];
00228 dst[2]= palette[src[i]*4+1];
00229 dst[1]= palette[src[i]*4+0];
00230 #else
00231
00232 dst[0]= palette[src[i]*4+2];
00233 dst[1]= palette[src[i]*4+1];
00234 dst[2]= palette[src[i]*4+0];
00235
00236 #endif
00237 dst+= 4;
00238 }
00239 }
00240
00241 void palette8tobgr32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00242 {
00243 long i;
00244 for (i=0; i<num_pixels; i++)
00245 {
00246 #ifdef WORDS_BIGENDIAN
00247 dst[3]= palette[src[i]*4+0];
00248 dst[2]= palette[src[i]*4+1];
00249 dst[1]= palette[src[i]*4+2];
00250 #else
00251
00252 dst[0]= palette[src[i]*4+0];
00253 dst[1]= palette[src[i]*4+1];
00254 dst[2]= palette[src[i]*4+2];
00255
00256 #endif
00257
00258 dst+= 4;
00259 }
00260 }
00261
00265 void palette8torgb24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00266 {
00267 long i;
00268
00269
00270
00271
00272
00273 for (i=0; i<num_pixels; i++)
00274 {
00275
00276 dst[0]= palette[src[i]*4+2];
00277 dst[1]= palette[src[i]*4+1];
00278 dst[2]= palette[src[i]*4+0];
00279 dst+= 3;
00280 }
00281 }
00282
00283 void palette8tobgr24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00284 {
00285 long i;
00286
00287
00288
00289
00290
00291 for (i=0; i<num_pixels; i++)
00292 {
00293
00294 dst[0]= palette[src[i]*4+0];
00295 dst[1]= palette[src[i]*4+1];
00296 dst[2]= palette[src[i]*4+2];
00297 dst+= 3;
00298 }
00299 }
00300
00304 void palette8torgb16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00305 {
00306 long i;
00307 for (i=0; i<num_pixels; i++)
00308 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[src[i]];
00309 }
00310 void palette8tobgr16(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00311 {
00312 long i;
00313 for (i=0; i<num_pixels; i++)
00314 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[src[i]]);
00315 }
00316
00320 void palette8torgb15(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00321 {
00322 long i;
00323 for (i=0; i<num_pixels; i++)
00324 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[src[i]];
00325 }
00326 void palette8tobgr15(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
00327 {
00328 long i;
00329 for (i=0; i<num_pixels; i++)
00330 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[src[i]]);
00331 }
00332
00333 void rgb32tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
00334 {
00335 long i;
00336 long num_pixels = src_size >> 2;
00337 for (i=0; i<num_pixels; i++)
00338 {
00339 #ifdef WORDS_BIGENDIAN
00340
00341 dst[3*i + 0] = src[4*i + 1];
00342 dst[3*i + 1] = src[4*i + 2];
00343 dst[3*i + 2] = src[4*i + 3];
00344 #else
00345 dst[3*i + 0] = src[4*i + 2];
00346 dst[3*i + 1] = src[4*i + 1];
00347 dst[3*i + 2] = src[4*i + 0];
00348 #endif
00349 }
00350 }
00351
00352 void rgb24tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
00353 {
00354 long i;
00355 for (i=0; 3*i<src_size; i++)
00356 {
00357 #ifdef WORDS_BIGENDIAN
00358
00359 dst[4*i + 0] = 0;
00360 dst[4*i + 1] = src[3*i + 0];
00361 dst[4*i + 2] = src[3*i + 1];
00362 dst[4*i + 3] = src[3*i + 2];
00363 #else
00364 dst[4*i + 0] = src[3*i + 2];
00365 dst[4*i + 1] = src[3*i + 1];
00366 dst[4*i + 2] = src[3*i + 0];
00367 dst[4*i + 3] = 0;
00368 #endif
00369 }
00370 }
00371
00372 void rgb16tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
00373 {
00374 const uint16_t *end;
00375 uint8_t *d = (uint8_t *)dst;
00376 const uint16_t *s = (uint16_t *)src;
00377 end = s + src_size/2;
00378 while (s < end)
00379 {
00380 register uint16_t bgr;
00381 bgr = *s++;
00382 #ifdef WORDS_BIGENDIAN
00383 *d++ = 0;
00384 *d++ = (bgr&0x1F)<<3;
00385 *d++ = (bgr&0x7E0)>>3;
00386 *d++ = (bgr&0xF800)>>8;
00387 #else
00388 *d++ = (bgr&0xF800)>>8;
00389 *d++ = (bgr&0x7E0)>>3;
00390 *d++ = (bgr&0x1F)<<3;
00391 *d++ = 0;
00392 #endif
00393 }
00394 }
00395
00396 void rgb16tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
00397 {
00398 const uint16_t *end;
00399 uint8_t *d = (uint8_t *)dst;
00400 const uint16_t *s = (const uint16_t *)src;
00401 end = s + src_size/2;
00402 while (s < end)
00403 {
00404 register uint16_t bgr;
00405 bgr = *s++;
00406 *d++ = (bgr&0xF800)>>8;
00407 *d++ = (bgr&0x7E0)>>3;
00408 *d++ = (bgr&0x1F)<<3;
00409 }
00410 }
00411
00412 void rgb16tobgr16(const uint8_t *src, uint8_t *dst, long src_size)
00413 {
00414 long i;
00415 long num_pixels = src_size >> 1;
00416
00417 for (i=0; i<num_pixels; i++)
00418 {
00419 unsigned b,g,r;
00420 register uint16_t rgb;
00421 rgb = src[2*i];
00422 r = rgb&0x1F;
00423 g = (rgb&0x7E0)>>5;
00424 b = (rgb&0xF800)>>11;
00425 dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
00426 }
00427 }
00428
00429 void rgb16tobgr15(const uint8_t *src, uint8_t *dst, long src_size)
00430 {
00431 long i;
00432 long num_pixels = src_size >> 1;
00433
00434 for (i=0; i<num_pixels; i++)
00435 {
00436 unsigned b,g,r;
00437 register uint16_t rgb;
00438 rgb = src[2*i];
00439 r = rgb&0x1F;
00440 g = (rgb&0x7E0)>>5;
00441 b = (rgb&0xF800)>>11;
00442 dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
00443 }
00444 }
00445
00446 void rgb15tobgr32(const uint8_t *src, uint8_t *dst, long src_size)
00447 {
00448 const uint16_t *end;
00449 uint8_t *d = (uint8_t *)dst;
00450 const uint16_t *s = (const uint16_t *)src;
00451 end = s + src_size/2;
00452 while (s < end)
00453 {
00454 register uint16_t bgr;
00455 bgr = *s++;
00456 #ifdef WORDS_BIGENDIAN
00457 *d++ = 0;
00458 *d++ = (bgr&0x1F)<<3;
00459 *d++ = (bgr&0x3E0)>>2;
00460 *d++ = (bgr&0x7C00)>>7;
00461 #else
00462 *d++ = (bgr&0x7C00)>>7;
00463 *d++ = (bgr&0x3E0)>>2;
00464 *d++ = (bgr&0x1F)<<3;
00465 *d++ = 0;
00466 #endif
00467 }
00468 }
00469
00470 void rgb15tobgr24(const uint8_t *src, uint8_t *dst, long src_size)
00471 {
00472 const uint16_t *end;
00473 uint8_t *d = (uint8_t *)dst;
00474 const uint16_t *s = (uint16_t *)src;
00475 end = s + src_size/2;
00476 while (s < end)
00477 {
00478 register uint16_t bgr;
00479 bgr = *s++;
00480 *d++ = (bgr&0x7C00)>>7;
00481 *d++ = (bgr&0x3E0)>>2;
00482 *d++ = (bgr&0x1F)<<3;
00483 }
00484 }
00485
00486 void rgb15tobgr16(const uint8_t *src, uint8_t *dst, long src_size)
00487 {
00488 long i;
00489 long num_pixels = src_size >> 1;
00490
00491 for (i=0; i<num_pixels; i++)
00492 {
00493 unsigned b,g,r;
00494 register uint16_t rgb;
00495 rgb = src[2*i];
00496 r = rgb&0x1F;
00497 g = (rgb&0x3E0)>>5;
00498 b = (rgb&0x7C00)>>10;
00499 dst[2*i] = (b&0x1F) | ((g&0x3F)<<5) | ((r&0x1F)<<11);
00500 }
00501 }
00502
00503 void rgb15tobgr15(const uint8_t *src, uint8_t *dst, long src_size)
00504 {
00505 long i;
00506 long num_pixels = src_size >> 1;
00507
00508 for (i=0; i<num_pixels; i++)
00509 {
00510 unsigned b,g,r;
00511 register uint16_t rgb;
00512 rgb = src[2*i];
00513 r = rgb&0x1F;
00514 g = (rgb&0x3E0)>>5;
00515 b = (rgb&0x7C00)>>10;
00516 dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
00517 }
00518 }
00519
00520 void rgb8tobgr8(const uint8_t *src, uint8_t *dst, long src_size)
00521 {
00522 long i;
00523 long num_pixels = src_size;
00524 for (i=0; i<num_pixels; i++)
00525 {
00526 unsigned b,g,r;
00527 register uint8_t rgb;
00528 rgb = src[i];
00529 r = (rgb&0x07);
00530 g = (rgb&0x38)>>3;
00531 b = (rgb&0xC0)>>6;
00532 dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
00533 }
00534 }