fix palette8tobgr32/palette8torgb32 on big endian
[ffmpeg.git] / 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 "../cpudetect.h"
15 #include "../mangle.h"
16 #include "../bswap.h"
17 #include "../libvo/fastmemcpy.h"
18
19 #define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit
20
21 void (*rgb24to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
22 void (*rgb24to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
23 void (*rgb24to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
24 void (*rgb32to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
25 void (*rgb32to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
26 void (*rgb32to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
27 void (*rgb15to16)(const uint8_t *src,uint8_t *dst,unsigned src_size);
28 void (*rgb15to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
29 void (*rgb15to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
30 void (*rgb16to15)(const uint8_t *src,uint8_t *dst,unsigned src_size);
31 void (*rgb16to24)(const uint8_t *src,uint8_t *dst,unsigned src_size);
32 void (*rgb16to32)(const uint8_t *src,uint8_t *dst,unsigned src_size);
33 //void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, unsigned src_size);
34 void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, unsigned src_size);
35 void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, unsigned src_size);
36 void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, unsigned src_size);
37 void (*rgb32tobgr32)(const uint8_t *src, uint8_t *dst, unsigned src_size);
38 //void (*rgb32tobgr24)(const uint8_t *src, uint8_t *dst, unsigned src_size);
39 void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, unsigned src_size);
40 void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, unsigned src_size);
41
42 void (*yv12toyuy2)(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 (*yv12touyvy)(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 (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc, const uint8_t *vsrc, uint8_t *dst,
49         unsigned int width, unsigned int height,
50         int lumStride, int chromStride, int dstStride);
51 void (*yuy2toyv12)(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 (*rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
55         unsigned int width, unsigned int height,
56         int lumStride, int chromStride, int srcStride);
57 void (*planar2x)(const uint8_t *src, uint8_t *dst, int width, int height, int srcStride, int dstStride);
58 void (*interleaveBytes)(uint8_t *src1, uint8_t *src2, uint8_t *dst,
59                             unsigned width, unsigned height, int src1Stride,
60                             int src2Stride, int dstStride);
61 void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
62                         uint8_t *dst1, uint8_t *dst2,
63                         unsigned width, unsigned height,
64                         int srcStride1, int srcStride2,
65                         int dstStride1, int dstStride2);
66 void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2, const uint8_t *src3,
67                         uint8_t *dst,
68                         unsigned width, unsigned height,
69                         int srcStride1, int srcStride2,
70                         int srcStride3, int dstStride);
71
72 #if defined(ARCH_X86) || defined(ARCH_X86_64)
73 static const uint64_t mmx_null  __attribute__((aligned(8))) = 0x0000000000000000ULL;
74 static const uint64_t mmx_one   __attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL;
75 static const uint64_t mask32b  attribute_used __attribute__((aligned(8))) = 0x000000FF000000FFULL;
76 static const uint64_t mask32g  attribute_used __attribute__((aligned(8))) = 0x0000FF000000FF00ULL;
77 static const uint64_t mask32r  attribute_used __attribute__((aligned(8))) = 0x00FF000000FF0000ULL;
78 static const uint64_t mask32   __attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL;
79 static const uint64_t mask3216br __attribute__((aligned(8)))=0x00F800F800F800F8ULL;
80 static const uint64_t mask3216g  __attribute__((aligned(8)))=0x0000FC000000FC00ULL;
81 static const uint64_t mask3215g  __attribute__((aligned(8)))=0x0000F8000000F800ULL;
82 static const uint64_t mul3216  __attribute__((aligned(8))) = 0x2000000420000004ULL;
83 static const uint64_t mul3215  __attribute__((aligned(8))) = 0x2000000820000008ULL;
84 static const uint64_t mask24b  attribute_used __attribute__((aligned(8))) = 0x00FF0000FF0000FFULL;
85 static const uint64_t mask24g  attribute_used __attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL;
86 static const uint64_t mask24r  attribute_used __attribute__((aligned(8))) = 0x0000FF0000FF0000ULL;
87 static const uint64_t mask24l  __attribute__((aligned(8))) = 0x0000000000FFFFFFULL;
88 static const uint64_t mask24h  __attribute__((aligned(8))) = 0x0000FFFFFF000000ULL;
89 static const uint64_t mask24hh  __attribute__((aligned(8))) = 0xffff000000000000ULL;
90 static const uint64_t mask24hhh  __attribute__((aligned(8))) = 0xffffffff00000000ULL;
91 static const uint64_t mask24hhhh  __attribute__((aligned(8))) = 0xffffffffffff0000ULL;
92 static const uint64_t mask15b  __attribute__((aligned(8))) = 0x001F001F001F001FULL; /* 00000000 00011111  xxB */
93 static const uint64_t mask15rg __attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL; /* 01111111 11100000  RGx */
94 static const uint64_t mask15s  __attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL;
95 static const uint64_t mask15g  __attribute__((aligned(8))) = 0x03E003E003E003E0ULL;
96 static const uint64_t mask15r  __attribute__((aligned(8))) = 0x7C007C007C007C00ULL;
97 #define mask16b mask15b
98 static const uint64_t mask16g  __attribute__((aligned(8))) = 0x07E007E007E007E0ULL;
99 static const uint64_t mask16r  __attribute__((aligned(8))) = 0xF800F800F800F800ULL;
100 static const uint64_t red_16mask  __attribute__((aligned(8))) = 0x0000f8000000f800ULL;
101 static const uint64_t green_16mask __attribute__((aligned(8)))= 0x000007e0000007e0ULL;
102 static const uint64_t blue_16mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
103 static const uint64_t red_15mask  __attribute__((aligned(8))) = 0x00007c000000f800ULL;
104 static const uint64_t green_15mask __attribute__((aligned(8)))= 0x000003e0000007e0ULL;
105 static const uint64_t blue_15mask __attribute__((aligned(8))) = 0x0000001f0000001fULL;
106
107 #ifdef FAST_BGR2YV12
108 static const uint64_t bgr2YCoeff  attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL;
109 static const uint64_t bgr2UCoeff  attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL;
110 static const uint64_t bgr2VCoeff  attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL;
111 #else
112 static const uint64_t bgr2YCoeff  attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL;
113 static const uint64_t bgr2UCoeff  attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL;
114 static const uint64_t bgr2VCoeff  attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL;
115 #endif
116 static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL;
117 static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8)))= 0x8080808080808080ULL;
118 static const uint64_t w1111       attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL;
119
120 #if 0
121 static volatile uint64_t __attribute__((aligned(8))) b5Dither;
122 static volatile uint64_t __attribute__((aligned(8))) g5Dither;
123 static volatile uint64_t __attribute__((aligned(8))) g6Dither;
124 static volatile uint64_t __attribute__((aligned(8))) r5Dither;
125
126 static uint64_t __attribute__((aligned(8))) dither4[2]={
127         0x0103010301030103LL,
128         0x0200020002000200LL,};
129
130 static uint64_t __attribute__((aligned(8))) dither8[2]={
131         0x0602060206020602LL,
132         0x0004000400040004LL,};
133 #endif
134 #endif
135
136 #define RGB2YUV_SHIFT 8
137 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
138 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
139 #define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
140 #define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
141 #define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
142 #define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
143 #define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
144 #define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
145 #define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
146
147 //Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
148 //Plain C versions
149 #undef HAVE_MMX
150 #undef HAVE_MMX2
151 #undef HAVE_3DNOW
152 #undef HAVE_SSE2
153 #define RENAME(a) a ## _C
154 #include "rgb2rgb_template.c"
155
156 #if defined(ARCH_X86) || defined(ARCH_X86_64)
157
158 //MMX versions
159 #undef RENAME
160 #define HAVE_MMX
161 #undef HAVE_MMX2
162 #undef HAVE_3DNOW
163 #undef HAVE_SSE2
164 #define RENAME(a) a ## _MMX
165 #include "rgb2rgb_template.c"
166
167 //MMX2 versions
168 #undef RENAME
169 #define HAVE_MMX
170 #define HAVE_MMX2
171 #undef HAVE_3DNOW
172 #undef HAVE_SSE2
173 #define RENAME(a) a ## _MMX2
174 #include "rgb2rgb_template.c"
175
176 //3DNOW versions
177 #undef RENAME
178 #define HAVE_MMX
179 #undef HAVE_MMX2
180 #define HAVE_3DNOW
181 #undef HAVE_SSE2
182 #define RENAME(a) a ## _3DNOW
183 #include "rgb2rgb_template.c"
184
185 #endif //ARCH_X86 || ARCH_X86_64
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 #if defined(ARCH_X86) || defined(ARCH_X86_64)
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                 yv12touyvy= yv12touyvy_MMX2;
217                 yuv422ptoyuy2= yuv422ptoyuy2_MMX2;
218                 yuy2toyv12= yuy2toyv12_MMX2;
219 //              uyvytoyv12= uyvytoyv12_MMX2;
220 //              yvu9toyv12= yvu9toyv12_MMX2;
221                 planar2x= planar2x_MMX2;
222                 rgb24toyv12= rgb24toyv12_MMX2;
223                 interleaveBytes= interleaveBytes_MMX2;
224                 vu9_to_vu12= vu9_to_vu12_MMX2;
225                 yvu9_to_yuy2= yvu9_to_yuy2_MMX2;
226         }else if(flags & SWS_CPU_CAPS_3DNOW){
227                 rgb15to16= rgb15to16_3DNOW;
228                 rgb15to24= rgb15to24_3DNOW;
229                 rgb15to32= rgb15to32_3DNOW;
230                 rgb16to24= rgb16to24_3DNOW;
231                 rgb16to32= rgb16to32_3DNOW;
232                 rgb16to15= rgb16to15_3DNOW;
233                 rgb24to16= rgb24to16_3DNOW;
234                 rgb24to15= rgb24to15_3DNOW;
235                 rgb24to32= rgb24to32_3DNOW;
236                 rgb32to16= rgb32to16_3DNOW;
237                 rgb32to15= rgb32to15_3DNOW;
238                 rgb32to24= rgb32to24_3DNOW;
239                 rgb24tobgr15= rgb24tobgr15_3DNOW;
240                 rgb24tobgr16= rgb24tobgr16_3DNOW;
241                 rgb24tobgr24= rgb24tobgr24_3DNOW;
242                 rgb32tobgr32= rgb32tobgr32_3DNOW;
243                 rgb32tobgr16= rgb32tobgr16_3DNOW;
244                 rgb32tobgr15= rgb32tobgr15_3DNOW;
245                 yv12toyuy2= yv12toyuy2_3DNOW;
246                 yv12touyvy= yv12touyvy_3DNOW;
247                 yuv422ptoyuy2= yuv422ptoyuy2_3DNOW;
248                 yuy2toyv12= yuy2toyv12_3DNOW;
249 //              uyvytoyv12= uyvytoyv12_3DNOW;
250 //              yvu9toyv12= yvu9toyv12_3DNOW;
251                 planar2x= planar2x_3DNOW;
252                 rgb24toyv12= rgb24toyv12_3DNOW;
253                 interleaveBytes= interleaveBytes_3DNOW;
254                 vu9_to_vu12= vu9_to_vu12_3DNOW;
255                 yvu9_to_yuy2= yvu9_to_yuy2_3DNOW;
256         }else if(flags & SWS_CPU_CAPS_MMX){
257                 rgb15to16= rgb15to16_MMX;
258                 rgb15to24= rgb15to24_MMX;
259                 rgb15to32= rgb15to32_MMX;
260                 rgb16to24= rgb16to24_MMX;
261                 rgb16to32= rgb16to32_MMX;
262                 rgb16to15= rgb16to15_MMX;
263                 rgb24to16= rgb24to16_MMX;
264                 rgb24to15= rgb24to15_MMX;
265                 rgb24to32= rgb24to32_MMX;
266                 rgb32to16= rgb32to16_MMX;
267                 rgb32to15= rgb32to15_MMX;
268                 rgb32to24= rgb32to24_MMX;
269                 rgb24tobgr15= rgb24tobgr15_MMX;
270                 rgb24tobgr16= rgb24tobgr16_MMX;
271                 rgb24tobgr24= rgb24tobgr24_MMX;
272                 rgb32tobgr32= rgb32tobgr32_MMX;
273                 rgb32tobgr16= rgb32tobgr16_MMX;
274                 rgb32tobgr15= rgb32tobgr15_MMX;
275                 yv12toyuy2= yv12toyuy2_MMX;
276                 yv12touyvy= yv12touyvy_MMX;
277                 yuv422ptoyuy2= yuv422ptoyuy2_MMX;
278                 yuy2toyv12= yuy2toyv12_MMX;
279 //              uyvytoyv12= uyvytoyv12_MMX;
280 //              yvu9toyv12= yvu9toyv12_MMX;
281                 planar2x= planar2x_MMX;
282                 rgb24toyv12= rgb24toyv12_MMX;
283                 interleaveBytes= interleaveBytes_MMX;
284                 vu9_to_vu12= vu9_to_vu12_MMX;
285                 yvu9_to_yuy2= yvu9_to_yuy2_MMX;
286         }else
287 #endif
288         {
289                 rgb15to16= rgb15to16_C;
290                 rgb15to24= rgb15to24_C;
291                 rgb15to32= rgb15to32_C;
292                 rgb16to24= rgb16to24_C;
293                 rgb16to32= rgb16to32_C;
294                 rgb16to15= rgb16to15_C;
295                 rgb24to16= rgb24to16_C;
296                 rgb24to15= rgb24to15_C;
297                 rgb24to32= rgb24to32_C;
298                 rgb32to16= rgb32to16_C;
299                 rgb32to15= rgb32to15_C;
300                 rgb32to24= rgb32to24_C;
301                 rgb24tobgr15= rgb24tobgr15_C;
302                 rgb24tobgr16= rgb24tobgr16_C;
303                 rgb24tobgr24= rgb24tobgr24_C;
304                 rgb32tobgr32= rgb32tobgr32_C;
305                 rgb32tobgr16= rgb32tobgr16_C;
306                 rgb32tobgr15= rgb32tobgr15_C;
307                 yv12toyuy2= yv12toyuy2_C;
308                 yv12touyvy= yv12touyvy_C;
309                 yuv422ptoyuy2= yuv422ptoyuy2_C;
310                 yuy2toyv12= yuy2toyv12_C;
311 //              uyvytoyv12= uyvytoyv12_C;
312 //              yvu9toyv12= yvu9toyv12_C;
313                 planar2x= planar2x_C;
314                 rgb24toyv12= rgb24toyv12_C;
315                 interleaveBytes= interleaveBytes_C;
316                 vu9_to_vu12= vu9_to_vu12_C;
317                 yvu9_to_yuy2= yvu9_to_yuy2_C;
318         }
319 }
320
321 /**
322  * Pallete is assumed to contain bgr32
323  */
324 void palette8torgb32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
325 {
326         unsigned i;
327
328 /*
329         for(i=0; i<num_pixels; i++)
330                 ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
331 */
332
333         for(i=0; i<num_pixels; i++)
334         {
335                 #ifdef WORDS_BIGENDIAN
336                 dst[3]= palette[ src[i]*4+2 ];
337                 dst[2]= palette[ src[i]*4+1 ];
338                 dst[1]= palette[ src[i]*4+0 ];
339                 #else
340                 //FIXME slow?
341                 dst[0]= palette[ src[i]*4+2 ];
342                 dst[1]= palette[ src[i]*4+1 ];
343                 dst[2]= palette[ src[i]*4+0 ];
344 //              dst[3]= 0; /* do we need this cleansing? */
345                 #endif
346                 dst+= 4;
347         }
348 }
349
350 void palette8tobgr32(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
351 {
352         unsigned i;
353         for(i=0; i<num_pixels; i++)
354         {
355                 #ifdef WORDS_BIGENDIAN
356                 dst[3]= palette[ src[i]*4+0 ];
357                 dst[2]= palette[ src[i]*4+1 ];
358                 dst[1]= palette[ src[i]*4+2 ];
359                 #else
360                 //FIXME slow?
361                 dst[0]= palette[ src[i]*4+0 ];
362                 dst[1]= palette[ src[i]*4+1 ];
363                 dst[2]= palette[ src[i]*4+2 ];
364 //              dst[3]= 0; /* do we need this cleansing? */
365                 #endif
366                 
367                 dst+= 4;
368         }
369 }
370
371 /**
372  * Pallete is assumed to contain bgr32
373  */
374 void palette8torgb24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
375 {
376         unsigned i;
377 /*
378         writes 1 byte o much and might cause alignment issues on some architectures?
379         for(i=0; i<num_pixels; i++)
380                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
381 */
382         for(i=0; i<num_pixels; i++)
383         {
384                 //FIXME slow?
385                 dst[0]= palette[ src[i]*4+2 ];
386                 dst[1]= palette[ src[i]*4+1 ];
387                 dst[2]= palette[ src[i]*4+0 ];
388                 dst+= 3;
389         }
390 }
391
392 void palette8tobgr24(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
393 {
394         unsigned i;
395 /*
396         writes 1 byte o much and might cause alignment issues on some architectures?
397         for(i=0; i<num_pixels; i++)
398                 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
399 */
400         for(i=0; i<num_pixels; i++)
401         {
402                 //FIXME slow?
403                 dst[0]= palette[ src[i]*4+0 ];
404                 dst[1]= palette[ src[i]*4+1 ];
405                 dst[2]= palette[ src[i]*4+2 ];
406                 dst+= 3;
407         }
408 }
409
410 /**
411  * Palette is assumed to contain bgr16, see rgb32to16 to convert the palette
412  */
413 void palette8torgb16(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 palette8tobgr16(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 /**
427  * Pallete is assumed to contain bgr15, see rgb32to15 to convert the palette
428  */
429 void palette8torgb15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
430 {
431         unsigned i;
432         for(i=0; i<num_pixels; i++)
433                 ((uint16_t *)dst)[i] = ((uint16_t *)palette)[ src[i] ];
434 }
435 void palette8tobgr15(const uint8_t *src, uint8_t *dst, unsigned num_pixels, const uint8_t *palette)
436 {
437         unsigned i;
438         for(i=0; i<num_pixels; i++)
439                 ((uint16_t *)dst)[i] = bswap_16(((uint16_t *)palette)[ src[i] ]);
440 }
441
442 void rgb32tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
443 {
444         unsigned i;
445         unsigned num_pixels = src_size >> 2;
446         for(i=0; i<num_pixels; i++)
447         {
448                 dst[3*i + 0] = src[4*i + 2];
449                 dst[3*i + 1] = src[4*i + 1];
450                 dst[3*i + 2] = src[4*i + 0];
451         }
452 }
453
454 void rgb24tobgr32(const uint8_t *src, uint8_t *dst, unsigned int src_size)
455 {
456         unsigned i;
457         for(i=0; 3*i<src_size; i++)
458         {
459                 dst[4*i + 0] = src[3*i + 2];
460                 dst[4*i + 1] = src[3*i + 1];
461                 dst[4*i + 2] = src[3*i + 0];
462                 dst[4*i + 3] = 0;
463         }
464 }
465
466 void rgb16tobgr32(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 = (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                 *d++ = 0;
480         }
481 }
482
483 void rgb16tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
484 {
485         const uint16_t *end;
486         uint8_t *d = (uint8_t *)dst;
487         const uint16_t *s = (const uint16_t *)src;
488         end = s + src_size/2;
489         while(s < end)
490         {
491                 register uint16_t bgr;
492                 bgr = *s++;
493                 *d++ = (bgr&0xF800)>>8;
494                 *d++ = (bgr&0x7E0)>>3;
495                 *d++ = (bgr&0x1F)<<3;
496         }
497 }
498
499 void rgb16tobgr16(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&0x3F)<<5) | ((r&0x1F)<<11);
513         }
514 }
515
516 void rgb16tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
517 {
518         unsigned i;
519         unsigned num_pixels = src_size >> 1;
520         
521         for(i=0; i<num_pixels; i++)
522         {
523             unsigned b,g,r;
524             register uint16_t rgb;
525             rgb = src[2*i];
526             r = rgb&0x1F;
527             g = (rgb&0x7E0)>>5;
528             b = (rgb&0xF800)>>11;
529             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
530         }
531 }
532
533 void rgb15tobgr32(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 = (const 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                 *d++ = 0;
547         }
548 }
549
550 void rgb15tobgr24(const uint8_t *src, uint8_t *dst, unsigned int src_size)
551 {
552         const uint16_t *end;
553         uint8_t *d = (uint8_t *)dst;
554         const uint16_t *s = (uint16_t *)src;
555         end = s + src_size/2;
556         while(s < end)
557         {
558                 register uint16_t bgr;
559                 bgr = *s++;
560                 *d++ = (bgr&0x7C00)>>7;
561                 *d++ = (bgr&0x3E0)>>2;
562                 *d++ = (bgr&0x1F)<<3;
563         }
564 }
565
566 void rgb15tobgr16(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&0x3F)<<5) | ((r&0x1F)<<11);
580         }
581 }
582
583 void rgb15tobgr15(const uint8_t *src, uint8_t *dst, unsigned int src_size)
584 {
585         unsigned i;
586         unsigned num_pixels = src_size >> 1;
587         
588         for(i=0; i<num_pixels; i++)
589         {
590             unsigned b,g,r;
591             register uint16_t rgb;
592             rgb = src[2*i];
593             r = rgb&0x1F;
594             g = (rgb&0x3E0)>>5;
595             b = (rgb&0x7C00)>>10;
596             dst[2*i] = (b&0x1F) | ((g&0x1F)<<5) | ((r&0x1F)<<10);
597         }
598 }
599
600 void rgb8tobgr8(const uint8_t *src, uint8_t *dst, unsigned int src_size)
601 {
602         unsigned i;
603         unsigned num_pixels = src_size;
604         for(i=0; i<num_pixels; i++)
605         {
606             unsigned b,g,r;
607             register uint8_t rgb;
608             rgb = src[i];
609             r = (rgb&0x07);
610             g = (rgb&0x38)>>3;
611             b = (rgb&0xC0)>>6;
612             dst[i] = ((b<<1)&0x07) | ((g&0x07)<<3) | ((r&0x03)<<6);
613         }
614 }