2 * rgb2rgb.c, Software RGB to RGB convertor
3 * pluralize by Software PAL8 to RGB convertor
4 * Software YUV to YUV convertor
5 * Software YUV to RGB convertor
6 * Written by Nick Kurshev.
7 * palette & YUV & runtime CPU stuff by Michael (michaelni@gmx.at)
9 * This file is part of FFmpeg.
11 * FFmpeg is free software; you can redistribute it and/or modify
12 * it under the terms of the GNU General Public License as published by
13 * the Free Software Foundation; either version 2 of the License, or
14 * (at your option) any later version.
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 * GNU General Public License for more details.
21 * You should have received a copy of the GNU General Public License
22 * along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 * the C code (not assembly, mmx, ...) of this file can be used
26 * under the LGPL license too
32 #include "swscale_internal.h"
36 #define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit
38 void (*rgb24to32
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
39 void (*rgb24to16
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
40 void (*rgb24to15
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
41 void (*rgb32to24
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
42 void (*rgb32to16
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
43 void (*rgb32to15
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
44 void (*rgb15to16
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
45 void (*rgb15to24
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
46 void (*rgb15to32
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
47 void (*rgb16to15
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
48 void (*rgb16to24
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
49 void (*rgb16to32
)(const uint8_t *src
,uint8_t *dst
,long src_size
);
50 //void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, long src_size);
51 void (*rgb24tobgr24
)(const uint8_t *src
, uint8_t *dst
, long src_size
);
52 void (*rgb24tobgr16
)(const uint8_t *src
, uint8_t *dst
, long src_size
);
53 void (*rgb24tobgr15
)(const uint8_t *src
, uint8_t *dst
, long src_size
);
54 void (*rgb32tobgr32
)(const uint8_t *src
, uint8_t *dst
, long src_size
);
55 //void (*rgb32tobgr24)(const uint8_t *src, uint8_t *dst, long src_size);
56 void (*rgb32tobgr16
)(const uint8_t *src
, uint8_t *dst
, long src_size
);
57 void (*rgb32tobgr15
)(const uint8_t *src
, uint8_t *dst
, long src_size
);
59 void (*yv12toyuy2
)(const uint8_t *ysrc
, const uint8_t *usrc
, const uint8_t *vsrc
, uint8_t *dst
,
60 long width
, long height
,
61 long lumStride
, long chromStride
, long dstStride
);
62 void (*yv12touyvy
)(const uint8_t *ysrc
, const uint8_t *usrc
, const uint8_t *vsrc
, uint8_t *dst
,
63 long width
, long height
,
64 long lumStride
, long chromStride
, long dstStride
);
65 void (*yuv422ptoyuy2
)(const uint8_t *ysrc
, const uint8_t *usrc
, const uint8_t *vsrc
, uint8_t *dst
,
66 long width
, long height
,
67 long lumStride
, long chromStride
, long dstStride
);
68 void (*yuy2toyv12
)(const uint8_t *src
, uint8_t *ydst
, uint8_t *udst
, uint8_t *vdst
,
69 long width
, long height
,
70 long lumStride
, long chromStride
, long srcStride
);
71 void (*rgb24toyv12
)(const uint8_t *src
, uint8_t *ydst
, uint8_t *udst
, uint8_t *vdst
,
72 long width
, long height
,
73 long lumStride
, long chromStride
, long srcStride
);
74 void (*planar2x
)(const uint8_t *src
, uint8_t *dst
, long width
, long height
,
75 long srcStride
, long dstStride
);
76 void (*interleaveBytes
)(uint8_t *src1
, uint8_t *src2
, uint8_t *dst
,
77 long width
, long height
, long src1Stride
,
78 long src2Stride
, long dstStride
);
79 void (*vu9_to_vu12
)(const uint8_t *src1
, const uint8_t *src2
,
80 uint8_t *dst1
, uint8_t *dst2
,
81 long width
, long height
,
82 long srcStride1
, long srcStride2
,
83 long dstStride1
, long dstStride2
);
84 void (*yvu9_to_yuy2
)(const uint8_t *src1
, const uint8_t *src2
, const uint8_t *src3
,
86 long width
, long height
,
87 long srcStride1
, long srcStride2
,
88 long srcStride3
, long dstStride
);
90 #if defined(ARCH_X86) && defined(CONFIG_GPL)
91 static const uint64_t mmx_null
__attribute__((aligned(8))) = 0x0000000000000000ULL
;
92 static const uint64_t mmx_one
__attribute__((aligned(8))) = 0xFFFFFFFFFFFFFFFFULL
;
93 static const uint64_t mask32b attribute_used
__attribute__((aligned(8))) = 0x000000FF000000FFULL
;
94 static const uint64_t mask32g attribute_used
__attribute__((aligned(8))) = 0x0000FF000000FF00ULL
;
95 static const uint64_t mask32r attribute_used
__attribute__((aligned(8))) = 0x00FF000000FF0000ULL
;
96 static const uint64_t mask32
__attribute__((aligned(8))) = 0x00FFFFFF00FFFFFFULL
;
97 static const uint64_t mask3216br
__attribute__((aligned(8))) = 0x00F800F800F800F8ULL
;
98 static const uint64_t mask3216g
__attribute__((aligned(8))) = 0x0000FC000000FC00ULL
;
99 static const uint64_t mask3215g
__attribute__((aligned(8))) = 0x0000F8000000F800ULL
;
100 static const uint64_t mul3216
__attribute__((aligned(8))) = 0x2000000420000004ULL
;
101 static const uint64_t mul3215
__attribute__((aligned(8))) = 0x2000000820000008ULL
;
102 static const uint64_t mask24b attribute_used
__attribute__((aligned(8))) = 0x00FF0000FF0000FFULL
;
103 static const uint64_t mask24g attribute_used
__attribute__((aligned(8))) = 0xFF0000FF0000FF00ULL
;
104 static const uint64_t mask24r attribute_used
__attribute__((aligned(8))) = 0x0000FF0000FF0000ULL
;
105 static const uint64_t mask24l
__attribute__((aligned(8))) = 0x0000000000FFFFFFULL
;
106 static const uint64_t mask24h
__attribute__((aligned(8))) = 0x0000FFFFFF000000ULL
;
107 static const uint64_t mask24hh
__attribute__((aligned(8))) = 0xffff000000000000ULL
;
108 static const uint64_t mask24hhh
__attribute__((aligned(8))) = 0xffffffff00000000ULL
;
109 static const uint64_t mask24hhhh
__attribute__((aligned(8))) = 0xffffffffffff0000ULL
;
110 static const uint64_t mask15b
__attribute__((aligned(8))) = 0x001F001F001F001FULL
; /* 00000000 00011111 xxB */
111 static const uint64_t mask15rg
__attribute__((aligned(8))) = 0x7FE07FE07FE07FE0ULL
; /* 01111111 11100000 RGx */
112 static const uint64_t mask15s
__attribute__((aligned(8))) = 0xFFE0FFE0FFE0FFE0ULL
;
113 static const uint64_t mask15g
__attribute__((aligned(8))) = 0x03E003E003E003E0ULL
;
114 static const uint64_t mask15r
__attribute__((aligned(8))) = 0x7C007C007C007C00ULL
;
115 #define mask16b mask15b
116 static const uint64_t mask16g
__attribute__((aligned(8))) = 0x07E007E007E007E0ULL
;
117 static const uint64_t mask16r
__attribute__((aligned(8))) = 0xF800F800F800F800ULL
;
118 static const uint64_t red_16mask
__attribute__((aligned(8))) = 0x0000f8000000f800ULL
;
119 static const uint64_t green_16mask
__attribute__((aligned(8))) = 0x000007e0000007e0ULL
;
120 static const uint64_t blue_16mask
__attribute__((aligned(8))) = 0x0000001f0000001fULL
;
121 static const uint64_t red_15mask
__attribute__((aligned(8))) = 0x00007c0000007c00ULL
;
122 static const uint64_t green_15mask
__attribute__((aligned(8))) = 0x000003e0000003e0ULL
;
123 static const uint64_t blue_15mask
__attribute__((aligned(8))) = 0x0000001f0000001fULL
;
126 static const uint64_t bgr2YCoeff attribute_used
__attribute__((aligned(8))) = 0x000000210041000DULL
;
127 static const uint64_t bgr2UCoeff attribute_used
__attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL
;
128 static const uint64_t bgr2VCoeff attribute_used
__attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL
;
130 static const uint64_t bgr2YCoeff attribute_used
__attribute__((aligned(8))) = 0x000020E540830C8BULL
;
131 static const uint64_t bgr2UCoeff attribute_used
__attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL
;
132 static const uint64_t bgr2VCoeff attribute_used
__attribute__((aligned(8))) = 0x00003831D0E6F6EAULL
;
134 static const uint64_t bgr2YOffset attribute_used
__attribute__((aligned(8))) = 0x1010101010101010ULL
;
135 static const uint64_t bgr2UVOffset attribute_used
__attribute__((aligned(8))) = 0x8080808080808080ULL
;
136 static const uint64_t w1111 attribute_used
__attribute__((aligned(8))) = 0x0001000100010001ULL
;
139 static volatile uint64_t __attribute__((aligned(8))) b5Dither
;
140 static volatile uint64_t __attribute__((aligned(8))) g5Dither
;
141 static volatile uint64_t __attribute__((aligned(8))) g6Dither
;
142 static volatile uint64_t __attribute__((aligned(8))) r5Dither
;
144 static uint64_t __attribute__((aligned(8))) dither4
[2]={
145 0x0103010301030103LL
,
146 0x0200020002000200LL
,};
148 static uint64_t __attribute__((aligned(8))) dither8
[2]={
149 0x0602060206020602LL
,
150 0x0004000400040004LL
,};
152 #endif /* defined(ARCH_X86) */
154 #define RGB2YUV_SHIFT 8
155 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))
156 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))
157 #define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
158 #define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))
159 #define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))
160 #define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))
161 #define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))
162 #define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))
163 #define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))
165 //Note: we have C, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one
171 #define RENAME(a) a ## _C
172 #include "rgb2rgb_template.c"
174 #if defined(ARCH_X86) && defined(CONFIG_GPL)
182 #define RENAME(a) a ## _MMX
183 #include "rgb2rgb_template.c"
191 #define RENAME(a) a ## _MMX2
192 #include "rgb2rgb_template.c"
200 #define RENAME(a) a ## _3DNOW
201 #include "rgb2rgb_template.c"
203 #endif //ARCH_X86 || ARCH_X86_64
206 rgb15->rgb16 Original by Strepto/Astral
207 ported to gcc & bugfixed : A'rpi
208 MMX2, 3DNOW optimization by Nick Kurshev
209 32bit c version, and and&add trick by Michael Niedermayer
212 void sws_rgb2rgb_init(int flags
){
213 #if (defined(HAVE_MMX2) || defined(HAVE_3DNOW) || defined(HAVE_MMX)) && defined(CONFIG_GPL)
214 if (flags
& SWS_CPU_CAPS_MMX2
)
216 else if (flags
& SWS_CPU_CAPS_3DNOW
)
217 rgb2rgb_init_3DNOW();
218 else if (flags
& SWS_CPU_CAPS_MMX
)
221 #endif /* defined(HAVE_MMX2) || defined(HAVE_3DNOW) || defined(HAVE_MMX) */
226 * Palette is assumed to contain BGR32.
228 void palette8torgb32(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
233 for (i=0; i<num_pixels; i++)
234 ((unsigned *)dst)[i] = ((unsigned *)palette)[ src[i] ];
237 for (i
=0; i
<num_pixels
; i
++)
239 #ifdef WORDS_BIGENDIAN
240 dst
[3]= palette
[ src
[i
]*4+2 ];
241 dst
[2]= palette
[ src
[i
]*4+1 ];
242 dst
[1]= palette
[ src
[i
]*4+0 ];
245 dst
[0]= palette
[ src
[i
]*4+2 ];
246 dst
[1]= palette
[ src
[i
]*4+1 ];
247 dst
[2]= palette
[ src
[i
]*4+0 ];
248 //dst[3]= 0; /* do we need this cleansing? */
254 void palette8tobgr32(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
257 for (i
=0; i
<num_pixels
; i
++)
259 #ifdef WORDS_BIGENDIAN
260 dst
[3]= palette
[ src
[i
]*4+0 ];
261 dst
[2]= palette
[ src
[i
]*4+1 ];
262 dst
[1]= palette
[ src
[i
]*4+2 ];
265 dst
[0]= palette
[ src
[i
]*4+0 ];
266 dst
[1]= palette
[ src
[i
]*4+1 ];
267 dst
[2]= palette
[ src
[i
]*4+2 ];
268 //dst[3]= 0; /* do we need this cleansing? */
276 * Palette is assumed to contain BGR32.
278 void palette8torgb24(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
282 writes 1 byte o much and might cause alignment issues on some architectures?
283 for (i=0; i<num_pixels; i++)
284 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
286 for (i
=0; i
<num_pixels
; i
++)
289 dst
[0]= palette
[ src
[i
]*4+2 ];
290 dst
[1]= palette
[ src
[i
]*4+1 ];
291 dst
[2]= palette
[ src
[i
]*4+0 ];
296 void palette8tobgr24(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
300 writes 1 byte o much and might cause alignment issues on some architectures?
301 for (i=0; i<num_pixels; i++)
302 ((unsigned *)(&dst[i*3])) = ((unsigned *)palette)[ src[i] ];
304 for (i
=0; i
<num_pixels
; i
++)
307 dst
[0]= palette
[ src
[i
]*4+0 ];
308 dst
[1]= palette
[ src
[i
]*4+1 ];
309 dst
[2]= palette
[ src
[i
]*4+2 ];
315 * Palette is assumed to contain bgr16, see rgb32to16 to convert the palette
317 void palette8torgb16(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
320 for (i
=0; i
<num_pixels
; i
++)
321 ((uint16_t *)dst
)[i
] = ((uint16_t *)palette
)[ src
[i
] ];
323 void palette8tobgr16(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
326 for (i
=0; i
<num_pixels
; i
++)
327 ((uint16_t *)dst
)[i
] = bswap_16(((uint16_t *)palette
)[ src
[i
] ]);
331 * Palette is assumed to contain BGR15, see rgb32to15 to convert the palette.
333 void palette8torgb15(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
336 for (i
=0; i
<num_pixels
; i
++)
337 ((uint16_t *)dst
)[i
] = ((uint16_t *)palette
)[ src
[i
] ];
339 void palette8tobgr15(const uint8_t *src
, uint8_t *dst
, long num_pixels
, const uint8_t *palette
)
342 for (i
=0; i
<num_pixels
; i
++)
343 ((uint16_t *)dst
)[i
] = bswap_16(((uint16_t *)palette
)[ src
[i
] ]);
346 void rgb32tobgr24(const uint8_t *src
, uint8_t *dst
, long src_size
)
349 long num_pixels
= src_size
>> 2;
350 for (i
=0; i
<num_pixels
; i
++)
352 #ifdef WORDS_BIGENDIAN
353 /* RGB32 (= A,B,G,R) -> BGR24 (= B,G,R) */
354 dst
[3*i
+ 0] = src
[4*i
+ 1];
355 dst
[3*i
+ 1] = src
[4*i
+ 2];
356 dst
[3*i
+ 2] = src
[4*i
+ 3];
358 dst
[3*i
+ 0] = src
[4*i
+ 2];
359 dst
[3*i
+ 1] = src
[4*i
+ 1];
360 dst
[3*i
+ 2] = src
[4*i
+ 0];
365 void rgb24tobgr32(const uint8_t *src
, uint8_t *dst
, long src_size
)
368 for (i
=0; 3*i
<src_size
; i
++)
370 #ifdef WORDS_BIGENDIAN
371 /* RGB24 (= R,G,B) -> BGR32 (= A,R,G,B) */
373 dst
[4*i
+ 1] = src
[3*i
+ 0];
374 dst
[4*i
+ 2] = src
[3*i
+ 1];
375 dst
[4*i
+ 3] = src
[3*i
+ 2];
377 dst
[4*i
+ 0] = src
[3*i
+ 2];
378 dst
[4*i
+ 1] = src
[3*i
+ 1];
379 dst
[4*i
+ 2] = src
[3*i
+ 0];
385 void rgb16tobgr32(const uint8_t *src
, uint8_t *dst
, long src_size
)
388 uint8_t *d
= (uint8_t *)dst
;
389 const uint16_t *s
= (uint16_t *)src
;
390 end
= s
+ src_size
/2;
393 register uint16_t bgr
;
395 #ifdef WORDS_BIGENDIAN
397 *d
++ = (bgr
&0x1F)<<3;
398 *d
++ = (bgr
&0x7E0)>>3;
399 *d
++ = (bgr
&0xF800)>>8;
401 *d
++ = (bgr
&0xF800)>>8;
402 *d
++ = (bgr
&0x7E0)>>3;
403 *d
++ = (bgr
&0x1F)<<3;
409 void rgb16tobgr24(const uint8_t *src
, uint8_t *dst
, long src_size
)
412 uint8_t *d
= (uint8_t *)dst
;
413 const uint16_t *s
= (const uint16_t *)src
;
414 end
= s
+ src_size
/2;
417 register uint16_t bgr
;
419 *d
++ = (bgr
&0xF800)>>8;
420 *d
++ = (bgr
&0x7E0)>>3;
421 *d
++ = (bgr
&0x1F)<<3;
425 void rgb16tobgr16(const uint8_t *src
, uint8_t *dst
, long src_size
)
428 long num_pixels
= src_size
>> 1;
430 for (i
=0; i
<num_pixels
; i
++)
433 register uint16_t rgb
;
437 b
= (rgb
&0xF800)>>11;
438 dst
[2*i
] = (b
&0x1F) | ((g
&0x3F)<<5) | ((r
&0x1F)<<11);
442 void rgb16tobgr15(const uint8_t *src
, uint8_t *dst
, long src_size
)
445 long num_pixels
= src_size
>> 1;
447 for (i
=0; i
<num_pixels
; i
++)
450 register uint16_t rgb
;
454 b
= (rgb
&0xF800)>>11;
455 dst
[2*i
] = (b
&0x1F) | ((g
&0x1F)<<5) | ((r
&0x1F)<<10);
459 void rgb15tobgr32(const uint8_t *src
, uint8_t *dst
, long src_size
)
462 uint8_t *d
= (uint8_t *)dst
;
463 const uint16_t *s
= (const uint16_t *)src
;
464 end
= s
+ src_size
/2;
467 register uint16_t bgr
;
469 #ifdef WORDS_BIGENDIAN
471 *d
++ = (bgr
&0x1F)<<3;
472 *d
++ = (bgr
&0x3E0)>>2;
473 *d
++ = (bgr
&0x7C00)>>7;
475 *d
++ = (bgr
&0x7C00)>>7;
476 *d
++ = (bgr
&0x3E0)>>2;
477 *d
++ = (bgr
&0x1F)<<3;
483 void rgb15tobgr24(const uint8_t *src
, uint8_t *dst
, long src_size
)
486 uint8_t *d
= (uint8_t *)dst
;
487 const uint16_t *s
= (uint16_t *)src
;
488 end
= s
+ src_size
/2;
491 register uint16_t bgr
;
493 *d
++ = (bgr
&0x7C00)>>7;
494 *d
++ = (bgr
&0x3E0)>>2;
495 *d
++ = (bgr
&0x1F)<<3;
499 void rgb15tobgr16(const uint8_t *src
, uint8_t *dst
, long src_size
)
502 long num_pixels
= src_size
>> 1;
504 for (i
=0; i
<num_pixels
; i
++)
507 register uint16_t rgb
;
511 b
= (rgb
&0x7C00)>>10;
512 dst
[2*i
] = (b
&0x1F) | ((g
&0x3F)<<5) | ((r
&0x1F)<<11);
516 void rgb15tobgr15(const uint8_t *src
, uint8_t *dst
, long src_size
)
519 long num_pixels
= src_size
>> 1;
521 for (i
=0; i
<num_pixels
; i
++)
524 register uint16_t rgb
;
528 b
= (rgb
&0x7C00)>>10;
529 dst
[2*i
] = (b
&0x1F) | ((g
&0x1F)<<5) | ((r
&0x1F)<<10);
533 void rgb8tobgr8(const uint8_t *src
, uint8_t *dst
, long src_size
)
536 long num_pixels
= src_size
;
537 for (i
=0; i
<num_pixels
; i
++)
540 register uint8_t rgb
;
545 dst
[i
] = ((b
<<1)&0x07) | ((g
&0x07)<<3) | ((r
&0x03)<<6);