2732
|
1 /*
|
|
2 * yuv2rgb.c, Software YUV to RGB coverter
|
|
3 *
|
|
4 * Copyright (C) 1999, Aaron Holtzman <aholtzma@ess.engr.uvic.ca>
|
|
5 * All Rights Reserved.
|
|
6 *
|
|
7 * Functions broken out from display_x11.c and several new modes
|
|
8 * added by H�kan Hjort <d95hjort@dtek.chalmers.se>
|
|
9 *
|
|
10 * 15 & 16 bpp support by Franck Sicard <Franck.Sicard@solsoft.fr>
|
|
11 *
|
|
12 * This file is part of mpeg2dec, a free MPEG-2 video decoder
|
|
13 *
|
|
14 * mpeg2dec is free software; you can redistribute it and/or modify
|
|
15 * it under the terms of the GNU General Public License as published by
|
|
16 * the Free Software Foundation; either version 2, or (at your option)
|
|
17 * any later version.
|
|
18 *
|
|
19 * mpeg2dec is distributed in the hope that it will be useful,
|
|
20 * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
21 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
22 * GNU General Public License for more details.
|
|
23 *
|
|
24 * You should have received a copy of the GNU General Public License
|
|
25 * along with GNU Make; see the file COPYING. If not, write to
|
|
26 * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
|
|
27 *
|
3143
|
28 * MMX/MMX2 Template stuff from Michael Niedermayer (michaelni@gmx.at) (needed for fast movntq support)
|
6456
|
29 * 1,4,8bpp support by Michael Niedermayer (michaelni@gmx.at)
|
9476
|
30 * context / deglobalize stuff by Michael Niedermayer
|
2732
|
31 */
|
|
32
|
|
33 #include <stdio.h>
|
|
34 #include <stdlib.h>
|
|
35 #include <inttypes.h>
|
9476
|
36 #include <assert.h>
|
2732
|
37
|
|
38 #include "config.h"
|
|
39 //#include "video_out.h"
|
|
40 #include "rgb2rgb.h"
|
9476
|
41 #include "swscale.h"
|
|
42 #include "swscale_internal.h"
|
4285
|
43 #include "../mangle.h"
|
9476
|
44 #include "../libvo/img_format.h" //FIXME try to reduce dependency of such stuff
|
2732
|
45
|
|
46 #ifdef HAVE_MLIB
|
|
47 #include "yuv2rgb_mlib.c"
|
|
48 #endif
|
|
49
|
3143
|
50 #define DITHER1XBPP // only for mmx
|
|
51
|
6577
|
52 const uint8_t __attribute__((aligned(8))) dither_2x2_4[2][8]={
|
|
53 { 1, 3, 1, 3, 1, 3, 1, 3, },
|
|
54 { 2, 0, 2, 0, 2, 0, 2, 0, },
|
|
55 };
|
|
56
|
|
57 const uint8_t __attribute__((aligned(8))) dither_2x2_8[2][8]={
|
|
58 { 6, 2, 6, 2, 6, 2, 6, 2, },
|
|
59 { 0, 4, 0, 4, 0, 4, 0, 4, },
|
|
60 };
|
|
61
|
|
62 const uint8_t __attribute__((aligned(8))) dither_8x8_32[8][8]={
|
6458
|
63 { 17, 9, 23, 15, 16, 8, 22, 14, },
|
|
64 { 5, 29, 3, 27, 4, 28, 2, 26, },
|
|
65 { 21, 13, 19, 11, 20, 12, 18, 10, },
|
|
66 { 0, 24, 6, 30, 1, 25, 7, 31, },
|
|
67 { 16, 8, 22, 14, 17, 9, 23, 15, },
|
|
68 { 4, 28, 2, 26, 5, 29, 3, 27, },
|
|
69 { 20, 12, 18, 10, 21, 13, 19, 11, },
|
|
70 { 1, 25, 7, 31, 0, 24, 6, 30, },
|
|
71 };
|
|
72
|
6577
|
73 #if 0
|
|
74 const uint8_t __attribute__((aligned(8))) dither_8x8_64[8][8]={
|
6458
|
75 { 0, 48, 12, 60, 3, 51, 15, 63, },
|
|
76 { 32, 16, 44, 28, 35, 19, 47, 31, },
|
|
77 { 8, 56, 4, 52, 11, 59, 7, 55, },
|
|
78 { 40, 24, 36, 20, 43, 27, 39, 23, },
|
|
79 { 2, 50, 14, 62, 1, 49, 13, 61, },
|
|
80 { 34, 18, 46, 30, 33, 17, 45, 29, },
|
|
81 { 10, 58, 6, 54, 9, 57, 5, 53, },
|
|
82 { 42, 26, 38, 22, 41, 25, 37, 21, },
|
|
83 };
|
6577
|
84 #endif
|
6458
|
85
|
6577
|
86 const uint8_t __attribute__((aligned(8))) dither_8x8_73[8][8]={
|
|
87 { 0, 55, 14, 68, 3, 58, 17, 72, },
|
|
88 { 37, 18, 50, 32, 40, 22, 54, 35, },
|
|
89 { 9, 64, 5, 59, 13, 67, 8, 63, },
|
|
90 { 46, 27, 41, 23, 49, 31, 44, 26, },
|
|
91 { 2, 57, 16, 71, 1, 56, 15, 70, },
|
|
92 { 39, 21, 52, 34, 38, 19, 51, 33, },
|
|
93 { 11, 66, 7, 62, 10, 65, 6, 60, },
|
|
94 { 48, 30, 43, 25, 47, 29, 42, 24, },
|
|
95 };
|
|
96
|
|
97 #if 0
|
|
98 const uint8_t __attribute__((aligned(8))) dither_8x8_128[8][8]={
|
6458
|
99 { 68, 36, 92, 60, 66, 34, 90, 58, },
|
|
100 { 20, 116, 12, 108, 18, 114, 10, 106, },
|
|
101 { 84, 52, 76, 44, 82, 50, 74, 42, },
|
|
102 { 0, 96, 24, 120, 6, 102, 30, 126, },
|
|
103 { 64, 32, 88, 56, 70, 38, 94, 62, },
|
|
104 { 16, 112, 8, 104, 22, 118, 14, 110, },
|
|
105 { 80, 48, 72, 40, 86, 54, 78, 46, },
|
|
106 { 4, 100, 28, 124, 2, 98, 26, 122, },
|
|
107 };
|
6577
|
108 #endif
|
|
109
|
|
110 #if 1
|
|
111 const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
|
|
112 {117, 62, 158, 103, 113, 58, 155, 100, },
|
|
113 { 34, 199, 21, 186, 31, 196, 17, 182, },
|
|
114 {144, 89, 131, 76, 141, 86, 127, 72, },
|
|
115 { 0, 165, 41, 206, 10, 175, 52, 217, },
|
|
116 {110, 55, 151, 96, 120, 65, 162, 107, },
|
|
117 { 28, 193, 14, 179, 38, 203, 24, 189, },
|
|
118 {138, 83, 124, 69, 148, 93, 134, 79, },
|
|
119 { 7, 172, 48, 213, 3, 168, 45, 210, },
|
|
120 };
|
|
121 #elif 1
|
|
122 // tries to correct a gamma of 1.5
|
|
123 const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
|
|
124 { 0, 143, 18, 200, 2, 156, 25, 215, },
|
|
125 { 78, 28, 125, 64, 89, 36, 138, 74, },
|
|
126 { 10, 180, 3, 161, 16, 195, 8, 175, },
|
|
127 {109, 51, 93, 38, 121, 60, 105, 47, },
|
|
128 { 1, 152, 23, 210, 0, 147, 20, 205, },
|
|
129 { 85, 33, 134, 71, 81, 30, 130, 67, },
|
|
130 { 14, 190, 6, 171, 12, 185, 5, 166, },
|
|
131 {117, 57, 101, 44, 113, 54, 97, 41, },
|
|
132 };
|
|
133 #elif 1
|
|
134 // tries to correct a gamma of 2.0
|
|
135 const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
|
|
136 { 0, 124, 8, 193, 0, 140, 12, 213, },
|
|
137 { 55, 14, 104, 42, 66, 19, 119, 52, },
|
|
138 { 3, 168, 1, 145, 6, 187, 3, 162, },
|
|
139 { 86, 31, 70, 21, 99, 39, 82, 28, },
|
|
140 { 0, 134, 11, 206, 0, 129, 9, 200, },
|
|
141 { 62, 17, 114, 48, 58, 16, 109, 45, },
|
|
142 { 5, 181, 2, 157, 4, 175, 1, 151, },
|
|
143 { 95, 36, 78, 26, 90, 34, 74, 24, },
|
|
144 };
|
|
145 #else
|
|
146 // tries to correct a gamma of 2.5
|
|
147 const uint8_t __attribute__((aligned(8))) dither_8x8_220[8][8]={
|
|
148 { 0, 107, 3, 187, 0, 125, 6, 212, },
|
|
149 { 39, 7, 86, 28, 49, 11, 102, 36, },
|
|
150 { 1, 158, 0, 131, 3, 180, 1, 151, },
|
|
151 { 68, 19, 52, 12, 81, 25, 64, 17, },
|
|
152 { 0, 119, 5, 203, 0, 113, 4, 195, },
|
|
153 { 45, 9, 96, 33, 42, 8, 91, 30, },
|
|
154 { 2, 172, 1, 144, 2, 165, 0, 137, },
|
|
155 { 77, 23, 60, 15, 72, 21, 56, 14, },
|
|
156 };
|
|
157 #endif
|
6458
|
158
|
9476
|
159 #ifdef ARCH_X86
|
3143
|
160
|
|
161 /* hope these constant values are cache line aligned */
|
12301
|
162 uint64_t attribute_used __attribute__((aligned(8))) mmx_00ffw = 0x00ff00ff00ff00ffULL;
|
|
163 uint64_t attribute_used __attribute__((aligned(8))) mmx_redmask = 0xf8f8f8f8f8f8f8f8ULL;
|
|
164 uint64_t attribute_used __attribute__((aligned(8))) mmx_grnmask = 0xfcfcfcfcfcfcfcfcULL;
|
3143
|
165
|
12301
|
166 uint64_t attribute_used __attribute__((aligned(8))) M24A= 0x00FF0000FF0000FFULL;
|
|
167 uint64_t attribute_used __attribute__((aligned(8))) M24B= 0xFF0000FF0000FF00ULL;
|
|
168 uint64_t attribute_used __attribute__((aligned(8))) M24C= 0x0000FF0000FF0000ULL;
|
3143
|
169
|
|
170 // the volatile is required because gcc otherwise optimizes some writes away not knowing that these
|
|
171 // are read in the asm block
|
12301
|
172 volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither;
|
|
173 volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither;
|
|
174 volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither;
|
|
175 volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither;
|
3143
|
176
|
|
177 uint64_t __attribute__((aligned(8))) dither4[2]={
|
|
178 0x0103010301030103LL,
|
|
179 0x0200020002000200LL,};
|
|
180
|
|
181 uint64_t __attribute__((aligned(8))) dither8[2]={
|
|
182 0x0602060206020602LL,
|
|
183 0x0004000400040004LL,};
|
|
184
|
|
185 #undef HAVE_MMX
|
|
186 #undef ARCH_X86
|
|
187
|
|
188 //MMX versions
|
|
189 #undef RENAME
|
|
190 #define HAVE_MMX
|
|
191 #undef HAVE_MMX2
|
|
192 #undef HAVE_3DNOW
|
|
193 #define ARCH_X86
|
|
194 #define RENAME(a) a ## _MMX
|
|
195 #include "yuv2rgb_template.c"
|
|
196
|
|
197 //MMX2 versions
|
|
198 #undef RENAME
|
|
199 #define HAVE_MMX
|
|
200 #define HAVE_MMX2
|
|
201 #undef HAVE_3DNOW
|
|
202 #define ARCH_X86
|
|
203 #define RENAME(a) a ## _MMX2
|
|
204 #include "yuv2rgb_template.c"
|
|
205
|
|
206 #endif // CAN_COMPILE_X86_ASM
|
2732
|
207
|
|
208 const int32_t Inverse_Table_6_9[8][4] = {
|
|
209 {117504, 138453, 13954, 34903}, /* no sequence_display_extension */
|
|
210 {117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */
|
|
211 {104597, 132201, 25675, 53279}, /* unspecified */
|
|
212 {104597, 132201, 25675, 53279}, /* reserved */
|
|
213 {104448, 132798, 24759, 53109}, /* FCC */
|
|
214 {104597, 132201, 25675, 53279}, /* ITU-R Rec. 624-4 System B, G */
|
|
215 {104597, 132201, 25675, 53279}, /* SMPTE 170M */
|
|
216 {117579, 136230, 16907, 35559} /* SMPTE 240M (1987) */
|
|
217 };
|
|
218
|
|
219 #define RGB(i) \
|
|
220 U = pu[i]; \
|
|
221 V = pv[i]; \
|
9476
|
222 r = c->table_rV[V]; \
|
|
223 g = c->table_gU[U] + c->table_gV[V]; \
|
|
224 b = c->table_bU[U];
|
2732
|
225
|
|
226 #define DST1(i) \
|
|
227 Y = py_1[2*i]; \
|
|
228 dst_1[2*i] = r[Y] + g[Y] + b[Y]; \
|
|
229 Y = py_1[2*i+1]; \
|
|
230 dst_1[2*i+1] = r[Y] + g[Y] + b[Y];
|
|
231
|
|
232 #define DST2(i) \
|
|
233 Y = py_2[2*i]; \
|
|
234 dst_2[2*i] = r[Y] + g[Y] + b[Y]; \
|
|
235 Y = py_2[2*i+1]; \
|
|
236 dst_2[2*i+1] = r[Y] + g[Y] + b[Y];
|
|
237
|
|
238 #define DST1RGB(i) \
|
|
239 Y = py_1[2*i]; \
|
|
240 dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y]; \
|
|
241 Y = py_1[2*i+1]; \
|
|
242 dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y];
|
|
243
|
|
244 #define DST2RGB(i) \
|
|
245 Y = py_2[2*i]; \
|
|
246 dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y]; \
|
|
247 Y = py_2[2*i+1]; \
|
|
248 dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y];
|
|
249
|
|
250 #define DST1BGR(i) \
|
|
251 Y = py_1[2*i]; \
|
|
252 dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y]; \
|
|
253 Y = py_1[2*i+1]; \
|
|
254 dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y];
|
|
255
|
|
256 #define DST2BGR(i) \
|
|
257 Y = py_2[2*i]; \
|
|
258 dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y]; \
|
|
259 Y = py_2[2*i+1]; \
|
|
260 dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y];
|
|
261
|
9476
|
262 #define PROLOG(func_name, dst_type) \
|
9499
|
263 static int func_name(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, \
|
9476
|
264 int srcSliceH, uint8_t* dst[], int dstStride[]){\
|
|
265 int y;\
|
|
266 \
|
|
267 if(c->srcFormat == IMGFMT_422P){\
|
|
268 srcStride[1] *= 2;\
|
|
269 srcStride[2] *= 2;\
|
|
270 }\
|
|
271 for(y=0; y<srcSliceH; y+=2){\
|
|
272 dst_type *dst_1= (dst_type*)(dst[0] + (y+srcSliceY )*dstStride[0]);\
|
|
273 dst_type *dst_2= (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
|
|
274 dst_type *r, *g, *b;\
|
|
275 uint8_t *py_1= src[0] + y*srcStride[0];\
|
|
276 uint8_t *py_2= py_1 + srcStride[0];\
|
|
277 uint8_t *pu= src[1] + (y>>1)*srcStride[1];\
|
|
278 uint8_t *pv= src[2] + (y>>1)*srcStride[2];\
|
|
279 unsigned int h_size= c->dstW>>3;\
|
|
280 while (h_size--) {\
|
|
281 int U, V, Y;\
|
2732
|
282
|
9476
|
283 #define EPILOG(dst_delta)\
|
|
284 pu += 4;\
|
|
285 pv += 4;\
|
|
286 py_1 += 8;\
|
|
287 py_2 += 8;\
|
|
288 dst_1 += dst_delta;\
|
|
289 dst_2 += dst_delta;\
|
|
290 }\
|
|
291 }\
|
9494
|
292 return srcSliceH;\
|
9476
|
293 }
|
2732
|
294
|
9476
|
295 PROLOG(yuv2rgb_c_32, uint32_t)
|
2732
|
296 RGB(0);
|
|
297 DST1(0);
|
|
298 DST2(0);
|
|
299
|
|
300 RGB(1);
|
|
301 DST2(1);
|
|
302 DST1(1);
|
|
303
|
|
304 RGB(2);
|
|
305 DST1(2);
|
|
306 DST2(2);
|
|
307
|
|
308 RGB(3);
|
|
309 DST2(3);
|
|
310 DST1(3);
|
9476
|
311 EPILOG(8)
|
2732
|
312
|
9476
|
313 PROLOG(yuv2rgb_c_24_rgb, uint8_t)
|
2732
|
314 RGB(0);
|
|
315 DST1RGB(0);
|
|
316 DST2RGB(0);
|
|
317
|
|
318 RGB(1);
|
|
319 DST2RGB(1);
|
|
320 DST1RGB(1);
|
|
321
|
|
322 RGB(2);
|
|
323 DST1RGB(2);
|
|
324 DST2RGB(2);
|
|
325
|
|
326 RGB(3);
|
|
327 DST2RGB(3);
|
|
328 DST1RGB(3);
|
9476
|
329 EPILOG(24)
|
2732
|
330
|
|
331 // only trivial mods from yuv2rgb_c_24_rgb
|
9476
|
332 PROLOG(yuv2rgb_c_24_bgr, uint8_t)
|
2732
|
333 RGB(0);
|
|
334 DST1BGR(0);
|
|
335 DST2BGR(0);
|
|
336
|
|
337 RGB(1);
|
|
338 DST2BGR(1);
|
|
339 DST1BGR(1);
|
|
340
|
|
341 RGB(2);
|
|
342 DST1BGR(2);
|
|
343 DST2BGR(2);
|
|
344
|
|
345 RGB(3);
|
|
346 DST2BGR(3);
|
|
347 DST1BGR(3);
|
9476
|
348 EPILOG(24)
|
2732
|
349
|
|
350 // This is exactly the same code as yuv2rgb_c_32 except for the types of
|
|
351 // r, g, b, dst_1, dst_2
|
9476
|
352 PROLOG(yuv2rgb_c_16, uint16_t)
|
2732
|
353 RGB(0);
|
|
354 DST1(0);
|
|
355 DST2(0);
|
|
356
|
|
357 RGB(1);
|
|
358 DST2(1);
|
|
359 DST1(1);
|
|
360
|
|
361 RGB(2);
|
|
362 DST1(2);
|
|
363 DST2(2);
|
|
364
|
|
365 RGB(3);
|
|
366 DST2(3);
|
|
367 DST1(3);
|
9476
|
368 EPILOG(8)
|
2732
|
369
|
6451
|
370 // This is exactly the same code as yuv2rgb_c_32 except for the types of
|
|
371 // r, g, b, dst_1, dst_2
|
9476
|
372 PROLOG(yuv2rgb_c_8, uint8_t)
|
6451
|
373 RGB(0);
|
|
374 DST1(0);
|
|
375 DST2(0);
|
|
376
|
|
377 RGB(1);
|
|
378 DST2(1);
|
|
379 DST1(1);
|
|
380
|
|
381 RGB(2);
|
|
382 DST1(2);
|
|
383 DST2(2);
|
|
384
|
|
385 RGB(3);
|
|
386 DST2(3);
|
|
387 DST1(3);
|
9476
|
388 EPILOG(8)
|
6451
|
389
|
6452
|
390 // r, g, b, dst_1, dst_2
|
9476
|
391 PROLOG(yuv2rgb_c_8_ordered_dither, uint8_t)
|
|
392 const uint8_t *d32= dither_8x8_32[y&7];
|
|
393 const uint8_t *d64= dither_8x8_73[y&7];
|
6456
|
394 #define DST1bpp8(i,o) \
|
6452
|
395 Y = py_1[2*i]; \
|
6456
|
396 dst_1[2*i] = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]]; \
|
6452
|
397 Y = py_1[2*i+1]; \
|
6456
|
398 dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
|
6452
|
399
|
6456
|
400 #define DST2bpp8(i,o) \
|
6452
|
401 Y = py_2[2*i]; \
|
6456
|
402 dst_2[2*i] = r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]]; \
|
6452
|
403 Y = py_2[2*i+1]; \
|
6456
|
404 dst_2[2*i+1] = r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
|
6452
|
405
|
|
406
|
|
407 RGB(0);
|
6456
|
408 DST1bpp8(0,0);
|
|
409 DST2bpp8(0,0);
|
6452
|
410
|
|
411 RGB(1);
|
6456
|
412 DST2bpp8(1,2);
|
|
413 DST1bpp8(1,2);
|
6452
|
414
|
|
415 RGB(2);
|
6456
|
416 DST1bpp8(2,4);
|
|
417 DST2bpp8(2,4);
|
6452
|
418
|
|
419 RGB(3);
|
6456
|
420 DST2bpp8(3,6);
|
|
421 DST1bpp8(3,6);
|
9476
|
422 EPILOG(8)
|
6451
|
423
|
6453
|
424
|
|
425 // This is exactly the same code as yuv2rgb_c_32 except for the types of
|
|
426 // r, g, b, dst_1, dst_2
|
9476
|
427 PROLOG(yuv2rgb_c_4, uint8_t)
|
9171
|
428 int acc;
|
|
429 #define DST1_4(i) \
|
|
430 Y = py_1[2*i]; \
|
|
431 acc = r[Y] + g[Y] + b[Y]; \
|
|
432 Y = py_1[2*i+1]; \
|
|
433 acc |= (r[Y] + g[Y] + b[Y])<<4;\
|
|
434 dst_1[i] = acc;
|
|
435
|
|
436 #define DST2_4(i) \
|
|
437 Y = py_2[2*i]; \
|
|
438 acc = r[Y] + g[Y] + b[Y]; \
|
|
439 Y = py_2[2*i+1]; \
|
|
440 acc |= (r[Y] + g[Y] + b[Y])<<4;\
|
|
441 dst_2[i] = acc;
|
|
442
|
|
443 RGB(0);
|
|
444 DST1_4(0);
|
|
445 DST2_4(0);
|
|
446
|
|
447 RGB(1);
|
|
448 DST2_4(1);
|
|
449 DST1_4(1);
|
|
450
|
|
451 RGB(2);
|
|
452 DST1_4(2);
|
|
453 DST2_4(2);
|
|
454
|
|
455 RGB(3);
|
|
456 DST2_4(3);
|
|
457 DST1_4(3);
|
9476
|
458 EPILOG(4)
|
9171
|
459
|
9476
|
460 PROLOG(yuv2rgb_c_4_ordered_dither, uint8_t)
|
|
461 const uint8_t *d64= dither_8x8_73[y&7];
|
|
462 const uint8_t *d128=dither_8x8_220[y&7];
|
9171
|
463 int acc;
|
|
464
|
|
465 #define DST1bpp4(i,o) \
|
|
466 Y = py_1[2*i]; \
|
|
467 acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]]; \
|
|
468 Y = py_1[2*i+1]; \
|
|
469 acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4;\
|
|
470 dst_1[i]= acc;
|
|
471
|
|
472 #define DST2bpp4(i,o) \
|
|
473 Y = py_2[2*i]; \
|
|
474 acc = r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]]; \
|
|
475 Y = py_2[2*i+1]; \
|
|
476 acc |= (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4;\
|
|
477 dst_2[i]= acc;
|
|
478
|
|
479
|
|
480 RGB(0);
|
|
481 DST1bpp4(0,0);
|
|
482 DST2bpp4(0,0);
|
|
483
|
|
484 RGB(1);
|
|
485 DST2bpp4(1,2);
|
|
486 DST1bpp4(1,2);
|
|
487
|
|
488 RGB(2);
|
|
489 DST1bpp4(2,4);
|
|
490 DST2bpp4(2,4);
|
|
491
|
|
492 RGB(3);
|
|
493 DST2bpp4(3,6);
|
|
494 DST1bpp4(3,6);
|
9476
|
495 EPILOG(4)
|
9171
|
496
|
|
497 // This is exactly the same code as yuv2rgb_c_32 except for the types of
|
|
498 // r, g, b, dst_1, dst_2
|
9476
|
499 PROLOG(yuv2rgb_c_4b, uint8_t)
|
6453
|
500 RGB(0);
|
|
501 DST1(0);
|
|
502 DST2(0);
|
|
503
|
|
504 RGB(1);
|
|
505 DST2(1);
|
|
506 DST1(1);
|
|
507
|
|
508 RGB(2);
|
|
509 DST1(2);
|
|
510 DST2(2);
|
|
511
|
|
512 RGB(3);
|
|
513 DST2(3);
|
|
514 DST1(3);
|
9476
|
515 EPILOG(8)
|
6453
|
516
|
9476
|
517 PROLOG(yuv2rgb_c_4b_ordered_dither, uint8_t)
|
|
518 const uint8_t *d64= dither_8x8_73[y&7];
|
|
519 const uint8_t *d128=dither_8x8_220[y&7];
|
6453
|
520
|
9171
|
521 #define DST1bpp4b(i,o) \
|
6453
|
522 Y = py_1[2*i]; \
|
|
523 dst_1[2*i] = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]]; \
|
|
524 Y = py_1[2*i+1]; \
|
|
525 dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
|
|
526
|
9171
|
527 #define DST2bpp4b(i,o) \
|
6453
|
528 Y = py_2[2*i]; \
|
|
529 dst_2[2*i] = r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]]; \
|
|
530 Y = py_2[2*i+1]; \
|
|
531 dst_2[2*i+1] = r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
|
|
532
|
|
533
|
|
534 RGB(0);
|
9171
|
535 DST1bpp4b(0,0);
|
|
536 DST2bpp4b(0,0);
|
6453
|
537
|
|
538 RGB(1);
|
9171
|
539 DST2bpp4b(1,2);
|
|
540 DST1bpp4b(1,2);
|
6453
|
541
|
|
542 RGB(2);
|
9171
|
543 DST1bpp4b(2,4);
|
|
544 DST2bpp4b(2,4);
|
6453
|
545
|
|
546 RGB(3);
|
9171
|
547 DST2bpp4b(3,6);
|
|
548 DST1bpp4b(3,6);
|
9476
|
549 EPILOG(8)
|
6453
|
550
|
9476
|
551 PROLOG(yuv2rgb_c_1_ordered_dither, uint8_t)
|
|
552 const uint8_t *d128=dither_8x8_220[y&7];
|
6456
|
553 char out_1=0, out_2=0;
|
9477
|
554 g= c->table_gU[128] + c->table_gV[128];
|
6456
|
555
|
|
556 #define DST1bpp1(i,o) \
|
|
557 Y = py_1[2*i]; \
|
|
558 out_1+= out_1 + g[Y+d128[0+o]]; \
|
|
559 Y = py_1[2*i+1]; \
|
|
560 out_1+= out_1 + g[Y+d128[1+o]];
|
|
561
|
|
562 #define DST2bpp1(i,o) \
|
|
563 Y = py_2[2*i]; \
|
|
564 out_2+= out_2 + g[Y+d128[8+o]]; \
|
|
565 Y = py_2[2*i+1]; \
|
|
566 out_2+= out_2 + g[Y+d128[9+o]];
|
|
567
|
|
568 DST1bpp1(0,0);
|
|
569 DST2bpp1(0,0);
|
|
570
|
|
571 DST2bpp1(1,2);
|
|
572 DST1bpp1(1,2);
|
|
573
|
|
574 DST1bpp1(2,4);
|
|
575 DST2bpp1(2,4);
|
|
576
|
|
577 DST2bpp1(3,6);
|
|
578 DST1bpp1(3,6);
|
|
579
|
|
580 dst_1[0]= out_1;
|
|
581 dst_2[0]= out_2;
|
9476
|
582 EPILOG(1)
|
6456
|
583
|
9476
|
584 SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
|
|
585 {
|
|
586 #ifdef ARCH_X86
|
9987
|
587 if(c->flags & SWS_CPU_CAPS_MMX2){
|
9476
|
588 switch(c->dstFormat){
|
|
589 case IMGFMT_BGR32: return yuv420_rgb32_MMX2;
|
|
590 case IMGFMT_BGR24: return yuv420_rgb24_MMX2;
|
|
591 case IMGFMT_BGR16: return yuv420_rgb16_MMX2;
|
|
592 case IMGFMT_BGR15: return yuv420_rgb15_MMX2;
|
|
593 }
|
|
594 }
|
9987
|
595 if(c->flags & SWS_CPU_CAPS_MMX){
|
9476
|
596 switch(c->dstFormat){
|
|
597 case IMGFMT_BGR32: return yuv420_rgb32_MMX;
|
|
598 case IMGFMT_BGR24: return yuv420_rgb24_MMX;
|
|
599 case IMGFMT_BGR16: return yuv420_rgb16_MMX;
|
|
600 case IMGFMT_BGR15: return yuv420_rgb15_MMX;
|
|
601 }
|
|
602 }
|
|
603 #endif
|
|
604 #ifdef HAVE_MLIB
|
|
605 {
|
|
606 SwsFunc t= yuv2rgb_init_mlib(c);
|
|
607 if(t) return t;
|
6456
|
608 }
|
9476
|
609 #endif
|
10046
|
610 MSG_WARN("No accelerated colorspace conversion found\n");
|
9476
|
611
|
|
612 switch(c->dstFormat){
|
|
613 case IMGFMT_RGB32:
|
|
614 case IMGFMT_BGR32: return yuv2rgb_c_32;
|
|
615 case IMGFMT_RGB24: return yuv2rgb_c_24_rgb;
|
|
616 case IMGFMT_BGR24: return yuv2rgb_c_24_bgr;
|
|
617 case IMGFMT_RGB16:
|
|
618 case IMGFMT_BGR16:
|
|
619 case IMGFMT_RGB15:
|
|
620 case IMGFMT_BGR15: return yuv2rgb_c_16;
|
|
621 case IMGFMT_RGB8:
|
|
622 case IMGFMT_BGR8: return yuv2rgb_c_8_ordered_dither;
|
|
623 case IMGFMT_RGB4:
|
|
624 case IMGFMT_BGR4: return yuv2rgb_c_4_ordered_dither;
|
|
625 case IMGFMT_RG4B:
|
|
626 case IMGFMT_BG4B: return yuv2rgb_c_4b_ordered_dither;
|
|
627 case IMGFMT_RGB1:
|
|
628 case IMGFMT_BGR1: return yuv2rgb_c_1_ordered_dither;
|
|
629 default:
|
|
630 assert(0);
|
|
631 }
|
|
632 return NULL;
|
6456
|
633 }
|
|
634
|
2732
|
635 static int div_round (int dividend, int divisor)
|
|
636 {
|
|
637 if (dividend > 0)
|
|
638 return (dividend + (divisor>>1)) / divisor;
|
|
639 else
|
|
640 return -((-dividend + (divisor>>1)) / divisor);
|
|
641 }
|
|
642
|
9476
|
643 int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange, int brightness, int contrast, int saturation)
|
2732
|
644 {
|
11691
|
645 const int isRgb = IMGFMT_IS_RGB(c->dstFormat);
|
|
646 const int bpp = isRgb?IMGFMT_RGB_DEPTH(c->dstFormat):IMGFMT_BGR_DEPTH(c->dstFormat);
|
2732
|
647 int i;
|
|
648 uint8_t table_Y[1024];
|
|
649 uint32_t *table_32 = 0;
|
|
650 uint16_t *table_16 = 0;
|
|
651 uint8_t *table_8 = 0;
|
6451
|
652 uint8_t *table_332 = 0;
|
6453
|
653 uint8_t *table_121 = 0;
|
|
654 uint8_t *table_1 = 0;
|
6140
|
655 int entry_size = 0;
|
2732
|
656 void *table_r = 0, *table_g = 0, *table_b = 0;
|
6577
|
657 void *table_start;
|
2732
|
658
|
9476
|
659 int64_t crv = inv_table[0];
|
|
660 int64_t cbu = inv_table[1];
|
|
661 int64_t cgu = -inv_table[2];
|
|
662 int64_t cgv = -inv_table[3];
|
|
663 int64_t cy = 1<<16;
|
|
664 int64_t oy = 0;
|
|
665
|
|
666 //printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
|
|
667 if(!fullRange){
|
|
668 cy= (cy*255) / 219;
|
|
669 oy= 16<<16;
|
|
670 }
|
|
671
|
|
672 cy = (cy *contrast )>>16;
|
|
673 crv= (crv*contrast * saturation)>>32;
|
|
674 cbu= (cbu*contrast * saturation)>>32;
|
|
675 cgu= (cgu*contrast * saturation)>>32;
|
|
676 cgv= (cgv*contrast * saturation)>>32;
|
|
677 //printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
|
|
678 oy -= 256*brightness;
|
2732
|
679
|
|
680 for (i = 0; i < 1024; i++) {
|
|
681 int j;
|
|
682
|
9476
|
683 j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
|
2732
|
684 j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
|
|
685 table_Y[i] = j;
|
|
686 }
|
|
687
|
|
688 switch (bpp) {
|
|
689 case 32:
|
6577
|
690 table_start= table_32 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
|
2732
|
691
|
|
692 entry_size = sizeof (uint32_t);
|
|
693 table_r = table_32 + 197;
|
|
694 table_b = table_32 + 197 + 685;
|
|
695 table_g = table_32 + 197 + 2*682;
|
|
696
|
|
697 for (i = -197; i < 256+197; i++)
|
9476
|
698 ((uint32_t *)table_r)[i] = table_Y[i+384] << (isRgb ? 16 : 0);
|
2732
|
699 for (i = -132; i < 256+132; i++)
|
|
700 ((uint32_t *)table_g)[i] = table_Y[i+384] << 8;
|
|
701 for (i = -232; i < 256+232; i++)
|
9476
|
702 ((uint32_t *)table_b)[i] = table_Y[i+384] << (isRgb ? 0 : 16);
|
2732
|
703 break;
|
|
704
|
|
705 case 24:
|
6577
|
706 table_start= table_8 = malloc ((256 + 2*232) * sizeof (uint8_t));
|
2732
|
707
|
|
708 entry_size = sizeof (uint8_t);
|
|
709 table_r = table_g = table_b = table_8 + 232;
|
|
710
|
|
711 for (i = -232; i < 256+232; i++)
|
|
712 ((uint8_t * )table_b)[i] = table_Y[i+384];
|
|
713 break;
|
|
714
|
|
715 case 15:
|
|
716 case 16:
|
6577
|
717 table_start= table_16 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
|
2732
|
718
|
|
719 entry_size = sizeof (uint16_t);
|
|
720 table_r = table_16 + 197;
|
|
721 table_b = table_16 + 197 + 685;
|
|
722 table_g = table_16 + 197 + 2*682;
|
|
723
|
|
724 for (i = -197; i < 256+197; i++) {
|
|
725 int j = table_Y[i+384] >> 3;
|
|
726
|
9476
|
727 if (isRgb)
|
2732
|
728 j <<= ((bpp==16) ? 11 : 10);
|
|
729
|
|
730 ((uint16_t *)table_r)[i] = j;
|
|
731 }
|
|
732 for (i = -132; i < 256+132; i++) {
|
|
733 int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
|
|
734
|
|
735 ((uint16_t *)table_g)[i] = j << 5;
|
|
736 }
|
|
737 for (i = -232; i < 256+232; i++) {
|
|
738 int j = table_Y[i+384] >> 3;
|
|
739
|
9476
|
740 if (!isRgb)
|
2732
|
741 j <<= ((bpp==16) ? 11 : 10);
|
|
742
|
|
743 ((uint16_t *)table_b)[i] = j;
|
|
744 }
|
|
745 break;
|
|
746
|
6451
|
747 case 8:
|
6577
|
748 table_start= table_332 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
|
6451
|
749
|
|
750 entry_size = sizeof (uint8_t);
|
|
751 table_r = table_332 + 197;
|
|
752 table_b = table_332 + 197 + 685;
|
|
753 table_g = table_332 + 197 + 2*682;
|
|
754
|
|
755 for (i = -197; i < 256+197; i++) {
|
6577
|
756 int j = (table_Y[i+384 - 16] + 18)/36;
|
6451
|
757
|
9476
|
758 if (isRgb)
|
6451
|
759 j <<= 5;
|
|
760
|
|
761 ((uint8_t *)table_r)[i] = j;
|
|
762 }
|
|
763 for (i = -132; i < 256+132; i++) {
|
6577
|
764 int j = (table_Y[i+384 - 16] + 18)/36;
|
6451
|
765
|
9476
|
766 if (!isRgb)
|
6451
|
767 j <<= 1;
|
|
768
|
|
769 ((uint8_t *)table_g)[i] = j << 2;
|
|
770 }
|
|
771 for (i = -232; i < 256+232; i++) {
|
6577
|
772 int j = (table_Y[i+384 - 37] + 43)/85;
|
6451
|
773
|
9476
|
774 if (!isRgb)
|
6451
|
775 j <<= 6;
|
|
776
|
|
777 ((uint8_t *)table_b)[i] = j;
|
|
778 }
|
|
779 break;
|
6453
|
780 case 4:
|
9171
|
781 case 4|128:
|
6577
|
782 table_start= table_121 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
|
6453
|
783
|
|
784 entry_size = sizeof (uint8_t);
|
|
785 table_r = table_121 + 197;
|
|
786 table_b = table_121 + 197 + 685;
|
|
787 table_g = table_121 + 197 + 2*682;
|
|
788
|
|
789 for (i = -197; i < 256+197; i++) {
|
6577
|
790 int j = table_Y[i+384 - 110] >> 7;
|
6453
|
791
|
9476
|
792 if (isRgb)
|
6453
|
793 j <<= 3;
|
|
794
|
|
795 ((uint8_t *)table_r)[i] = j;
|
|
796 }
|
|
797 for (i = -132; i < 256+132; i++) {
|
6577
|
798 int j = (table_Y[i+384 - 37]+ 43)/85;
|
6453
|
799
|
|
800 ((uint8_t *)table_g)[i] = j << 1;
|
|
801 }
|
|
802 for (i = -232; i < 256+232; i++) {
|
6577
|
803 int j =table_Y[i+384 - 110] >> 7;
|
6453
|
804
|
9476
|
805 if (!isRgb)
|
6453
|
806 j <<= 3;
|
|
807
|
|
808 ((uint8_t *)table_b)[i] = j;
|
|
809 }
|
|
810 break;
|
6451
|
811
|
6456
|
812 case 1:
|
6577
|
813 table_start= table_1 = malloc (256*2 * sizeof (uint8_t));
|
6456
|
814
|
|
815 entry_size = sizeof (uint8_t);
|
6577
|
816 table_g = table_1;
|
6456
|
817 table_r = table_b = NULL;
|
|
818
|
6577
|
819 for (i = 0; i < 256+256; i++) {
|
|
820 int j = table_Y[i + 384 - 110]>>7;
|
6456
|
821
|
|
822 ((uint8_t *)table_g)[i] = j;
|
|
823 }
|
|
824 break;
|
|
825
|
2732
|
826 default:
|
6577
|
827 table_start= NULL;
|
10046
|
828 MSG_ERR("%ibpp not supported by yuv2rgb\n", bpp);
|
9476
|
829 //free mem?
|
|
830 return -1;
|
2732
|
831 }
|
|
832
|
|
833 for (i = 0; i < 256; i++) {
|
9476
|
834 c->table_rV[i] = table_r + entry_size * div_round (crv * (i-128), 76309);
|
|
835 c->table_gU[i] = table_g + entry_size * div_round (cgu * (i-128), 76309);
|
|
836 c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
|
|
837 c->table_bU[i] = table_b + entry_size * div_round (cbu * (i-128), 76309);
|
2732
|
838 }
|
6577
|
839
|
9476
|
840 if(c->yuvTable) free(c->yuvTable);
|
|
841 c->yuvTable= table_start;
|
|
842 return 0;
|
2732
|
843 }
|