Mercurial > mplayer.hg
comparison postproc/yuv2rgb.c @ 2732:ae79207a3055
Move yuv2rgb to postprocess
author | nick |
---|---|
date | Tue, 06 Nov 2001 11:22:40 +0000 |
parents | |
children | 86910f54c391 |
comparison
equal
deleted
inserted
replaced
2731:214f79969a80 | 2732:ae79207a3055 |
---|---|
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 Håkan 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 * | |
28 */ | |
29 | |
30 #include <stdio.h> | |
31 #include <stdlib.h> | |
32 #include <inttypes.h> | |
33 | |
34 #include "config.h" | |
35 //#include "video_out.h" | |
36 #include "rgb2rgb.h" | |
37 | |
38 #ifdef HAVE_MLIB | |
39 #include "yuv2rgb_mlib.c" | |
40 #endif | |
41 | |
42 extern yuv2rgb_fun yuv2rgb_init_mmx (int bpp, int mode); | |
43 | |
44 | |
45 uint32_t matrix_coefficients = 6; | |
46 | |
47 const int32_t Inverse_Table_6_9[8][4] = { | |
48 {117504, 138453, 13954, 34903}, /* no sequence_display_extension */ | |
49 {117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */ | |
50 {104597, 132201, 25675, 53279}, /* unspecified */ | |
51 {104597, 132201, 25675, 53279}, /* reserved */ | |
52 {104448, 132798, 24759, 53109}, /* FCC */ | |
53 {104597, 132201, 25675, 53279}, /* ITU-R Rec. 624-4 System B, G */ | |
54 {104597, 132201, 25675, 53279}, /* SMPTE 170M */ | |
55 {117579, 136230, 16907, 35559} /* SMPTE 240M (1987) */ | |
56 }; | |
57 | |
58 static void yuv2rgb_c_init (int bpp, int mode); | |
59 | |
60 yuv2rgb_fun yuv2rgb; | |
61 | |
62 static void (* yuv2rgb_c_internal) (uint8_t *, uint8_t *, | |
63 uint8_t *, uint8_t *, | |
64 void *, void *, int); | |
65 | |
66 static void yuv2rgb_c (void * dst, uint8_t * py, | |
67 uint8_t * pu, uint8_t * pv, | |
68 int h_size, int v_size, | |
69 int rgb_stride, int y_stride, int uv_stride) | |
70 { | |
71 v_size >>= 1; | |
72 | |
73 while (v_size--) { | |
74 yuv2rgb_c_internal (py, py + y_stride, pu, pv, dst, dst + rgb_stride, | |
75 h_size); | |
76 | |
77 py += 2 * y_stride; | |
78 pu += uv_stride; | |
79 pv += uv_stride; | |
80 dst += 2 * rgb_stride; | |
81 } | |
82 } | |
83 | |
84 void yuv2rgb_init (int bpp, int mode) | |
85 { | |
86 yuv2rgb = NULL; | |
87 #ifdef HAVE_MMX | |
88 if (yuv2rgb == NULL /*&& (config.flags & VO_MMX_ENABLE)*/) { | |
89 yuv2rgb = yuv2rgb_init_mmx (bpp, mode); | |
90 if (yuv2rgb != NULL) | |
91 printf ("Using MMX for colorspace transform\n"); | |
92 else | |
93 printf ("Cannot init MMX colorspace transform\n"); | |
94 } | |
95 #endif | |
96 #ifdef HAVE_MLIB | |
97 if (yuv2rgb == NULL /*&& (config.flags & VO_MLIB_ENABLE)*/) { | |
98 yuv2rgb = yuv2rgb_init_mlib (bpp, mode); | |
99 if (yuv2rgb != NULL) | |
100 printf ("Using mlib for colorspace transform\n"); | |
101 } | |
102 #endif | |
103 if (yuv2rgb == NULL) { | |
104 printf ("No accelerated colorspace conversion found\n"); | |
105 yuv2rgb_c_init (bpp, mode); | |
106 yuv2rgb = (yuv2rgb_fun)yuv2rgb_c; | |
107 } | |
108 } | |
109 | |
110 void * table_rV[256]; | |
111 void * table_gU[256]; | |
112 int table_gV[256]; | |
113 void * table_bU[256]; | |
114 | |
115 #define RGB(i) \ | |
116 U = pu[i]; \ | |
117 V = pv[i]; \ | |
118 r = table_rV[V]; \ | |
119 g = table_gU[U] + table_gV[V]; \ | |
120 b = table_bU[U]; | |
121 | |
122 #define DST1(i) \ | |
123 Y = py_1[2*i]; \ | |
124 dst_1[2*i] = r[Y] + g[Y] + b[Y]; \ | |
125 Y = py_1[2*i+1]; \ | |
126 dst_1[2*i+1] = r[Y] + g[Y] + b[Y]; | |
127 | |
128 #define DST2(i) \ | |
129 Y = py_2[2*i]; \ | |
130 dst_2[2*i] = r[Y] + g[Y] + b[Y]; \ | |
131 Y = py_2[2*i+1]; \ | |
132 dst_2[2*i+1] = r[Y] + g[Y] + b[Y]; | |
133 | |
134 #define DST1RGB(i) \ | |
135 Y = py_1[2*i]; \ | |
136 dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y]; \ | |
137 Y = py_1[2*i+1]; \ | |
138 dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y]; | |
139 | |
140 #define DST2RGB(i) \ | |
141 Y = py_2[2*i]; \ | |
142 dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y]; \ | |
143 Y = py_2[2*i+1]; \ | |
144 dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y]; | |
145 | |
146 #define DST1BGR(i) \ | |
147 Y = py_1[2*i]; \ | |
148 dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y]; \ | |
149 Y = py_1[2*i+1]; \ | |
150 dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y]; | |
151 | |
152 #define DST2BGR(i) \ | |
153 Y = py_2[2*i]; \ | |
154 dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y]; \ | |
155 Y = py_2[2*i+1]; \ | |
156 dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y]; | |
157 | |
158 static void yuv2rgb_c_32 (uint8_t * py_1, uint8_t * py_2, | |
159 uint8_t * pu, uint8_t * pv, | |
160 void * _dst_1, void * _dst_2, int h_size) | |
161 { | |
162 int U, V, Y; | |
163 uint32_t * r, * g, * b; | |
164 uint32_t * dst_1, * dst_2; | |
165 | |
166 h_size >>= 3; | |
167 dst_1 = _dst_1; | |
168 dst_2 = _dst_2; | |
169 | |
170 while (h_size--) { | |
171 RGB(0); | |
172 DST1(0); | |
173 DST2(0); | |
174 | |
175 RGB(1); | |
176 DST2(1); | |
177 DST1(1); | |
178 | |
179 RGB(2); | |
180 DST1(2); | |
181 DST2(2); | |
182 | |
183 RGB(3); | |
184 DST2(3); | |
185 DST1(3); | |
186 | |
187 pu += 4; | |
188 pv += 4; | |
189 py_1 += 8; | |
190 py_2 += 8; | |
191 dst_1 += 8; | |
192 dst_2 += 8; | |
193 } | |
194 } | |
195 | |
196 // This is very near from the yuv2rgb_c_32 code | |
197 static void yuv2rgb_c_24_rgb (uint8_t * py_1, uint8_t * py_2, | |
198 uint8_t * pu, uint8_t * pv, | |
199 void * _dst_1, void * _dst_2, int h_size) | |
200 { | |
201 int U, V, Y; | |
202 uint8_t * r, * g, * b; | |
203 uint8_t * dst_1, * dst_2; | |
204 | |
205 h_size >>= 3; | |
206 dst_1 = _dst_1; | |
207 dst_2 = _dst_2; | |
208 | |
209 while (h_size--) { | |
210 RGB(0); | |
211 DST1RGB(0); | |
212 DST2RGB(0); | |
213 | |
214 RGB(1); | |
215 DST2RGB(1); | |
216 DST1RGB(1); | |
217 | |
218 RGB(2); | |
219 DST1RGB(2); | |
220 DST2RGB(2); | |
221 | |
222 RGB(3); | |
223 DST2RGB(3); | |
224 DST1RGB(3); | |
225 | |
226 pu += 4; | |
227 pv += 4; | |
228 py_1 += 8; | |
229 py_2 += 8; | |
230 dst_1 += 24; | |
231 dst_2 += 24; | |
232 } | |
233 } | |
234 | |
235 // only trivial mods from yuv2rgb_c_24_rgb | |
236 static void yuv2rgb_c_24_bgr (uint8_t * py_1, uint8_t * py_2, | |
237 uint8_t * pu, uint8_t * pv, | |
238 void * _dst_1, void * _dst_2, int h_size) | |
239 { | |
240 int U, V, Y; | |
241 uint8_t * r, * g, * b; | |
242 uint8_t * dst_1, * dst_2; | |
243 | |
244 h_size >>= 3; | |
245 dst_1 = _dst_1; | |
246 dst_2 = _dst_2; | |
247 | |
248 while (h_size--) { | |
249 RGB(0); | |
250 DST1BGR(0); | |
251 DST2BGR(0); | |
252 | |
253 RGB(1); | |
254 DST2BGR(1); | |
255 DST1BGR(1); | |
256 | |
257 RGB(2); | |
258 DST1BGR(2); | |
259 DST2BGR(2); | |
260 | |
261 RGB(3); | |
262 DST2BGR(3); | |
263 DST1BGR(3); | |
264 | |
265 pu += 4; | |
266 pv += 4; | |
267 py_1 += 8; | |
268 py_2 += 8; | |
269 dst_1 += 24; | |
270 dst_2 += 24; | |
271 } | |
272 } | |
273 | |
274 // This is exactly the same code as yuv2rgb_c_32 except for the types of | |
275 // r, g, b, dst_1, dst_2 | |
276 static void yuv2rgb_c_16 (uint8_t * py_1, uint8_t * py_2, | |
277 uint8_t * pu, uint8_t * pv, | |
278 void * _dst_1, void * _dst_2, int h_size) | |
279 { | |
280 int U, V, Y; | |
281 uint16_t * r, * g, * b; | |
282 uint16_t * dst_1, * dst_2; | |
283 | |
284 h_size >>= 3; | |
285 dst_1 = _dst_1; | |
286 dst_2 = _dst_2; | |
287 | |
288 while (h_size--) { | |
289 RGB(0); | |
290 DST1(0); | |
291 DST2(0); | |
292 | |
293 RGB(1); | |
294 DST2(1); | |
295 DST1(1); | |
296 | |
297 RGB(2); | |
298 DST1(2); | |
299 DST2(2); | |
300 | |
301 RGB(3); | |
302 DST2(3); | |
303 DST1(3); | |
304 | |
305 pu += 4; | |
306 pv += 4; | |
307 py_1 += 8; | |
308 py_2 += 8; | |
309 dst_1 += 8; | |
310 dst_2 += 8; | |
311 } | |
312 } | |
313 | |
314 static int div_round (int dividend, int divisor) | |
315 { | |
316 if (dividend > 0) | |
317 return (dividend + (divisor>>1)) / divisor; | |
318 else | |
319 return -((-dividend + (divisor>>1)) / divisor); | |
320 } | |
321 | |
322 static void yuv2rgb_c_init (int bpp, int mode) | |
323 { | |
324 int i; | |
325 uint8_t table_Y[1024]; | |
326 uint32_t *table_32 = 0; | |
327 uint16_t *table_16 = 0; | |
328 uint8_t *table_8 = 0; | |
329 uint32_t entry_size = 0; | |
330 void *table_r = 0, *table_g = 0, *table_b = 0; | |
331 | |
332 int crv = Inverse_Table_6_9[matrix_coefficients][0]; | |
333 int cbu = Inverse_Table_6_9[matrix_coefficients][1]; | |
334 int cgu = -Inverse_Table_6_9[matrix_coefficients][2]; | |
335 int cgv = -Inverse_Table_6_9[matrix_coefficients][3]; | |
336 | |
337 for (i = 0; i < 1024; i++) { | |
338 int j; | |
339 | |
340 j = (76309 * (i - 384 - 16) + 32768) >> 16; | |
341 j = (j < 0) ? 0 : ((j > 255) ? 255 : j); | |
342 table_Y[i] = j; | |
343 } | |
344 | |
345 switch (bpp) { | |
346 case 32: | |
347 yuv2rgb_c_internal = yuv2rgb_c_32; | |
348 | |
349 table_32 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t)); | |
350 | |
351 entry_size = sizeof (uint32_t); | |
352 table_r = table_32 + 197; | |
353 table_b = table_32 + 197 + 685; | |
354 table_g = table_32 + 197 + 2*682; | |
355 | |
356 for (i = -197; i < 256+197; i++) | |
357 ((uint32_t *)table_r)[i] = table_Y[i+384] << ((mode==MODE_RGB) ? 16 : 0); | |
358 for (i = -132; i < 256+132; i++) | |
359 ((uint32_t *)table_g)[i] = table_Y[i+384] << 8; | |
360 for (i = -232; i < 256+232; i++) | |
361 ((uint32_t *)table_b)[i] = table_Y[i+384] << ((mode==MODE_RGB) ? 0 : 16); | |
362 break; | |
363 | |
364 case 24: | |
365 // yuv2rgb_c_internal = (mode==MODE_RGB) ? yuv2rgb_c_24_rgb : yuv2rgb_c_24_bgr; | |
366 yuv2rgb_c_internal = (mode!=MODE_RGB) ? yuv2rgb_c_24_rgb : yuv2rgb_c_24_bgr; | |
367 | |
368 table_8 = malloc ((256 + 2*232) * sizeof (uint8_t)); | |
369 | |
370 entry_size = sizeof (uint8_t); | |
371 table_r = table_g = table_b = table_8 + 232; | |
372 | |
373 for (i = -232; i < 256+232; i++) | |
374 ((uint8_t * )table_b)[i] = table_Y[i+384]; | |
375 break; | |
376 | |
377 case 15: | |
378 case 16: | |
379 yuv2rgb_c_internal = yuv2rgb_c_16; | |
380 | |
381 table_16 = malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t)); | |
382 | |
383 entry_size = sizeof (uint16_t); | |
384 table_r = table_16 + 197; | |
385 table_b = table_16 + 197 + 685; | |
386 table_g = table_16 + 197 + 2*682; | |
387 | |
388 for (i = -197; i < 256+197; i++) { | |
389 int j = table_Y[i+384] >> 3; | |
390 | |
391 if (mode == MODE_RGB) | |
392 j <<= ((bpp==16) ? 11 : 10); | |
393 | |
394 ((uint16_t *)table_r)[i] = j; | |
395 } | |
396 for (i = -132; i < 256+132; i++) { | |
397 int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3); | |
398 | |
399 ((uint16_t *)table_g)[i] = j << 5; | |
400 } | |
401 for (i = -232; i < 256+232; i++) { | |
402 int j = table_Y[i+384] >> 3; | |
403 | |
404 if (mode == MODE_BGR) | |
405 j <<= ((bpp==16) ? 11 : 10); | |
406 | |
407 ((uint16_t *)table_b)[i] = j; | |
408 } | |
409 break; | |
410 | |
411 default: | |
412 printf ("%ibpp not supported by yuv2rgb\n", bpp); | |
413 //exit (1); | |
414 } | |
415 | |
416 for (i = 0; i < 256; i++) { | |
417 table_rV[i] = table_r + entry_size * div_round (crv * (i-128), 76309); | |
418 table_gU[i] = table_g + entry_size * div_round (cgu * (i-128), 76309); | |
419 table_gV[i] = entry_size * div_round (cgv * (i-128), 76309); | |
420 table_bU[i] = table_b + entry_size * div_round (cbu * (i-128), 76309); | |
421 } | |
422 } |