Mercurial > mplayer.hg
comparison libswscale/swscale.c @ 23155:1befc8d767cd
cosmetics attack, part II: Remove all tabs and prettyprint/reindent the code.
author | diego |
---|---|
date | Sun, 29 Apr 2007 13:39:27 +0000 |
parents | 9528d1ebe68f |
children | 6431b46104b7 |
comparison
equal
deleted
inserted
replaced
23154:e564b9cd7290 | 23155:1befc8d767cd |
---|---|
100 #define PI M_PI | 100 #define PI M_PI |
101 #else | 101 #else |
102 #define PI 3.14159265358979323846 | 102 #define PI 3.14159265358979323846 |
103 #endif | 103 #endif |
104 | 104 |
105 #define isSupportedIn(x) ((x)==PIX_FMT_YUV420P || (x)==PIX_FMT_YUYV422 || (x)==PIX_FMT_UYVY422\ | 105 #define isSupportedIn(x) ((x)==PIX_FMT_YUV420P || (x)==PIX_FMT_YUYV422 || (x)==PIX_FMT_UYVY422 \ |
106 || (x)==PIX_FMT_RGB32|| (x)==PIX_FMT_BGR24|| (x)==PIX_FMT_BGR565|| (x)==PIX_FMT_BGR555\ | 106 || (x)==PIX_FMT_RGB32 || (x)==PIX_FMT_BGR24 || (x)==PIX_FMT_BGR565 || (x)==PIX_FMT_BGR555 \ |
107 || (x)==PIX_FMT_BGR32|| (x)==PIX_FMT_RGB24|| (x)==PIX_FMT_RGB565|| (x)==PIX_FMT_RGB555\ | 107 || (x)==PIX_FMT_BGR32 || (x)==PIX_FMT_RGB24 || (x)==PIX_FMT_RGB565 || (x)==PIX_FMT_RGB555 \ |
108 || (x)==PIX_FMT_GRAY8 || (x)==PIX_FMT_YUV410P\ | 108 || (x)==PIX_FMT_GRAY8 || (x)==PIX_FMT_YUV410P \ |
109 || (x)==PIX_FMT_GRAY16BE || (x)==PIX_FMT_GRAY16LE\ | 109 || (x)==PIX_FMT_GRAY16BE || (x)==PIX_FMT_GRAY16LE \ |
110 || (x)==PIX_FMT_YUV444P || (x)==PIX_FMT_YUV422P || (x)==PIX_FMT_YUV411P\ | 110 || (x)==PIX_FMT_YUV444P || (x)==PIX_FMT_YUV422P || (x)==PIX_FMT_YUV411P \ |
111 || (x)==PIX_FMT_PAL8 || (x)==PIX_FMT_BGR8 || (x)==PIX_FMT_RGB8\ | 111 || (x)==PIX_FMT_PAL8 || (x)==PIX_FMT_BGR8 || (x)==PIX_FMT_RGB8 \ |
112 || (x)==PIX_FMT_BGR4_BYTE || (x)==PIX_FMT_RGB4_BYTE) | 112 || (x)==PIX_FMT_BGR4_BYTE || (x)==PIX_FMT_RGB4_BYTE) |
113 #define isSupportedOut(x) ((x)==PIX_FMT_YUV420P || (x)==PIX_FMT_YUYV422 || (x)==PIX_FMT_UYVY422\ | 113 #define isSupportedOut(x) ((x)==PIX_FMT_YUV420P || (x)==PIX_FMT_YUYV422 || (x)==PIX_FMT_UYVY422 \ |
114 || (x)==PIX_FMT_YUV444P || (x)==PIX_FMT_YUV422P || (x)==PIX_FMT_YUV411P\ | 114 || (x)==PIX_FMT_YUV444P || (x)==PIX_FMT_YUV422P || (x)==PIX_FMT_YUV411P \ |
115 || isRGB(x) || isBGR(x)\ | 115 || isRGB(x) || isBGR(x) \ |
116 || (x)==PIX_FMT_NV12 || (x)==PIX_FMT_NV21\ | 116 || (x)==PIX_FMT_NV12 || (x)==PIX_FMT_NV21 \ |
117 || (x)==PIX_FMT_GRAY16BE || (x)==PIX_FMT_GRAY16LE\ | 117 || (x)==PIX_FMT_GRAY16BE || (x)==PIX_FMT_GRAY16LE \ |
118 || (x)==PIX_FMT_GRAY8 || (x)==PIX_FMT_YUV410P) | 118 || (x)==PIX_FMT_GRAY8 || (x)==PIX_FMT_YUV410P) |
119 #define isPacked(x) ((x)==PIX_FMT_PAL8 || (x)==PIX_FMT_YUYV422 ||\ | 119 #define isPacked(x) ((x)==PIX_FMT_PAL8 || (x)==PIX_FMT_YUYV422 || \ |
120 (x)==PIX_FMT_UYVY422 || isRGB(x) || isBGR(x)) | 120 (x)==PIX_FMT_UYVY422 || isRGB(x) || isBGR(x)) |
121 | 121 |
122 #define RGB2YUV_SHIFT 16 | 122 #define RGB2YUV_SHIFT 16 |
123 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5)) | 123 #define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5)) |
124 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5)) | 124 #define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5)) |
148 */ | 148 */ |
149 | 149 |
150 #if defined(ARCH_X86) && defined (CONFIG_GPL) | 150 #if defined(ARCH_X86) && defined (CONFIG_GPL) |
151 static uint64_t attribute_used __attribute__((aligned(8))) bF8= 0xF8F8F8F8F8F8F8F8LL; | 151 static uint64_t attribute_used __attribute__((aligned(8))) bF8= 0xF8F8F8F8F8F8F8F8LL; |
152 static uint64_t attribute_used __attribute__((aligned(8))) bFC= 0xFCFCFCFCFCFCFCFCLL; | 152 static uint64_t attribute_used __attribute__((aligned(8))) bFC= 0xFCFCFCFCFCFCFCFCLL; |
153 static uint64_t __attribute__((aligned(8))) w10= 0x0010001000100010LL; | 153 static uint64_t __attribute__((aligned(8))) w10= 0x0010001000100010LL; |
154 static uint64_t attribute_used __attribute__((aligned(8))) w02= 0x0002000200020002LL; | 154 static uint64_t attribute_used __attribute__((aligned(8))) w02= 0x0002000200020002LL; |
155 static uint64_t attribute_used __attribute__((aligned(8))) bm00001111=0x00000000FFFFFFFFLL; | 155 static uint64_t attribute_used __attribute__((aligned(8))) bm00001111=0x00000000FFFFFFFFLL; |
156 static uint64_t attribute_used __attribute__((aligned(8))) bm00000111=0x0000000000FFFFFFLL; | 156 static uint64_t attribute_used __attribute__((aligned(8))) bm00000111=0x0000000000FFFFFFLL; |
157 static uint64_t attribute_used __attribute__((aligned(8))) bm11111000=0xFFFFFFFFFF000000LL; | 157 static uint64_t attribute_used __attribute__((aligned(8))) bm11111000=0xFFFFFFFFFF000000LL; |
158 static uint64_t attribute_used __attribute__((aligned(8))) bm01010101=0x00FF00FF00FF00FFLL; | 158 static uint64_t attribute_used __attribute__((aligned(8))) bm01010101=0x00FF00FF00FF00FFLL; |
161 static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither; | 161 static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither; |
162 static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither; | 162 static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither; |
163 static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither; | 163 static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither; |
164 | 164 |
165 static uint64_t __attribute__((aligned(8))) dither4[2]={ | 165 static uint64_t __attribute__((aligned(8))) dither4[2]={ |
166 0x0103010301030103LL, | 166 0x0103010301030103LL, |
167 0x0200020002000200LL,}; | 167 0x0200020002000200LL,}; |
168 | 168 |
169 static uint64_t __attribute__((aligned(8))) dither8[2]={ | 169 static uint64_t __attribute__((aligned(8))) dither8[2]={ |
170 0x0602060206020602LL, | 170 0x0602060206020602LL, |
171 0x0004000400040004LL,}; | 171 0x0004000400040004LL,}; |
172 | 172 |
173 static uint64_t __attribute__((aligned(8))) b16Mask= 0x001F001F001F001FLL; | 173 static uint64_t __attribute__((aligned(8))) b16Mask= 0x001F001F001F001FLL; |
174 static uint64_t attribute_used __attribute__((aligned(8))) g16Mask= 0x07E007E007E007E0LL; | 174 static uint64_t attribute_used __attribute__((aligned(8))) g16Mask= 0x07E007E007E007E0LL; |
175 static uint64_t attribute_used __attribute__((aligned(8))) r16Mask= 0xF800F800F800F800LL; | 175 static uint64_t attribute_used __attribute__((aligned(8))) r16Mask= 0xF800F800F800F800LL; |
176 static uint64_t __attribute__((aligned(8))) b15Mask= 0x001F001F001F001FLL; | 176 static uint64_t __attribute__((aligned(8))) b15Mask= 0x001F001F001F001FLL; |
177 static uint64_t attribute_used __attribute__((aligned(8))) g15Mask= 0x03E003E003E003E0LL; | 177 static uint64_t attribute_used __attribute__((aligned(8))) g15Mask= 0x03E003E003E003E0LL; |
178 static uint64_t attribute_used __attribute__((aligned(8))) r15Mask= 0x7C007C007C007C00LL; | 178 static uint64_t attribute_used __attribute__((aligned(8))) r15Mask= 0x7C007C007C007C00LL; |
179 | 179 |
180 static uint64_t attribute_used __attribute__((aligned(8))) M24A= 0x00FF0000FF0000FFLL; | 180 static uint64_t attribute_used __attribute__((aligned(8))) M24A= 0x00FF0000FF0000FFLL; |
181 static uint64_t attribute_used __attribute__((aligned(8))) M24B= 0xFF0000FF0000FF00LL; | 181 static uint64_t attribute_used __attribute__((aligned(8))) M24B= 0xFF0000FF0000FF00LL; |
182 static uint64_t attribute_used __attribute__((aligned(8))) M24C= 0x0000FF0000FF0000LL; | 182 static uint64_t attribute_used __attribute__((aligned(8))) M24C= 0x0000FF0000FF0000LL; |
183 | 183 |
184 #ifdef FAST_BGR2YV12 | 184 #ifdef FAST_BGR2YV12 |
185 static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL; | 185 static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000000210041000DULL; |
186 static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL; | 186 static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000FFEEFFDC0038ULL; |
187 static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL; | 187 static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00000038FFD2FFF8ULL; |
188 #else | 188 #else |
189 static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL; | 189 static const uint64_t bgr2YCoeff attribute_used __attribute__((aligned(8))) = 0x000020E540830C8BULL; |
190 static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL; | 190 static const uint64_t bgr2UCoeff attribute_used __attribute__((aligned(8))) = 0x0000ED0FDAC23831ULL; |
191 static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL; | 191 static const uint64_t bgr2VCoeff attribute_used __attribute__((aligned(8))) = 0x00003831D0E6F6EAULL; |
192 #endif /* FAST_BGR2YV12 */ | 192 #endif /* FAST_BGR2YV12 */ |
193 static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL; | 193 static const uint64_t bgr2YOffset attribute_used __attribute__((aligned(8))) = 0x1010101010101010ULL; |
194 static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8)))= 0x8080808080808080ULL; | 194 static const uint64_t bgr2UVOffset attribute_used __attribute__((aligned(8))) = 0x8080808080808080ULL; |
195 static const uint64_t w1111 attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL; | 195 static const uint64_t w1111 attribute_used __attribute__((aligned(8))) = 0x0001000100010001ULL; |
196 #endif /* defined(ARCH_X86) */ | 196 #endif /* defined(ARCH_X86) */ |
197 | 197 |
198 // clipping helper table for C implementations: | 198 // clipping helper table for C implementations: |
199 static unsigned char clip_table[768]; | 199 static unsigned char clip_table[768]; |
200 | 200 |
295 } | 295 } |
296 | 296 |
297 #if defined(ARCH_X86) && defined (CONFIG_GPL) | 297 #if defined(ARCH_X86) && defined (CONFIG_GPL) |
298 void in_asm_used_var_warning_killer() | 298 void in_asm_used_var_warning_killer() |
299 { | 299 { |
300 volatile int i= bF8+bFC+w10+ | 300 volatile int i= bF8+bFC+w10+ |
301 bm00001111+bm00000111+bm11111000+b16Mask+g16Mask+r16Mask+b15Mask+g15Mask+r15Mask+ | 301 bm00001111+bm00000111+bm11111000+b16Mask+g16Mask+r16Mask+b15Mask+g15Mask+r15Mask+ |
302 M24A+M24B+M24C+w02 + b5Dither+g5Dither+r5Dither+g6Dither+dither4[0]+dither8[0]+bm01010101; | 302 M24A+M24B+M24C+w02 + b5Dither+g5Dither+r5Dither+g6Dither+dither4[0]+dither8[0]+bm01010101; |
303 if(i) i=0; | 303 if (i) i=0; |
304 } | 304 } |
305 #endif | 305 #endif |
306 | 306 |
307 static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize, | 307 static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize, |
308 int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize, | 308 int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize, |
309 uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW) | 309 uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW) |
310 { | 310 { |
311 //FIXME Optimize (just quickly writen not opti..) | 311 //FIXME Optimize (just quickly writen not opti..) |
312 int i; | 312 int i; |
313 for(i=0; i<dstW; i++) | 313 for (i=0; i<dstW; i++) |
314 { | 314 { |
315 int val=1<<18; | 315 int val=1<<18; |
316 int j; | 316 int j; |
317 for(j=0; j<lumFilterSize; j++) | 317 for (j=0; j<lumFilterSize; j++) |
318 val += lumSrc[j][i] * lumFilter[j]; | 318 val += lumSrc[j][i] * lumFilter[j]; |
319 | 319 |
320 dest[i]= av_clip_uint8(val>>19); | 320 dest[i]= av_clip_uint8(val>>19); |
321 } | 321 } |
322 | 322 |
323 if(uDest != NULL) | 323 if (uDest != NULL) |
324 for(i=0; i<chrDstW; i++) | 324 for (i=0; i<chrDstW; i++) |
325 { | 325 { |
326 int u=1<<18; | 326 int u=1<<18; |
327 int v=1<<18; | 327 int v=1<<18; |
328 int j; | 328 int j; |
329 for(j=0; j<chrFilterSize; j++) | 329 for (j=0; j<chrFilterSize; j++) |
330 { | 330 { |
331 u += chrSrc[j][i] * chrFilter[j]; | 331 u += chrSrc[j][i] * chrFilter[j]; |
332 v += chrSrc[j][i + 2048] * chrFilter[j]; | 332 v += chrSrc[j][i + 2048] * chrFilter[j]; |
333 } | 333 } |
334 | 334 |
335 uDest[i]= av_clip_uint8(u>>19); | 335 uDest[i]= av_clip_uint8(u>>19); |
336 vDest[i]= av_clip_uint8(v>>19); | 336 vDest[i]= av_clip_uint8(v>>19); |
337 } | 337 } |
338 } | 338 } |
339 | 339 |
340 static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize, | 340 static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize, |
341 int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize, | 341 int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize, |
342 uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat) | 342 uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat) |
343 { | 343 { |
344 //FIXME Optimize (just quickly writen not opti..) | 344 //FIXME Optimize (just quickly writen not opti..) |
345 int i; | 345 int i; |
346 for(i=0; i<dstW; i++) | 346 for (i=0; i<dstW; i++) |
347 { | 347 { |
348 int val=1<<18; | 348 int val=1<<18; |
349 int j; | 349 int j; |
350 for(j=0; j<lumFilterSize; j++) | 350 for (j=0; j<lumFilterSize; j++) |
351 val += lumSrc[j][i] * lumFilter[j]; | 351 val += lumSrc[j][i] * lumFilter[j]; |
352 | 352 |
353 dest[i]= av_clip_uint8(val>>19); | 353 dest[i]= av_clip_uint8(val>>19); |
354 } | 354 } |
355 | 355 |
356 if(uDest == NULL) | 356 if (uDest == NULL) |
357 return; | 357 return; |
358 | 358 |
359 if(dstFormat == PIX_FMT_NV12) | 359 if (dstFormat == PIX_FMT_NV12) |
360 for(i=0; i<chrDstW; i++) | 360 for (i=0; i<chrDstW; i++) |
361 { | 361 { |
362 int u=1<<18; | 362 int u=1<<18; |
363 int v=1<<18; | 363 int v=1<<18; |
364 int j; | 364 int j; |
365 for(j=0; j<chrFilterSize; j++) | 365 for (j=0; j<chrFilterSize; j++) |
366 { | 366 { |
367 u += chrSrc[j][i] * chrFilter[j]; | 367 u += chrSrc[j][i] * chrFilter[j]; |
368 v += chrSrc[j][i + 2048] * chrFilter[j]; | 368 v += chrSrc[j][i + 2048] * chrFilter[j]; |
369 } | 369 } |
370 | 370 |
371 uDest[2*i]= av_clip_uint8(u>>19); | 371 uDest[2*i]= av_clip_uint8(u>>19); |
372 uDest[2*i+1]= av_clip_uint8(v>>19); | 372 uDest[2*i+1]= av_clip_uint8(v>>19); |
373 } | 373 } |
374 else | 374 else |
375 for(i=0; i<chrDstW; i++) | 375 for (i=0; i<chrDstW; i++) |
376 { | 376 { |
377 int u=1<<18; | 377 int u=1<<18; |
378 int v=1<<18; | 378 int v=1<<18; |
379 int j; | 379 int j; |
380 for(j=0; j<chrFilterSize; j++) | 380 for (j=0; j<chrFilterSize; j++) |
381 { | 381 { |
382 u += chrSrc[j][i] * chrFilter[j]; | 382 u += chrSrc[j][i] * chrFilter[j]; |
383 v += chrSrc[j][i + 2048] * chrFilter[j]; | 383 v += chrSrc[j][i + 2048] * chrFilter[j]; |
384 } | 384 } |
385 | 385 |
386 uDest[2*i]= av_clip_uint8(v>>19); | 386 uDest[2*i]= av_clip_uint8(v>>19); |
387 uDest[2*i+1]= av_clip_uint8(u>>19); | 387 uDest[2*i+1]= av_clip_uint8(u>>19); |
388 } | 388 } |
389 } | 389 } |
390 | 390 |
391 #define YSCALE_YUV_2_PACKEDX_C(type) \ | 391 #define YSCALE_YUV_2_PACKEDX_C(type) \ |
392 for(i=0; i<(dstW>>1); i++){\ | 392 for (i=0; i<(dstW>>1); i++){\ |
393 int j;\ | 393 int j;\ |
394 int Y1=1<<18;\ | 394 int Y1 = 1<<18;\ |
395 int Y2=1<<18;\ | 395 int Y2 = 1<<18;\ |
396 int U=1<<18;\ | 396 int U = 1<<18;\ |
397 int V=1<<18;\ | 397 int V = 1<<18;\ |
398 type attribute_unused *r, *b, *g;\ | 398 type attribute_unused *r, *b, *g;\ |
399 const int i2= 2*i;\ | 399 const int i2= 2*i;\ |
400 \ | 400 \ |
401 for(j=0; j<lumFilterSize; j++)\ | 401 for (j=0; j<lumFilterSize; j++)\ |
402 {\ | 402 {\ |
403 Y1 += lumSrc[j][i2] * lumFilter[j];\ | 403 Y1 += lumSrc[j][i2] * lumFilter[j];\ |
404 Y2 += lumSrc[j][i2+1] * lumFilter[j];\ | 404 Y2 += lumSrc[j][i2+1] * lumFilter[j];\ |
405 }\ | 405 }\ |
406 for(j=0; j<chrFilterSize; j++)\ | 406 for (j=0; j<chrFilterSize; j++)\ |
407 {\ | 407 {\ |
408 U += chrSrc[j][i] * chrFilter[j];\ | 408 U += chrSrc[j][i] * chrFilter[j];\ |
409 V += chrSrc[j][i+2048] * chrFilter[j];\ | 409 V += chrSrc[j][i+2048] * chrFilter[j];\ |
410 }\ | 410 }\ |
411 Y1>>=19;\ | 411 Y1>>=19;\ |
412 Y2>>=19;\ | 412 Y2>>=19;\ |
413 U >>=19;\ | 413 U >>=19;\ |
414 V >>=19;\ | 414 V >>=19;\ |
415 if((Y1|Y2|U|V)&256)\ | 415 if ((Y1|Y2|U|V)&256)\ |
416 {\ | 416 {\ |
417 if(Y1>255) Y1=255;\ | 417 if (Y1>255) Y1=255; \ |
418 else if(Y1<0)Y1=0;\ | 418 else if (Y1<0)Y1=0; \ |
419 if(Y2>255) Y2=255;\ | 419 if (Y2>255) Y2=255; \ |
420 else if(Y2<0)Y2=0;\ | 420 else if (Y2<0)Y2=0; \ |
421 if(U>255) U=255;\ | 421 if (U>255) U=255; \ |
422 else if(U<0) U=0;\ | 422 else if (U<0) U=0; \ |
423 if(V>255) V=255;\ | 423 if (V>255) V=255; \ |
424 else if(V<0) V=0;\ | 424 else if (V<0) V=0; \ |
425 } | 425 } |
426 | 426 |
427 #define YSCALE_YUV_2_RGBX_C(type) \ | 427 #define YSCALE_YUV_2_RGBX_C(type) \ |
428 YSCALE_YUV_2_PACKEDX_C(type)\ | 428 YSCALE_YUV_2_PACKEDX_C(type) \ |
429 r = (type *)c->table_rV[V];\ | 429 r = (type *)c->table_rV[V]; \ |
430 g = (type *)(c->table_gU[U] + c->table_gV[V]);\ | 430 g = (type *)(c->table_gU[U] + c->table_gV[V]); \ |
431 b = (type *)c->table_bU[U];\ | 431 b = (type *)c->table_bU[U]; \ |
432 | 432 |
433 #define YSCALE_YUV_2_PACKED2_C \ | 433 #define YSCALE_YUV_2_PACKED2_C \ |
434 for(i=0; i<(dstW>>1); i++){\ | 434 for (i=0; i<(dstW>>1); i++){ \ |
435 const int i2= 2*i;\ | 435 const int i2= 2*i; \ |
436 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>19;\ | 436 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>19; \ |
437 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19;\ | 437 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19; \ |
438 int U= (uvbuf0[i ]*uvalpha1+uvbuf1[i ]*uvalpha)>>19;\ | 438 int U= (uvbuf0[i ]*uvalpha1+uvbuf1[i ]*uvalpha)>>19; \ |
439 int V= (uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19;\ | 439 int V= (uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19; \ |
440 | 440 |
441 #define YSCALE_YUV_2_RGB2_C(type) \ | 441 #define YSCALE_YUV_2_RGB2_C(type) \ |
442 YSCALE_YUV_2_PACKED2_C\ | 442 YSCALE_YUV_2_PACKED2_C\ |
443 type *r, *b, *g;\ | 443 type *r, *b, *g;\ |
444 r = (type *)c->table_rV[V];\ | 444 r = (type *)c->table_rV[V];\ |
445 g = (type *)(c->table_gU[U] + c->table_gV[V]);\ | 445 g = (type *)(c->table_gU[U] + c->table_gV[V]);\ |
446 b = (type *)c->table_bU[U];\ | 446 b = (type *)c->table_bU[U];\ |
447 | 447 |
448 #define YSCALE_YUV_2_PACKED1_C \ | 448 #define YSCALE_YUV_2_PACKED1_C \ |
449 for(i=0; i<(dstW>>1); i++){\ | 449 for (i=0; i<(dstW>>1); i++){\ |
450 const int i2= 2*i;\ | 450 const int i2= 2*i;\ |
451 int Y1= buf0[i2 ]>>7;\ | 451 int Y1= buf0[i2 ]>>7;\ |
452 int Y2= buf0[i2+1]>>7;\ | 452 int Y2= buf0[i2+1]>>7;\ |
453 int U= (uvbuf1[i ])>>7;\ | 453 int U= (uvbuf1[i ])>>7;\ |
454 int V= (uvbuf1[i+2048])>>7;\ | 454 int V= (uvbuf1[i+2048])>>7;\ |
455 | 455 |
456 #define YSCALE_YUV_2_RGB1_C(type) \ | 456 #define YSCALE_YUV_2_RGB1_C(type) \ |
457 YSCALE_YUV_2_PACKED1_C\ | 457 YSCALE_YUV_2_PACKED1_C\ |
458 type *r, *b, *g;\ | 458 type *r, *b, *g;\ |
459 r = (type *)c->table_rV[V];\ | 459 r = (type *)c->table_rV[V];\ |
460 g = (type *)(c->table_gU[U] + c->table_gV[V]);\ | 460 g = (type *)(c->table_gU[U] + c->table_gV[V]);\ |
461 b = (type *)c->table_bU[U];\ | 461 b = (type *)c->table_bU[U];\ |
462 | 462 |
463 #define YSCALE_YUV_2_PACKED1B_C \ | 463 #define YSCALE_YUV_2_PACKED1B_C \ |
464 for(i=0; i<(dstW>>1); i++){\ | 464 for (i=0; i<(dstW>>1); i++){\ |
465 const int i2= 2*i;\ | 465 const int i2= 2*i;\ |
466 int Y1= buf0[i2 ]>>7;\ | 466 int Y1= buf0[i2 ]>>7;\ |
467 int Y2= buf0[i2+1]>>7;\ | 467 int Y2= buf0[i2+1]>>7;\ |
468 int U= (uvbuf0[i ] + uvbuf1[i ])>>8;\ | 468 int U= (uvbuf0[i ] + uvbuf1[i ])>>8;\ |
469 int V= (uvbuf0[i+2048] + uvbuf1[i+2048])>>8;\ | 469 int V= (uvbuf0[i+2048] + uvbuf1[i+2048])>>8;\ |
470 | 470 |
471 #define YSCALE_YUV_2_RGB1B_C(type) \ | 471 #define YSCALE_YUV_2_RGB1B_C(type) \ |
472 YSCALE_YUV_2_PACKED1B_C\ | 472 YSCALE_YUV_2_PACKED1B_C\ |
473 type *r, *b, *g;\ | 473 type *r, *b, *g;\ |
474 r = (type *)c->table_rV[V];\ | 474 r = (type *)c->table_rV[V];\ |
475 g = (type *)(c->table_gU[U] + c->table_gV[V]);\ | 475 g = (type *)(c->table_gU[U] + c->table_gV[V]);\ |
476 b = (type *)c->table_bU[U];\ | 476 b = (type *)c->table_bU[U];\ |
477 | 477 |
478 #define YSCALE_YUV_2_ANYRGB_C(func, func2)\ | 478 #define YSCALE_YUV_2_ANYRGB_C(func, func2)\ |
479 switch(c->dstFormat)\ | 479 switch(c->dstFormat)\ |
480 {\ | 480 {\ |
481 case PIX_FMT_RGB32:\ | 481 case PIX_FMT_RGB32:\ |
482 case PIX_FMT_BGR32:\ | 482 case PIX_FMT_BGR32:\ |
483 func(uint32_t)\ | 483 func(uint32_t)\ |
484 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\ | 484 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\ |
485 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\ | 485 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\ |
486 } \ | 486 } \ |
487 break;\ | 487 break;\ |
488 case PIX_FMT_RGB24:\ | 488 case PIX_FMT_RGB24:\ |
489 func(uint8_t)\ | 489 func(uint8_t)\ |
490 ((uint8_t*)dest)[0]= r[Y1];\ | 490 ((uint8_t*)dest)[0]= r[Y1];\ |
491 ((uint8_t*)dest)[1]= g[Y1];\ | 491 ((uint8_t*)dest)[1]= g[Y1];\ |
492 ((uint8_t*)dest)[2]= b[Y1];\ | 492 ((uint8_t*)dest)[2]= b[Y1];\ |
493 ((uint8_t*)dest)[3]= r[Y2];\ | 493 ((uint8_t*)dest)[3]= r[Y2];\ |
494 ((uint8_t*)dest)[4]= g[Y2];\ | 494 ((uint8_t*)dest)[4]= g[Y2];\ |
495 ((uint8_t*)dest)[5]= b[Y2];\ | 495 ((uint8_t*)dest)[5]= b[Y2];\ |
496 dest+=6;\ | 496 dest+=6;\ |
497 }\ | 497 }\ |
498 break;\ | 498 break;\ |
499 case PIX_FMT_BGR24:\ | 499 case PIX_FMT_BGR24:\ |
500 func(uint8_t)\ | 500 func(uint8_t)\ |
501 ((uint8_t*)dest)[0]= b[Y1];\ | 501 ((uint8_t*)dest)[0]= b[Y1];\ |
502 ((uint8_t*)dest)[1]= g[Y1];\ | 502 ((uint8_t*)dest)[1]= g[Y1];\ |
503 ((uint8_t*)dest)[2]= r[Y1];\ | 503 ((uint8_t*)dest)[2]= r[Y1];\ |
504 ((uint8_t*)dest)[3]= b[Y2];\ | 504 ((uint8_t*)dest)[3]= b[Y2];\ |
505 ((uint8_t*)dest)[4]= g[Y2];\ | 505 ((uint8_t*)dest)[4]= g[Y2];\ |
506 ((uint8_t*)dest)[5]= r[Y2];\ | 506 ((uint8_t*)dest)[5]= r[Y2];\ |
507 dest+=6;\ | 507 dest+=6;\ |
508 }\ | 508 }\ |
509 break;\ | 509 break;\ |
510 case PIX_FMT_RGB565:\ | 510 case PIX_FMT_RGB565:\ |
511 case PIX_FMT_BGR565:\ | 511 case PIX_FMT_BGR565:\ |
512 {\ | 512 {\ |
513 const int dr1= dither_2x2_8[y&1 ][0];\ | 513 const int dr1= dither_2x2_8[y&1 ][0];\ |
514 const int dg1= dither_2x2_4[y&1 ][0];\ | 514 const int dg1= dither_2x2_4[y&1 ][0];\ |
515 const int db1= dither_2x2_8[(y&1)^1][0];\ | 515 const int db1= dither_2x2_8[(y&1)^1][0];\ |
516 const int dr2= dither_2x2_8[y&1 ][1];\ | 516 const int dr2= dither_2x2_8[y&1 ][1];\ |
517 const int dg2= dither_2x2_4[y&1 ][1];\ | 517 const int dg2= dither_2x2_4[y&1 ][1];\ |
518 const int db2= dither_2x2_8[(y&1)^1][1];\ | 518 const int db2= dither_2x2_8[(y&1)^1][1];\ |
519 func(uint16_t)\ | 519 func(uint16_t)\ |
520 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\ | 520 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\ |
521 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\ | 521 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\ |
522 }\ | 522 }\ |
523 }\ | 523 }\ |
524 break;\ | 524 break;\ |
525 case PIX_FMT_RGB555:\ | 525 case PIX_FMT_RGB555:\ |
526 case PIX_FMT_BGR555:\ | 526 case PIX_FMT_BGR555:\ |
527 {\ | 527 {\ |
528 const int dr1= dither_2x2_8[y&1 ][0];\ | 528 const int dr1= dither_2x2_8[y&1 ][0];\ |
529 const int dg1= dither_2x2_8[y&1 ][1];\ | 529 const int dg1= dither_2x2_8[y&1 ][1];\ |
530 const int db1= dither_2x2_8[(y&1)^1][0];\ | 530 const int db1= dither_2x2_8[(y&1)^1][0];\ |
531 const int dr2= dither_2x2_8[y&1 ][1];\ | 531 const int dr2= dither_2x2_8[y&1 ][1];\ |
532 const int dg2= dither_2x2_8[y&1 ][0];\ | 532 const int dg2= dither_2x2_8[y&1 ][0];\ |
533 const int db2= dither_2x2_8[(y&1)^1][1];\ | 533 const int db2= dither_2x2_8[(y&1)^1][1];\ |
534 func(uint16_t)\ | 534 func(uint16_t)\ |
535 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\ | 535 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\ |
536 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\ | 536 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\ |
537 }\ | 537 }\ |
538 }\ | 538 }\ |
539 break;\ | 539 break;\ |
540 case PIX_FMT_RGB8:\ | 540 case PIX_FMT_RGB8:\ |
541 case PIX_FMT_BGR8:\ | 541 case PIX_FMT_BGR8:\ |
542 {\ | 542 {\ |
543 const uint8_t * const d64= dither_8x8_73[y&7];\ | 543 const uint8_t * const d64= dither_8x8_73[y&7];\ |
544 const uint8_t * const d32= dither_8x8_32[y&7];\ | 544 const uint8_t * const d32= dither_8x8_32[y&7];\ |
545 func(uint8_t)\ | 545 func(uint8_t)\ |
546 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\ | 546 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\ |
547 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\ | 547 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\ |
548 }\ | 548 }\ |
549 }\ | 549 }\ |
550 break;\ | 550 break;\ |
551 case PIX_FMT_RGB4:\ | 551 case PIX_FMT_RGB4:\ |
552 case PIX_FMT_BGR4:\ | 552 case PIX_FMT_BGR4:\ |
553 {\ | 553 {\ |
554 const uint8_t * const d64= dither_8x8_73 [y&7];\ | 554 const uint8_t * const d64= dither_8x8_73 [y&7];\ |
555 const uint8_t * const d128=dither_8x8_220[y&7];\ | 555 const uint8_t * const d128=dither_8x8_220[y&7];\ |
556 func(uint8_t)\ | 556 func(uint8_t)\ |
557 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\ | 557 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\ |
558 + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\ | 558 + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\ |
559 }\ | 559 }\ |
560 }\ | 560 }\ |
561 break;\ | 561 break;\ |
562 case PIX_FMT_RGB4_BYTE:\ | 562 case PIX_FMT_RGB4_BYTE:\ |
563 case PIX_FMT_BGR4_BYTE:\ | 563 case PIX_FMT_BGR4_BYTE:\ |
564 {\ | 564 {\ |
565 const uint8_t * const d64= dither_8x8_73 [y&7];\ | 565 const uint8_t * const d64= dither_8x8_73 [y&7];\ |
566 const uint8_t * const d128=dither_8x8_220[y&7];\ | 566 const uint8_t * const d128=dither_8x8_220[y&7];\ |
567 func(uint8_t)\ | 567 func(uint8_t)\ |
568 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\ | 568 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\ |
569 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\ | 569 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\ |
570 }\ | 570 }\ |
571 }\ | 571 }\ |
572 break;\ | 572 break;\ |
573 case PIX_FMT_MONOBLACK:\ | 573 case PIX_FMT_MONOBLACK:\ |
574 {\ | 574 {\ |
575 const uint8_t * const d128=dither_8x8_220[y&7];\ | 575 const uint8_t * const d128=dither_8x8_220[y&7];\ |
576 uint8_t *g= c->table_gU[128] + c->table_gV[128];\ | 576 uint8_t *g= c->table_gU[128] + c->table_gV[128];\ |
577 for(i=0; i<dstW-7; i+=8){\ | 577 for (i=0; i<dstW-7; i+=8){\ |
578 int acc;\ | 578 int acc;\ |
579 acc = g[((buf0[i ]*yalpha1+buf1[i ]*yalpha)>>19) + d128[0]];\ | 579 acc = g[((buf0[i ]*yalpha1+buf1[i ]*yalpha)>>19) + d128[0]];\ |
580 acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\ | 580 acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\ |
581 acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\ | 581 acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\ |
582 acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\ | 582 acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\ |
583 acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\ | 583 acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\ |
584 acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\ | 584 acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\ |
585 acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\ | 585 acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\ |
586 acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\ | 586 acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\ |
587 ((uint8_t*)dest)[0]= acc;\ | 587 ((uint8_t*)dest)[0]= acc;\ |
588 dest++;\ | 588 dest++;\ |
589 }\ | 589 }\ |
590 \ | 590 \ |
591 /*\ | 591 /*\ |
592 ((uint8_t*)dest)-= dstW>>4;\ | 592 ((uint8_t*)dest)-= dstW>>4;\ |
593 {\ | 593 {\ |
594 int acc=0;\ | 594 int acc=0;\ |
595 int left=0;\ | 595 int left=0;\ |
596 static int top[1024];\ | 596 static int top[1024];\ |
597 static int last_new[1024][1024];\ | 597 static int last_new[1024][1024];\ |
598 static int last_in3[1024][1024];\ | 598 static int last_in3[1024][1024];\ |
599 static int drift[1024][1024];\ | 599 static int drift[1024][1024];\ |
600 int topLeft=0;\ | 600 int topLeft=0;\ |
601 int shift=0;\ | 601 int shift=0;\ |
602 int count=0;\ | 602 int count=0;\ |
603 const uint8_t * const d128=dither_8x8_220[y&7];\ | 603 const uint8_t * const d128=dither_8x8_220[y&7];\ |
604 int error_new=0;\ | 604 int error_new=0;\ |
605 int error_in3=0;\ | 605 int error_in3=0;\ |
606 int f=0;\ | 606 int f=0;\ |
607 \ | 607 \ |
608 for(i=dstW>>1; i<dstW; i++){\ | 608 for (i=dstW>>1; i<dstW; i++){\ |
609 int in= ((buf0[i ]*yalpha1+buf1[i ]*yalpha)>>19);\ | 609 int in= ((buf0[i ]*yalpha1+buf1[i ]*yalpha)>>19);\ |
610 int in2 = (76309 * (in - 16) + 32768) >> 16;\ | 610 int in2 = (76309 * (in - 16) + 32768) >> 16;\ |
611 int in3 = (in2 < 0) ? 0 : ((in2 > 255) ? 255 : in2);\ | 611 int in3 = (in2 < 0) ? 0 : ((in2 > 255) ? 255 : in2);\ |
612 int old= (left*7 + topLeft + top[i]*5 + top[i+1]*3)/20 + in3\ | 612 int old= (left*7 + topLeft + top[i]*5 + top[i+1]*3)/20 + in3\ |
613 + (last_new[y][i] - in3)*f/256;\ | 613 + (last_new[y][i] - in3)*f/256;\ |
614 int new= old> 128 ? 255 : 0;\ | 614 int new= old> 128 ? 255 : 0;\ |
615 \ | 615 \ |
616 error_new+= FFABS(last_new[y][i] - new);\ | 616 error_new+= FFABS(last_new[y][i] - new);\ |
617 error_in3+= FFABS(last_in3[y][i] - in3);\ | 617 error_in3+= FFABS(last_in3[y][i] - in3);\ |
618 f= error_new - error_in3*4;\ | 618 f= error_new - error_in3*4;\ |
619 if(f<0) f=0;\ | 619 if (f<0) f=0;\ |
620 if(f>256) f=256;\ | 620 if (f>256) f=256;\ |
621 \ | 621 \ |
622 topLeft= top[i];\ | 622 topLeft= top[i];\ |
623 left= top[i]= old - new;\ | 623 left= top[i]= old - new;\ |
624 last_new[y][i]= new;\ | 624 last_new[y][i]= new;\ |
625 last_in3[y][i]= in3;\ | 625 last_in3[y][i]= in3;\ |
626 \ | 626 \ |
627 acc+= acc + (new&1);\ | 627 acc+= acc + (new&1);\ |
628 if((i&7)==6){\ | 628 if ((i&7)==6){\ |
629 ((uint8_t*)dest)[0]= acc;\ | 629 ((uint8_t*)dest)[0]= acc;\ |
630 ((uint8_t*)dest)++;\ | 630 ((uint8_t*)dest)++;\ |
631 }\ | 631 }\ |
632 }\ | 632 }\ |
633 }\ | 633 }\ |
634 */\ | 634 */\ |
635 }\ | 635 }\ |
636 break;\ | 636 break;\ |
637 case PIX_FMT_YUYV422:\ | 637 case PIX_FMT_YUYV422:\ |
638 func2\ | 638 func2\ |
639 ((uint8_t*)dest)[2*i2+0]= Y1;\ | 639 ((uint8_t*)dest)[2*i2+0]= Y1;\ |
640 ((uint8_t*)dest)[2*i2+1]= U;\ | 640 ((uint8_t*)dest)[2*i2+1]= U;\ |
641 ((uint8_t*)dest)[2*i2+2]= Y2;\ | 641 ((uint8_t*)dest)[2*i2+2]= Y2;\ |
642 ((uint8_t*)dest)[2*i2+3]= V;\ | 642 ((uint8_t*)dest)[2*i2+3]= V;\ |
643 } \ | 643 } \ |
644 break;\ | 644 break;\ |
645 case PIX_FMT_UYVY422:\ | 645 case PIX_FMT_UYVY422:\ |
646 func2\ | 646 func2\ |
647 ((uint8_t*)dest)[2*i2+0]= U;\ | 647 ((uint8_t*)dest)[2*i2+0]= U;\ |
648 ((uint8_t*)dest)[2*i2+1]= Y1;\ | 648 ((uint8_t*)dest)[2*i2+1]= Y1;\ |
649 ((uint8_t*)dest)[2*i2+2]= V;\ | 649 ((uint8_t*)dest)[2*i2+2]= V;\ |
650 ((uint8_t*)dest)[2*i2+3]= Y2;\ | 650 ((uint8_t*)dest)[2*i2+3]= Y2;\ |
651 } \ | 651 } \ |
652 break;\ | 652 break;\ |
653 }\ | 653 }\ |
654 | 654 |
655 | 655 |
656 static inline void yuv2packedXinC(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize, | 656 static inline void yuv2packedXinC(SwsContext *c, int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize, |
657 int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize, | 657 int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize, |
658 uint8_t *dest, int dstW, int y) | 658 uint8_t *dest, int dstW, int y) |
659 { | 659 { |
660 int i; | 660 int i; |
661 switch(c->dstFormat) | 661 switch(c->dstFormat) |
662 { | 662 { |
663 case PIX_FMT_BGR32: | 663 case PIX_FMT_BGR32: |
664 case PIX_FMT_RGB32: | 664 case PIX_FMT_RGB32: |
665 YSCALE_YUV_2_RGBX_C(uint32_t) | 665 YSCALE_YUV_2_RGBX_C(uint32_t) |
666 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1]; | 666 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1]; |
667 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2]; | 667 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2]; |
668 } | 668 } |
669 break; | 669 break; |
670 case PIX_FMT_RGB24: | 670 case PIX_FMT_RGB24: |
671 YSCALE_YUV_2_RGBX_C(uint8_t) | 671 YSCALE_YUV_2_RGBX_C(uint8_t) |
672 ((uint8_t*)dest)[0]= r[Y1]; | 672 ((uint8_t*)dest)[0]= r[Y1]; |
673 ((uint8_t*)dest)[1]= g[Y1]; | 673 ((uint8_t*)dest)[1]= g[Y1]; |
674 ((uint8_t*)dest)[2]= b[Y1]; | 674 ((uint8_t*)dest)[2]= b[Y1]; |
675 ((uint8_t*)dest)[3]= r[Y2]; | 675 ((uint8_t*)dest)[3]= r[Y2]; |
676 ((uint8_t*)dest)[4]= g[Y2]; | 676 ((uint8_t*)dest)[4]= g[Y2]; |
677 ((uint8_t*)dest)[5]= b[Y2]; | 677 ((uint8_t*)dest)[5]= b[Y2]; |
678 dest+=6; | 678 dest+=6; |
679 } | 679 } |
680 break; | 680 break; |
681 case PIX_FMT_BGR24: | 681 case PIX_FMT_BGR24: |
682 YSCALE_YUV_2_RGBX_C(uint8_t) | 682 YSCALE_YUV_2_RGBX_C(uint8_t) |
683 ((uint8_t*)dest)[0]= b[Y1]; | 683 ((uint8_t*)dest)[0]= b[Y1]; |
684 ((uint8_t*)dest)[1]= g[Y1]; | 684 ((uint8_t*)dest)[1]= g[Y1]; |
685 ((uint8_t*)dest)[2]= r[Y1]; | 685 ((uint8_t*)dest)[2]= r[Y1]; |
686 ((uint8_t*)dest)[3]= b[Y2]; | 686 ((uint8_t*)dest)[3]= b[Y2]; |
687 ((uint8_t*)dest)[4]= g[Y2]; | 687 ((uint8_t*)dest)[4]= g[Y2]; |
688 ((uint8_t*)dest)[5]= r[Y2]; | 688 ((uint8_t*)dest)[5]= r[Y2]; |
689 dest+=6; | 689 dest+=6; |
690 } | 690 } |
691 break; | 691 break; |
692 case PIX_FMT_RGB565: | 692 case PIX_FMT_RGB565: |
693 case PIX_FMT_BGR565: | 693 case PIX_FMT_BGR565: |
694 { | 694 { |
695 const int dr1= dither_2x2_8[y&1 ][0]; | 695 const int dr1= dither_2x2_8[y&1 ][0]; |
696 const int dg1= dither_2x2_4[y&1 ][0]; | 696 const int dg1= dither_2x2_4[y&1 ][0]; |
697 const int db1= dither_2x2_8[(y&1)^1][0]; | 697 const int db1= dither_2x2_8[(y&1)^1][0]; |
698 const int dr2= dither_2x2_8[y&1 ][1]; | 698 const int dr2= dither_2x2_8[y&1 ][1]; |
699 const int dg2= dither_2x2_4[y&1 ][1]; | 699 const int dg2= dither_2x2_4[y&1 ][1]; |
700 const int db2= dither_2x2_8[(y&1)^1][1]; | 700 const int db2= dither_2x2_8[(y&1)^1][1]; |
701 YSCALE_YUV_2_RGBX_C(uint16_t) | 701 YSCALE_YUV_2_RGBX_C(uint16_t) |
702 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1]; | 702 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1]; |
703 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2]; | 703 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2]; |
704 } | 704 } |
705 } | 705 } |
706 break; | 706 break; |
707 case PIX_FMT_RGB555: | 707 case PIX_FMT_RGB555: |
708 case PIX_FMT_BGR555: | 708 case PIX_FMT_BGR555: |
709 { | 709 { |
710 const int dr1= dither_2x2_8[y&1 ][0]; | 710 const int dr1= dither_2x2_8[y&1 ][0]; |
711 const int dg1= dither_2x2_8[y&1 ][1]; | 711 const int dg1= dither_2x2_8[y&1 ][1]; |
712 const int db1= dither_2x2_8[(y&1)^1][0]; | 712 const int db1= dither_2x2_8[(y&1)^1][0]; |
713 const int dr2= dither_2x2_8[y&1 ][1]; | 713 const int dr2= dither_2x2_8[y&1 ][1]; |
714 const int dg2= dither_2x2_8[y&1 ][0]; | 714 const int dg2= dither_2x2_8[y&1 ][0]; |
715 const int db2= dither_2x2_8[(y&1)^1][1]; | 715 const int db2= dither_2x2_8[(y&1)^1][1]; |
716 YSCALE_YUV_2_RGBX_C(uint16_t) | 716 YSCALE_YUV_2_RGBX_C(uint16_t) |
717 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1]; | 717 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1]; |
718 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2]; | 718 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2]; |
719 } | 719 } |
720 } | 720 } |
721 break; | 721 break; |
722 case PIX_FMT_RGB8: | 722 case PIX_FMT_RGB8: |
723 case PIX_FMT_BGR8: | 723 case PIX_FMT_BGR8: |
724 { | 724 { |
725 const uint8_t * const d64= dither_8x8_73[y&7]; | 725 const uint8_t * const d64= dither_8x8_73[y&7]; |
726 const uint8_t * const d32= dither_8x8_32[y&7]; | 726 const uint8_t * const d32= dither_8x8_32[y&7]; |
727 YSCALE_YUV_2_RGBX_C(uint8_t) | 727 YSCALE_YUV_2_RGBX_C(uint8_t) |
728 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]]; | 728 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]]; |
729 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]]; | 729 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]]; |
730 } | 730 } |
731 } | 731 } |
732 break; | 732 break; |
733 case PIX_FMT_RGB4: | 733 case PIX_FMT_RGB4: |
734 case PIX_FMT_BGR4: | 734 case PIX_FMT_BGR4: |
735 { | 735 { |
736 const uint8_t * const d64= dither_8x8_73 [y&7]; | 736 const uint8_t * const d64= dither_8x8_73 [y&7]; |
737 const uint8_t * const d128=dither_8x8_220[y&7]; | 737 const uint8_t * const d128=dither_8x8_220[y&7]; |
738 YSCALE_YUV_2_RGBX_C(uint8_t) | 738 YSCALE_YUV_2_RGBX_C(uint8_t) |
739 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]] | 739 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]] |
740 +((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4); | 740 +((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4); |
741 } | 741 } |
742 } | 742 } |
743 break; | 743 break; |
744 case PIX_FMT_RGB4_BYTE: | 744 case PIX_FMT_RGB4_BYTE: |
745 case PIX_FMT_BGR4_BYTE: | 745 case PIX_FMT_BGR4_BYTE: |
746 { | 746 { |
747 const uint8_t * const d64= dither_8x8_73 [y&7]; | 747 const uint8_t * const d64= dither_8x8_73 [y&7]; |
748 const uint8_t * const d128=dither_8x8_220[y&7]; | 748 const uint8_t * const d128=dither_8x8_220[y&7]; |
749 YSCALE_YUV_2_RGBX_C(uint8_t) | 749 YSCALE_YUV_2_RGBX_C(uint8_t) |
750 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]; | 750 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]; |
751 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]]; | 751 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]]; |
752 } | 752 } |
753 } | 753 } |
754 break; | 754 break; |
755 case PIX_FMT_MONOBLACK: | 755 case PIX_FMT_MONOBLACK: |
756 { | 756 { |
757 const uint8_t * const d128=dither_8x8_220[y&7]; | 757 const uint8_t * const d128=dither_8x8_220[y&7]; |
758 uint8_t *g= c->table_gU[128] + c->table_gV[128]; | 758 uint8_t *g= c->table_gU[128] + c->table_gV[128]; |
759 int acc=0; | 759 int acc=0; |
760 for(i=0; i<dstW-1; i+=2){ | 760 for (i=0; i<dstW-1; i+=2){ |
761 int j; | 761 int j; |
762 int Y1=1<<18; | 762 int Y1=1<<18; |
763 int Y2=1<<18; | 763 int Y2=1<<18; |
764 | 764 |
765 for(j=0; j<lumFilterSize; j++) | 765 for (j=0; j<lumFilterSize; j++) |
766 { | 766 { |
767 Y1 += lumSrc[j][i] * lumFilter[j]; | 767 Y1 += lumSrc[j][i] * lumFilter[j]; |
768 Y2 += lumSrc[j][i+1] * lumFilter[j]; | 768 Y2 += lumSrc[j][i+1] * lumFilter[j]; |
769 } | 769 } |
770 Y1>>=19; | 770 Y1>>=19; |
771 Y2>>=19; | 771 Y2>>=19; |
772 if((Y1|Y2)&256) | 772 if ((Y1|Y2)&256) |
773 { | 773 { |
774 if(Y1>255) Y1=255; | 774 if (Y1>255) Y1=255; |
775 else if(Y1<0)Y1=0; | 775 else if (Y1<0)Y1=0; |
776 if(Y2>255) Y2=255; | 776 if (Y2>255) Y2=255; |
777 else if(Y2<0)Y2=0; | 777 else if (Y2<0)Y2=0; |
778 } | 778 } |
779 acc+= acc + g[Y1+d128[(i+0)&7]]; | 779 acc+= acc + g[Y1+d128[(i+0)&7]]; |
780 acc+= acc + g[Y2+d128[(i+1)&7]]; | 780 acc+= acc + g[Y2+d128[(i+1)&7]]; |
781 if((i&7)==6){ | 781 if ((i&7)==6){ |
782 ((uint8_t*)dest)[0]= acc; | 782 ((uint8_t*)dest)[0]= acc; |
783 dest++; | 783 dest++; |
784 } | 784 } |
785 } | 785 } |
786 } | 786 } |
787 break; | 787 break; |
788 case PIX_FMT_YUYV422: | 788 case PIX_FMT_YUYV422: |
789 YSCALE_YUV_2_PACKEDX_C(void) | 789 YSCALE_YUV_2_PACKEDX_C(void) |
790 ((uint8_t*)dest)[2*i2+0]= Y1; | 790 ((uint8_t*)dest)[2*i2+0]= Y1; |
791 ((uint8_t*)dest)[2*i2+1]= U; | 791 ((uint8_t*)dest)[2*i2+1]= U; |
792 ((uint8_t*)dest)[2*i2+2]= Y2; | 792 ((uint8_t*)dest)[2*i2+2]= Y2; |
793 ((uint8_t*)dest)[2*i2+3]= V; | 793 ((uint8_t*)dest)[2*i2+3]= V; |
794 } | 794 } |
795 break; | 795 break; |
796 case PIX_FMT_UYVY422: | 796 case PIX_FMT_UYVY422: |
797 YSCALE_YUV_2_PACKEDX_C(void) | 797 YSCALE_YUV_2_PACKEDX_C(void) |
798 ((uint8_t*)dest)[2*i2+0]= U; | 798 ((uint8_t*)dest)[2*i2+0]= U; |
799 ((uint8_t*)dest)[2*i2+1]= Y1; | 799 ((uint8_t*)dest)[2*i2+1]= Y1; |
800 ((uint8_t*)dest)[2*i2+2]= V; | 800 ((uint8_t*)dest)[2*i2+2]= V; |
801 ((uint8_t*)dest)[2*i2+3]= Y2; | 801 ((uint8_t*)dest)[2*i2+3]= Y2; |
802 } | 802 } |
803 break; | 803 break; |
804 } | 804 } |
805 } | 805 } |
806 | 806 |
807 | 807 |
808 //Note: we have C, X86, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one | 808 //Note: we have C, X86, MMX, MMX2, 3DNOW version therse no 3DNOW+MMX2 one |
809 //Plain C versions | 809 //Plain C versions |
900 | 900 |
901 // minor note: the HAVE_xyz is messed up after that line so don't use it | 901 // minor note: the HAVE_xyz is messed up after that line so don't use it |
902 | 902 |
903 static double getSplineCoeff(double a, double b, double c, double d, double dist) | 903 static double getSplineCoeff(double a, double b, double c, double d, double dist) |
904 { | 904 { |
905 // printf("%f %f %f %f %f\n", a,b,c,d,dist); | 905 // printf("%f %f %f %f %f\n", a,b,c,d,dist); |
906 if(dist<=1.0) return ((d*dist + c)*dist + b)*dist +a; | 906 if (dist<=1.0) return ((d*dist + c)*dist + b)*dist +a; |
907 else return getSplineCoeff( 0.0, | 907 else return getSplineCoeff( 0.0, |
908 b+ 2.0*c + 3.0*d, | 908 b+ 2.0*c + 3.0*d, |
909 c + 3.0*d, | 909 c + 3.0*d, |
910 -b- 3.0*c - 6.0*d, | 910 -b- 3.0*c - 6.0*d, |
911 dist-1.0); | 911 dist-1.0); |
912 } | 912 } |
913 | 913 |
914 static inline int initFilter(int16_t **outFilter, int16_t **filterPos, int *outFilterSize, int xInc, | 914 static inline int initFilter(int16_t **outFilter, int16_t **filterPos, int *outFilterSize, int xInc, |
915 int srcW, int dstW, int filterAlign, int one, int flags, | 915 int srcW, int dstW, int filterAlign, int one, int flags, |
916 SwsVector *srcFilter, SwsVector *dstFilter, double param[2]) | 916 SwsVector *srcFilter, SwsVector *dstFilter, double param[2]) |
917 { | 917 { |
918 int i; | 918 int i; |
919 int filterSize; | 919 int filterSize; |
920 int filter2Size; | 920 int filter2Size; |
921 int minFilterSize; | 921 int minFilterSize; |
922 double *filter=NULL; | 922 double *filter=NULL; |
923 double *filter2=NULL; | 923 double *filter2=NULL; |
924 #if defined(ARCH_X86) | 924 #if defined(ARCH_X86) |
925 if(flags & SWS_CPU_CAPS_MMX) | 925 if (flags & SWS_CPU_CAPS_MMX) |
926 asm volatile("emms\n\t"::: "memory"); //FIXME this shouldnt be required but it IS (even for non mmx versions) | 926 asm volatile("emms\n\t"::: "memory"); //FIXME this shouldnt be required but it IS (even for non mmx versions) |
927 #endif | 927 #endif |
928 | 928 |
929 // Note the +1 is for the MMXscaler which reads over the end | 929 // Note the +1 is for the MMXscaler which reads over the end |
930 *filterPos = av_malloc((dstW+1)*sizeof(int16_t)); | 930 *filterPos = av_malloc((dstW+1)*sizeof(int16_t)); |
931 | 931 |
932 if(FFABS(xInc - 0x10000) <10) // unscaled | 932 if (FFABS(xInc - 0x10000) <10) // unscaled |
933 { | 933 { |
934 int i; | 934 int i; |
935 filterSize= 1; | 935 filterSize= 1; |
936 filter= av_malloc(dstW*sizeof(double)*filterSize); | 936 filter= av_malloc(dstW*sizeof(double)*filterSize); |
937 for(i=0; i<dstW*filterSize; i++) filter[i]=0; | 937 for (i=0; i<dstW*filterSize; i++) filter[i]=0; |
938 | 938 |
939 for(i=0; i<dstW; i++) | 939 for (i=0; i<dstW; i++) |
940 { | 940 { |
941 filter[i*filterSize]=1; | 941 filter[i*filterSize]=1; |
942 (*filterPos)[i]=i; | 942 (*filterPos)[i]=i; |
943 } | 943 } |
944 | 944 |
945 } | 945 } |
946 else if(flags&SWS_POINT) // lame looking point sampling mode | 946 else if (flags&SWS_POINT) // lame looking point sampling mode |
947 { | 947 { |
948 int i; | 948 int i; |
949 int xDstInSrc; | 949 int xDstInSrc; |
950 filterSize= 1; | 950 filterSize= 1; |
951 filter= av_malloc(dstW*sizeof(double)*filterSize); | 951 filter= av_malloc(dstW*sizeof(double)*filterSize); |
952 | 952 |
953 xDstInSrc= xInc/2 - 0x8000; | 953 xDstInSrc= xInc/2 - 0x8000; |
954 for(i=0; i<dstW; i++) | 954 for (i=0; i<dstW; i++) |
955 { | 955 { |
956 int xx= (xDstInSrc - ((filterSize-1)<<15) + (1<<15))>>16; | 956 int xx= (xDstInSrc - ((filterSize-1)<<15) + (1<<15))>>16; |
957 | 957 |
958 (*filterPos)[i]= xx; | 958 (*filterPos)[i]= xx; |
959 filter[i]= 1.0; | 959 filter[i]= 1.0; |
960 xDstInSrc+= xInc; | 960 xDstInSrc+= xInc; |
961 } | 961 } |
962 } | 962 } |
963 else if((xInc <= (1<<16) && (flags&SWS_AREA)) || (flags&SWS_FAST_BILINEAR)) // bilinear upscale | 963 else if ((xInc <= (1<<16) && (flags&SWS_AREA)) || (flags&SWS_FAST_BILINEAR)) // bilinear upscale |
964 { | 964 { |
965 int i; | 965 int i; |
966 int xDstInSrc; | 966 int xDstInSrc; |
967 if (flags&SWS_BICUBIC) filterSize= 4; | 967 if (flags&SWS_BICUBIC) filterSize= 4; |
968 else if(flags&SWS_X ) filterSize= 4; | 968 else if (flags&SWS_X ) filterSize= 4; |
969 else filterSize= 2; // SWS_BILINEAR / SWS_AREA | 969 else filterSize= 2; // SWS_BILINEAR / SWS_AREA |
970 filter= av_malloc(dstW*sizeof(double)*filterSize); | 970 filter= av_malloc(dstW*sizeof(double)*filterSize); |
971 | 971 |
972 xDstInSrc= xInc/2 - 0x8000; | 972 xDstInSrc= xInc/2 - 0x8000; |
973 for(i=0; i<dstW; i++) | 973 for (i=0; i<dstW; i++) |
974 { | 974 { |
975 int xx= (xDstInSrc - ((filterSize-1)<<15) + (1<<15))>>16; | 975 int xx= (xDstInSrc - ((filterSize-1)<<15) + (1<<15))>>16; |
976 int j; | 976 int j; |
977 | 977 |
978 (*filterPos)[i]= xx; | 978 (*filterPos)[i]= xx; |
979 //Bilinear upscale / linear interpolate / Area averaging | 979 //Bilinear upscale / linear interpolate / Area averaging |
980 for(j=0; j<filterSize; j++) | 980 for (j=0; j<filterSize; j++) |
981 { | 981 { |
982 double d= FFABS((xx<<16) - xDstInSrc)/(double)(1<<16); | 982 double d= FFABS((xx<<16) - xDstInSrc)/(double)(1<<16); |
983 double coeff= 1.0 - d; | 983 double coeff= 1.0 - d; |
984 if(coeff<0) coeff=0; | 984 if (coeff<0) coeff=0; |
985 filter[i*filterSize + j]= coeff; | 985 filter[i*filterSize + j]= coeff; |
986 xx++; | 986 xx++; |
987 } | 987 } |
988 xDstInSrc+= xInc; | 988 xDstInSrc+= xInc; |
989 } | 989 } |
990 } | 990 } |
991 else | 991 else |
992 { | 992 { |
993 double xDstInSrc; | 993 double xDstInSrc; |
994 double sizeFactor, filterSizeInSrc; | 994 double sizeFactor, filterSizeInSrc; |
995 const double xInc1= (double)xInc / (double)(1<<16); | 995 const double xInc1= (double)xInc / (double)(1<<16); |
996 | 996 |
997 if (flags&SWS_BICUBIC) sizeFactor= 4.0; | 997 if (flags&SWS_BICUBIC) sizeFactor= 4.0; |
998 else if(flags&SWS_X) sizeFactor= 8.0; | 998 else if (flags&SWS_X) sizeFactor= 8.0; |
999 else if(flags&SWS_AREA) sizeFactor= 1.0; //downscale only, for upscale it is bilinear | 999 else if (flags&SWS_AREA) sizeFactor= 1.0; //downscale only, for upscale it is bilinear |
1000 else if(flags&SWS_GAUSS) sizeFactor= 8.0; // infinite ;) | 1000 else if (flags&SWS_GAUSS) sizeFactor= 8.0; // infinite ;) |
1001 else if(flags&SWS_LANCZOS) sizeFactor= param[0] != SWS_PARAM_DEFAULT ? 2.0*param[0] : 6.0; | 1001 else if (flags&SWS_LANCZOS) sizeFactor= param[0] != SWS_PARAM_DEFAULT ? 2.0*param[0] : 6.0; |
1002 else if(flags&SWS_SINC) sizeFactor= 20.0; // infinite ;) | 1002 else if (flags&SWS_SINC) sizeFactor= 20.0; // infinite ;) |
1003 else if(flags&SWS_SPLINE) sizeFactor= 20.0; // infinite ;) | 1003 else if (flags&SWS_SPLINE) sizeFactor= 20.0; // infinite ;) |
1004 else if(flags&SWS_BILINEAR) sizeFactor= 2.0; | 1004 else if (flags&SWS_BILINEAR) sizeFactor= 2.0; |
1005 else { | 1005 else { |
1006 sizeFactor= 0.0; //GCC warning killer | 1006 sizeFactor= 0.0; //GCC warning killer |
1007 ASSERT(0) | 1007 ASSERT(0) |
1008 } | 1008 } |
1009 | 1009 |
1010 if(xInc1 <= 1.0) filterSizeInSrc= sizeFactor; // upscale | 1010 if (xInc1 <= 1.0) filterSizeInSrc= sizeFactor; // upscale |
1011 else filterSizeInSrc= sizeFactor*srcW / (double)dstW; | 1011 else filterSizeInSrc= sizeFactor*srcW / (double)dstW; |
1012 | 1012 |
1013 filterSize= (int)ceil(1 + filterSizeInSrc); // will be reduced later if possible | 1013 filterSize= (int)ceil(1 + filterSizeInSrc); // will be reduced later if possible |
1014 if(filterSize > srcW-2) filterSize=srcW-2; | 1014 if (filterSize > srcW-2) filterSize=srcW-2; |
1015 | 1015 |
1016 filter= av_malloc(dstW*sizeof(double)*filterSize); | 1016 filter= av_malloc(dstW*sizeof(double)*filterSize); |
1017 | 1017 |
1018 xDstInSrc= xInc1 / 2.0 - 0.5; | 1018 xDstInSrc= xInc1 / 2.0 - 0.5; |
1019 for(i=0; i<dstW; i++) | 1019 for (i=0; i<dstW; i++) |
1020 { | 1020 { |
1021 int xx= (int)(xDstInSrc - (filterSize-1)*0.5 + 0.5); | 1021 int xx= (int)(xDstInSrc - (filterSize-1)*0.5 + 0.5); |
1022 int j; | 1022 int j; |
1023 (*filterPos)[i]= xx; | 1023 (*filterPos)[i]= xx; |
1024 for(j=0; j<filterSize; j++) | 1024 for (j=0; j<filterSize; j++) |
1025 { | 1025 { |
1026 double d= FFABS(xx - xDstInSrc)/filterSizeInSrc*sizeFactor; | 1026 double d= FFABS(xx - xDstInSrc)/filterSizeInSrc*sizeFactor; |
1027 double coeff; | 1027 double coeff; |
1028 if(flags & SWS_BICUBIC) | 1028 if (flags & SWS_BICUBIC) |
1029 { | 1029 { |
1030 double B= param[0] != SWS_PARAM_DEFAULT ? param[0] : 0.0; | 1030 double B= param[0] != SWS_PARAM_DEFAULT ? param[0] : 0.0; |
1031 double C= param[1] != SWS_PARAM_DEFAULT ? param[1] : 0.6; | 1031 double C= param[1] != SWS_PARAM_DEFAULT ? param[1] : 0.6; |
1032 | 1032 |
1033 if(d<1.0) | 1033 if (d<1.0) |
1034 coeff = (12-9*B-6*C)*d*d*d + (-18+12*B+6*C)*d*d + 6-2*B; | 1034 coeff = (12-9*B-6*C)*d*d*d + (-18+12*B+6*C)*d*d + 6-2*B; |
1035 else if(d<2.0) | 1035 else if (d<2.0) |
1036 coeff = (-B-6*C)*d*d*d + (6*B+30*C)*d*d + (-12*B-48*C)*d +8*B+24*C; | 1036 coeff = (-B-6*C)*d*d*d + (6*B+30*C)*d*d + (-12*B-48*C)*d +8*B+24*C; |
1037 else | 1037 else |
1038 coeff=0.0; | 1038 coeff=0.0; |
1039 } | 1039 } |
1040 /* else if(flags & SWS_X) | 1040 /* else if (flags & SWS_X) |
1041 { | 1041 { |
1042 double p= param ? param*0.01 : 0.3; | 1042 double p= param ? param*0.01 : 0.3; |
1043 coeff = d ? sin(d*PI)/(d*PI) : 1.0; | 1043 coeff = d ? sin(d*PI)/(d*PI) : 1.0; |
1044 coeff*= pow(2.0, - p*d*d); | 1044 coeff*= pow(2.0, - p*d*d); |
1045 }*/ | 1045 }*/ |
1046 else if(flags & SWS_X) | 1046 else if (flags & SWS_X) |
1047 { | 1047 { |
1048 double A= param[0] != SWS_PARAM_DEFAULT ? param[0] : 1.0; | 1048 double A= param[0] != SWS_PARAM_DEFAULT ? param[0] : 1.0; |
1049 | 1049 |
1050 if(d<1.0) | 1050 if (d<1.0) |
1051 coeff = cos(d*PI); | 1051 coeff = cos(d*PI); |
1052 else | 1052 else |
1053 coeff=-1.0; | 1053 coeff=-1.0; |
1054 if(coeff<0.0) coeff= -pow(-coeff, A); | 1054 if (coeff<0.0) coeff= -pow(-coeff, A); |
1055 else coeff= pow( coeff, A); | 1055 else coeff= pow( coeff, A); |
1056 coeff= coeff*0.5 + 0.5; | 1056 coeff= coeff*0.5 + 0.5; |
1057 } | 1057 } |
1058 else if(flags & SWS_AREA) | 1058 else if (flags & SWS_AREA) |
1059 { | 1059 { |
1060 double srcPixelSize= 1.0/xInc1; | 1060 double srcPixelSize= 1.0/xInc1; |
1061 if(d + srcPixelSize/2 < 0.5) coeff= 1.0; | 1061 if (d + srcPixelSize/2 < 0.5) coeff= 1.0; |
1062 else if(d - srcPixelSize/2 < 0.5) coeff= (0.5-d)/srcPixelSize + 0.5; | 1062 else if (d - srcPixelSize/2 < 0.5) coeff= (0.5-d)/srcPixelSize + 0.5; |
1063 else coeff=0.0; | 1063 else coeff=0.0; |
1064 } | 1064 } |
1065 else if(flags & SWS_GAUSS) | 1065 else if (flags & SWS_GAUSS) |
1066 { | 1066 { |
1067 double p= param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0; | 1067 double p= param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0; |
1068 coeff = pow(2.0, - p*d*d); | 1068 coeff = pow(2.0, - p*d*d); |
1069 } | 1069 } |
1070 else if(flags & SWS_SINC) | 1070 else if (flags & SWS_SINC) |
1071 { | 1071 { |
1072 coeff = d ? sin(d*PI)/(d*PI) : 1.0; | 1072 coeff = d ? sin(d*PI)/(d*PI) : 1.0; |
1073 } | 1073 } |
1074 else if(flags & SWS_LANCZOS) | 1074 else if (flags & SWS_LANCZOS) |
1075 { | 1075 { |
1076 double p= param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0; | 1076 double p= param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0; |
1077 coeff = d ? sin(d*PI)*sin(d*PI/p)/(d*d*PI*PI/p) : 1.0; | 1077 coeff = d ? sin(d*PI)*sin(d*PI/p)/(d*d*PI*PI/p) : 1.0; |
1078 if(d>p) coeff=0; | 1078 if (d>p) coeff=0; |
1079 } | 1079 } |
1080 else if(flags & SWS_BILINEAR) | 1080 else if (flags & SWS_BILINEAR) |
1081 { | 1081 { |
1082 coeff= 1.0 - d; | 1082 coeff= 1.0 - d; |
1083 if(coeff<0) coeff=0; | 1083 if (coeff<0) coeff=0; |
1084 } | 1084 } |
1085 else if(flags & SWS_SPLINE) | 1085 else if (flags & SWS_SPLINE) |
1086 { | 1086 { |
1087 double p=-2.196152422706632; | 1087 double p=-2.196152422706632; |
1088 coeff = getSplineCoeff(1.0, 0.0, p, -p-1.0, d); | 1088 coeff = getSplineCoeff(1.0, 0.0, p, -p-1.0, d); |
1089 } | 1089 } |
1090 else { | 1090 else { |
1091 coeff= 0.0; //GCC warning killer | 1091 coeff= 0.0; //GCC warning killer |
1092 ASSERT(0) | 1092 ASSERT(0) |
1093 } | 1093 } |
1094 | 1094 |
1095 filter[i*filterSize + j]= coeff; | 1095 filter[i*filterSize + j]= coeff; |
1096 xx++; | 1096 xx++; |
1097 } | 1097 } |
1098 xDstInSrc+= xInc1; | 1098 xDstInSrc+= xInc1; |
1099 } | 1099 } |
1100 } | 1100 } |
1101 | 1101 |
1102 /* apply src & dst Filter to filter -> filter2 | 1102 /* apply src & dst Filter to filter -> filter2 |
1103 av_free(filter); | 1103 av_free(filter); |
1104 */ | 1104 */ |
1105 ASSERT(filterSize>0) | 1105 ASSERT(filterSize>0) |
1106 filter2Size= filterSize; | 1106 filter2Size= filterSize; |
1107 if(srcFilter) filter2Size+= srcFilter->length - 1; | 1107 if (srcFilter) filter2Size+= srcFilter->length - 1; |
1108 if(dstFilter) filter2Size+= dstFilter->length - 1; | 1108 if (dstFilter) filter2Size+= dstFilter->length - 1; |
1109 ASSERT(filter2Size>0) | 1109 ASSERT(filter2Size>0) |
1110 filter2= av_malloc(filter2Size*dstW*sizeof(double)); | 1110 filter2= av_malloc(filter2Size*dstW*sizeof(double)); |
1111 | 1111 |
1112 for(i=0; i<dstW; i++) | 1112 for (i=0; i<dstW; i++) |
1113 { | 1113 { |
1114 int j; | 1114 int j; |
1115 SwsVector scaleFilter; | 1115 SwsVector scaleFilter; |
1116 SwsVector *outVec; | 1116 SwsVector *outVec; |
1117 | 1117 |
1118 scaleFilter.coeff= filter + i*filterSize; | 1118 scaleFilter.coeff= filter + i*filterSize; |
1119 scaleFilter.length= filterSize; | 1119 scaleFilter.length= filterSize; |
1120 | 1120 |
1121 if(srcFilter) outVec= sws_getConvVec(srcFilter, &scaleFilter); | 1121 if (srcFilter) outVec= sws_getConvVec(srcFilter, &scaleFilter); |
1122 else outVec= &scaleFilter; | 1122 else outVec= &scaleFilter; |
1123 | 1123 |
1124 ASSERT(outVec->length == filter2Size) | 1124 ASSERT(outVec->length == filter2Size) |
1125 //FIXME dstFilter | 1125 //FIXME dstFilter |
1126 | 1126 |
1127 for(j=0; j<outVec->length; j++) | 1127 for (j=0; j<outVec->length; j++) |
1128 { | 1128 { |
1129 filter2[i*filter2Size + j]= outVec->coeff[j]; | 1129 filter2[i*filter2Size + j]= outVec->coeff[j]; |
1130 } | 1130 } |
1131 | 1131 |
1132 (*filterPos)[i]+= (filterSize-1)/2 - (filter2Size-1)/2; | 1132 (*filterPos)[i]+= (filterSize-1)/2 - (filter2Size-1)/2; |
1133 | 1133 |
1134 if(outVec != &scaleFilter) sws_freeVec(outVec); | 1134 if (outVec != &scaleFilter) sws_freeVec(outVec); |
1135 } | 1135 } |
1136 av_free(filter); filter=NULL; | 1136 av_free(filter); filter=NULL; |
1137 | 1137 |
1138 /* try to reduce the filter-size (step1 find size and shift left) */ | 1138 /* try to reduce the filter-size (step1 find size and shift left) */ |
1139 // Assume its near normalized (*0.5 or *2.0 is ok but * 0.001 is not) | 1139 // Assume its near normalized (*0.5 or *2.0 is ok but * 0.001 is not) |
1140 minFilterSize= 0; | 1140 minFilterSize= 0; |
1141 for(i=dstW-1; i>=0; i--) | 1141 for (i=dstW-1; i>=0; i--) |
1142 { | 1142 { |
1143 int min= filter2Size; | 1143 int min= filter2Size; |
1144 int j; | 1144 int j; |
1145 double cutOff=0.0; | 1145 double cutOff=0.0; |
1146 | 1146 |
1147 /* get rid off near zero elements on the left by shifting left */ | 1147 /* get rid off near zero elements on the left by shifting left */ |
1148 for(j=0; j<filter2Size; j++) | 1148 for (j=0; j<filter2Size; j++) |
1149 { | 1149 { |
1150 int k; | 1150 int k; |
1151 cutOff += FFABS(filter2[i*filter2Size]); | 1151 cutOff += FFABS(filter2[i*filter2Size]); |
1152 | 1152 |
1153 if(cutOff > SWS_MAX_REDUCE_CUTOFF) break; | 1153 if (cutOff > SWS_MAX_REDUCE_CUTOFF) break; |
1154 | 1154 |
1155 /* preserve Monotonicity because the core can't handle the filter otherwise */ | 1155 /* preserve Monotonicity because the core can't handle the filter otherwise */ |
1156 if(i<dstW-1 && (*filterPos)[i] >= (*filterPos)[i+1]) break; | 1156 if (i<dstW-1 && (*filterPos)[i] >= (*filterPos)[i+1]) break; |
1157 | 1157 |
1158 // Move filter coeffs left | 1158 // Move filter coeffs left |
1159 for(k=1; k<filter2Size; k++) | 1159 for (k=1; k<filter2Size; k++) |
1160 filter2[i*filter2Size + k - 1]= filter2[i*filter2Size + k]; | 1160 filter2[i*filter2Size + k - 1]= filter2[i*filter2Size + k]; |
1161 filter2[i*filter2Size + k - 1]= 0.0; | 1161 filter2[i*filter2Size + k - 1]= 0.0; |
1162 (*filterPos)[i]++; | 1162 (*filterPos)[i]++; |
1163 } | 1163 } |
1164 | 1164 |
1165 cutOff=0.0; | 1165 cutOff=0.0; |
1166 /* count near zeros on the right */ | 1166 /* count near zeros on the right */ |
1167 for(j=filter2Size-1; j>0; j--) | 1167 for (j=filter2Size-1; j>0; j--) |
1168 { | 1168 { |
1169 cutOff += FFABS(filter2[i*filter2Size + j]); | 1169 cutOff += FFABS(filter2[i*filter2Size + j]); |
1170 | 1170 |
1171 if(cutOff > SWS_MAX_REDUCE_CUTOFF) break; | 1171 if (cutOff > SWS_MAX_REDUCE_CUTOFF) break; |
1172 min--; | 1172 min--; |
1173 } | 1173 } |
1174 | 1174 |
1175 if(min>minFilterSize) minFilterSize= min; | 1175 if (min>minFilterSize) minFilterSize= min; |
1176 } | 1176 } |
1177 | 1177 |
1178 if (flags & SWS_CPU_CAPS_ALTIVEC) { | 1178 if (flags & SWS_CPU_CAPS_ALTIVEC) { |
1179 // we can handle the special case 4, | 1179 // we can handle the special case 4, |
1180 // so we don't want to go to the full 8 | 1180 // so we don't want to go to the full 8 |
1181 if (minFilterSize < 5) | 1181 if (minFilterSize < 5) |
1182 filterAlign = 4; | 1182 filterAlign = 4; |
1183 | 1183 |
1184 // we really don't want to waste our time | 1184 // we really don't want to waste our time |
1185 // doing useless computation, so fall-back on | 1185 // doing useless computation, so fall-back on |
1186 // the scalar C code for very small filter. | 1186 // the scalar C code for very small filter. |
1187 // vectorizing is worth it only if you have | 1187 // vectorizing is worth it only if you have |
1188 // decent-sized vector. | 1188 // decent-sized vector. |
1189 if (minFilterSize < 3) | 1189 if (minFilterSize < 3) |
1190 filterAlign = 1; | 1190 filterAlign = 1; |
1191 } | 1191 } |
1192 | 1192 |
1193 if (flags & SWS_CPU_CAPS_MMX) { | 1193 if (flags & SWS_CPU_CAPS_MMX) { |
1194 // special case for unscaled vertical filtering | 1194 // special case for unscaled vertical filtering |
1195 if(minFilterSize == 1 && filterAlign == 2) | 1195 if (minFilterSize == 1 && filterAlign == 2) |
1196 filterAlign= 1; | 1196 filterAlign= 1; |
1197 } | 1197 } |
1198 | 1198 |
1199 ASSERT(minFilterSize > 0) | 1199 ASSERT(minFilterSize > 0) |
1200 filterSize= (minFilterSize +(filterAlign-1)) & (~(filterAlign-1)); | 1200 filterSize= (minFilterSize +(filterAlign-1)) & (~(filterAlign-1)); |
1201 ASSERT(filterSize > 0) | 1201 ASSERT(filterSize > 0) |
1202 filter= av_malloc(filterSize*dstW*sizeof(double)); | 1202 filter= av_malloc(filterSize*dstW*sizeof(double)); |
1203 if(filterSize >= MAX_FILTER_SIZE) | 1203 if (filterSize >= MAX_FILTER_SIZE) |
1204 return -1; | 1204 return -1; |
1205 *outFilterSize= filterSize; | 1205 *outFilterSize= filterSize; |
1206 | 1206 |
1207 if(flags&SWS_PRINT_INFO) | 1207 if (flags&SWS_PRINT_INFO) |
1208 av_log(NULL, AV_LOG_VERBOSE, "SwScaler: reducing / aligning filtersize %d -> %d\n", filter2Size, filterSize); | 1208 av_log(NULL, AV_LOG_VERBOSE, "SwScaler: reducing / aligning filtersize %d -> %d\n", filter2Size, filterSize); |
1209 /* try to reduce the filter-size (step2 reduce it) */ | 1209 /* try to reduce the filter-size (step2 reduce it) */ |
1210 for(i=0; i<dstW; i++) | 1210 for (i=0; i<dstW; i++) |
1211 { | 1211 { |
1212 int j; | 1212 int j; |
1213 | 1213 |
1214 for(j=0; j<filterSize; j++) | 1214 for (j=0; j<filterSize; j++) |
1215 { | 1215 { |
1216 if(j>=filter2Size) filter[i*filterSize + j]= 0.0; | 1216 if (j>=filter2Size) filter[i*filterSize + j]= 0.0; |
1217 else filter[i*filterSize + j]= filter2[i*filter2Size + j]; | 1217 else filter[i*filterSize + j]= filter2[i*filter2Size + j]; |
1218 } | 1218 } |
1219 } | 1219 } |
1220 av_free(filter2); filter2=NULL; | 1220 av_free(filter2); filter2=NULL; |
1221 | 1221 |
1222 | 1222 |
1223 //FIXME try to align filterpos if possible | 1223 //FIXME try to align filterpos if possible |
1224 | 1224 |
1225 //fix borders | 1225 //fix borders |
1226 for(i=0; i<dstW; i++) | 1226 for (i=0; i<dstW; i++) |
1227 { | 1227 { |
1228 int j; | 1228 int j; |
1229 if((*filterPos)[i] < 0) | 1229 if ((*filterPos)[i] < 0) |
1230 { | 1230 { |
1231 // Move filter coeffs left to compensate for filterPos | 1231 // Move filter coeffs left to compensate for filterPos |
1232 for(j=1; j<filterSize; j++) | 1232 for (j=1; j<filterSize; j++) |
1233 { | 1233 { |
1234 int left= FFMAX(j + (*filterPos)[i], 0); | 1234 int left= FFMAX(j + (*filterPos)[i], 0); |
1235 filter[i*filterSize + left] += filter[i*filterSize + j]; | 1235 filter[i*filterSize + left] += filter[i*filterSize + j]; |
1236 filter[i*filterSize + j]=0; | 1236 filter[i*filterSize + j]=0; |
1237 } | 1237 } |
1238 (*filterPos)[i]= 0; | 1238 (*filterPos)[i]= 0; |
1239 } | 1239 } |
1240 | 1240 |
1241 if((*filterPos)[i] + filterSize > srcW) | 1241 if ((*filterPos)[i] + filterSize > srcW) |
1242 { | 1242 { |
1243 int shift= (*filterPos)[i] + filterSize - srcW; | 1243 int shift= (*filterPos)[i] + filterSize - srcW; |
1244 // Move filter coeffs right to compensate for filterPos | 1244 // Move filter coeffs right to compensate for filterPos |
1245 for(j=filterSize-2; j>=0; j--) | 1245 for (j=filterSize-2; j>=0; j--) |
1246 { | 1246 { |
1247 int right= FFMIN(j + shift, filterSize-1); | 1247 int right= FFMIN(j + shift, filterSize-1); |
1248 filter[i*filterSize +right] += filter[i*filterSize +j]; | 1248 filter[i*filterSize +right] += filter[i*filterSize +j]; |
1249 filter[i*filterSize +j]=0; | 1249 filter[i*filterSize +j]=0; |
1250 } | 1250 } |
1251 (*filterPos)[i]= srcW - filterSize; | 1251 (*filterPos)[i]= srcW - filterSize; |
1252 } | 1252 } |
1253 } | 1253 } |
1254 | 1254 |
1255 // Note the +1 is for the MMXscaler which reads over the end | 1255 // Note the +1 is for the MMXscaler which reads over the end |
1256 /* align at 16 for AltiVec (needed by hScale_altivec_real) */ | 1256 /* align at 16 for AltiVec (needed by hScale_altivec_real) */ |
1257 *outFilter= av_mallocz(*outFilterSize*(dstW+1)*sizeof(int16_t)); | 1257 *outFilter= av_mallocz(*outFilterSize*(dstW+1)*sizeof(int16_t)); |
1258 | 1258 |
1259 /* Normalize & Store in outFilter */ | 1259 /* Normalize & Store in outFilter */ |
1260 for(i=0; i<dstW; i++) | 1260 for (i=0; i<dstW; i++) |
1261 { | 1261 { |
1262 int j; | 1262 int j; |
1263 double error=0; | 1263 double error=0; |
1264 double sum=0; | 1264 double sum=0; |
1265 double scale= one; | 1265 double scale= one; |
1266 | 1266 |
1267 for(j=0; j<filterSize; j++) | 1267 for (j=0; j<filterSize; j++) |
1268 { | 1268 { |
1269 sum+= filter[i*filterSize + j]; | 1269 sum+= filter[i*filterSize + j]; |
1270 } | 1270 } |
1271 scale/= sum; | 1271 scale/= sum; |
1272 for(j=0; j<*outFilterSize; j++) | 1272 for (j=0; j<*outFilterSize; j++) |
1273 { | 1273 { |
1274 double v= filter[i*filterSize + j]*scale + error; | 1274 double v= filter[i*filterSize + j]*scale + error; |
1275 int intV= floor(v + 0.5); | 1275 int intV= floor(v + 0.5); |
1276 (*outFilter)[i*(*outFilterSize) + j]= intV; | 1276 (*outFilter)[i*(*outFilterSize) + j]= intV; |
1277 error = v - intV; | 1277 error = v - intV; |
1278 } | 1278 } |
1279 } | 1279 } |
1280 | 1280 |
1281 (*filterPos)[dstW]= (*filterPos)[dstW-1]; // the MMX scaler will read over the end | 1281 (*filterPos)[dstW]= (*filterPos)[dstW-1]; // the MMX scaler will read over the end |
1282 for(i=0; i<*outFilterSize; i++) | 1282 for (i=0; i<*outFilterSize; i++) |
1283 { | 1283 { |
1284 int j= dstW*(*outFilterSize); | 1284 int j= dstW*(*outFilterSize); |
1285 (*outFilter)[j + i]= (*outFilter)[j + i - (*outFilterSize)]; | 1285 (*outFilter)[j + i]= (*outFilter)[j + i - (*outFilterSize)]; |
1286 } | 1286 } |
1287 | 1287 |
1288 av_free(filter); | 1288 av_free(filter); |
1289 return 0; | 1289 return 0; |
1290 } | 1290 } |
1291 | 1291 |
1292 #ifdef COMPILE_MMX2 | 1292 #ifdef COMPILE_MMX2 |
1293 static void initMMX2HScaler(int dstW, int xInc, uint8_t *funnyCode, int16_t *filter, int32_t *filterPos, int numSplits) | 1293 static void initMMX2HScaler(int dstW, int xInc, uint8_t *funnyCode, int16_t *filter, int32_t *filterPos, int numSplits) |
1294 { | 1294 { |
1295 uint8_t *fragmentA; | 1295 uint8_t *fragmentA; |
1296 long imm8OfPShufW1A; | 1296 long imm8OfPShufW1A; |
1297 long imm8OfPShufW2A; | 1297 long imm8OfPShufW2A; |
1298 long fragmentLengthA; | 1298 long fragmentLengthA; |
1299 uint8_t *fragmentB; | 1299 uint8_t *fragmentB; |
1300 long imm8OfPShufW1B; | 1300 long imm8OfPShufW1B; |
1301 long imm8OfPShufW2B; | 1301 long imm8OfPShufW2B; |
1302 long fragmentLengthB; | 1302 long fragmentLengthB; |
1303 int fragmentPos; | 1303 int fragmentPos; |
1304 | 1304 |
1305 int xpos, i; | 1305 int xpos, i; |
1306 | 1306 |
1307 // create an optimized horizontal scaling routine | 1307 // create an optimized horizontal scaling routine |
1308 | 1308 |
1309 //code fragment | 1309 //code fragment |
1310 | 1310 |
1311 asm volatile( | 1311 asm volatile( |
1312 "jmp 9f \n\t" | 1312 "jmp 9f \n\t" |
1313 // Begin | 1313 // Begin |
1314 "0: \n\t" | 1314 "0: \n\t" |
1315 "movq (%%"REG_d", %%"REG_a"), %%mm3\n\t" | 1315 "movq (%%"REG_d", %%"REG_a"), %%mm3 \n\t" |
1316 "movd (%%"REG_c", %%"REG_S"), %%mm0\n\t" | 1316 "movd (%%"REG_c", %%"REG_S"), %%mm0 \n\t" |
1317 "movd 1(%%"REG_c", %%"REG_S"), %%mm1\n\t" | 1317 "movd 1(%%"REG_c", %%"REG_S"), %%mm1 \n\t" |
1318 "punpcklbw %%mm7, %%mm1 \n\t" | 1318 "punpcklbw %%mm7, %%mm1 \n\t" |
1319 "punpcklbw %%mm7, %%mm0 \n\t" | 1319 "punpcklbw %%mm7, %%mm0 \n\t" |
1320 "pshufw $0xFF, %%mm1, %%mm1 \n\t" | 1320 "pshufw $0xFF, %%mm1, %%mm1 \n\t" |
1321 "1: \n\t" | 1321 "1: \n\t" |
1322 "pshufw $0xFF, %%mm0, %%mm0 \n\t" | 1322 "pshufw $0xFF, %%mm0, %%mm0 \n\t" |
1323 "2: \n\t" | 1323 "2: \n\t" |
1324 "psubw %%mm1, %%mm0 \n\t" | 1324 "psubw %%mm1, %%mm0 \n\t" |
1325 "movl 8(%%"REG_b", %%"REG_a"), %%esi\n\t" | 1325 "movl 8(%%"REG_b", %%"REG_a"), %%esi \n\t" |
1326 "pmullw %%mm3, %%mm0 \n\t" | 1326 "pmullw %%mm3, %%mm0 \n\t" |
1327 "psllw $7, %%mm1 \n\t" | 1327 "psllw $7, %%mm1 \n\t" |
1328 "paddw %%mm1, %%mm0 \n\t" | 1328 "paddw %%mm1, %%mm0 \n\t" |
1329 | 1329 |
1330 "movq %%mm0, (%%"REG_D", %%"REG_a")\n\t" | 1330 "movq %%mm0, (%%"REG_D", %%"REG_a") \n\t" |
1331 | 1331 |
1332 "add $8, %%"REG_a" \n\t" | 1332 "add $8, %%"REG_a" \n\t" |
1333 // End | 1333 // End |
1334 "9: \n\t" | 1334 "9: \n\t" |
1335 // "int $3\n\t" | 1335 // "int $3 \n\t" |
1336 "lea 0b, %0 \n\t" | 1336 "lea 0b, %0 \n\t" |
1337 "lea 1b, %1 \n\t" | 1337 "lea 1b, %1 \n\t" |
1338 "lea 2b, %2 \n\t" | 1338 "lea 2b, %2 \n\t" |
1339 "dec %1 \n\t" | 1339 "dec %1 \n\t" |
1340 "dec %2 \n\t" | 1340 "dec %2 \n\t" |
1341 "sub %0, %1 \n\t" | 1341 "sub %0, %1 \n\t" |
1342 "sub %0, %2 \n\t" | 1342 "sub %0, %2 \n\t" |
1343 "lea 9b, %3 \n\t" | 1343 "lea 9b, %3 \n\t" |
1344 "sub %0, %3 \n\t" | 1344 "sub %0, %3 \n\t" |
1345 | 1345 |
1346 | 1346 |
1347 :"=r" (fragmentA), "=r" (imm8OfPShufW1A), "=r" (imm8OfPShufW2A), | 1347 :"=r" (fragmentA), "=r" (imm8OfPShufW1A), "=r" (imm8OfPShufW2A), |
1348 "=r" (fragmentLengthA) | 1348 "=r" (fragmentLengthA) |
1349 ); | 1349 ); |
1350 | 1350 |
1351 asm volatile( | 1351 asm volatile( |
1352 "jmp 9f \n\t" | 1352 "jmp 9f \n\t" |
1353 // Begin | 1353 // Begin |
1354 "0: \n\t" | 1354 "0: \n\t" |
1355 "movq (%%"REG_d", %%"REG_a"), %%mm3\n\t" | 1355 "movq (%%"REG_d", %%"REG_a"), %%mm3 \n\t" |
1356 "movd (%%"REG_c", %%"REG_S"), %%mm0\n\t" | 1356 "movd (%%"REG_c", %%"REG_S"), %%mm0 \n\t" |
1357 "punpcklbw %%mm7, %%mm0 \n\t" | 1357 "punpcklbw %%mm7, %%mm0 \n\t" |
1358 "pshufw $0xFF, %%mm0, %%mm1 \n\t" | 1358 "pshufw $0xFF, %%mm0, %%mm1 \n\t" |
1359 "1: \n\t" | 1359 "1: \n\t" |
1360 "pshufw $0xFF, %%mm0, %%mm0 \n\t" | 1360 "pshufw $0xFF, %%mm0, %%mm0 \n\t" |
1361 "2: \n\t" | 1361 "2: \n\t" |
1362 "psubw %%mm1, %%mm0 \n\t" | 1362 "psubw %%mm1, %%mm0 \n\t" |
1363 "movl 8(%%"REG_b", %%"REG_a"), %%esi\n\t" | 1363 "movl 8(%%"REG_b", %%"REG_a"), %%esi \n\t" |
1364 "pmullw %%mm3, %%mm0 \n\t" | 1364 "pmullw %%mm3, %%mm0 \n\t" |
1365 "psllw $7, %%mm1 \n\t" | 1365 "psllw $7, %%mm1 \n\t" |
1366 "paddw %%mm1, %%mm0 \n\t" | 1366 "paddw %%mm1, %%mm0 \n\t" |
1367 | 1367 |
1368 "movq %%mm0, (%%"REG_D", %%"REG_a")\n\t" | 1368 "movq %%mm0, (%%"REG_D", %%"REG_a") \n\t" |
1369 | 1369 |
1370 "add $8, %%"REG_a" \n\t" | 1370 "add $8, %%"REG_a" \n\t" |
1371 // End | 1371 // End |
1372 "9: \n\t" | 1372 "9: \n\t" |
1373 // "int $3\n\t" | 1373 // "int $3 \n\t" |
1374 "lea 0b, %0 \n\t" | 1374 "lea 0b, %0 \n\t" |
1375 "lea 1b, %1 \n\t" | 1375 "lea 1b, %1 \n\t" |
1376 "lea 2b, %2 \n\t" | 1376 "lea 2b, %2 \n\t" |
1377 "dec %1 \n\t" | 1377 "dec %1 \n\t" |
1378 "dec %2 \n\t" | 1378 "dec %2 \n\t" |
1379 "sub %0, %1 \n\t" | 1379 "sub %0, %1 \n\t" |
1380 "sub %0, %2 \n\t" | 1380 "sub %0, %2 \n\t" |
1381 "lea 9b, %3 \n\t" | 1381 "lea 9b, %3 \n\t" |
1382 "sub %0, %3 \n\t" | 1382 "sub %0, %3 \n\t" |
1383 | 1383 |
1384 | 1384 |
1385 :"=r" (fragmentB), "=r" (imm8OfPShufW1B), "=r" (imm8OfPShufW2B), | 1385 :"=r" (fragmentB), "=r" (imm8OfPShufW1B), "=r" (imm8OfPShufW2B), |
1386 "=r" (fragmentLengthB) | 1386 "=r" (fragmentLengthB) |
1387 ); | 1387 ); |
1388 | 1388 |
1389 xpos= 0; //lumXInc/2 - 0x8000; // difference between pixel centers | 1389 xpos= 0; //lumXInc/2 - 0x8000; // difference between pixel centers |
1390 fragmentPos=0; | 1390 fragmentPos=0; |
1391 | 1391 |
1392 for(i=0; i<dstW/numSplits; i++) | 1392 for (i=0; i<dstW/numSplits; i++) |
1393 { | 1393 { |
1394 int xx=xpos>>16; | 1394 int xx=xpos>>16; |
1395 | 1395 |
1396 if((i&3) == 0) | 1396 if ((i&3) == 0) |
1397 { | 1397 { |
1398 int a=0; | 1398 int a=0; |
1399 int b=((xpos+xInc)>>16) - xx; | 1399 int b=((xpos+xInc)>>16) - xx; |
1400 int c=((xpos+xInc*2)>>16) - xx; | 1400 int c=((xpos+xInc*2)>>16) - xx; |
1401 int d=((xpos+xInc*3)>>16) - xx; | 1401 int d=((xpos+xInc*3)>>16) - xx; |
1402 | 1402 |
1403 filter[i ] = (( xpos & 0xFFFF) ^ 0xFFFF)>>9; | 1403 filter[i ] = (( xpos & 0xFFFF) ^ 0xFFFF)>>9; |
1404 filter[i+1] = (((xpos+xInc ) & 0xFFFF) ^ 0xFFFF)>>9; | 1404 filter[i+1] = (((xpos+xInc ) & 0xFFFF) ^ 0xFFFF)>>9; |
1405 filter[i+2] = (((xpos+xInc*2) & 0xFFFF) ^ 0xFFFF)>>9; | 1405 filter[i+2] = (((xpos+xInc*2) & 0xFFFF) ^ 0xFFFF)>>9; |
1406 filter[i+3] = (((xpos+xInc*3) & 0xFFFF) ^ 0xFFFF)>>9; | 1406 filter[i+3] = (((xpos+xInc*3) & 0xFFFF) ^ 0xFFFF)>>9; |
1407 filterPos[i/2]= xx; | 1407 filterPos[i/2]= xx; |
1408 | 1408 |
1409 if(d+1<4) | 1409 if (d+1<4) |
1410 { | 1410 { |
1411 int maxShift= 3-(d+1); | 1411 int maxShift= 3-(d+1); |
1412 int shift=0; | 1412 int shift=0; |
1413 | 1413 |
1414 memcpy(funnyCode + fragmentPos, fragmentB, fragmentLengthB); | 1414 memcpy(funnyCode + fragmentPos, fragmentB, fragmentLengthB); |
1415 | 1415 |
1416 funnyCode[fragmentPos + imm8OfPShufW1B]= | 1416 funnyCode[fragmentPos + imm8OfPShufW1B]= |
1417 (a+1) | ((b+1)<<2) | ((c+1)<<4) | ((d+1)<<6); | 1417 (a+1) | ((b+1)<<2) | ((c+1)<<4) | ((d+1)<<6); |
1418 funnyCode[fragmentPos + imm8OfPShufW2B]= | 1418 funnyCode[fragmentPos + imm8OfPShufW2B]= |
1419 a | (b<<2) | (c<<4) | (d<<6); | 1419 a | (b<<2) | (c<<4) | (d<<6); |
1420 | 1420 |
1421 if(i+3>=dstW) shift=maxShift; //avoid overread | 1421 if (i+3>=dstW) shift=maxShift; //avoid overread |
1422 else if((filterPos[i/2]&3) <= maxShift) shift=filterPos[i/2]&3; //Align | 1422 else if ((filterPos[i/2]&3) <= maxShift) shift=filterPos[i/2]&3; //Align |
1423 | 1423 |
1424 if(shift && i>=shift) | 1424 if (shift && i>=shift) |
1425 { | 1425 { |
1426 funnyCode[fragmentPos + imm8OfPShufW1B]+= 0x55*shift; | 1426 funnyCode[fragmentPos + imm8OfPShufW1B]+= 0x55*shift; |
1427 funnyCode[fragmentPos + imm8OfPShufW2B]+= 0x55*shift; | 1427 funnyCode[fragmentPos + imm8OfPShufW2B]+= 0x55*shift; |
1428 filterPos[i/2]-=shift; | 1428 filterPos[i/2]-=shift; |
1429 } | 1429 } |
1430 | 1430 |
1431 fragmentPos+= fragmentLengthB; | 1431 fragmentPos+= fragmentLengthB; |
1432 } | 1432 } |
1433 else | 1433 else |
1434 { | 1434 { |
1435 int maxShift= 3-d; | 1435 int maxShift= 3-d; |
1436 int shift=0; | 1436 int shift=0; |
1437 | 1437 |
1438 memcpy(funnyCode + fragmentPos, fragmentA, fragmentLengthA); | 1438 memcpy(funnyCode + fragmentPos, fragmentA, fragmentLengthA); |
1439 | 1439 |
1440 funnyCode[fragmentPos + imm8OfPShufW1A]= | 1440 funnyCode[fragmentPos + imm8OfPShufW1A]= |
1441 funnyCode[fragmentPos + imm8OfPShufW2A]= | 1441 funnyCode[fragmentPos + imm8OfPShufW2A]= |
1442 a | (b<<2) | (c<<4) | (d<<6); | 1442 a | (b<<2) | (c<<4) | (d<<6); |
1443 | 1443 |
1444 if(i+4>=dstW) shift=maxShift; //avoid overread | 1444 if (i+4>=dstW) shift=maxShift; //avoid overread |
1445 else if((filterPos[i/2]&3) <= maxShift) shift=filterPos[i/2]&3; //partial align | 1445 else if ((filterPos[i/2]&3) <= maxShift) shift=filterPos[i/2]&3; //partial align |
1446 | 1446 |
1447 if(shift && i>=shift) | 1447 if (shift && i>=shift) |
1448 { | 1448 { |
1449 funnyCode[fragmentPos + imm8OfPShufW1A]+= 0x55*shift; | 1449 funnyCode[fragmentPos + imm8OfPShufW1A]+= 0x55*shift; |
1450 funnyCode[fragmentPos + imm8OfPShufW2A]+= 0x55*shift; | 1450 funnyCode[fragmentPos + imm8OfPShufW2A]+= 0x55*shift; |
1451 filterPos[i/2]-=shift; | 1451 filterPos[i/2]-=shift; |
1452 } | 1452 } |
1453 | 1453 |
1454 fragmentPos+= fragmentLengthA; | 1454 fragmentPos+= fragmentLengthA; |
1455 } | 1455 } |
1456 | 1456 |
1457 funnyCode[fragmentPos]= RET; | 1457 funnyCode[fragmentPos]= RET; |
1458 } | 1458 } |
1459 xpos+=xInc; | 1459 xpos+=xInc; |
1460 } | 1460 } |
1461 filterPos[i/2]= xpos>>16; // needed to jump to the next part | 1461 filterPos[i/2]= xpos>>16; // needed to jump to the next part |
1462 } | 1462 } |
1463 #endif /* COMPILE_MMX2 */ | 1463 #endif /* COMPILE_MMX2 */ |
1464 | 1464 |
1465 static void globalInit(void){ | 1465 static void globalInit(void){ |
1466 // generating tables: | 1466 // generating tables: |
1467 int i; | 1467 int i; |
1468 for(i=0; i<768; i++){ | 1468 for (i=0; i<768; i++){ |
1469 int c= av_clip_uint8(i-256); | 1469 int c= av_clip_uint8(i-256); |
1470 clip_table[i]=c; | 1470 clip_table[i]=c; |
1471 } | 1471 } |
1472 } | 1472 } |
1473 | 1473 |
1474 static SwsFunc getSwsFunc(int flags){ | 1474 static SwsFunc getSwsFunc(int flags){ |
1475 | 1475 |
1476 #if defined(RUNTIME_CPUDETECT) && defined (CONFIG_GPL) | 1476 #if defined(RUNTIME_CPUDETECT) && defined (CONFIG_GPL) |
1477 #if defined(ARCH_X86) | 1477 #if defined(ARCH_X86) |
1478 // ordered per speed fasterst first | 1478 // ordered per speed fasterst first |
1479 if(flags & SWS_CPU_CAPS_MMX2) | 1479 if (flags & SWS_CPU_CAPS_MMX2) |
1480 return swScale_MMX2; | 1480 return swScale_MMX2; |
1481 else if(flags & SWS_CPU_CAPS_3DNOW) | 1481 else if (flags & SWS_CPU_CAPS_3DNOW) |
1482 return swScale_3DNow; | 1482 return swScale_3DNow; |
1483 else if(flags & SWS_CPU_CAPS_MMX) | 1483 else if (flags & SWS_CPU_CAPS_MMX) |
1484 return swScale_MMX; | 1484 return swScale_MMX; |
1485 else | 1485 else |
1486 return swScale_C; | 1486 return swScale_C; |
1487 | 1487 |
1488 #else | 1488 #else |
1489 #ifdef ARCH_POWERPC | 1489 #ifdef ARCH_POWERPC |
1490 if(flags & SWS_CPU_CAPS_ALTIVEC) | 1490 if (flags & SWS_CPU_CAPS_ALTIVEC) |
1491 return swScale_altivec; | 1491 return swScale_altivec; |
1492 else | 1492 else |
1493 return swScale_C; | 1493 return swScale_C; |
1494 #endif | 1494 #endif |
1495 return swScale_C; | 1495 return swScale_C; |
1496 #endif /* defined(ARCH_X86) */ | 1496 #endif /* defined(ARCH_X86) */ |
1497 #else //RUNTIME_CPUDETECT | 1497 #else //RUNTIME_CPUDETECT |
1498 #ifdef HAVE_MMX2 | 1498 #ifdef HAVE_MMX2 |
1499 return swScale_MMX2; | 1499 return swScale_MMX2; |
1500 #elif defined (HAVE_3DNOW) | 1500 #elif defined (HAVE_3DNOW) |
1501 return swScale_3DNow; | 1501 return swScale_3DNow; |
1502 #elif defined (HAVE_MMX) | 1502 #elif defined (HAVE_MMX) |
1503 return swScale_MMX; | 1503 return swScale_MMX; |
1504 #elif defined (HAVE_ALTIVEC) | 1504 #elif defined (HAVE_ALTIVEC) |
1505 return swScale_altivec; | 1505 return swScale_altivec; |
1506 #else | 1506 #else |
1507 return swScale_C; | 1507 return swScale_C; |
1508 #endif | 1508 #endif |
1509 #endif //!RUNTIME_CPUDETECT | 1509 #endif //!RUNTIME_CPUDETECT |
1510 } | 1510 } |
1511 | 1511 |
1512 static int PlanarToNV12Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1512 static int PlanarToNV12Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1513 int srcSliceH, uint8_t* dstParam[], int dstStride[]){ | 1513 int srcSliceH, uint8_t* dstParam[], int dstStride[]){ |
1514 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY; | 1514 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY; |
1515 /* Copy Y plane */ | 1515 /* Copy Y plane */ |
1516 if(dstStride[0]==srcStride[0] && srcStride[0] > 0) | 1516 if (dstStride[0]==srcStride[0] && srcStride[0] > 0) |
1517 memcpy(dst, src[0], srcSliceH*dstStride[0]); | 1517 memcpy(dst, src[0], srcSliceH*dstStride[0]); |
1518 else | 1518 else |
1519 { | 1519 { |
1520 int i; | 1520 int i; |
1521 uint8_t *srcPtr= src[0]; | 1521 uint8_t *srcPtr= src[0]; |
1522 uint8_t *dstPtr= dst; | 1522 uint8_t *dstPtr= dst; |
1523 for(i=0; i<srcSliceH; i++) | 1523 for (i=0; i<srcSliceH; i++) |
1524 { | 1524 { |
1525 memcpy(dstPtr, srcPtr, c->srcW); | 1525 memcpy(dstPtr, srcPtr, c->srcW); |
1526 srcPtr+= srcStride[0]; | 1526 srcPtr+= srcStride[0]; |
1527 dstPtr+= dstStride[0]; | 1527 dstPtr+= dstStride[0]; |
1528 } | 1528 } |
1529 } | 1529 } |
1530 dst = dstParam[1] + dstStride[1]*srcSliceY/2; | 1530 dst = dstParam[1] + dstStride[1]*srcSliceY/2; |
1531 if (c->dstFormat == PIX_FMT_NV12) | 1531 if (c->dstFormat == PIX_FMT_NV12) |
1532 interleaveBytes( src[1],src[2],dst,c->srcW/2,srcSliceH/2,srcStride[1],srcStride[2],dstStride[0] ); | 1532 interleaveBytes( src[1],src[2],dst,c->srcW/2,srcSliceH/2,srcStride[1],srcStride[2],dstStride[0] ); |
1533 else | 1533 else |
1534 interleaveBytes( src[2],src[1],dst,c->srcW/2,srcSliceH/2,srcStride[2],srcStride[1],dstStride[0] ); | 1534 interleaveBytes( src[2],src[1],dst,c->srcW/2,srcSliceH/2,srcStride[2],srcStride[1],dstStride[0] ); |
1535 | 1535 |
1536 return srcSliceH; | 1536 return srcSliceH; |
1537 } | 1537 } |
1538 | 1538 |
1539 static int PlanarToYuy2Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1539 static int PlanarToYuy2Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1540 int srcSliceH, uint8_t* dstParam[], int dstStride[]){ | 1540 int srcSliceH, uint8_t* dstParam[], int dstStride[]){ |
1541 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY; | 1541 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY; |
1542 | 1542 |
1543 yv12toyuy2( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] ); | 1543 yv12toyuy2( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] ); |
1544 | 1544 |
1545 return srcSliceH; | 1545 return srcSliceH; |
1546 } | 1546 } |
1547 | 1547 |
1548 static int PlanarToUyvyWrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1548 static int PlanarToUyvyWrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1549 int srcSliceH, uint8_t* dstParam[], int dstStride[]){ | 1549 int srcSliceH, uint8_t* dstParam[], int dstStride[]){ |
1550 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY; | 1550 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY; |
1551 | 1551 |
1552 yv12touyvy( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] ); | 1552 yv12touyvy( src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0] ); |
1553 | 1553 |
1554 return srcSliceH; | 1554 return srcSliceH; |
1555 } | 1555 } |
1556 | 1556 |
1557 /* {RGB,BGR}{15,16,24,32} -> {RGB,BGR}{15,16,24,32} */ | 1557 /* {RGB,BGR}{15,16,24,32} -> {RGB,BGR}{15,16,24,32} */ |
1558 static int rgb2rgbWrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1558 static int rgb2rgbWrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1559 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 1559 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
1560 const int srcFormat= c->srcFormat; | 1560 const int srcFormat= c->srcFormat; |
1561 const int dstFormat= c->dstFormat; | 1561 const int dstFormat= c->dstFormat; |
1562 const int srcBpp= (fmt_depth(srcFormat) + 7) >> 3; | 1562 const int srcBpp= (fmt_depth(srcFormat) + 7) >> 3; |
1563 const int dstBpp= (fmt_depth(dstFormat) + 7) >> 3; | 1563 const int dstBpp= (fmt_depth(dstFormat) + 7) >> 3; |
1564 const int srcId= fmt_depth(srcFormat) >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */ | 1564 const int srcId= fmt_depth(srcFormat) >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */ |
1565 const int dstId= fmt_depth(dstFormat) >> 2; | 1565 const int dstId= fmt_depth(dstFormat) >> 2; |
1566 void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL; | 1566 void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL; |
1567 | 1567 |
1568 /* BGR -> BGR */ | 1568 /* BGR -> BGR */ |
1569 if( (isBGR(srcFormat) && isBGR(dstFormat)) | 1569 if ( (isBGR(srcFormat) && isBGR(dstFormat)) |
1570 || (isRGB(srcFormat) && isRGB(dstFormat))){ | 1570 || (isRGB(srcFormat) && isRGB(dstFormat))){ |
1571 switch(srcId | (dstId<<4)){ | 1571 switch(srcId | (dstId<<4)){ |
1572 case 0x34: conv= rgb16to15; break; | 1572 case 0x34: conv= rgb16to15; break; |
1573 case 0x36: conv= rgb24to15; break; | 1573 case 0x36: conv= rgb24to15; break; |
1574 case 0x38: conv= rgb32to15; break; | 1574 case 0x38: conv= rgb32to15; break; |
1575 case 0x43: conv= rgb15to16; break; | 1575 case 0x43: conv= rgb15to16; break; |
1576 case 0x46: conv= rgb24to16; break; | 1576 case 0x46: conv= rgb24to16; break; |
1577 case 0x48: conv= rgb32to16; break; | 1577 case 0x48: conv= rgb32to16; break; |
1578 case 0x63: conv= rgb15to24; break; | 1578 case 0x63: conv= rgb15to24; break; |
1579 case 0x64: conv= rgb16to24; break; | 1579 case 0x64: conv= rgb16to24; break; |
1580 case 0x68: conv= rgb32to24; break; | 1580 case 0x68: conv= rgb32to24; break; |
1581 case 0x83: conv= rgb15to32; break; | 1581 case 0x83: conv= rgb15to32; break; |
1582 case 0x84: conv= rgb16to32; break; | 1582 case 0x84: conv= rgb16to32; break; |
1583 case 0x86: conv= rgb24to32; break; | 1583 case 0x86: conv= rgb24to32; break; |
1584 default: av_log(c, AV_LOG_ERROR, "swScaler: internal error %s -> %s converter\n", | 1584 default: av_log(c, AV_LOG_ERROR, "swScaler: internal error %s -> %s converter\n", |
1585 sws_format_name(srcFormat), sws_format_name(dstFormat)); break; | 1585 sws_format_name(srcFormat), sws_format_name(dstFormat)); break; |
1586 } | 1586 } |
1587 }else if( (isBGR(srcFormat) && isRGB(dstFormat)) | 1587 }else if ( (isBGR(srcFormat) && isRGB(dstFormat)) |
1588 || (isRGB(srcFormat) && isBGR(dstFormat))){ | 1588 || (isRGB(srcFormat) && isBGR(dstFormat))){ |
1589 switch(srcId | (dstId<<4)){ | 1589 switch(srcId | (dstId<<4)){ |
1590 case 0x33: conv= rgb15tobgr15; break; | 1590 case 0x33: conv= rgb15tobgr15; break; |
1591 case 0x34: conv= rgb16tobgr15; break; | 1591 case 0x34: conv= rgb16tobgr15; break; |
1592 case 0x36: conv= rgb24tobgr15; break; | 1592 case 0x36: conv= rgb24tobgr15; break; |
1593 case 0x38: conv= rgb32tobgr15; break; | 1593 case 0x38: conv= rgb32tobgr15; break; |
1594 case 0x43: conv= rgb15tobgr16; break; | 1594 case 0x43: conv= rgb15tobgr16; break; |
1595 case 0x44: conv= rgb16tobgr16; break; | 1595 case 0x44: conv= rgb16tobgr16; break; |
1596 case 0x46: conv= rgb24tobgr16; break; | 1596 case 0x46: conv= rgb24tobgr16; break; |
1597 case 0x48: conv= rgb32tobgr16; break; | 1597 case 0x48: conv= rgb32tobgr16; break; |
1598 case 0x63: conv= rgb15tobgr24; break; | 1598 case 0x63: conv= rgb15tobgr24; break; |
1599 case 0x64: conv= rgb16tobgr24; break; | 1599 case 0x64: conv= rgb16tobgr24; break; |
1600 case 0x66: conv= rgb24tobgr24; break; | 1600 case 0x66: conv= rgb24tobgr24; break; |
1601 case 0x68: conv= rgb32tobgr24; break; | 1601 case 0x68: conv= rgb32tobgr24; break; |
1602 case 0x83: conv= rgb15tobgr32; break; | 1602 case 0x83: conv= rgb15tobgr32; break; |
1603 case 0x84: conv= rgb16tobgr32; break; | 1603 case 0x84: conv= rgb16tobgr32; break; |
1604 case 0x86: conv= rgb24tobgr32; break; | 1604 case 0x86: conv= rgb24tobgr32; break; |
1605 case 0x88: conv= rgb32tobgr32; break; | 1605 case 0x88: conv= rgb32tobgr32; break; |
1606 default: av_log(c, AV_LOG_ERROR, "swScaler: internal error %s -> %s converter\n", | 1606 default: av_log(c, AV_LOG_ERROR, "swScaler: internal error %s -> %s converter\n", |
1607 sws_format_name(srcFormat), sws_format_name(dstFormat)); break; | 1607 sws_format_name(srcFormat), sws_format_name(dstFormat)); break; |
1608 } | 1608 } |
1609 }else{ | 1609 }else{ |
1610 av_log(c, AV_LOG_ERROR, "swScaler: internal error %s -> %s converter\n", | 1610 av_log(c, AV_LOG_ERROR, "swScaler: internal error %s -> %s converter\n", |
1611 sws_format_name(srcFormat), sws_format_name(dstFormat)); | 1611 sws_format_name(srcFormat), sws_format_name(dstFormat)); |
1612 } | 1612 } |
1613 | 1613 |
1614 if(dstStride[0]*srcBpp == srcStride[0]*dstBpp) | 1614 if (dstStride[0]*srcBpp == srcStride[0]*dstBpp) |
1615 conv(src[0], dst[0] + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]); | 1615 conv(src[0], dst[0] + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]); |
1616 else | 1616 else |
1617 { | 1617 { |
1618 int i; | 1618 int i; |
1619 uint8_t *srcPtr= src[0]; | 1619 uint8_t *srcPtr= src[0]; |
1620 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY; | 1620 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY; |
1621 | 1621 |
1622 for(i=0; i<srcSliceH; i++) | 1622 for (i=0; i<srcSliceH; i++) |
1623 { | 1623 { |
1624 conv(srcPtr, dstPtr, c->srcW*srcBpp); | 1624 conv(srcPtr, dstPtr, c->srcW*srcBpp); |
1625 srcPtr+= srcStride[0]; | 1625 srcPtr+= srcStride[0]; |
1626 dstPtr+= dstStride[0]; | 1626 dstPtr+= dstStride[0]; |
1627 } | 1627 } |
1628 } | 1628 } |
1629 return srcSliceH; | 1629 return srcSliceH; |
1630 } | 1630 } |
1631 | 1631 |
1632 static int bgr24toyv12Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1632 static int bgr24toyv12Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1633 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 1633 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
1634 | 1634 |
1635 rgb24toyv12( | 1635 rgb24toyv12( |
1636 src[0], | 1636 src[0], |
1637 dst[0]+ srcSliceY *dstStride[0], | 1637 dst[0]+ srcSliceY *dstStride[0], |
1638 dst[1]+(srcSliceY>>1)*dstStride[1], | 1638 dst[1]+(srcSliceY>>1)*dstStride[1], |
1639 dst[2]+(srcSliceY>>1)*dstStride[2], | 1639 dst[2]+(srcSliceY>>1)*dstStride[2], |
1640 c->srcW, srcSliceH, | 1640 c->srcW, srcSliceH, |
1641 dstStride[0], dstStride[1], srcStride[0]); | 1641 dstStride[0], dstStride[1], srcStride[0]); |
1642 return srcSliceH; | 1642 return srcSliceH; |
1643 } | 1643 } |
1644 | 1644 |
1645 static int yvu9toyv12Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1645 static int yvu9toyv12Wrapper(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1646 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 1646 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
1647 int i; | 1647 int i; |
1648 | 1648 |
1649 /* copy Y */ | 1649 /* copy Y */ |
1650 if(srcStride[0]==dstStride[0] && srcStride[0] > 0) | 1650 if (srcStride[0]==dstStride[0] && srcStride[0] > 0) |
1651 memcpy(dst[0]+ srcSliceY*dstStride[0], src[0], srcStride[0]*srcSliceH); | 1651 memcpy(dst[0]+ srcSliceY*dstStride[0], src[0], srcStride[0]*srcSliceH); |
1652 else{ | 1652 else{ |
1653 uint8_t *srcPtr= src[0]; | 1653 uint8_t *srcPtr= src[0]; |
1654 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY; | 1654 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY; |
1655 | 1655 |
1656 for(i=0; i<srcSliceH; i++) | 1656 for (i=0; i<srcSliceH; i++) |
1657 { | 1657 { |
1658 memcpy(dstPtr, srcPtr, c->srcW); | 1658 memcpy(dstPtr, srcPtr, c->srcW); |
1659 srcPtr+= srcStride[0]; | 1659 srcPtr+= srcStride[0]; |
1660 dstPtr+= dstStride[0]; | 1660 dstPtr+= dstStride[0]; |
1661 } | 1661 } |
1662 } | 1662 } |
1663 | 1663 |
1664 if(c->dstFormat==PIX_FMT_YUV420P){ | 1664 if (c->dstFormat==PIX_FMT_YUV420P){ |
1665 planar2x(src[1], dst[1], c->chrSrcW, c->chrSrcH, srcStride[1], dstStride[1]); | 1665 planar2x(src[1], dst[1], c->chrSrcW, c->chrSrcH, srcStride[1], dstStride[1]); |
1666 planar2x(src[2], dst[2], c->chrSrcW, c->chrSrcH, srcStride[2], dstStride[2]); | 1666 planar2x(src[2], dst[2], c->chrSrcW, c->chrSrcH, srcStride[2], dstStride[2]); |
1667 }else{ | 1667 }else{ |
1668 planar2x(src[1], dst[2], c->chrSrcW, c->chrSrcH, srcStride[1], dstStride[2]); | 1668 planar2x(src[1], dst[2], c->chrSrcW, c->chrSrcH, srcStride[1], dstStride[2]); |
1669 planar2x(src[2], dst[1], c->chrSrcW, c->chrSrcH, srcStride[2], dstStride[1]); | 1669 planar2x(src[2], dst[1], c->chrSrcW, c->chrSrcH, srcStride[2], dstStride[1]); |
1670 } | 1670 } |
1671 return srcSliceH; | 1671 return srcSliceH; |
1672 } | 1672 } |
1673 | 1673 |
1674 /* unscaled copy like stuff (assumes nearly identical formats) */ | 1674 /* unscaled copy like stuff (assumes nearly identical formats) */ |
1675 static int simpleCopy(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1675 static int simpleCopy(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1676 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 1676 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
1677 | 1677 |
1678 if(isPacked(c->srcFormat)) | 1678 if (isPacked(c->srcFormat)) |
1679 { | 1679 { |
1680 if(dstStride[0]==srcStride[0] && srcStride[0] > 0) | 1680 if (dstStride[0]==srcStride[0] && srcStride[0] > 0) |
1681 memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]); | 1681 memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]); |
1682 else | 1682 else |
1683 { | 1683 { |
1684 int i; | 1684 int i; |
1685 uint8_t *srcPtr= src[0]; | 1685 uint8_t *srcPtr= src[0]; |
1686 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY; | 1686 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY; |
1687 int length=0; | 1687 int length=0; |
1688 | 1688 |
1689 /* universal length finder */ | 1689 /* universal length finder */ |
1690 while(length+c->srcW <= FFABS(dstStride[0]) | 1690 while(length+c->srcW <= FFABS(dstStride[0]) |
1691 && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW; | 1691 && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW; |
1692 ASSERT(length!=0); | 1692 ASSERT(length!=0); |
1693 | 1693 |
1694 for(i=0; i<srcSliceH; i++) | 1694 for (i=0; i<srcSliceH; i++) |
1695 { | 1695 { |
1696 memcpy(dstPtr, srcPtr, length); | 1696 memcpy(dstPtr, srcPtr, length); |
1697 srcPtr+= srcStride[0]; | 1697 srcPtr+= srcStride[0]; |
1698 dstPtr+= dstStride[0]; | 1698 dstPtr+= dstStride[0]; |
1699 } | 1699 } |
1700 } | 1700 } |
1701 } | 1701 } |
1702 else | 1702 else |
1703 { /* Planar YUV or gray */ | 1703 { /* Planar YUV or gray */ |
1704 int plane; | 1704 int plane; |
1705 for(plane=0; plane<3; plane++) | 1705 for (plane=0; plane<3; plane++) |
1706 { | 1706 { |
1707 int length= plane==0 ? c->srcW : -((-c->srcW )>>c->chrDstHSubSample); | 1707 int length= plane==0 ? c->srcW : -((-c->srcW )>>c->chrDstHSubSample); |
1708 int y= plane==0 ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample); | 1708 int y= plane==0 ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample); |
1709 int height= plane==0 ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample); | 1709 int height= plane==0 ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample); |
1710 | 1710 |
1711 if((isGray(c->srcFormat) || isGray(c->dstFormat)) && plane>0) | 1711 if ((isGray(c->srcFormat) || isGray(c->dstFormat)) && plane>0) |
1712 { | 1712 { |
1713 if(!isGray(c->dstFormat)) | 1713 if (!isGray(c->dstFormat)) |
1714 memset(dst[plane], 128, dstStride[plane]*height); | 1714 memset(dst[plane], 128, dstStride[plane]*height); |
1715 } | 1715 } |
1716 else | 1716 else |
1717 { | 1717 { |
1718 if(dstStride[plane]==srcStride[plane] && srcStride[plane] > 0) | 1718 if (dstStride[plane]==srcStride[plane] && srcStride[plane] > 0) |
1719 memcpy(dst[plane] + dstStride[plane]*y, src[plane], height*dstStride[plane]); | 1719 memcpy(dst[plane] + dstStride[plane]*y, src[plane], height*dstStride[plane]); |
1720 else | 1720 else |
1721 { | 1721 { |
1722 int i; | 1722 int i; |
1723 uint8_t *srcPtr= src[plane]; | 1723 uint8_t *srcPtr= src[plane]; |
1724 uint8_t *dstPtr= dst[plane] + dstStride[plane]*y; | 1724 uint8_t *dstPtr= dst[plane] + dstStride[plane]*y; |
1725 for(i=0; i<height; i++) | 1725 for (i=0; i<height; i++) |
1726 { | 1726 { |
1727 memcpy(dstPtr, srcPtr, length); | 1727 memcpy(dstPtr, srcPtr, length); |
1728 srcPtr+= srcStride[plane]; | 1728 srcPtr+= srcStride[plane]; |
1729 dstPtr+= dstStride[plane]; | 1729 dstPtr+= dstStride[plane]; |
1730 } | 1730 } |
1731 } | 1731 } |
1732 } | 1732 } |
1733 } | 1733 } |
1734 } | 1734 } |
1735 return srcSliceH; | 1735 return srcSliceH; |
1736 } | 1736 } |
1737 | 1737 |
1738 static int gray16togray(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1738 static int gray16togray(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1739 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 1739 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
1740 | 1740 |
1741 int length= c->srcW; | 1741 int length= c->srcW; |
1742 int y= srcSliceY; | 1742 int y= srcSliceY; |
1743 int height= srcSliceH; | 1743 int height= srcSliceH; |
1744 int i, j; | 1744 int i, j; |
1745 uint8_t *srcPtr= src[0]; | 1745 uint8_t *srcPtr= src[0]; |
1746 uint8_t *dstPtr= dst[0] + dstStride[0]*y; | 1746 uint8_t *dstPtr= dst[0] + dstStride[0]*y; |
1747 | 1747 |
1748 if(!isGray(c->dstFormat)){ | 1748 if (!isGray(c->dstFormat)){ |
1749 int height= -((-srcSliceH)>>c->chrDstVSubSample); | 1749 int height= -((-srcSliceH)>>c->chrDstVSubSample); |
1750 memset(dst[1], 128, dstStride[1]*height); | 1750 memset(dst[1], 128, dstStride[1]*height); |
1751 memset(dst[2], 128, dstStride[2]*height); | 1751 memset(dst[2], 128, dstStride[2]*height); |
1752 } | 1752 } |
1753 if(c->srcFormat == PIX_FMT_GRAY16LE) srcPtr++; | 1753 if (c->srcFormat == PIX_FMT_GRAY16LE) srcPtr++; |
1754 for(i=0; i<height; i++) | 1754 for (i=0; i<height; i++) |
1755 { | 1755 { |
1756 for(j=0; j<length; j++) dstPtr[j] = srcPtr[j<<1]; | 1756 for (j=0; j<length; j++) dstPtr[j] = srcPtr[j<<1]; |
1757 srcPtr+= srcStride[0]; | 1757 srcPtr+= srcStride[0]; |
1758 dstPtr+= dstStride[0]; | 1758 dstPtr+= dstStride[0]; |
1759 } | 1759 } |
1760 return srcSliceH; | 1760 return srcSliceH; |
1761 } | 1761 } |
1762 | 1762 |
1763 static int graytogray16(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1763 static int graytogray16(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1764 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 1764 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
1765 | 1765 |
1766 int length= c->srcW; | 1766 int length= c->srcW; |
1767 int y= srcSliceY; | 1767 int y= srcSliceY; |
1768 int height= srcSliceH; | 1768 int height= srcSliceH; |
1769 int i, j; | 1769 int i, j; |
1770 uint8_t *srcPtr= src[0]; | 1770 uint8_t *srcPtr= src[0]; |
1771 uint8_t *dstPtr= dst[0] + dstStride[0]*y; | 1771 uint8_t *dstPtr= dst[0] + dstStride[0]*y; |
1772 for(i=0; i<height; i++) | 1772 for (i=0; i<height; i++) |
1773 { | 1773 { |
1774 for(j=0; j<length; j++) | 1774 for (j=0; j<length; j++) |
1775 { | 1775 { |
1776 dstPtr[j<<1] = srcPtr[j]; | 1776 dstPtr[j<<1] = srcPtr[j]; |
1777 dstPtr[(j<<1)+1] = srcPtr[j]; | 1777 dstPtr[(j<<1)+1] = srcPtr[j]; |
1778 } | 1778 } |
1779 srcPtr+= srcStride[0]; | 1779 srcPtr+= srcStride[0]; |
1780 dstPtr+= dstStride[0]; | 1780 dstPtr+= dstStride[0]; |
1781 } | 1781 } |
1782 return srcSliceH; | 1782 return srcSliceH; |
1783 } | 1783 } |
1784 | 1784 |
1785 static int gray16swap(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 1785 static int gray16swap(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
1786 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 1786 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
1787 | 1787 |
1788 int length= c->srcW; | 1788 int length= c->srcW; |
1789 int y= srcSliceY; | 1789 int y= srcSliceY; |
1790 int height= srcSliceH; | 1790 int height= srcSliceH; |
1791 int i, j; | 1791 int i, j; |
1792 uint16_t *srcPtr= src[0]; | 1792 uint16_t *srcPtr= src[0]; |
1793 uint16_t *dstPtr= dst[0] + dstStride[0]*y/2; | 1793 uint16_t *dstPtr= dst[0] + dstStride[0]*y/2; |
1794 for(i=0; i<height; i++) | 1794 for (i=0; i<height; i++) |
1795 { | 1795 { |
1796 for(j=0; j<length; j++) dstPtr[j] = bswap_16(srcPtr[j]); | 1796 for (j=0; j<length; j++) dstPtr[j] = bswap_16(srcPtr[j]); |
1797 srcPtr+= srcStride[0]/2; | 1797 srcPtr+= srcStride[0]/2; |
1798 dstPtr+= dstStride[0]/2; | 1798 dstPtr+= dstStride[0]/2; |
1799 } | 1799 } |
1800 return srcSliceH; | 1800 return srcSliceH; |
1801 } | 1801 } |
1802 | 1802 |
1803 | 1803 |
1804 static void getSubSampleFactors(int *h, int *v, int format){ | 1804 static void getSubSampleFactors(int *h, int *v, int format){ |
1805 switch(format){ | 1805 switch(format){ |
1806 case PIX_FMT_UYVY422: | 1806 case PIX_FMT_UYVY422: |
1807 case PIX_FMT_YUYV422: | 1807 case PIX_FMT_YUYV422: |
1808 *h=1; | 1808 *h=1; |
1809 *v=0; | 1809 *v=0; |
1810 break; | 1810 break; |
1811 case PIX_FMT_YUV420P: | 1811 case PIX_FMT_YUV420P: |
1812 case PIX_FMT_GRAY16BE: | 1812 case PIX_FMT_GRAY16BE: |
1813 case PIX_FMT_GRAY16LE: | 1813 case PIX_FMT_GRAY16LE: |
1814 case PIX_FMT_GRAY8: //FIXME remove after different subsamplings are fully implemented | 1814 case PIX_FMT_GRAY8: //FIXME remove after different subsamplings are fully implemented |
1815 case PIX_FMT_NV12: | 1815 case PIX_FMT_NV12: |
1816 case PIX_FMT_NV21: | 1816 case PIX_FMT_NV21: |
1817 *h=1; | 1817 *h=1; |
1818 *v=1; | 1818 *v=1; |
1819 break; | 1819 break; |
1820 case PIX_FMT_YUV410P: | 1820 case PIX_FMT_YUV410P: |
1821 *h=2; | 1821 *h=2; |
1822 *v=2; | 1822 *v=2; |
1823 break; | 1823 break; |
1824 case PIX_FMT_YUV444P: | 1824 case PIX_FMT_YUV444P: |
1825 *h=0; | 1825 *h=0; |
1826 *v=0; | 1826 *v=0; |
1827 break; | 1827 break; |
1828 case PIX_FMT_YUV422P: | 1828 case PIX_FMT_YUV422P: |
1829 *h=1; | 1829 *h=1; |
1830 *v=0; | 1830 *v=0; |
1831 break; | 1831 break; |
1832 case PIX_FMT_YUV411P: | 1832 case PIX_FMT_YUV411P: |
1833 *h=2; | 1833 *h=2; |
1834 *v=0; | 1834 *v=0; |
1835 break; | 1835 break; |
1836 default: | 1836 default: |
1837 *h=0; | 1837 *h=0; |
1838 *v=0; | 1838 *v=0; |
1839 break; | 1839 break; |
1840 } | 1840 } |
1841 } | 1841 } |
1842 | 1842 |
1843 static uint16_t roundToInt16(int64_t f){ | 1843 static uint16_t roundToInt16(int64_t f){ |
1844 int r= (f + (1<<15))>>16; | 1844 int r= (f + (1<<15))>>16; |
1845 if(r<-0x7FFF) return 0x8000; | 1845 if (r<-0x7FFF) return 0x8000; |
1846 else if(r> 0x7FFF) return 0x7FFF; | 1846 else if (r> 0x7FFF) return 0x7FFF; |
1847 else return r; | 1847 else return r; |
1848 } | 1848 } |
1849 | 1849 |
1850 /** | 1850 /** |
1851 * @param inv_table the yuv2rgb coeffs, normally Inverse_Table_6_9[x] | 1851 * @param inv_table the yuv2rgb coeffs, normally Inverse_Table_6_9[x] |
1852 * @param fullRange if 1 then the luma range is 0..255 if 0 its 16..235 | 1852 * @param fullRange if 1 then the luma range is 0..255 if 0 its 16..235 |
1853 * @return -1 if not supported | 1853 * @return -1 if not supported |
1854 */ | 1854 */ |
1855 int sws_setColorspaceDetails(SwsContext *c, const int inv_table[4], int srcRange, const int table[4], int dstRange, int brightness, int contrast, int saturation){ | 1855 int sws_setColorspaceDetails(SwsContext *c, const int inv_table[4], int srcRange, const int table[4], int dstRange, int brightness, int contrast, int saturation){ |
1856 int64_t crv = inv_table[0]; | 1856 int64_t crv = inv_table[0]; |
1857 int64_t cbu = inv_table[1]; | 1857 int64_t cbu = inv_table[1]; |
1858 int64_t cgu = -inv_table[2]; | 1858 int64_t cgu = -inv_table[2]; |
1859 int64_t cgv = -inv_table[3]; | 1859 int64_t cgv = -inv_table[3]; |
1860 int64_t cy = 1<<16; | 1860 int64_t cy = 1<<16; |
1861 int64_t oy = 0; | 1861 int64_t oy = 0; |
1862 | 1862 |
1863 if(isYUV(c->dstFormat) || isGray(c->dstFormat)) return -1; | 1863 if (isYUV(c->dstFormat) || isGray(c->dstFormat)) return -1; |
1864 memcpy(c->srcColorspaceTable, inv_table, sizeof(int)*4); | 1864 memcpy(c->srcColorspaceTable, inv_table, sizeof(int)*4); |
1865 memcpy(c->dstColorspaceTable, table, sizeof(int)*4); | 1865 memcpy(c->dstColorspaceTable, table, sizeof(int)*4); |
1866 | 1866 |
1867 c->brightness= brightness; | 1867 c->brightness= brightness; |
1868 c->contrast = contrast; | 1868 c->contrast = contrast; |
1869 c->saturation= saturation; | 1869 c->saturation= saturation; |
1870 c->srcRange = srcRange; | 1870 c->srcRange = srcRange; |
1871 c->dstRange = dstRange; | 1871 c->dstRange = dstRange; |
1872 | 1872 |
1873 c->uOffset= 0x0400040004000400LL; | 1873 c->uOffset= 0x0400040004000400LL; |
1874 c->vOffset= 0x0400040004000400LL; | 1874 c->vOffset= 0x0400040004000400LL; |
1875 | 1875 |
1876 if(!srcRange){ | 1876 if (!srcRange){ |
1877 cy= (cy*255) / 219; | 1877 cy= (cy*255) / 219; |
1878 oy= 16<<16; | 1878 oy= 16<<16; |
1879 }else{ | 1879 }else{ |
1880 crv= (crv*224) / 255; | 1880 crv= (crv*224) / 255; |
1881 cbu= (cbu*224) / 255; | 1881 cbu= (cbu*224) / 255; |
1882 cgu= (cgu*224) / 255; | 1882 cgu= (cgu*224) / 255; |
1883 cgv= (cgv*224) / 255; | 1883 cgv= (cgv*224) / 255; |
1884 } | 1884 } |
1885 | 1885 |
1886 cy = (cy *contrast )>>16; | 1886 cy = (cy *contrast )>>16; |
1887 crv= (crv*contrast * saturation)>>32; | 1887 crv= (crv*contrast * saturation)>>32; |
1888 cbu= (cbu*contrast * saturation)>>32; | 1888 cbu= (cbu*contrast * saturation)>>32; |
1889 cgu= (cgu*contrast * saturation)>>32; | 1889 cgu= (cgu*contrast * saturation)>>32; |
1890 cgv= (cgv*contrast * saturation)>>32; | 1890 cgv= (cgv*contrast * saturation)>>32; |
1891 | 1891 |
1892 oy -= 256*brightness; | 1892 oy -= 256*brightness; |
1893 | 1893 |
1894 c->yCoeff= roundToInt16(cy *8192) * 0x0001000100010001ULL; | 1894 c->yCoeff= roundToInt16(cy *8192) * 0x0001000100010001ULL; |
1895 c->vrCoeff= roundToInt16(crv*8192) * 0x0001000100010001ULL; | 1895 c->vrCoeff= roundToInt16(crv*8192) * 0x0001000100010001ULL; |
1896 c->ubCoeff= roundToInt16(cbu*8192) * 0x0001000100010001ULL; | 1896 c->ubCoeff= roundToInt16(cbu*8192) * 0x0001000100010001ULL; |
1897 c->vgCoeff= roundToInt16(cgv*8192) * 0x0001000100010001ULL; | 1897 c->vgCoeff= roundToInt16(cgv*8192) * 0x0001000100010001ULL; |
1898 c->ugCoeff= roundToInt16(cgu*8192) * 0x0001000100010001ULL; | 1898 c->ugCoeff= roundToInt16(cgu*8192) * 0x0001000100010001ULL; |
1899 c->yOffset= roundToInt16(oy * 8) * 0x0001000100010001ULL; | 1899 c->yOffset= roundToInt16(oy * 8) * 0x0001000100010001ULL; |
1900 | 1900 |
1901 yuv2rgb_c_init_tables(c, inv_table, srcRange, brightness, contrast, saturation); | 1901 yuv2rgb_c_init_tables(c, inv_table, srcRange, brightness, contrast, saturation); |
1902 //FIXME factorize | 1902 //FIXME factorize |
1903 | 1903 |
1904 #ifdef COMPILE_ALTIVEC | 1904 #ifdef COMPILE_ALTIVEC |
1905 if (c->flags & SWS_CPU_CAPS_ALTIVEC) | 1905 if (c->flags & SWS_CPU_CAPS_ALTIVEC) |
1906 yuv2rgb_altivec_init_tables (c, inv_table, brightness, contrast, saturation); | 1906 yuv2rgb_altivec_init_tables (c, inv_table, brightness, contrast, saturation); |
1907 #endif | 1907 #endif |
1908 return 0; | 1908 return 0; |
1909 } | 1909 } |
1910 | 1910 |
1911 /** | 1911 /** |
1912 * @return -1 if not supported | 1912 * @return -1 if not supported |
1913 */ | 1913 */ |
1914 int sws_getColorspaceDetails(SwsContext *c, int **inv_table, int *srcRange, int **table, int *dstRange, int *brightness, int *contrast, int *saturation){ | 1914 int sws_getColorspaceDetails(SwsContext *c, int **inv_table, int *srcRange, int **table, int *dstRange, int *brightness, int *contrast, int *saturation){ |
1915 if(isYUV(c->dstFormat) || isGray(c->dstFormat)) return -1; | 1915 if (isYUV(c->dstFormat) || isGray(c->dstFormat)) return -1; |
1916 | 1916 |
1917 *inv_table = c->srcColorspaceTable; | 1917 *inv_table = c->srcColorspaceTable; |
1918 *table = c->dstColorspaceTable; | 1918 *table = c->dstColorspaceTable; |
1919 *srcRange = c->srcRange; | 1919 *srcRange = c->srcRange; |
1920 *dstRange = c->dstRange; | 1920 *dstRange = c->dstRange; |
1921 *brightness= c->brightness; | 1921 *brightness= c->brightness; |
1922 *contrast = c->contrast; | 1922 *contrast = c->contrast; |
1923 *saturation= c->saturation; | 1923 *saturation= c->saturation; |
1924 | 1924 |
1925 return 0; | 1925 return 0; |
1926 } | 1926 } |
1927 | 1927 |
1928 static int handle_jpeg(int *format) | 1928 static int handle_jpeg(int *format) |
1929 { | 1929 { |
1930 switch (*format) { | 1930 switch (*format) { |
1931 case PIX_FMT_YUVJ420P: | 1931 case PIX_FMT_YUVJ420P: |
1932 *format = PIX_FMT_YUV420P; | 1932 *format = PIX_FMT_YUV420P; |
1933 return 1; | 1933 return 1; |
1934 case PIX_FMT_YUVJ422P: | 1934 case PIX_FMT_YUVJ422P: |
1935 *format = PIX_FMT_YUV422P; | 1935 *format = PIX_FMT_YUV422P; |
1936 return 1; | 1936 return 1; |
1937 case PIX_FMT_YUVJ444P: | 1937 case PIX_FMT_YUVJ444P: |
1938 *format = PIX_FMT_YUV444P; | 1938 *format = PIX_FMT_YUV444P; |
1939 return 1; | 1939 return 1; |
1940 default: | 1940 default: |
1941 return 0; | 1941 return 0; |
1942 } | 1942 } |
1943 } | 1943 } |
1944 | 1944 |
1945 SwsContext *sws_getContext(int srcW, int srcH, int srcFormat, int dstW, int dstH, int dstFormat, int flags, | 1945 SwsContext *sws_getContext(int srcW, int srcH, int srcFormat, int dstW, int dstH, int dstFormat, int flags, |
1946 SwsFilter *srcFilter, SwsFilter *dstFilter, double *param){ | 1946 SwsFilter *srcFilter, SwsFilter *dstFilter, double *param){ |
1947 | 1947 |
1948 SwsContext *c; | 1948 SwsContext *c; |
1949 int i; | 1949 int i; |
1950 int usesVFilter, usesHFilter; | 1950 int usesVFilter, usesHFilter; |
1951 int unscaled, needsDither; | 1951 int unscaled, needsDither; |
1952 int srcRange, dstRange; | 1952 int srcRange, dstRange; |
1953 SwsFilter dummyFilter= {NULL, NULL, NULL, NULL}; | 1953 SwsFilter dummyFilter= {NULL, NULL, NULL, NULL}; |
1954 #if defined(ARCH_X86) | 1954 #if defined(ARCH_X86) |
1955 if(flags & SWS_CPU_CAPS_MMX) | 1955 if (flags & SWS_CPU_CAPS_MMX) |
1956 asm volatile("emms\n\t"::: "memory"); | 1956 asm volatile("emms\n\t"::: "memory"); |
1957 #endif | 1957 #endif |
1958 | 1958 |
1959 #if !defined(RUNTIME_CPUDETECT) || !defined (CONFIG_GPL) //ensure that the flags match the compiled variant if cpudetect is off | 1959 #if !defined(RUNTIME_CPUDETECT) || !defined (CONFIG_GPL) //ensure that the flags match the compiled variant if cpudetect is off |
1960 flags &= ~(SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2|SWS_CPU_CAPS_3DNOW|SWS_CPU_CAPS_ALTIVEC); | 1960 flags &= ~(SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2|SWS_CPU_CAPS_3DNOW|SWS_CPU_CAPS_ALTIVEC); |
1961 #ifdef HAVE_MMX2 | 1961 #ifdef HAVE_MMX2 |
1962 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2; | 1962 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2; |
1963 #elif defined (HAVE_3DNOW) | 1963 #elif defined (HAVE_3DNOW) |
1964 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW; | 1964 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW; |
1965 #elif defined (HAVE_MMX) | 1965 #elif defined (HAVE_MMX) |
1966 flags |= SWS_CPU_CAPS_MMX; | 1966 flags |= SWS_CPU_CAPS_MMX; |
1967 #elif defined (HAVE_ALTIVEC) | 1967 #elif defined (HAVE_ALTIVEC) |
1968 flags |= SWS_CPU_CAPS_ALTIVEC; | 1968 flags |= SWS_CPU_CAPS_ALTIVEC; |
1969 #endif | 1969 #endif |
1970 #endif /* RUNTIME_CPUDETECT */ | 1970 #endif /* RUNTIME_CPUDETECT */ |
1971 if(clip_table[512] != 255) globalInit(); | 1971 if (clip_table[512] != 255) globalInit(); |
1972 if(rgb15to16 == NULL) sws_rgb2rgb_init(flags); | 1972 if (rgb15to16 == NULL) sws_rgb2rgb_init(flags); |
1973 | 1973 |
1974 unscaled = (srcW == dstW && srcH == dstH); | 1974 unscaled = (srcW == dstW && srcH == dstH); |
1975 needsDither= (isBGR(dstFormat) || isRGB(dstFormat)) | 1975 needsDither= (isBGR(dstFormat) || isRGB(dstFormat)) |
1976 && (fmt_depth(dstFormat))<24 | 1976 && (fmt_depth(dstFormat))<24 |
1977 && ((fmt_depth(dstFormat))<(fmt_depth(srcFormat)) || (!(isRGB(srcFormat) || isBGR(srcFormat)))); | 1977 && ((fmt_depth(dstFormat))<(fmt_depth(srcFormat)) || (!(isRGB(srcFormat) || isBGR(srcFormat)))); |
1978 | 1978 |
1979 srcRange = handle_jpeg(&srcFormat); | 1979 srcRange = handle_jpeg(&srcFormat); |
1980 dstRange = handle_jpeg(&dstFormat); | 1980 dstRange = handle_jpeg(&dstFormat); |
1981 | 1981 |
1982 if(!isSupportedIn(srcFormat)) | 1982 if (!isSupportedIn(srcFormat)) |
1983 { | 1983 { |
1984 av_log(NULL, AV_LOG_ERROR, "swScaler: %s is not supported as input format\n", sws_format_name(srcFormat)); | 1984 av_log(NULL, AV_LOG_ERROR, "swScaler: %s is not supported as input format\n", sws_format_name(srcFormat)); |
1985 return NULL; | 1985 return NULL; |
1986 } | 1986 } |
1987 if(!isSupportedOut(dstFormat)) | 1987 if (!isSupportedOut(dstFormat)) |
1988 { | 1988 { |
1989 av_log(NULL, AV_LOG_ERROR, "swScaler: %s is not supported as output format\n", sws_format_name(dstFormat)); | 1989 av_log(NULL, AV_LOG_ERROR, "swScaler: %s is not supported as output format\n", sws_format_name(dstFormat)); |
1990 return NULL; | 1990 return NULL; |
1991 } | 1991 } |
1992 | 1992 |
1993 /* sanity check */ | 1993 /* sanity check */ |
1994 if(srcW<4 || srcH<1 || dstW<8 || dstH<1) //FIXME check if these are enough and try to lowwer them after fixing the relevant parts of the code | 1994 if (srcW<4 || srcH<1 || dstW<8 || dstH<1) //FIXME check if these are enough and try to lowwer them after fixing the relevant parts of the code |
1995 { | 1995 { |
1996 av_log(NULL, AV_LOG_ERROR, "swScaler: %dx%d -> %dx%d is invalid scaling dimension\n", | 1996 av_log(NULL, AV_LOG_ERROR, "swScaler: %dx%d -> %dx%d is invalid scaling dimension\n", |
1997 srcW, srcH, dstW, dstH); | 1997 srcW, srcH, dstW, dstH); |
1998 return NULL; | 1998 return NULL; |
1999 } | 1999 } |
2000 | 2000 |
2001 if(!dstFilter) dstFilter= &dummyFilter; | 2001 if (!dstFilter) dstFilter= &dummyFilter; |
2002 if(!srcFilter) srcFilter= &dummyFilter; | 2002 if (!srcFilter) srcFilter= &dummyFilter; |
2003 | 2003 |
2004 c= av_mallocz(sizeof(SwsContext)); | 2004 c= av_mallocz(sizeof(SwsContext)); |
2005 | 2005 |
2006 c->av_class = &sws_context_class; | 2006 c->av_class = &sws_context_class; |
2007 c->srcW= srcW; | 2007 c->srcW= srcW; |
2008 c->srcH= srcH; | 2008 c->srcH= srcH; |
2009 c->dstW= dstW; | 2009 c->dstW= dstW; |
2010 c->dstH= dstH; | 2010 c->dstH= dstH; |
2011 c->lumXInc= ((srcW<<16) + (dstW>>1))/dstW; | 2011 c->lumXInc= ((srcW<<16) + (dstW>>1))/dstW; |
2012 c->lumYInc= ((srcH<<16) + (dstH>>1))/dstH; | 2012 c->lumYInc= ((srcH<<16) + (dstH>>1))/dstH; |
2013 c->flags= flags; | 2013 c->flags= flags; |
2014 c->dstFormat= dstFormat; | 2014 c->dstFormat= dstFormat; |
2015 c->srcFormat= srcFormat; | 2015 c->srcFormat= srcFormat; |
2016 c->vRounder= 4* 0x0001000100010001ULL; | 2016 c->vRounder= 4* 0x0001000100010001ULL; |
2017 | 2017 |
2018 usesHFilter= usesVFilter= 0; | 2018 usesHFilter= usesVFilter= 0; |
2019 if(dstFilter->lumV!=NULL && dstFilter->lumV->length>1) usesVFilter=1; | 2019 if (dstFilter->lumV!=NULL && dstFilter->lumV->length>1) usesVFilter=1; |
2020 if(dstFilter->lumH!=NULL && dstFilter->lumH->length>1) usesHFilter=1; | 2020 if (dstFilter->lumH!=NULL && dstFilter->lumH->length>1) usesHFilter=1; |
2021 if(dstFilter->chrV!=NULL && dstFilter->chrV->length>1) usesVFilter=1; | 2021 if (dstFilter->chrV!=NULL && dstFilter->chrV->length>1) usesVFilter=1; |
2022 if(dstFilter->chrH!=NULL && dstFilter->chrH->length>1) usesHFilter=1; | 2022 if (dstFilter->chrH!=NULL && dstFilter->chrH->length>1) usesHFilter=1; |
2023 if(srcFilter->lumV!=NULL && srcFilter->lumV->length>1) usesVFilter=1; | 2023 if (srcFilter->lumV!=NULL && srcFilter->lumV->length>1) usesVFilter=1; |
2024 if(srcFilter->lumH!=NULL && srcFilter->lumH->length>1) usesHFilter=1; | 2024 if (srcFilter->lumH!=NULL && srcFilter->lumH->length>1) usesHFilter=1; |
2025 if(srcFilter->chrV!=NULL && srcFilter->chrV->length>1) usesVFilter=1; | 2025 if (srcFilter->chrV!=NULL && srcFilter->chrV->length>1) usesVFilter=1; |
2026 if(srcFilter->chrH!=NULL && srcFilter->chrH->length>1) usesHFilter=1; | 2026 if (srcFilter->chrH!=NULL && srcFilter->chrH->length>1) usesHFilter=1; |
2027 | 2027 |
2028 getSubSampleFactors(&c->chrSrcHSubSample, &c->chrSrcVSubSample, srcFormat); | 2028 getSubSampleFactors(&c->chrSrcHSubSample, &c->chrSrcVSubSample, srcFormat); |
2029 getSubSampleFactors(&c->chrDstHSubSample, &c->chrDstVSubSample, dstFormat); | 2029 getSubSampleFactors(&c->chrDstHSubSample, &c->chrDstVSubSample, dstFormat); |
2030 | 2030 |
2031 // reuse chroma for 2 pixles rgb/bgr unless user wants full chroma interpolation | 2031 // reuse chroma for 2 pixles rgb/bgr unless user wants full chroma interpolation |
2032 if((isBGR(dstFormat) || isRGB(dstFormat)) && !(flags&SWS_FULL_CHR_H_INT)) c->chrDstHSubSample=1; | 2032 if ((isBGR(dstFormat) || isRGB(dstFormat)) && !(flags&SWS_FULL_CHR_H_INT)) c->chrDstHSubSample=1; |
2033 | 2033 |
2034 // drop some chroma lines if the user wants it | 2034 // drop some chroma lines if the user wants it |
2035 c->vChrDrop= (flags&SWS_SRC_V_CHR_DROP_MASK)>>SWS_SRC_V_CHR_DROP_SHIFT; | 2035 c->vChrDrop= (flags&SWS_SRC_V_CHR_DROP_MASK)>>SWS_SRC_V_CHR_DROP_SHIFT; |
2036 c->chrSrcVSubSample+= c->vChrDrop; | 2036 c->chrSrcVSubSample+= c->vChrDrop; |
2037 | 2037 |
2038 // drop every 2. pixel for chroma calculation unless user wants full chroma | 2038 // drop every 2. pixel for chroma calculation unless user wants full chroma |
2039 if((isBGR(srcFormat) || isRGB(srcFormat)) && !(flags&SWS_FULL_CHR_H_INP) | 2039 if ((isBGR(srcFormat) || isRGB(srcFormat)) && !(flags&SWS_FULL_CHR_H_INP) |
2040 && srcFormat!=PIX_FMT_RGB8 && srcFormat!=PIX_FMT_BGR8 | 2040 && srcFormat!=PIX_FMT_RGB8 && srcFormat!=PIX_FMT_BGR8 |
2041 && srcFormat!=PIX_FMT_RGB4 && srcFormat!=PIX_FMT_BGR4 | 2041 && srcFormat!=PIX_FMT_RGB4 && srcFormat!=PIX_FMT_BGR4 |
2042 && srcFormat!=PIX_FMT_RGB4_BYTE && srcFormat!=PIX_FMT_BGR4_BYTE) | 2042 && srcFormat!=PIX_FMT_RGB4_BYTE && srcFormat!=PIX_FMT_BGR4_BYTE) |
2043 c->chrSrcHSubSample=1; | 2043 c->chrSrcHSubSample=1; |
2044 | 2044 |
2045 if(param){ | 2045 if (param){ |
2046 c->param[0] = param[0]; | 2046 c->param[0] = param[0]; |
2047 c->param[1] = param[1]; | 2047 c->param[1] = param[1]; |
2048 }else{ | 2048 }else{ |
2049 c->param[0] = | 2049 c->param[0] = |
2050 c->param[1] = SWS_PARAM_DEFAULT; | 2050 c->param[1] = SWS_PARAM_DEFAULT; |
2051 } | 2051 } |
2052 | 2052 |
2053 c->chrIntHSubSample= c->chrDstHSubSample; | 2053 c->chrIntHSubSample= c->chrDstHSubSample; |
2054 c->chrIntVSubSample= c->chrSrcVSubSample; | 2054 c->chrIntVSubSample= c->chrSrcVSubSample; |
2055 | 2055 |
2056 // Note the -((-x)>>y) is so that we always round toward +inf. | 2056 // Note the -((-x)>>y) is so that we always round toward +inf. |
2057 c->chrSrcW= -((-srcW) >> c->chrSrcHSubSample); | 2057 c->chrSrcW= -((-srcW) >> c->chrSrcHSubSample); |
2058 c->chrSrcH= -((-srcH) >> c->chrSrcVSubSample); | 2058 c->chrSrcH= -((-srcH) >> c->chrSrcVSubSample); |
2059 c->chrDstW= -((-dstW) >> c->chrDstHSubSample); | 2059 c->chrDstW= -((-dstW) >> c->chrDstHSubSample); |
2060 c->chrDstH= -((-dstH) >> c->chrDstVSubSample); | 2060 c->chrDstH= -((-dstH) >> c->chrDstVSubSample); |
2061 | 2061 |
2062 sws_setColorspaceDetails(c, Inverse_Table_6_9[SWS_CS_DEFAULT], srcRange, Inverse_Table_6_9[SWS_CS_DEFAULT] /* FIXME*/, dstRange, 0, 1<<16, 1<<16); | 2062 sws_setColorspaceDetails(c, Inverse_Table_6_9[SWS_CS_DEFAULT], srcRange, Inverse_Table_6_9[SWS_CS_DEFAULT] /* FIXME*/, dstRange, 0, 1<<16, 1<<16); |
2063 | 2063 |
2064 /* unscaled special Cases */ | 2064 /* unscaled special Cases */ |
2065 if(unscaled && !usesHFilter && !usesVFilter) | 2065 if (unscaled && !usesHFilter && !usesVFilter) |
2066 { | 2066 { |
2067 /* yv12_to_nv12 */ | 2067 /* yv12_to_nv12 */ |
2068 if(srcFormat == PIX_FMT_YUV420P && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) | 2068 if (srcFormat == PIX_FMT_YUV420P && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) |
2069 { | 2069 { |
2070 c->swScale= PlanarToNV12Wrapper; | 2070 c->swScale= PlanarToNV12Wrapper; |
2071 } | 2071 } |
2072 #ifdef CONFIG_GPL | 2072 #ifdef CONFIG_GPL |
2073 /* yuv2bgr */ | 2073 /* yuv2bgr */ |
2074 if((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P) && (isBGR(dstFormat) || isRGB(dstFormat))) | 2074 if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P) && (isBGR(dstFormat) || isRGB(dstFormat))) |
2075 { | 2075 { |
2076 c->swScale= yuv2rgb_get_func_ptr(c); | 2076 c->swScale= yuv2rgb_get_func_ptr(c); |
2077 } | 2077 } |
2078 #endif | 2078 #endif |
2079 | 2079 |
2080 if( srcFormat==PIX_FMT_YUV410P && dstFormat==PIX_FMT_YUV420P ) | 2080 if ( srcFormat==PIX_FMT_YUV410P && dstFormat==PIX_FMT_YUV420P ) |
2081 { | 2081 { |
2082 c->swScale= yvu9toyv12Wrapper; | 2082 c->swScale= yvu9toyv12Wrapper; |
2083 } | 2083 } |
2084 | 2084 |
2085 /* bgr24toYV12 */ | 2085 /* bgr24toYV12 */ |
2086 if(srcFormat==PIX_FMT_BGR24 && dstFormat==PIX_FMT_YUV420P) | 2086 if (srcFormat==PIX_FMT_BGR24 && dstFormat==PIX_FMT_YUV420P) |
2087 c->swScale= bgr24toyv12Wrapper; | 2087 c->swScale= bgr24toyv12Wrapper; |
2088 | 2088 |
2089 /* rgb/bgr -> rgb/bgr (no dither needed forms) */ | 2089 /* rgb/bgr -> rgb/bgr (no dither needed forms) */ |
2090 if( (isBGR(srcFormat) || isRGB(srcFormat)) | 2090 if ( (isBGR(srcFormat) || isRGB(srcFormat)) |
2091 && (isBGR(dstFormat) || isRGB(dstFormat)) | 2091 && (isBGR(dstFormat) || isRGB(dstFormat)) |
2092 && srcFormat != PIX_FMT_BGR8 && dstFormat != PIX_FMT_BGR8 | 2092 && srcFormat != PIX_FMT_BGR8 && dstFormat != PIX_FMT_BGR8 |
2093 && srcFormat != PIX_FMT_RGB8 && dstFormat != PIX_FMT_RGB8 | 2093 && srcFormat != PIX_FMT_RGB8 && dstFormat != PIX_FMT_RGB8 |
2094 && srcFormat != PIX_FMT_BGR4 && dstFormat != PIX_FMT_BGR4 | 2094 && srcFormat != PIX_FMT_BGR4 && dstFormat != PIX_FMT_BGR4 |
2095 && srcFormat != PIX_FMT_RGB4 && dstFormat != PIX_FMT_RGB4 | 2095 && srcFormat != PIX_FMT_RGB4 && dstFormat != PIX_FMT_RGB4 |
2096 && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE | 2096 && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE |
2097 && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE | 2097 && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE |
2098 && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK | 2098 && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK |
2099 && !needsDither) | 2099 && !needsDither) |
2100 c->swScale= rgb2rgbWrapper; | 2100 c->swScale= rgb2rgbWrapper; |
2101 | 2101 |
2102 /* LQ converters if -sws 0 or -sws 4*/ | 2102 /* LQ converters if -sws 0 or -sws 4*/ |
2103 if(c->flags&(SWS_FAST_BILINEAR|SWS_POINT)){ | 2103 if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)){ |
2104 /* rgb/bgr -> rgb/bgr (dither needed forms) */ | 2104 /* rgb/bgr -> rgb/bgr (dither needed forms) */ |
2105 if( (isBGR(srcFormat) || isRGB(srcFormat)) | 2105 if ( (isBGR(srcFormat) || isRGB(srcFormat)) |
2106 && (isBGR(dstFormat) || isRGB(dstFormat)) | 2106 && (isBGR(dstFormat) || isRGB(dstFormat)) |
2107 && needsDither) | 2107 && needsDither) |
2108 c->swScale= rgb2rgbWrapper; | 2108 c->swScale= rgb2rgbWrapper; |
2109 | 2109 |
2110 /* yv12_to_yuy2 */ | 2110 /* yv12_to_yuy2 */ |
2111 if(srcFormat == PIX_FMT_YUV420P && | 2111 if (srcFormat == PIX_FMT_YUV420P && |
2112 (dstFormat == PIX_FMT_YUYV422 || dstFormat == PIX_FMT_UYVY422)) | 2112 (dstFormat == PIX_FMT_YUYV422 || dstFormat == PIX_FMT_UYVY422)) |
2113 { | 2113 { |
2114 if (dstFormat == PIX_FMT_YUYV422) | 2114 if (dstFormat == PIX_FMT_YUYV422) |
2115 c->swScale= PlanarToYuy2Wrapper; | 2115 c->swScale= PlanarToYuy2Wrapper; |
2116 else | 2116 else |
2117 c->swScale= PlanarToUyvyWrapper; | 2117 c->swScale= PlanarToUyvyWrapper; |
2118 } | 2118 } |
2119 } | 2119 } |
2120 | 2120 |
2121 #ifdef COMPILE_ALTIVEC | 2121 #ifdef COMPILE_ALTIVEC |
2122 if ((c->flags & SWS_CPU_CAPS_ALTIVEC) && | 2122 if ((c->flags & SWS_CPU_CAPS_ALTIVEC) && |
2123 ((srcFormat == PIX_FMT_YUV420P && | 2123 ((srcFormat == PIX_FMT_YUV420P && |
2124 (dstFormat == PIX_FMT_YUYV422 || dstFormat == PIX_FMT_UYVY422)))) { | 2124 (dstFormat == PIX_FMT_YUYV422 || dstFormat == PIX_FMT_UYVY422)))) { |
2125 // unscaled YV12 -> packed YUV, we want speed | 2125 // unscaled YV12 -> packed YUV, we want speed |
2126 if (dstFormat == PIX_FMT_YUYV422) | 2126 if (dstFormat == PIX_FMT_YUYV422) |
2127 c->swScale= yv12toyuy2_unscaled_altivec; | 2127 c->swScale= yv12toyuy2_unscaled_altivec; |
2128 else | 2128 else |
2129 c->swScale= yv12touyvy_unscaled_altivec; | 2129 c->swScale= yv12touyvy_unscaled_altivec; |
2130 } | 2130 } |
2131 #endif | 2131 #endif |
2132 | 2132 |
2133 /* simple copy */ | 2133 /* simple copy */ |
2134 if( srcFormat == dstFormat | 2134 if ( srcFormat == dstFormat |
2135 || (isPlanarYUV(srcFormat) && isGray(dstFormat)) | 2135 || (isPlanarYUV(srcFormat) && isGray(dstFormat)) |
2136 || (isPlanarYUV(dstFormat) && isGray(srcFormat)) | 2136 || (isPlanarYUV(dstFormat) && isGray(srcFormat)) ) |
2137 ) | 2137 { |
2138 { | 2138 c->swScale= simpleCopy; |
2139 c->swScale= simpleCopy; | 2139 } |
2140 } | 2140 |
2141 | 2141 /* gray16{le,be} conversions */ |
2142 /* gray16{le,be} conversions */ | 2142 if (isGray16(srcFormat) && (isPlanarYUV(dstFormat) || (dstFormat == PIX_FMT_GRAY8))) |
2143 if(isGray16(srcFormat) && (isPlanarYUV(dstFormat) || (dstFormat == PIX_FMT_GRAY8))) | 2143 { |
2144 { | 2144 c->swScale= gray16togray; |
2145 c->swScale= gray16togray; | 2145 } |
2146 } | 2146 if ((isPlanarYUV(srcFormat) || (srcFormat == PIX_FMT_GRAY8)) && isGray16(dstFormat)) |
2147 if((isPlanarYUV(srcFormat) || (srcFormat == PIX_FMT_GRAY8)) && isGray16(dstFormat)) | 2147 { |
2148 { | 2148 c->swScale= graytogray16; |
2149 c->swScale= graytogray16; | 2149 } |
2150 } | 2150 if (srcFormat != dstFormat && isGray16(srcFormat) && isGray16(dstFormat)) |
2151 if(srcFormat != dstFormat && isGray16(srcFormat) && isGray16(dstFormat)) | 2151 { |
2152 { | 2152 c->swScale= gray16swap; |
2153 c->swScale= gray16swap; | 2153 } |
2154 } | 2154 |
2155 | 2155 if (c->swScale){ |
2156 if(c->swScale){ | 2156 if (flags&SWS_PRINT_INFO) |
2157 if(flags&SWS_PRINT_INFO) | 2157 av_log(c, AV_LOG_INFO, "SwScaler: using unscaled %s -> %s special converter\n", |
2158 av_log(c, AV_LOG_INFO, "SwScaler: using unscaled %s -> %s special converter\n", | 2158 sws_format_name(srcFormat), sws_format_name(dstFormat)); |
2159 sws_format_name(srcFormat), sws_format_name(dstFormat)); | 2159 return c; |
2160 return c; | 2160 } |
2161 } | 2161 } |
2162 } | 2162 |
2163 | 2163 if (flags & SWS_CPU_CAPS_MMX2) |
2164 if(flags & SWS_CPU_CAPS_MMX2) | 2164 { |
2165 { | 2165 c->canMMX2BeUsed= (dstW >=srcW && (dstW&31)==0 && (srcW&15)==0) ? 1 : 0; |
2166 c->canMMX2BeUsed= (dstW >=srcW && (dstW&31)==0 && (srcW&15)==0) ? 1 : 0; | 2166 if (!c->canMMX2BeUsed && dstW >=srcW && (srcW&15)==0 && (flags&SWS_FAST_BILINEAR)) |
2167 if(!c->canMMX2BeUsed && dstW >=srcW && (srcW&15)==0 && (flags&SWS_FAST_BILINEAR)) | 2167 { |
2168 { | 2168 if (flags&SWS_PRINT_INFO) |
2169 if(flags&SWS_PRINT_INFO) | 2169 av_log(c, AV_LOG_INFO, "SwScaler: output Width is not a multiple of 32 -> no MMX2 scaler\n"); |
2170 av_log(c, AV_LOG_INFO, "SwScaler: output Width is not a multiple of 32 -> no MMX2 scaler\n"); | 2170 } |
2171 } | 2171 if (usesHFilter) c->canMMX2BeUsed=0; |
2172 if(usesHFilter) c->canMMX2BeUsed=0; | 2172 } |
2173 } | 2173 else |
2174 else | 2174 c->canMMX2BeUsed=0; |
2175 c->canMMX2BeUsed=0; | 2175 |
2176 | 2176 c->chrXInc= ((c->chrSrcW<<16) + (c->chrDstW>>1))/c->chrDstW; |
2177 c->chrXInc= ((c->chrSrcW<<16) + (c->chrDstW>>1))/c->chrDstW; | 2177 c->chrYInc= ((c->chrSrcH<<16) + (c->chrDstH>>1))/c->chrDstH; |
2178 c->chrYInc= ((c->chrSrcH<<16) + (c->chrDstH>>1))/c->chrDstH; | 2178 |
2179 | 2179 // match pixel 0 of the src to pixel 0 of dst and match pixel n-2 of src to pixel n-2 of dst |
2180 // match pixel 0 of the src to pixel 0 of dst and match pixel n-2 of src to pixel n-2 of dst | 2180 // but only for the FAST_BILINEAR mode otherwise do correct scaling |
2181 // but only for the FAST_BILINEAR mode otherwise do correct scaling | 2181 // n-2 is the last chrominance sample available |
2182 // n-2 is the last chrominance sample available | 2182 // this is not perfect, but noone shuld notice the difference, the more correct variant |
2183 // this is not perfect, but noone shuld notice the difference, the more correct variant | 2183 // would be like the vertical one, but that would require some special code for the |
2184 // would be like the vertical one, but that would require some special code for the | 2184 // first and last pixel |
2185 // first and last pixel | 2185 if (flags&SWS_FAST_BILINEAR) |
2186 if(flags&SWS_FAST_BILINEAR) | 2186 { |
2187 { | 2187 if (c->canMMX2BeUsed) |
2188 if(c->canMMX2BeUsed) | 2188 { |
2189 { | 2189 c->lumXInc+= 20; |
2190 c->lumXInc+= 20; | 2190 c->chrXInc+= 20; |
2191 c->chrXInc+= 20; | 2191 } |
2192 } | 2192 //we don't use the x86asm scaler if mmx is available |
2193 //we don't use the x86asm scaler if mmx is available | 2193 else if (flags & SWS_CPU_CAPS_MMX) |
2194 else if(flags & SWS_CPU_CAPS_MMX) | 2194 { |
2195 { | 2195 c->lumXInc = ((srcW-2)<<16)/(dstW-2) - 20; |
2196 c->lumXInc = ((srcW-2)<<16)/(dstW-2) - 20; | 2196 c->chrXInc = ((c->chrSrcW-2)<<16)/(c->chrDstW-2) - 20; |
2197 c->chrXInc = ((c->chrSrcW-2)<<16)/(c->chrDstW-2) - 20; | 2197 } |
2198 } | 2198 } |
2199 } | 2199 |
2200 | 2200 /* precalculate horizontal scaler filter coefficients */ |
2201 /* precalculate horizontal scaler filter coefficients */ | 2201 { |
2202 { | 2202 const int filterAlign= |
2203 const int filterAlign= | 2203 (flags & SWS_CPU_CAPS_MMX) ? 4 : |
2204 (flags & SWS_CPU_CAPS_MMX) ? 4 : | 2204 (flags & SWS_CPU_CAPS_ALTIVEC) ? 8 : |
2205 (flags & SWS_CPU_CAPS_ALTIVEC) ? 8 : | 2205 1; |
2206 1; | 2206 |
2207 | 2207 initFilter(&c->hLumFilter, &c->hLumFilterPos, &c->hLumFilterSize, c->lumXInc, |
2208 initFilter(&c->hLumFilter, &c->hLumFilterPos, &c->hLumFilterSize, c->lumXInc, | 2208 srcW , dstW, filterAlign, 1<<14, |
2209 srcW , dstW, filterAlign, 1<<14, | 2209 (flags&SWS_BICUBLIN) ? (flags|SWS_BICUBIC) : flags, |
2210 (flags&SWS_BICUBLIN) ? (flags|SWS_BICUBIC) : flags, | 2210 srcFilter->lumH, dstFilter->lumH, c->param); |
2211 srcFilter->lumH, dstFilter->lumH, c->param); | 2211 initFilter(&c->hChrFilter, &c->hChrFilterPos, &c->hChrFilterSize, c->chrXInc, |
2212 initFilter(&c->hChrFilter, &c->hChrFilterPos, &c->hChrFilterSize, c->chrXInc, | 2212 c->chrSrcW, c->chrDstW, filterAlign, 1<<14, |
2213 c->chrSrcW, c->chrDstW, filterAlign, 1<<14, | 2213 (flags&SWS_BICUBLIN) ? (flags|SWS_BILINEAR) : flags, |
2214 (flags&SWS_BICUBLIN) ? (flags|SWS_BILINEAR) : flags, | 2214 srcFilter->chrH, dstFilter->chrH, c->param); |
2215 srcFilter->chrH, dstFilter->chrH, c->param); | |
2216 | 2215 |
2217 #define MAX_FUNNY_CODE_SIZE 10000 | 2216 #define MAX_FUNNY_CODE_SIZE 10000 |
2218 #if defined(COMPILE_MMX2) | 2217 #if defined(COMPILE_MMX2) |
2219 // can't downscale !!! | 2218 // can't downscale !!! |
2220 if(c->canMMX2BeUsed && (flags & SWS_FAST_BILINEAR)) | 2219 if (c->canMMX2BeUsed && (flags & SWS_FAST_BILINEAR)) |
2221 { | 2220 { |
2222 #ifdef MAP_ANONYMOUS | 2221 #ifdef MAP_ANONYMOUS |
2223 c->funnyYCode = (uint8_t*)mmap(NULL, MAX_FUNNY_CODE_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, 0, 0); | 2222 c->funnyYCode = (uint8_t*)mmap(NULL, MAX_FUNNY_CODE_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, 0, 0); |
2224 c->funnyUVCode = (uint8_t*)mmap(NULL, MAX_FUNNY_CODE_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, 0, 0); | 2223 c->funnyUVCode = (uint8_t*)mmap(NULL, MAX_FUNNY_CODE_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, 0, 0); |
2225 #else | 2224 #else |
2226 c->funnyYCode = av_malloc(MAX_FUNNY_CODE_SIZE); | 2225 c->funnyYCode = av_malloc(MAX_FUNNY_CODE_SIZE); |
2227 c->funnyUVCode = av_malloc(MAX_FUNNY_CODE_SIZE); | 2226 c->funnyUVCode = av_malloc(MAX_FUNNY_CODE_SIZE); |
2228 #endif | 2227 #endif |
2229 | 2228 |
2230 c->lumMmx2Filter = av_malloc((dstW /8+8)*sizeof(int16_t)); | 2229 c->lumMmx2Filter = av_malloc((dstW /8+8)*sizeof(int16_t)); |
2231 c->chrMmx2Filter = av_malloc((c->chrDstW /4+8)*sizeof(int16_t)); | 2230 c->chrMmx2Filter = av_malloc((c->chrDstW /4+8)*sizeof(int16_t)); |
2232 c->lumMmx2FilterPos= av_malloc((dstW /2/8+8)*sizeof(int32_t)); | 2231 c->lumMmx2FilterPos= av_malloc((dstW /2/8+8)*sizeof(int32_t)); |
2233 c->chrMmx2FilterPos= av_malloc((c->chrDstW/2/4+8)*sizeof(int32_t)); | 2232 c->chrMmx2FilterPos= av_malloc((c->chrDstW/2/4+8)*sizeof(int32_t)); |
2234 | 2233 |
2235 initMMX2HScaler( dstW, c->lumXInc, c->funnyYCode , c->lumMmx2Filter, c->lumMmx2FilterPos, 8); | 2234 initMMX2HScaler( dstW, c->lumXInc, c->funnyYCode , c->lumMmx2Filter, c->lumMmx2FilterPos, 8); |
2236 initMMX2HScaler(c->chrDstW, c->chrXInc, c->funnyUVCode, c->chrMmx2Filter, c->chrMmx2FilterPos, 4); | 2235 initMMX2HScaler(c->chrDstW, c->chrXInc, c->funnyUVCode, c->chrMmx2Filter, c->chrMmx2FilterPos, 4); |
2237 } | 2236 } |
2238 #endif /* defined(COMPILE_MMX2) */ | 2237 #endif /* defined(COMPILE_MMX2) */ |
2239 } // Init Horizontal stuff | 2238 } // Init Horizontal stuff |
2240 | 2239 |
2241 | 2240 |
2242 | 2241 |
2243 /* precalculate vertical scaler filter coefficients */ | 2242 /* precalculate vertical scaler filter coefficients */ |
2244 { | 2243 { |
2245 const int filterAlign= | 2244 const int filterAlign= |
2246 (flags & SWS_CPU_CAPS_MMX) && (flags & SWS_ACCURATE_RND) ? 2 : | 2245 (flags & SWS_CPU_CAPS_MMX) && (flags & SWS_ACCURATE_RND) ? 2 : |
2247 (flags & SWS_CPU_CAPS_ALTIVEC) ? 8 : | 2246 (flags & SWS_CPU_CAPS_ALTIVEC) ? 8 : |
2248 1; | 2247 1; |
2249 | 2248 |
2250 initFilter(&c->vLumFilter, &c->vLumFilterPos, &c->vLumFilterSize, c->lumYInc, | 2249 initFilter(&c->vLumFilter, &c->vLumFilterPos, &c->vLumFilterSize, c->lumYInc, |
2251 srcH , dstH, filterAlign, (1<<12)-4, | 2250 srcH , dstH, filterAlign, (1<<12)-4, |
2252 (flags&SWS_BICUBLIN) ? (flags|SWS_BICUBIC) : flags, | 2251 (flags&SWS_BICUBLIN) ? (flags|SWS_BICUBIC) : flags, |
2253 srcFilter->lumV, dstFilter->lumV, c->param); | 2252 srcFilter->lumV, dstFilter->lumV, c->param); |
2254 initFilter(&c->vChrFilter, &c->vChrFilterPos, &c->vChrFilterSize, c->chrYInc, | 2253 initFilter(&c->vChrFilter, &c->vChrFilterPos, &c->vChrFilterSize, c->chrYInc, |
2255 c->chrSrcH, c->chrDstH, filterAlign, (1<<12)-4, | 2254 c->chrSrcH, c->chrDstH, filterAlign, (1<<12)-4, |
2256 (flags&SWS_BICUBLIN) ? (flags|SWS_BILINEAR) : flags, | 2255 (flags&SWS_BICUBLIN) ? (flags|SWS_BILINEAR) : flags, |
2257 srcFilter->chrV, dstFilter->chrV, c->param); | 2256 srcFilter->chrV, dstFilter->chrV, c->param); |
2258 | 2257 |
2259 #ifdef HAVE_ALTIVEC | 2258 #ifdef HAVE_ALTIVEC |
2260 c->vYCoeffsBank = av_malloc(sizeof (vector signed short)*c->vLumFilterSize*c->dstH); | 2259 c->vYCoeffsBank = av_malloc(sizeof (vector signed short)*c->vLumFilterSize*c->dstH); |
2261 c->vCCoeffsBank = av_malloc(sizeof (vector signed short)*c->vChrFilterSize*c->chrDstH); | 2260 c->vCCoeffsBank = av_malloc(sizeof (vector signed short)*c->vChrFilterSize*c->chrDstH); |
2262 | 2261 |
2263 for (i=0;i<c->vLumFilterSize*c->dstH;i++) { | 2262 for (i=0;i<c->vLumFilterSize*c->dstH;i++) { |
2264 int j; | 2263 int j; |
2265 short *p = (short *)&c->vYCoeffsBank[i]; | 2264 short *p = (short *)&c->vYCoeffsBank[i]; |
2266 for (j=0;j<8;j++) | 2265 for (j=0;j<8;j++) |
2267 p[j] = c->vLumFilter[i]; | 2266 p[j] = c->vLumFilter[i]; |
2268 } | 2267 } |
2269 | 2268 |
2270 for (i=0;i<c->vChrFilterSize*c->chrDstH;i++) { | 2269 for (i=0;i<c->vChrFilterSize*c->chrDstH;i++) { |
2271 int j; | 2270 int j; |
2272 short *p = (short *)&c->vCCoeffsBank[i]; | 2271 short *p = (short *)&c->vCCoeffsBank[i]; |
2273 for (j=0;j<8;j++) | 2272 for (j=0;j<8;j++) |
2274 p[j] = c->vChrFilter[i]; | 2273 p[j] = c->vChrFilter[i]; |
2275 } | 2274 } |
2276 #endif | 2275 #endif |
2277 } | 2276 } |
2278 | 2277 |
2279 // Calculate Buffer Sizes so that they won't run out while handling these damn slices | 2278 // Calculate Buffer Sizes so that they won't run out while handling these damn slices |
2280 c->vLumBufSize= c->vLumFilterSize; | 2279 c->vLumBufSize= c->vLumFilterSize; |
2281 c->vChrBufSize= c->vChrFilterSize; | 2280 c->vChrBufSize= c->vChrFilterSize; |
2282 for(i=0; i<dstH; i++) | 2281 for (i=0; i<dstH; i++) |
2283 { | 2282 { |
2284 int chrI= i*c->chrDstH / dstH; | 2283 int chrI= i*c->chrDstH / dstH; |
2285 int nextSlice= FFMAX(c->vLumFilterPos[i ] + c->vLumFilterSize - 1, | 2284 int nextSlice= FFMAX(c->vLumFilterPos[i ] + c->vLumFilterSize - 1, |
2286 ((c->vChrFilterPos[chrI] + c->vChrFilterSize - 1)<<c->chrSrcVSubSample)); | 2285 ((c->vChrFilterPos[chrI] + c->vChrFilterSize - 1)<<c->chrSrcVSubSample)); |
2287 | 2286 |
2288 nextSlice>>= c->chrSrcVSubSample; | 2287 nextSlice>>= c->chrSrcVSubSample; |
2289 nextSlice<<= c->chrSrcVSubSample; | 2288 nextSlice<<= c->chrSrcVSubSample; |
2290 if(c->vLumFilterPos[i ] + c->vLumBufSize < nextSlice) | 2289 if (c->vLumFilterPos[i ] + c->vLumBufSize < nextSlice) |
2291 c->vLumBufSize= nextSlice - c->vLumFilterPos[i ]; | 2290 c->vLumBufSize= nextSlice - c->vLumFilterPos[i ]; |
2292 if(c->vChrFilterPos[chrI] + c->vChrBufSize < (nextSlice>>c->chrSrcVSubSample)) | 2291 if (c->vChrFilterPos[chrI] + c->vChrBufSize < (nextSlice>>c->chrSrcVSubSample)) |
2293 c->vChrBufSize= (nextSlice>>c->chrSrcVSubSample) - c->vChrFilterPos[chrI]; | 2292 c->vChrBufSize= (nextSlice>>c->chrSrcVSubSample) - c->vChrFilterPos[chrI]; |
2294 } | 2293 } |
2295 | 2294 |
2296 // allocate pixbufs (we use dynamic allocation because otherwise we would need to | 2295 // allocate pixbufs (we use dynamic allocation because otherwise we would need to |
2297 c->lumPixBuf= av_malloc(c->vLumBufSize*2*sizeof(int16_t*)); | 2296 c->lumPixBuf= av_malloc(c->vLumBufSize*2*sizeof(int16_t*)); |
2298 c->chrPixBuf= av_malloc(c->vChrBufSize*2*sizeof(int16_t*)); | 2297 c->chrPixBuf= av_malloc(c->vChrBufSize*2*sizeof(int16_t*)); |
2299 //Note we need at least one pixel more at the end because of the mmx code (just in case someone wanna replace the 4000/8000) | 2298 //Note we need at least one pixel more at the end because of the mmx code (just in case someone wanna replace the 4000/8000) |
2300 /* align at 16 bytes for AltiVec */ | 2299 /* align at 16 bytes for AltiVec */ |
2301 for(i=0; i<c->vLumBufSize; i++) | 2300 for (i=0; i<c->vLumBufSize; i++) |
2302 c->lumPixBuf[i]= c->lumPixBuf[i+c->vLumBufSize]= av_mallocz(4000); | 2301 c->lumPixBuf[i]= c->lumPixBuf[i+c->vLumBufSize]= av_mallocz(4000); |
2303 for(i=0; i<c->vChrBufSize; i++) | 2302 for (i=0; i<c->vChrBufSize; i++) |
2304 c->chrPixBuf[i]= c->chrPixBuf[i+c->vChrBufSize]= av_malloc(8000); | 2303 c->chrPixBuf[i]= c->chrPixBuf[i+c->vChrBufSize]= av_malloc(8000); |
2305 | 2304 |
2306 //try to avoid drawing green stuff between the right end and the stride end | 2305 //try to avoid drawing green stuff between the right end and the stride end |
2307 for(i=0; i<c->vChrBufSize; i++) memset(c->chrPixBuf[i], 64, 8000); | 2306 for (i=0; i<c->vChrBufSize; i++) memset(c->chrPixBuf[i], 64, 8000); |
2308 | 2307 |
2309 ASSERT(c->chrDstH <= dstH) | 2308 ASSERT(c->chrDstH <= dstH) |
2310 | 2309 |
2311 if(flags&SWS_PRINT_INFO) | 2310 if (flags&SWS_PRINT_INFO) |
2312 { | 2311 { |
2313 #ifdef DITHER1XBPP | 2312 #ifdef DITHER1XBPP |
2314 char *dither= " dithered"; | 2313 char *dither= " dithered"; |
2315 #else | 2314 #else |
2316 char *dither= ""; | 2315 char *dither= ""; |
2317 #endif | 2316 #endif |
2318 if(flags&SWS_FAST_BILINEAR) | 2317 if (flags&SWS_FAST_BILINEAR) |
2319 av_log(c, AV_LOG_INFO, "SwScaler: FAST_BILINEAR scaler, "); | 2318 av_log(c, AV_LOG_INFO, "SwScaler: FAST_BILINEAR scaler, "); |
2320 else if(flags&SWS_BILINEAR) | 2319 else if (flags&SWS_BILINEAR) |
2321 av_log(c, AV_LOG_INFO, "SwScaler: BILINEAR scaler, "); | 2320 av_log(c, AV_LOG_INFO, "SwScaler: BILINEAR scaler, "); |
2322 else if(flags&SWS_BICUBIC) | 2321 else if (flags&SWS_BICUBIC) |
2323 av_log(c, AV_LOG_INFO, "SwScaler: BICUBIC scaler, "); | 2322 av_log(c, AV_LOG_INFO, "SwScaler: BICUBIC scaler, "); |
2324 else if(flags&SWS_X) | 2323 else if (flags&SWS_X) |
2325 av_log(c, AV_LOG_INFO, "SwScaler: Experimental scaler, "); | 2324 av_log(c, AV_LOG_INFO, "SwScaler: Experimental scaler, "); |
2326 else if(flags&SWS_POINT) | 2325 else if (flags&SWS_POINT) |
2327 av_log(c, AV_LOG_INFO, "SwScaler: Nearest Neighbor / POINT scaler, "); | 2326 av_log(c, AV_LOG_INFO, "SwScaler: Nearest Neighbor / POINT scaler, "); |
2328 else if(flags&SWS_AREA) | 2327 else if (flags&SWS_AREA) |
2329 av_log(c, AV_LOG_INFO, "SwScaler: Area Averageing scaler, "); | 2328 av_log(c, AV_LOG_INFO, "SwScaler: Area Averageing scaler, "); |
2330 else if(flags&SWS_BICUBLIN) | 2329 else if (flags&SWS_BICUBLIN) |
2331 av_log(c, AV_LOG_INFO, "SwScaler: luma BICUBIC / chroma BILINEAR scaler, "); | 2330 av_log(c, AV_LOG_INFO, "SwScaler: luma BICUBIC / chroma BILINEAR scaler, "); |
2332 else if(flags&SWS_GAUSS) | 2331 else if (flags&SWS_GAUSS) |
2333 av_log(c, AV_LOG_INFO, "SwScaler: Gaussian scaler, "); | 2332 av_log(c, AV_LOG_INFO, "SwScaler: Gaussian scaler, "); |
2334 else if(flags&SWS_SINC) | 2333 else if (flags&SWS_SINC) |
2335 av_log(c, AV_LOG_INFO, "SwScaler: Sinc scaler, "); | 2334 av_log(c, AV_LOG_INFO, "SwScaler: Sinc scaler, "); |
2336 else if(flags&SWS_LANCZOS) | 2335 else if (flags&SWS_LANCZOS) |
2337 av_log(c, AV_LOG_INFO, "SwScaler: Lanczos scaler, "); | 2336 av_log(c, AV_LOG_INFO, "SwScaler: Lanczos scaler, "); |
2338 else if(flags&SWS_SPLINE) | 2337 else if (flags&SWS_SPLINE) |
2339 av_log(c, AV_LOG_INFO, "SwScaler: Bicubic spline scaler, "); | 2338 av_log(c, AV_LOG_INFO, "SwScaler: Bicubic spline scaler, "); |
2340 else | 2339 else |
2341 av_log(c, AV_LOG_INFO, "SwScaler: ehh flags invalid?! "); | 2340 av_log(c, AV_LOG_INFO, "SwScaler: ehh flags invalid?! "); |
2342 | 2341 |
2343 if(dstFormat==PIX_FMT_BGR555 || dstFormat==PIX_FMT_BGR565) | 2342 if (dstFormat==PIX_FMT_BGR555 || dstFormat==PIX_FMT_BGR565) |
2344 av_log(c, AV_LOG_INFO, "from %s to%s %s ", | 2343 av_log(c, AV_LOG_INFO, "from %s to%s %s ", |
2345 sws_format_name(srcFormat), dither, sws_format_name(dstFormat)); | 2344 sws_format_name(srcFormat), dither, sws_format_name(dstFormat)); |
2346 else | 2345 else |
2347 av_log(c, AV_LOG_INFO, "from %s to %s ", | 2346 av_log(c, AV_LOG_INFO, "from %s to %s ", |
2348 sws_format_name(srcFormat), sws_format_name(dstFormat)); | 2347 sws_format_name(srcFormat), sws_format_name(dstFormat)); |
2349 | 2348 |
2350 if(flags & SWS_CPU_CAPS_MMX2) | 2349 if (flags & SWS_CPU_CAPS_MMX2) |
2351 av_log(c, AV_LOG_INFO, "using MMX2\n"); | 2350 av_log(c, AV_LOG_INFO, "using MMX2\n"); |
2352 else if(flags & SWS_CPU_CAPS_3DNOW) | 2351 else if (flags & SWS_CPU_CAPS_3DNOW) |
2353 av_log(c, AV_LOG_INFO, "using 3DNOW\n"); | 2352 av_log(c, AV_LOG_INFO, "using 3DNOW\n"); |
2354 else if(flags & SWS_CPU_CAPS_MMX) | 2353 else if (flags & SWS_CPU_CAPS_MMX) |
2355 av_log(c, AV_LOG_INFO, "using MMX\n"); | 2354 av_log(c, AV_LOG_INFO, "using MMX\n"); |
2356 else if(flags & SWS_CPU_CAPS_ALTIVEC) | 2355 else if (flags & SWS_CPU_CAPS_ALTIVEC) |
2357 av_log(c, AV_LOG_INFO, "using AltiVec\n"); | 2356 av_log(c, AV_LOG_INFO, "using AltiVec\n"); |
2358 else | 2357 else |
2359 av_log(c, AV_LOG_INFO, "using C\n"); | 2358 av_log(c, AV_LOG_INFO, "using C\n"); |
2360 } | 2359 } |
2361 | 2360 |
2362 if(flags & SWS_PRINT_INFO) | 2361 if (flags & SWS_PRINT_INFO) |
2363 { | 2362 { |
2364 if(flags & SWS_CPU_CAPS_MMX) | 2363 if (flags & SWS_CPU_CAPS_MMX) |
2365 { | 2364 { |
2366 if(c->canMMX2BeUsed && (flags&SWS_FAST_BILINEAR)) | 2365 if (c->canMMX2BeUsed && (flags&SWS_FAST_BILINEAR)) |
2367 av_log(c, AV_LOG_VERBOSE, "SwScaler: using FAST_BILINEAR MMX2 scaler for horizontal scaling\n"); | 2366 av_log(c, AV_LOG_VERBOSE, "SwScaler: using FAST_BILINEAR MMX2 scaler for horizontal scaling\n"); |
2368 else | 2367 else |
2369 { | 2368 { |
2370 if(c->hLumFilterSize==4) | 2369 if (c->hLumFilterSize==4) |
2371 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 4-tap MMX scaler for horizontal luminance scaling\n"); | 2370 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 4-tap MMX scaler for horizontal luminance scaling\n"); |
2372 else if(c->hLumFilterSize==8) | 2371 else if (c->hLumFilterSize==8) |
2373 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 8-tap MMX scaler for horizontal luminance scaling\n"); | 2372 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 8-tap MMX scaler for horizontal luminance scaling\n"); |
2374 else | 2373 else |
2375 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap MMX scaler for horizontal luminance scaling\n"); | 2374 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap MMX scaler for horizontal luminance scaling\n"); |
2376 | 2375 |
2377 if(c->hChrFilterSize==4) | 2376 if (c->hChrFilterSize==4) |
2378 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 4-tap MMX scaler for horizontal chrominance scaling\n"); | 2377 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 4-tap MMX scaler for horizontal chrominance scaling\n"); |
2379 else if(c->hChrFilterSize==8) | 2378 else if (c->hChrFilterSize==8) |
2380 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 8-tap MMX scaler for horizontal chrominance scaling\n"); | 2379 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 8-tap MMX scaler for horizontal chrominance scaling\n"); |
2381 else | 2380 else |
2382 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap MMX scaler for horizontal chrominance scaling\n"); | 2381 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap MMX scaler for horizontal chrominance scaling\n"); |
2383 } | 2382 } |
2384 } | 2383 } |
2385 else | 2384 else |
2386 { | 2385 { |
2387 #if defined(ARCH_X86) | 2386 #if defined(ARCH_X86) |
2388 av_log(c, AV_LOG_VERBOSE, "SwScaler: using X86-Asm scaler for horizontal scaling\n"); | 2387 av_log(c, AV_LOG_VERBOSE, "SwScaler: using X86-Asm scaler for horizontal scaling\n"); |
2389 #else | 2388 #else |
2390 if(flags & SWS_FAST_BILINEAR) | 2389 if (flags & SWS_FAST_BILINEAR) |
2391 av_log(c, AV_LOG_VERBOSE, "SwScaler: using FAST_BILINEAR C scaler for horizontal scaling\n"); | 2390 av_log(c, AV_LOG_VERBOSE, "SwScaler: using FAST_BILINEAR C scaler for horizontal scaling\n"); |
2392 else | 2391 else |
2393 av_log(c, AV_LOG_VERBOSE, "SwScaler: using C scaler for horizontal scaling\n"); | 2392 av_log(c, AV_LOG_VERBOSE, "SwScaler: using C scaler for horizontal scaling\n"); |
2394 #endif | 2393 #endif |
2395 } | 2394 } |
2396 if(isPlanarYUV(dstFormat)) | 2395 if (isPlanarYUV(dstFormat)) |
2397 { | 2396 { |
2398 if(c->vLumFilterSize==1) | 2397 if (c->vLumFilterSize==1) |
2399 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 1-tap %s \"scaler\" for vertical scaling (YV12 like)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2398 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 1-tap %s \"scaler\" for vertical scaling (YV12 like)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2400 else | 2399 else |
2401 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap %s scaler for vertical scaling (YV12 like)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2400 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap %s scaler for vertical scaling (YV12 like)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2402 } | 2401 } |
2403 else | 2402 else |
2404 { | 2403 { |
2405 if(c->vLumFilterSize==1 && c->vChrFilterSize==2) | 2404 if (c->vLumFilterSize==1 && c->vChrFilterSize==2) |
2406 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 1-tap %s \"scaler\" for vertical luminance scaling (BGR)\n" | 2405 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 1-tap %s \"scaler\" for vertical luminance scaling (BGR)\n" |
2407 "SwScaler: 2-tap scaler for vertical chrominance scaling (BGR)\n",(flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2406 "SwScaler: 2-tap scaler for vertical chrominance scaling (BGR)\n",(flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2408 else if(c->vLumFilterSize==2 && c->vChrFilterSize==2) | 2407 else if (c->vLumFilterSize==2 && c->vChrFilterSize==2) |
2409 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 2-tap linear %s scaler for vertical scaling (BGR)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2408 av_log(c, AV_LOG_VERBOSE, "SwScaler: using 2-tap linear %s scaler for vertical scaling (BGR)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2410 else | 2409 else |
2411 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap %s scaler for vertical scaling (BGR)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2410 av_log(c, AV_LOG_VERBOSE, "SwScaler: using n-tap %s scaler for vertical scaling (BGR)\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2412 } | 2411 } |
2413 | 2412 |
2414 if(dstFormat==PIX_FMT_BGR24) | 2413 if (dstFormat==PIX_FMT_BGR24) |
2415 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR24 Converter\n", | 2414 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR24 Converter\n", |
2416 (flags & SWS_CPU_CAPS_MMX2) ? "MMX2" : ((flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C")); | 2415 (flags & SWS_CPU_CAPS_MMX2) ? "MMX2" : ((flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C")); |
2417 else if(dstFormat==PIX_FMT_RGB32) | 2416 else if (dstFormat==PIX_FMT_RGB32) |
2418 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR32 Converter\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2417 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR32 Converter\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2419 else if(dstFormat==PIX_FMT_BGR565) | 2418 else if (dstFormat==PIX_FMT_BGR565) |
2420 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR16 Converter\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2419 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR16 Converter\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2421 else if(dstFormat==PIX_FMT_BGR555) | 2420 else if (dstFormat==PIX_FMT_BGR555) |
2422 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR15 Converter\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); | 2421 av_log(c, AV_LOG_VERBOSE, "SwScaler: using %s YV12->BGR15 Converter\n", (flags & SWS_CPU_CAPS_MMX) ? "MMX" : "C"); |
2423 | 2422 |
2424 av_log(c, AV_LOG_VERBOSE, "SwScaler: %dx%d -> %dx%d\n", srcW, srcH, dstW, dstH); | 2423 av_log(c, AV_LOG_VERBOSE, "SwScaler: %dx%d -> %dx%d\n", srcW, srcH, dstW, dstH); |
2425 } | 2424 } |
2426 if(flags & SWS_PRINT_INFO) | 2425 if (flags & SWS_PRINT_INFO) |
2427 { | 2426 { |
2428 av_log(c, AV_LOG_DEBUG, "SwScaler:Lum srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n", | 2427 av_log(c, AV_LOG_DEBUG, "SwScaler:Lum srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n", |
2429 c->srcW, c->srcH, c->dstW, c->dstH, c->lumXInc, c->lumYInc); | 2428 c->srcW, c->srcH, c->dstW, c->dstH, c->lumXInc, c->lumYInc); |
2430 av_log(c, AV_LOG_DEBUG, "SwScaler:Chr srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n", | 2429 av_log(c, AV_LOG_DEBUG, "SwScaler:Chr srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n", |
2431 c->chrSrcW, c->chrSrcH, c->chrDstW, c->chrDstH, c->chrXInc, c->chrYInc); | 2430 c->chrSrcW, c->chrSrcH, c->chrDstW, c->chrDstH, c->chrXInc, c->chrYInc); |
2432 } | 2431 } |
2433 | 2432 |
2434 c->swScale= getSwsFunc(flags); | 2433 c->swScale= getSwsFunc(flags); |
2435 return c; | 2434 return c; |
2436 } | 2435 } |
2437 | 2436 |
2438 /** | 2437 /** |
2439 * swscale warper, so we don't need to export the SwsContext. | 2438 * swscale warper, so we don't need to export the SwsContext. |
2440 * assumes planar YUV to be in YUV order instead of YVU | 2439 * assumes planar YUV to be in YUV order instead of YVU |
2441 */ | 2440 */ |
2442 int sws_scale(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 2441 int sws_scale(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
2443 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 2442 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
2444 int i; | 2443 int i; |
2445 uint8_t* src2[4]= {src[0], src[1], src[2]}; | 2444 uint8_t* src2[4]= {src[0], src[1], src[2]}; |
2446 uint32_t pal[256]; | 2445 uint32_t pal[256]; |
2447 if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) { | 2446 if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) { |
2448 av_log(c, AV_LOG_ERROR, "swScaler: slices start in the middle!\n"); | 2447 av_log(c, AV_LOG_ERROR, "swScaler: slices start in the middle!\n"); |
2449 return 0; | 2448 return 0; |
2450 } | 2449 } |
2451 if (c->sliceDir == 0) { | 2450 if (c->sliceDir == 0) { |
2452 if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1; | 2451 if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1; |
2453 } | 2452 } |
2454 | 2453 |
2455 if(c->srcFormat == PIX_FMT_PAL8){ | 2454 if (c->srcFormat == PIX_FMT_PAL8){ |
2456 for(i=0; i<256; i++){ | 2455 for (i=0; i<256; i++){ |
2457 int p= ((uint32_t*)(src[1]))[i]; | 2456 int p= ((uint32_t*)(src[1]))[i]; |
2458 int r= (p>>16)&0xFF; | 2457 int r= (p>>16)&0xFF; |
2459 int g= (p>> 8)&0xFF; | 2458 int g= (p>> 8)&0xFF; |
2460 int b= p &0xFF; | 2459 int b= p &0xFF; |
2461 int y= av_clip_uint8(((RY*r + GY*g + BY*b)>>RGB2YUV_SHIFT) + 16 ); | 2460 int y= av_clip_uint8(((RY*r + GY*g + BY*b)>>RGB2YUV_SHIFT) + 16 ); |
2462 int u= av_clip_uint8(((RU*r + GU*g + BU*b)>>RGB2YUV_SHIFT) + 128); | 2461 int u= av_clip_uint8(((RU*r + GU*g + BU*b)>>RGB2YUV_SHIFT) + 128); |
2463 int v= av_clip_uint8(((RV*r + GV*g + BV*b)>>RGB2YUV_SHIFT) + 128); | 2462 int v= av_clip_uint8(((RV*r + GV*g + BV*b)>>RGB2YUV_SHIFT) + 128); |
2464 pal[i]= y + (u<<8) + (v<<16); | 2463 pal[i]= y + (u<<8) + (v<<16); |
2465 } | 2464 } |
2466 src2[1]= pal; | 2465 src2[1]= pal; |
2467 } | 2466 } |
2468 | 2467 |
2469 // copy strides, so they can safely be modified | 2468 // copy strides, so they can safely be modified |
2470 if (c->sliceDir == 1) { | 2469 if (c->sliceDir == 1) { |
2471 // slices go from top to bottom | 2470 // slices go from top to bottom |
2472 int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2]}; | 2471 int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2]}; |
2473 int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2]}; | 2472 int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2]}; |
2474 return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst, dstStride2); | 2473 return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst, dstStride2); |
2475 } else { | 2474 } else { |
2476 // slices go from bottom to top => we flip the image internally | 2475 // slices go from bottom to top => we flip the image internally |
2477 uint8_t* dst2[4]= {dst[0] + (c->dstH-1)*dstStride[0], | 2476 uint8_t* dst2[4]= {dst[0] + (c->dstH-1)*dstStride[0], |
2478 dst[1] + ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1], | 2477 dst[1] + ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1], |
2479 dst[2] + ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2]}; | 2478 dst[2] + ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2]}; |
2480 int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2]}; | 2479 int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2]}; |
2481 int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2]}; | 2480 int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2]}; |
2482 | 2481 |
2483 src2[0] += (srcSliceH-1)*srcStride[0]; | 2482 src2[0] += (srcSliceH-1)*srcStride[0]; |
2484 if(c->srcFormat != PIX_FMT_PAL8) | 2483 if (c->srcFormat != PIX_FMT_PAL8) |
2485 src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1]; | 2484 src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1]; |
2486 src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2]; | 2485 src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2]; |
2487 | 2486 |
2488 return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2); | 2487 return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2); |
2489 } | 2488 } |
2490 } | 2489 } |
2491 | 2490 |
2492 /** | 2491 /** |
2493 * swscale warper, so we don't need to export the SwsContext | 2492 * swscale warper, so we don't need to export the SwsContext |
2494 */ | 2493 */ |
2495 int sws_scale_ordered(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, | 2494 int sws_scale_ordered(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, |
2496 int srcSliceH, uint8_t* dst[], int dstStride[]){ | 2495 int srcSliceH, uint8_t* dst[], int dstStride[]){ |
2497 return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride); | 2496 return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride); |
2498 } | 2497 } |
2499 | 2498 |
2500 SwsFilter *sws_getDefaultFilter(float lumaGBlur, float chromaGBlur, | 2499 SwsFilter *sws_getDefaultFilter(float lumaGBlur, float chromaGBlur, |
2501 float lumaSharpen, float chromaSharpen, | 2500 float lumaSharpen, float chromaSharpen, |
2502 float chromaHShift, float chromaVShift, | 2501 float chromaHShift, float chromaVShift, |
2503 int verbose) | 2502 int verbose) |
2504 { | 2503 { |
2505 SwsFilter *filter= av_malloc(sizeof(SwsFilter)); | 2504 SwsFilter *filter= av_malloc(sizeof(SwsFilter)); |
2506 | 2505 |
2507 if(lumaGBlur!=0.0){ | 2506 if (lumaGBlur!=0.0){ |
2508 filter->lumH= sws_getGaussianVec(lumaGBlur, 3.0); | 2507 filter->lumH= sws_getGaussianVec(lumaGBlur, 3.0); |
2509 filter->lumV= sws_getGaussianVec(lumaGBlur, 3.0); | 2508 filter->lumV= sws_getGaussianVec(lumaGBlur, 3.0); |
2510 }else{ | 2509 }else{ |
2511 filter->lumH= sws_getIdentityVec(); | 2510 filter->lumH= sws_getIdentityVec(); |
2512 filter->lumV= sws_getIdentityVec(); | 2511 filter->lumV= sws_getIdentityVec(); |
2513 } | 2512 } |
2514 | 2513 |
2515 if(chromaGBlur!=0.0){ | 2514 if (chromaGBlur!=0.0){ |
2516 filter->chrH= sws_getGaussianVec(chromaGBlur, 3.0); | 2515 filter->chrH= sws_getGaussianVec(chromaGBlur, 3.0); |
2517 filter->chrV= sws_getGaussianVec(chromaGBlur, 3.0); | 2516 filter->chrV= sws_getGaussianVec(chromaGBlur, 3.0); |
2518 }else{ | 2517 }else{ |
2519 filter->chrH= sws_getIdentityVec(); | 2518 filter->chrH= sws_getIdentityVec(); |
2520 filter->chrV= sws_getIdentityVec(); | 2519 filter->chrV= sws_getIdentityVec(); |
2521 } | 2520 } |
2522 | 2521 |
2523 if(chromaSharpen!=0.0){ | 2522 if (chromaSharpen!=0.0){ |
2524 SwsVector *id= sws_getIdentityVec(); | 2523 SwsVector *id= sws_getIdentityVec(); |
2525 sws_scaleVec(filter->chrH, -chromaSharpen); | 2524 sws_scaleVec(filter->chrH, -chromaSharpen); |
2526 sws_scaleVec(filter->chrV, -chromaSharpen); | 2525 sws_scaleVec(filter->chrV, -chromaSharpen); |
2527 sws_addVec(filter->chrH, id); | 2526 sws_addVec(filter->chrH, id); |
2528 sws_addVec(filter->chrV, id); | 2527 sws_addVec(filter->chrV, id); |
2529 sws_freeVec(id); | 2528 sws_freeVec(id); |
2530 } | 2529 } |
2531 | 2530 |
2532 if(lumaSharpen!=0.0){ | 2531 if (lumaSharpen!=0.0){ |
2533 SwsVector *id= sws_getIdentityVec(); | 2532 SwsVector *id= sws_getIdentityVec(); |
2534 sws_scaleVec(filter->lumH, -lumaSharpen); | 2533 sws_scaleVec(filter->lumH, -lumaSharpen); |
2535 sws_scaleVec(filter->lumV, -lumaSharpen); | 2534 sws_scaleVec(filter->lumV, -lumaSharpen); |
2536 sws_addVec(filter->lumH, id); | 2535 sws_addVec(filter->lumH, id); |
2537 sws_addVec(filter->lumV, id); | 2536 sws_addVec(filter->lumV, id); |
2538 sws_freeVec(id); | 2537 sws_freeVec(id); |
2539 } | 2538 } |
2540 | 2539 |
2541 if(chromaHShift != 0.0) | 2540 if (chromaHShift != 0.0) |
2542 sws_shiftVec(filter->chrH, (int)(chromaHShift+0.5)); | 2541 sws_shiftVec(filter->chrH, (int)(chromaHShift+0.5)); |
2543 | 2542 |
2544 if(chromaVShift != 0.0) | 2543 if (chromaVShift != 0.0) |
2545 sws_shiftVec(filter->chrV, (int)(chromaVShift+0.5)); | 2544 sws_shiftVec(filter->chrV, (int)(chromaVShift+0.5)); |
2546 | 2545 |
2547 sws_normalizeVec(filter->chrH, 1.0); | 2546 sws_normalizeVec(filter->chrH, 1.0); |
2548 sws_normalizeVec(filter->chrV, 1.0); | 2547 sws_normalizeVec(filter->chrV, 1.0); |
2549 sws_normalizeVec(filter->lumH, 1.0); | 2548 sws_normalizeVec(filter->lumH, 1.0); |
2550 sws_normalizeVec(filter->lumV, 1.0); | 2549 sws_normalizeVec(filter->lumV, 1.0); |
2551 | 2550 |
2552 if(verbose) sws_printVec(filter->chrH); | 2551 if (verbose) sws_printVec(filter->chrH); |
2553 if(verbose) sws_printVec(filter->lumH); | 2552 if (verbose) sws_printVec(filter->lumH); |
2554 | 2553 |
2555 return filter; | 2554 return filter; |
2556 } | 2555 } |
2557 | 2556 |
2558 /** | 2557 /** |
2559 * returns a normalized gaussian curve used to filter stuff | 2558 * returns a normalized gaussian curve used to filter stuff |
2560 * quality=3 is high quality, lowwer is lowwer quality | 2559 * quality=3 is high quality, lowwer is lowwer quality |
2561 */ | 2560 */ |
2562 SwsVector *sws_getGaussianVec(double variance, double quality){ | 2561 SwsVector *sws_getGaussianVec(double variance, double quality){ |
2563 const int length= (int)(variance*quality + 0.5) | 1; | 2562 const int length= (int)(variance*quality + 0.5) | 1; |
2564 int i; | 2563 int i; |
2565 double *coeff= av_malloc(length*sizeof(double)); | 2564 double *coeff= av_malloc(length*sizeof(double)); |
2566 double middle= (length-1)*0.5; | 2565 double middle= (length-1)*0.5; |
2567 SwsVector *vec= av_malloc(sizeof(SwsVector)); | 2566 SwsVector *vec= av_malloc(sizeof(SwsVector)); |
2568 | 2567 |
2569 vec->coeff= coeff; | 2568 vec->coeff= coeff; |
2570 vec->length= length; | 2569 vec->length= length; |
2571 | 2570 |
2572 for(i=0; i<length; i++) | 2571 for (i=0; i<length; i++) |
2573 { | 2572 { |
2574 double dist= i-middle; | 2573 double dist= i-middle; |
2575 coeff[i]= exp( -dist*dist/(2*variance*variance) ) / sqrt(2*variance*PI); | 2574 coeff[i]= exp( -dist*dist/(2*variance*variance) ) / sqrt(2*variance*PI); |
2576 } | 2575 } |
2577 | 2576 |
2578 sws_normalizeVec(vec, 1.0); | 2577 sws_normalizeVec(vec, 1.0); |
2579 | 2578 |
2580 return vec; | 2579 return vec; |
2581 } | 2580 } |
2582 | 2581 |
2583 SwsVector *sws_getConstVec(double c, int length){ | 2582 SwsVector *sws_getConstVec(double c, int length){ |
2584 int i; | 2583 int i; |
2585 double *coeff= av_malloc(length*sizeof(double)); | 2584 double *coeff= av_malloc(length*sizeof(double)); |
2586 SwsVector *vec= av_malloc(sizeof(SwsVector)); | 2585 SwsVector *vec= av_malloc(sizeof(SwsVector)); |
2587 | 2586 |
2588 vec->coeff= coeff; | 2587 vec->coeff= coeff; |
2589 vec->length= length; | 2588 vec->length= length; |
2590 | 2589 |
2591 for(i=0; i<length; i++) | 2590 for (i=0; i<length; i++) |
2592 coeff[i]= c; | 2591 coeff[i]= c; |
2593 | 2592 |
2594 return vec; | 2593 return vec; |
2595 } | 2594 } |
2596 | 2595 |
2597 | 2596 |
2598 SwsVector *sws_getIdentityVec(void){ | 2597 SwsVector *sws_getIdentityVec(void){ |
2599 return sws_getConstVec(1.0, 1); | 2598 return sws_getConstVec(1.0, 1); |
2600 } | 2599 } |
2601 | 2600 |
2602 double sws_dcVec(SwsVector *a){ | 2601 double sws_dcVec(SwsVector *a){ |
2603 int i; | 2602 int i; |
2604 double sum=0; | 2603 double sum=0; |
2605 | 2604 |
2606 for(i=0; i<a->length; i++) | 2605 for (i=0; i<a->length; i++) |
2607 sum+= a->coeff[i]; | 2606 sum+= a->coeff[i]; |
2608 | 2607 |
2609 return sum; | 2608 return sum; |
2610 } | 2609 } |
2611 | 2610 |
2612 void sws_scaleVec(SwsVector *a, double scalar){ | 2611 void sws_scaleVec(SwsVector *a, double scalar){ |
2613 int i; | 2612 int i; |
2614 | 2613 |
2615 for(i=0; i<a->length; i++) | 2614 for (i=0; i<a->length; i++) |
2616 a->coeff[i]*= scalar; | 2615 a->coeff[i]*= scalar; |
2617 } | 2616 } |
2618 | 2617 |
2619 void sws_normalizeVec(SwsVector *a, double height){ | 2618 void sws_normalizeVec(SwsVector *a, double height){ |
2620 sws_scaleVec(a, height/sws_dcVec(a)); | 2619 sws_scaleVec(a, height/sws_dcVec(a)); |
2621 } | 2620 } |
2622 | 2621 |
2623 static SwsVector *sws_getConvVec(SwsVector *a, SwsVector *b){ | 2622 static SwsVector *sws_getConvVec(SwsVector *a, SwsVector *b){ |
2624 int length= a->length + b->length - 1; | 2623 int length= a->length + b->length - 1; |
2625 double *coeff= av_malloc(length*sizeof(double)); | 2624 double *coeff= av_malloc(length*sizeof(double)); |
2626 int i, j; | 2625 int i, j; |
2627 SwsVector *vec= av_malloc(sizeof(SwsVector)); | 2626 SwsVector *vec= av_malloc(sizeof(SwsVector)); |
2628 | 2627 |
2629 vec->coeff= coeff; | 2628 vec->coeff= coeff; |
2630 vec->length= length; | 2629 vec->length= length; |
2631 | 2630 |
2632 for(i=0; i<length; i++) coeff[i]= 0.0; | 2631 for (i=0; i<length; i++) coeff[i]= 0.0; |
2633 | 2632 |
2634 for(i=0; i<a->length; i++) | 2633 for (i=0; i<a->length; i++) |
2635 { | 2634 { |
2636 for(j=0; j<b->length; j++) | 2635 for (j=0; j<b->length; j++) |
2637 { | 2636 { |
2638 coeff[i+j]+= a->coeff[i]*b->coeff[j]; | 2637 coeff[i+j]+= a->coeff[i]*b->coeff[j]; |
2639 } | 2638 } |
2640 } | 2639 } |
2641 | 2640 |
2642 return vec; | 2641 return vec; |
2643 } | 2642 } |
2644 | 2643 |
2645 static SwsVector *sws_sumVec(SwsVector *a, SwsVector *b){ | 2644 static SwsVector *sws_sumVec(SwsVector *a, SwsVector *b){ |
2646 int length= FFMAX(a->length, b->length); | 2645 int length= FFMAX(a->length, b->length); |
2647 double *coeff= av_malloc(length*sizeof(double)); | 2646 double *coeff= av_malloc(length*sizeof(double)); |
2648 int i; | 2647 int i; |
2649 SwsVector *vec= av_malloc(sizeof(SwsVector)); | 2648 SwsVector *vec= av_malloc(sizeof(SwsVector)); |
2650 | 2649 |
2651 vec->coeff= coeff; | 2650 vec->coeff= coeff; |
2652 vec->length= length; | 2651 vec->length= length; |
2653 | 2652 |
2654 for(i=0; i<length; i++) coeff[i]= 0.0; | 2653 for (i=0; i<length; i++) coeff[i]= 0.0; |
2655 | 2654 |
2656 for(i=0; i<a->length; i++) coeff[i + (length-1)/2 - (a->length-1)/2]+= a->coeff[i]; | 2655 for (i=0; i<a->length; i++) coeff[i + (length-1)/2 - (a->length-1)/2]+= a->coeff[i]; |
2657 for(i=0; i<b->length; i++) coeff[i + (length-1)/2 - (b->length-1)/2]+= b->coeff[i]; | 2656 for (i=0; i<b->length; i++) coeff[i + (length-1)/2 - (b->length-1)/2]+= b->coeff[i]; |
2658 | 2657 |
2659 return vec; | 2658 return vec; |
2660 } | 2659 } |
2661 | 2660 |
2662 static SwsVector *sws_diffVec(SwsVector *a, SwsVector *b){ | 2661 static SwsVector *sws_diffVec(SwsVector *a, SwsVector *b){ |
2663 int length= FFMAX(a->length, b->length); | 2662 int length= FFMAX(a->length, b->length); |
2664 double *coeff= av_malloc(length*sizeof(double)); | 2663 double *coeff= av_malloc(length*sizeof(double)); |
2665 int i; | 2664 int i; |
2666 SwsVector *vec= av_malloc(sizeof(SwsVector)); | 2665 SwsVector *vec= av_malloc(sizeof(SwsVector)); |
2667 | 2666 |
2668 vec->coeff= coeff; | 2667 vec->coeff= coeff; |
2669 vec->length= length; | 2668 vec->length= length; |
2670 | 2669 |
2671 for(i=0; i<length; i++) coeff[i]= 0.0; | 2670 for (i=0; i<length; i++) coeff[i]= 0.0; |
2672 | 2671 |
2673 for(i=0; i<a->length; i++) coeff[i + (length-1)/2 - (a->length-1)/2]+= a->coeff[i]; | 2672 for (i=0; i<a->length; i++) coeff[i + (length-1)/2 - (a->length-1)/2]+= a->coeff[i]; |
2674 for(i=0; i<b->length; i++) coeff[i + (length-1)/2 - (b->length-1)/2]-= b->coeff[i]; | 2673 for (i=0; i<b->length; i++) coeff[i + (length-1)/2 - (b->length-1)/2]-= b->coeff[i]; |
2675 | 2674 |
2676 return vec; | 2675 return vec; |
2677 } | 2676 } |
2678 | 2677 |
2679 /* shift left / or right if "shift" is negative */ | 2678 /* shift left / or right if "shift" is negative */ |
2680 static SwsVector *sws_getShiftedVec(SwsVector *a, int shift){ | 2679 static SwsVector *sws_getShiftedVec(SwsVector *a, int shift){ |
2681 int length= a->length + FFABS(shift)*2; | 2680 int length= a->length + FFABS(shift)*2; |
2682 double *coeff= av_malloc(length*sizeof(double)); | 2681 double *coeff= av_malloc(length*sizeof(double)); |
2683 int i; | 2682 int i; |
2684 SwsVector *vec= av_malloc(sizeof(SwsVector)); | 2683 SwsVector *vec= av_malloc(sizeof(SwsVector)); |
2685 | 2684 |
2686 vec->coeff= coeff; | 2685 vec->coeff= coeff; |
2687 vec->length= length; | 2686 vec->length= length; |
2688 | 2687 |
2689 for(i=0; i<length; i++) coeff[i]= 0.0; | 2688 for (i=0; i<length; i++) coeff[i]= 0.0; |
2690 | 2689 |
2691 for(i=0; i<a->length; i++) | 2690 for (i=0; i<a->length; i++) |
2692 { | 2691 { |
2693 coeff[i + (length-1)/2 - (a->length-1)/2 - shift]= a->coeff[i]; | 2692 coeff[i + (length-1)/2 - (a->length-1)/2 - shift]= a->coeff[i]; |
2694 } | 2693 } |
2695 | 2694 |
2696 return vec; | 2695 return vec; |
2697 } | 2696 } |
2698 | 2697 |
2699 void sws_shiftVec(SwsVector *a, int shift){ | 2698 void sws_shiftVec(SwsVector *a, int shift){ |
2700 SwsVector *shifted= sws_getShiftedVec(a, shift); | 2699 SwsVector *shifted= sws_getShiftedVec(a, shift); |
2701 av_free(a->coeff); | 2700 av_free(a->coeff); |
2702 a->coeff= shifted->coeff; | 2701 a->coeff= shifted->coeff; |
2703 a->length= shifted->length; | 2702 a->length= shifted->length; |
2704 av_free(shifted); | 2703 av_free(shifted); |
2705 } | 2704 } |
2706 | 2705 |
2707 void sws_addVec(SwsVector *a, SwsVector *b){ | 2706 void sws_addVec(SwsVector *a, SwsVector *b){ |
2708 SwsVector *sum= sws_sumVec(a, b); | 2707 SwsVector *sum= sws_sumVec(a, b); |
2709 av_free(a->coeff); | 2708 av_free(a->coeff); |
2710 a->coeff= sum->coeff; | 2709 a->coeff= sum->coeff; |
2711 a->length= sum->length; | 2710 a->length= sum->length; |
2712 av_free(sum); | 2711 av_free(sum); |
2713 } | 2712 } |
2714 | 2713 |
2715 void sws_subVec(SwsVector *a, SwsVector *b){ | 2714 void sws_subVec(SwsVector *a, SwsVector *b){ |
2716 SwsVector *diff= sws_diffVec(a, b); | 2715 SwsVector *diff= sws_diffVec(a, b); |
2717 av_free(a->coeff); | 2716 av_free(a->coeff); |
2718 a->coeff= diff->coeff; | 2717 a->coeff= diff->coeff; |
2719 a->length= diff->length; | 2718 a->length= diff->length; |
2720 av_free(diff); | 2719 av_free(diff); |
2721 } | 2720 } |
2722 | 2721 |
2723 void sws_convVec(SwsVector *a, SwsVector *b){ | 2722 void sws_convVec(SwsVector *a, SwsVector *b){ |
2724 SwsVector *conv= sws_getConvVec(a, b); | 2723 SwsVector *conv= sws_getConvVec(a, b); |
2725 av_free(a->coeff); | 2724 av_free(a->coeff); |
2726 a->coeff= conv->coeff; | 2725 a->coeff= conv->coeff; |
2727 a->length= conv->length; | 2726 a->length= conv->length; |
2728 av_free(conv); | 2727 av_free(conv); |
2729 } | 2728 } |
2730 | 2729 |
2731 SwsVector *sws_cloneVec(SwsVector *a){ | 2730 SwsVector *sws_cloneVec(SwsVector *a){ |
2732 double *coeff= av_malloc(a->length*sizeof(double)); | 2731 double *coeff= av_malloc(a->length*sizeof(double)); |
2733 int i; | 2732 int i; |
2734 SwsVector *vec= av_malloc(sizeof(SwsVector)); | 2733 SwsVector *vec= av_malloc(sizeof(SwsVector)); |
2735 | 2734 |
2736 vec->coeff= coeff; | 2735 vec->coeff= coeff; |
2737 vec->length= a->length; | 2736 vec->length= a->length; |
2738 | 2737 |
2739 for(i=0; i<a->length; i++) coeff[i]= a->coeff[i]; | 2738 for (i=0; i<a->length; i++) coeff[i]= a->coeff[i]; |
2740 | 2739 |
2741 return vec; | 2740 return vec; |
2742 } | 2741 } |
2743 | 2742 |
2744 void sws_printVec(SwsVector *a){ | 2743 void sws_printVec(SwsVector *a){ |
2745 int i; | 2744 int i; |
2746 double max=0; | 2745 double max=0; |
2747 double min=0; | 2746 double min=0; |
2748 double range; | 2747 double range; |
2749 | 2748 |
2750 for(i=0; i<a->length; i++) | 2749 for (i=0; i<a->length; i++) |
2751 if(a->coeff[i]>max) max= a->coeff[i]; | 2750 if (a->coeff[i]>max) max= a->coeff[i]; |
2752 | 2751 |
2753 for(i=0; i<a->length; i++) | 2752 for (i=0; i<a->length; i++) |
2754 if(a->coeff[i]<min) min= a->coeff[i]; | 2753 if (a->coeff[i]<min) min= a->coeff[i]; |
2755 | 2754 |
2756 range= max - min; | 2755 range= max - min; |
2757 | 2756 |
2758 for(i=0; i<a->length; i++) | 2757 for (i=0; i<a->length; i++) |
2759 { | 2758 { |
2760 int x= (int)((a->coeff[i]-min)*60.0/range +0.5); | 2759 int x= (int)((a->coeff[i]-min)*60.0/range +0.5); |
2761 av_log(NULL, AV_LOG_DEBUG, "%1.3f ", a->coeff[i]); | 2760 av_log(NULL, AV_LOG_DEBUG, "%1.3f ", a->coeff[i]); |
2762 for(;x>0; x--) av_log(NULL, AV_LOG_DEBUG, " "); | 2761 for (;x>0; x--) av_log(NULL, AV_LOG_DEBUG, " "); |
2763 av_log(NULL, AV_LOG_DEBUG, "|\n"); | 2762 av_log(NULL, AV_LOG_DEBUG, "|\n"); |
2764 } | 2763 } |
2765 } | 2764 } |
2766 | 2765 |
2767 void sws_freeVec(SwsVector *a){ | 2766 void sws_freeVec(SwsVector *a){ |
2768 if(!a) return; | 2767 if (!a) return; |
2769 av_free(a->coeff); | 2768 av_free(a->coeff); |
2770 a->coeff=NULL; | 2769 a->coeff=NULL; |
2771 a->length=0; | 2770 a->length=0; |
2772 av_free(a); | 2771 av_free(a); |
2773 } | 2772 } |
2774 | 2773 |
2775 void sws_freeFilter(SwsFilter *filter){ | 2774 void sws_freeFilter(SwsFilter *filter){ |
2776 if(!filter) return; | 2775 if (!filter) return; |
2777 | 2776 |
2778 if(filter->lumH) sws_freeVec(filter->lumH); | 2777 if (filter->lumH) sws_freeVec(filter->lumH); |
2779 if(filter->lumV) sws_freeVec(filter->lumV); | 2778 if (filter->lumV) sws_freeVec(filter->lumV); |
2780 if(filter->chrH) sws_freeVec(filter->chrH); | 2779 if (filter->chrH) sws_freeVec(filter->chrH); |
2781 if(filter->chrV) sws_freeVec(filter->chrV); | 2780 if (filter->chrV) sws_freeVec(filter->chrV); |
2782 av_free(filter); | 2781 av_free(filter); |
2783 } | 2782 } |
2784 | 2783 |
2785 | 2784 |
2786 void sws_freeContext(SwsContext *c){ | 2785 void sws_freeContext(SwsContext *c){ |
2787 int i; | 2786 int i; |
2788 if(!c) return; | 2787 if (!c) return; |
2789 | 2788 |
2790 if(c->lumPixBuf) | 2789 if (c->lumPixBuf) |
2791 { | 2790 { |
2792 for(i=0; i<c->vLumBufSize; i++) | 2791 for (i=0; i<c->vLumBufSize; i++) |
2793 { | 2792 { |
2794 av_free(c->lumPixBuf[i]); | 2793 av_free(c->lumPixBuf[i]); |
2795 c->lumPixBuf[i]=NULL; | 2794 c->lumPixBuf[i]=NULL; |
2796 } | 2795 } |
2797 av_free(c->lumPixBuf); | 2796 av_free(c->lumPixBuf); |
2798 c->lumPixBuf=NULL; | 2797 c->lumPixBuf=NULL; |
2799 } | 2798 } |
2800 | 2799 |
2801 if(c->chrPixBuf) | 2800 if (c->chrPixBuf) |
2802 { | 2801 { |
2803 for(i=0; i<c->vChrBufSize; i++) | 2802 for (i=0; i<c->vChrBufSize; i++) |
2804 { | 2803 { |
2805 av_free(c->chrPixBuf[i]); | 2804 av_free(c->chrPixBuf[i]); |
2806 c->chrPixBuf[i]=NULL; | 2805 c->chrPixBuf[i]=NULL; |
2807 } | 2806 } |
2808 av_free(c->chrPixBuf); | 2807 av_free(c->chrPixBuf); |
2809 c->chrPixBuf=NULL; | 2808 c->chrPixBuf=NULL; |
2810 } | 2809 } |
2811 | 2810 |
2812 av_free(c->vLumFilter); | 2811 av_free(c->vLumFilter); |
2813 c->vLumFilter = NULL; | 2812 c->vLumFilter = NULL; |
2814 av_free(c->vChrFilter); | 2813 av_free(c->vChrFilter); |
2815 c->vChrFilter = NULL; | 2814 c->vChrFilter = NULL; |
2816 av_free(c->hLumFilter); | 2815 av_free(c->hLumFilter); |
2817 c->hLumFilter = NULL; | 2816 c->hLumFilter = NULL; |
2818 av_free(c->hChrFilter); | 2817 av_free(c->hChrFilter); |
2819 c->hChrFilter = NULL; | 2818 c->hChrFilter = NULL; |
2820 #ifdef HAVE_ALTIVEC | 2819 #ifdef HAVE_ALTIVEC |
2821 av_free(c->vYCoeffsBank); | 2820 av_free(c->vYCoeffsBank); |
2822 c->vYCoeffsBank = NULL; | 2821 c->vYCoeffsBank = NULL; |
2823 av_free(c->vCCoeffsBank); | 2822 av_free(c->vCCoeffsBank); |
2824 c->vCCoeffsBank = NULL; | 2823 c->vCCoeffsBank = NULL; |
2825 #endif | 2824 #endif |
2826 | 2825 |
2827 av_free(c->vLumFilterPos); | 2826 av_free(c->vLumFilterPos); |
2828 c->vLumFilterPos = NULL; | 2827 c->vLumFilterPos = NULL; |
2829 av_free(c->vChrFilterPos); | 2828 av_free(c->vChrFilterPos); |
2830 c->vChrFilterPos = NULL; | 2829 c->vChrFilterPos = NULL; |
2831 av_free(c->hLumFilterPos); | 2830 av_free(c->hLumFilterPos); |
2832 c->hLumFilterPos = NULL; | 2831 c->hLumFilterPos = NULL; |
2833 av_free(c->hChrFilterPos); | 2832 av_free(c->hChrFilterPos); |
2834 c->hChrFilterPos = NULL; | 2833 c->hChrFilterPos = NULL; |
2835 | 2834 |
2836 #if defined(ARCH_X86) && defined(CONFIG_GPL) | 2835 #if defined(ARCH_X86) && defined(CONFIG_GPL) |
2837 #ifdef MAP_ANONYMOUS | 2836 #ifdef MAP_ANONYMOUS |
2838 if(c->funnyYCode) munmap(c->funnyYCode, MAX_FUNNY_CODE_SIZE); | 2837 if (c->funnyYCode) munmap(c->funnyYCode, MAX_FUNNY_CODE_SIZE); |
2839 if(c->funnyUVCode) munmap(c->funnyUVCode, MAX_FUNNY_CODE_SIZE); | 2838 if (c->funnyUVCode) munmap(c->funnyUVCode, MAX_FUNNY_CODE_SIZE); |
2840 #else | 2839 #else |
2841 av_free(c->funnyYCode); | 2840 av_free(c->funnyYCode); |
2842 av_free(c->funnyUVCode); | 2841 av_free(c->funnyUVCode); |
2843 #endif | 2842 #endif |
2844 c->funnyYCode=NULL; | 2843 c->funnyYCode=NULL; |
2845 c->funnyUVCode=NULL; | 2844 c->funnyUVCode=NULL; |
2846 #endif /* defined(ARCH_X86) */ | 2845 #endif /* defined(ARCH_X86) */ |
2847 | 2846 |
2848 av_free(c->lumMmx2Filter); | 2847 av_free(c->lumMmx2Filter); |
2849 c->lumMmx2Filter=NULL; | 2848 c->lumMmx2Filter=NULL; |
2850 av_free(c->chrMmx2Filter); | 2849 av_free(c->chrMmx2Filter); |
2851 c->chrMmx2Filter=NULL; | 2850 c->chrMmx2Filter=NULL; |
2852 av_free(c->lumMmx2FilterPos); | 2851 av_free(c->lumMmx2FilterPos); |
2853 c->lumMmx2FilterPos=NULL; | 2852 c->lumMmx2FilterPos=NULL; |
2854 av_free(c->chrMmx2FilterPos); | 2853 av_free(c->chrMmx2FilterPos); |
2855 c->chrMmx2FilterPos=NULL; | 2854 c->chrMmx2FilterPos=NULL; |
2856 av_free(c->yuvTable); | 2855 av_free(c->yuvTable); |
2857 c->yuvTable=NULL; | 2856 c->yuvTable=NULL; |
2858 | 2857 |
2859 av_free(c); | 2858 av_free(c); |
2860 } | 2859 } |
2861 | 2860 |
2862 /** | 2861 /** |
2863 * Checks if context is valid or reallocs a new one instead. | 2862 * Checks if context is valid or reallocs a new one instead. |
2864 * If context is NULL, just calls sws_getContext() to get a new one. | 2863 * If context is NULL, just calls sws_getContext() to get a new one. |
2868 * | 2867 * |
2869 * Be warned that srcFilter, dstFilter are not checked, they are | 2868 * Be warned that srcFilter, dstFilter are not checked, they are |
2870 * asumed to remain valid. | 2869 * asumed to remain valid. |
2871 */ | 2870 */ |
2872 struct SwsContext *sws_getCachedContext(struct SwsContext *context, | 2871 struct SwsContext *sws_getCachedContext(struct SwsContext *context, |
2873 int srcW, int srcH, int srcFormat, | 2872 int srcW, int srcH, int srcFormat, |
2874 int dstW, int dstH, int dstFormat, int flags, | 2873 int dstW, int dstH, int dstFormat, int flags, |
2875 SwsFilter *srcFilter, SwsFilter *dstFilter, double *param) | 2874 SwsFilter *srcFilter, SwsFilter *dstFilter, double *param) |
2876 { | 2875 { |
2877 if (context != NULL) { | 2876 if (context != NULL) { |
2878 if ((context->srcW != srcW) || (context->srcH != srcH) || | 2877 if ((context->srcW != srcW) || (context->srcH != srcH) || |
2879 (context->srcFormat != srcFormat) || | 2878 (context->srcFormat != srcFormat) || |
2880 (context->dstW != dstW) || (context->dstH != dstH) || | 2879 (context->dstW != dstW) || (context->dstH != dstH) || |
2885 context = NULL; | 2884 context = NULL; |
2886 } | 2885 } |
2887 } | 2886 } |
2888 if (context == NULL) { | 2887 if (context == NULL) { |
2889 return sws_getContext(srcW, srcH, srcFormat, | 2888 return sws_getContext(srcW, srcH, srcFormat, |
2890 dstW, dstH, dstFormat, flags, | 2889 dstW, dstH, dstFormat, flags, |
2891 srcFilter, dstFilter, param); | 2890 srcFilter, dstFilter, param); |
2892 } | 2891 } |
2893 return context; | 2892 return context; |
2894 } | 2893 } |
2895 | 2894 |