#if 0 //64*cubic
static const uint8_t obmc32[1024]={
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,
- 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 2, 2, 1, 1, 1, 1, 0, 0, 0,
- 0, 0, 1, 1, 2, 2, 3, 4, 4, 5, 6, 6, 7, 7, 8, 8, 8, 8, 7, 7, 6, 6, 5, 4, 4, 3, 2, 2, 1, 1, 0, 0,
- 0, 0, 1, 2, 2, 3, 4, 6, 7, 8, 9,10,11,12,12,12,12,12,12,11,10, 9, 8, 7, 6, 4, 3, 2, 2, 1, 0, 0,
- 0, 1, 1, 2, 3, 5, 6, 8,10,11,13,14,15,16,17,18,18,17,16,15,14,13,11,10, 8, 6, 5, 3, 2, 1, 1, 0,
- 0, 1, 1, 3, 4, 6, 8,10,13,15,17,19,20,22,22,23,23,22,22,20,19,17,15,13,10, 8, 6, 4, 3, 1, 1, 0,
- 0, 1, 2, 4, 6, 8,10,13,16,19,21,23,25,27,28,29,29,28,27,25,23,21,19,16,13,10, 8, 6, 4, 2, 1, 0,
- 0, 1, 2, 4, 7,10,13,16,19,22,25,28,31,33,34,35,35,34,33,31,28,25,22,19,16,13,10, 7, 4, 2, 1, 0,
- 0, 1, 3, 5, 8,11,15,19,22,26,30,33,36,38,40,41,41,40,38,36,33,30,26,22,19,15,11, 8, 5, 3, 1, 0,
- 0, 1, 3, 6, 9,12,17,21,25,30,34,38,41,44,45,46,46,45,44,41,38,34,30,25,21,17,12, 9, 6, 3, 1, 0,
- 0, 1, 3, 6,10,14,19,23,28,33,38,42,45,48,51,52,52,51,48,45,42,38,33,28,23,19,14,10, 6, 3, 1, 0,
- 0, 1, 4, 7,11,15,20,25,31,36,41,45,49,52,55,56,56,55,52,49,45,41,36,31,25,20,15,11, 7, 4, 1, 0,
- 0, 2, 4, 7,12,16,22,27,33,38,44,48,52,56,58,60,60,58,56,52,48,44,38,33,27,22,16,12, 7, 4, 2, 0,
- 0, 1, 4, 8,12,17,22,28,34,40,45,51,55,58,61,62,62,61,58,55,51,45,40,34,28,22,17,12, 8, 4, 1, 0,
- 0, 2, 4, 8,12,18,23,29,35,41,46,52,56,60,62,64,64,62,60,56,52,46,41,35,29,23,18,12, 8, 4, 2, 0,
- 0, 2, 4, 8,12,18,23,29,35,41,46,52,56,60,62,64,64,62,60,56,52,46,41,35,29,23,18,12, 8, 4, 2, 0,
- 0, 1, 4, 8,12,17,22,28,34,40,45,51,55,58,61,62,62,61,58,55,51,45,40,34,28,22,17,12, 8, 4, 1, 0,
- 0, 2, 4, 7,12,16,22,27,33,38,44,48,52,56,58,60,60,58,56,52,48,44,38,33,27,22,16,12, 7, 4, 2, 0,
- 0, 1, 4, 7,11,15,20,25,31,36,41,45,49,52,55,56,56,55,52,49,45,41,36,31,25,20,15,11, 7, 4, 1, 0,
- 0, 1, 3, 6,10,14,19,23,28,33,38,42,45,48,51,52,52,51,48,45,42,38,33,28,23,19,14,10, 6, 3, 1, 0,
- 0, 1, 3, 6, 9,12,17,21,25,30,34,38,41,44,45,46,46,45,44,41,38,34,30,25,21,17,12, 9, 6, 3, 1, 0,
- 0, 1, 3, 5, 8,11,15,19,22,26,30,33,36,38,40,41,41,40,38,36,33,30,26,22,19,15,11, 8, 5, 3, 1, 0,
- 0, 1, 2, 4, 7,10,13,16,19,22,25,28,31,33,34,35,35,34,33,31,28,25,22,19,16,13,10, 7, 4, 2, 1, 0,
- 0, 1, 2, 4, 6, 8,10,13,16,19,21,23,25,27,28,29,29,28,27,25,23,21,19,16,13,10, 8, 6, 4, 2, 1, 0,
- 0, 1, 1, 3, 4, 6, 8,10,13,15,17,19,20,22,22,23,23,22,22,20,19,17,15,13,10, 8, 6, 4, 3, 1, 1, 0,
- 0, 1, 1, 2, 3, 5, 6, 8,10,11,13,14,15,16,17,18,18,17,16,15,14,13,11,10, 8, 6, 5, 3, 2, 1, 1, 0,
- 0, 0, 1, 2, 2, 3, 4, 6, 7, 8, 9,10,11,12,12,12,12,12,12,11,10, 9, 8, 7, 6, 4, 3, 2, 2, 1, 0, 0,
- 0, 0, 1, 1, 2, 2, 3, 4, 4, 5, 6, 6, 7, 7, 8, 8, 8, 8, 7, 7, 6, 6, 5, 4, 4, 3, 2, 2, 1, 1, 0, 0,
- 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 2, 2, 1, 1, 1, 1, 0, 0, 0,
- 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 4, 4, 4, 4, 4, 4, 4, 4, 0, 0, 0, 0, 0,
+ 0, 0, 0, 4, 4, 4, 4, 8, 8, 12, 12, 12, 16, 16, 16, 16, 16, 16, 16, 16, 12, 12, 12, 8, 8, 4, 4, 4, 4, 0, 0, 0,
+ 0, 0, 4, 4, 8, 8, 12, 16, 16, 20, 24, 24, 28, 28, 32, 32, 32, 32, 28, 28, 24, 24, 20, 16, 16, 12, 8, 8, 4, 4, 0, 0,
+ 0, 0, 4, 8, 8, 12, 16, 24, 28, 32, 36, 40, 44, 48, 48, 48, 48, 48, 48, 44, 40, 36, 32, 28, 24, 16, 12, 8, 8, 4, 0, 0,
+ 0, 4, 4, 8, 12, 20, 24, 32, 40, 44, 52, 56, 60, 64, 68, 72, 72, 68, 64, 60, 56, 52, 44, 40, 32, 24, 20, 12, 8, 4, 4, 0,
+ 0, 4, 4, 12, 16, 24, 32, 40, 52, 60, 68, 76, 80, 88, 88, 92, 92, 88, 88, 80, 76, 68, 60, 52, 40, 32, 24, 16, 12, 4, 4, 0,
+ 0, 4, 8, 16, 24, 32, 40, 52, 64, 76, 84, 92,100,108,112,116,116,112,108,100, 92, 84, 76, 64, 52, 40, 32, 24, 16, 8, 4, 0,
+ 0, 4, 8, 16, 28, 40, 52, 64, 76, 88,100,112,124,132,136,140,140,136,132,124,112,100, 88, 76, 64, 52, 40, 28, 16, 8, 4, 0,
+ 0, 4, 12, 20, 32, 44, 60, 76, 88,104,120,132,144,152,160,164,164,160,152,144,132,120,104, 88, 76, 60, 44, 32, 20, 12, 4, 0,
+ 0, 4, 12, 24, 36, 48, 68, 84,100,120,136,152,164,176,180,184,184,180,176,164,152,136,120,100, 84, 68, 48, 36, 24, 12, 4, 0,
+ 0, 4, 12, 24, 40, 56, 76, 92,112,132,152,168,180,192,204,208,208,204,192,180,168,152,132,112, 92, 76, 56, 40, 24, 12, 4, 0,
+ 0, 4, 16, 28, 44, 60, 80,100,124,144,164,180,196,208,220,224,224,220,208,196,180,164,144,124,100, 80, 60, 44, 28, 16, 4, 0,
+ 0, 8, 16, 28, 48, 64, 88,108,132,152,176,192,208,224,232,240,240,232,224,208,192,176,152,132,108, 88, 64, 48, 28, 16, 8, 0,
+ 0, 4, 16, 32, 48, 68, 88,112,136,160,180,204,220,232,244,248,248,244,232,220,204,180,160,136,112, 88, 68, 48, 32, 16, 4, 0,
+ 1, 8, 16, 32, 48, 72, 92,116,140,164,184,208,224,240,248,255,255,248,240,224,208,184,164,140,116, 92, 72, 48, 32, 16, 8, 1,
+ 1, 8, 16, 32, 48, 72, 92,116,140,164,184,208,224,240,248,255,255,248,240,224,208,184,164,140,116, 92, 72, 48, 32, 16, 8, 1,
+ 0, 4, 16, 32, 48, 68, 88,112,136,160,180,204,220,232,244,248,248,244,232,220,204,180,160,136,112, 88, 68, 48, 32, 16, 4, 0,
+ 0, 8, 16, 28, 48, 64, 88,108,132,152,176,192,208,224,232,240,240,232,224,208,192,176,152,132,108, 88, 64, 48, 28, 16, 8, 0,
+ 0, 4, 16, 28, 44, 60, 80,100,124,144,164,180,196,208,220,224,224,220,208,196,180,164,144,124,100, 80, 60, 44, 28, 16, 4, 0,
+ 0, 4, 12, 24, 40, 56, 76, 92,112,132,152,168,180,192,204,208,208,204,192,180,168,152,132,112, 92, 76, 56, 40, 24, 12, 4, 0,
+ 0, 4, 12, 24, 36, 48, 68, 84,100,120,136,152,164,176,180,184,184,180,176,164,152,136,120,100, 84, 68, 48, 36, 24, 12, 4, 0,
+ 0, 4, 12, 20, 32, 44, 60, 76, 88,104,120,132,144,152,160,164,164,160,152,144,132,120,104, 88, 76, 60, 44, 32, 20, 12, 4, 0,
+ 0, 4, 8, 16, 28, 40, 52, 64, 76, 88,100,112,124,132,136,140,140,136,132,124,112,100, 88, 76, 64, 52, 40, 28, 16, 8, 4, 0,
+ 0, 4, 8, 16, 24, 32, 40, 52, 64, 76, 84, 92,100,108,112,116,116,112,108,100, 92, 84, 76, 64, 52, 40, 32, 24, 16, 8, 4, 0,
+ 0, 4, 4, 12, 16, 24, 32, 40, 52, 60, 68, 76, 80, 88, 88, 92, 92, 88, 88, 80, 76, 68, 60, 52, 40, 32, 24, 16, 12, 4, 4, 0,
+ 0, 4, 4, 8, 12, 20, 24, 32, 40, 44, 52, 56, 60, 64, 68, 72, 72, 68, 64, 60, 56, 52, 44, 40, 32, 24, 20, 12, 8, 4, 4, 0,
+ 0, 0, 4, 8, 8, 12, 16, 24, 28, 32, 36, 40, 44, 48, 48, 48, 48, 48, 48, 44, 40, 36, 32, 28, 24, 16, 12, 8, 8, 4, 0, 0,
+ 0, 0, 4, 4, 8, 8, 12, 16, 16, 20, 24, 24, 28, 28, 32, 32, 32, 32, 28, 28, 24, 24, 20, 16, 16, 12, 8, 8, 4, 4, 0, 0,
+ 0, 0, 0, 4, 4, 4, 4, 8, 8, 12, 12, 12, 16, 16, 16, 16, 16, 16, 16, 16, 12, 12, 12, 8, 8, 4, 4, 4, 4, 0, 0, 0,
+ 0, 0, 0, 0, 0, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 4, 4, 4, 4, 4, 4, 4, 4, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
//error:0.000022
};
static const uint8_t obmc16[256]={
- 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,
- 0, 1, 1, 2, 4, 5, 5, 6, 6, 5, 5, 4, 2, 1, 1, 0,
- 0, 1, 4, 6, 9,11,13,15,15,13,11, 9, 6, 4, 1, 0,
- 0, 2, 6,11,15,20,24,26,26,24,20,15,11, 6, 2, 0,
- 0, 4, 9,15,23,29,34,38,38,34,29,23,15, 9, 4, 0,
- 0, 5,11,20,29,38,45,49,49,45,38,29,20,11, 5, 0,
- 1, 5,13,24,34,45,53,57,57,53,45,34,24,13, 5, 1,
- 1, 6,15,26,38,49,57,62,62,57,49,38,26,15, 6, 1,
- 1, 6,15,26,38,49,57,62,62,57,49,38,26,15, 6, 1,
- 1, 5,13,24,34,45,53,57,57,53,45,34,24,13, 5, 1,
- 0, 5,11,20,29,38,45,49,49,45,38,29,20,11, 5, 0,
- 0, 4, 9,15,23,29,34,38,38,34,29,23,15, 9, 4, 0,
- 0, 2, 6,11,15,20,24,26,26,24,20,15,11, 6, 2, 0,
- 0, 1, 4, 6, 9,11,13,15,15,13,11, 9, 6, 4, 1, 0,
- 0, 1, 1, 2, 4, 5, 5, 6, 6, 5, 5, 4, 2, 1, 1, 0,
- 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0,
+ 0, 4, 4, 8, 16, 20, 20, 24, 24, 20, 20, 16, 8, 4, 4, 0,
+ 0, 4, 16, 24, 36, 44, 52, 60, 60, 52, 44, 36, 24, 16, 4, 0,
+ 0, 8, 24, 44, 60, 80, 96,104,104, 96, 80, 60, 44, 24, 8, 0,
+ 0, 16, 36, 60, 92,116,136,152,152,136,116, 92, 60, 36, 16, 0,
+ 0, 20, 44, 80,116,152,180,196,196,180,152,116, 80, 44, 20, 0,
+ 4, 20, 52, 96,136,180,212,228,228,212,180,136, 96, 52, 20, 4,
+ 4, 24, 60,104,152,196,228,248,248,228,196,152,104, 60, 24, 4,
+ 4, 24, 60,104,152,196,228,248,248,228,196,152,104, 60, 24, 4,
+ 4, 20, 52, 96,136,180,212,228,228,212,180,136, 96, 52, 20, 4,
+ 0, 20, 44, 80,116,152,180,196,196,180,152,116, 80, 44, 20, 0,
+ 0, 16, 36, 60, 92,116,136,152,152,136,116, 92, 60, 36, 16, 0,
+ 0, 8, 24, 44, 60, 80, 96,104,104, 96, 80, 60, 44, 24, 8, 0,
+ 0, 4, 16, 24, 36, 44, 52, 60, 60, 52, 44, 36, 24, 16, 4, 0,
+ 0, 4, 4, 8, 16, 20, 20, 24, 24, 20, 20, 16, 8, 4, 4, 0,
+ 0, 0, 0, 0, 0, 0, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0,
//error:0.000033
};
#elif 1 // 64*linear
};
#else //64*cos
static const uint8_t obmc32[1024]={
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 2, 2, 1, 1, 1, 1, 0, 0, 0,
- 0, 0, 1, 1, 1, 2, 2, 3, 4, 5, 5, 6, 7, 7, 7, 7, 7, 7, 7, 7, 6, 5, 5, 4, 3, 2, 2, 1, 1, 1, 0, 0,
- 0, 0, 1, 1, 2, 3, 4, 5, 6, 7, 9,10,11,11,12,12,12,12,11,11,10, 9, 7, 6, 5, 4, 3, 2, 1, 1, 0, 0,
- 0, 0, 1, 2, 3, 5, 6, 8, 9,11,12,14,15,16,17,17,17,17,16,15,14,12,11, 9, 8, 6, 5, 3, 2, 1, 0, 0,
- 0, 1, 1, 2, 4, 6, 8,10,12,15,17,19,20,21,22,23,23,22,21,20,19,17,15,12,10, 8, 6, 4, 2, 1, 1, 0,
- 0, 1, 2, 3, 5, 8,10,13,16,19,21,24,26,27,28,29,29,28,27,26,24,21,19,16,13,10, 8, 5, 3, 2, 1, 0,
- 0, 1, 2, 4, 6, 9,12,16,19,23,26,29,31,33,34,35,35,34,33,31,29,26,23,19,16,12, 9, 6, 4, 2, 1, 0,
- 0, 1, 3, 5, 7,11,15,19,23,26,30,34,37,39,40,41,41,40,39,37,34,30,26,23,19,15,11, 7, 5, 3, 1, 0,
- 0, 1, 3, 5, 9,12,17,21,26,30,35,38,42,44,46,47,47,46,44,42,38,35,30,26,21,17,12, 9, 5, 3, 1, 0,
- 0, 1, 3, 6, 9,14,19,24,29,34,38,43,46,49,51,52,52,51,49,46,43,38,34,29,24,19,14, 9, 6, 3, 1, 0,
- 0, 1, 3, 6,11,15,20,26,31,37,42,46,50,53,56,57,57,56,53,50,46,42,37,31,26,20,15,11, 6, 3, 1, 0,
- 0, 1, 3, 7,11,16,21,27,33,39,44,49,53,57,59,60,60,59,57,53,49,44,39,33,27,21,16,11, 7, 3, 1, 0,
- 0, 1, 4, 7,12,17,22,28,34,40,46,51,56,59,61,63,63,61,59,56,51,46,40,34,28,22,17,12, 7, 4, 1, 0,
- 0, 1, 4, 7,12,17,23,29,35,41,47,52,57,60,63,64,64,63,60,57,52,47,41,35,29,23,17,12, 7, 4, 1, 0,
- 0, 1, 4, 7,12,17,23,29,35,41,47,52,57,60,63,64,64,63,60,57,52,47,41,35,29,23,17,12, 7, 4, 1, 0,
- 0, 1, 4, 7,12,17,22,28,34,40,46,51,56,59,61,63,63,61,59,56,51,46,40,34,28,22,17,12, 7, 4, 1, 0,
- 0, 1, 3, 7,11,16,21,27,33,39,44,49,53,57,59,60,60,59,57,53,49,44,39,33,27,21,16,11, 7, 3, 1, 0,
- 0, 1, 3, 6,11,15,20,26,31,37,42,46,50,53,56,57,57,56,53,50,46,42,37,31,26,20,15,11, 6, 3, 1, 0,
- 0, 1, 3, 6, 9,14,19,24,29,34,38,43,46,49,51,52,52,51,49,46,43,38,34,29,24,19,14, 9, 6, 3, 1, 0,
- 0, 1, 3, 5, 9,12,17,21,26,30,35,38,42,44,46,47,47,46,44,42,38,35,30,26,21,17,12, 9, 5, 3, 1, 0,
- 0, 1, 3, 5, 7,11,15,19,23,26,30,34,37,39,40,41,41,40,39,37,34,30,26,23,19,15,11, 7, 5, 3, 1, 0,
- 0, 1, 2, 4, 6, 9,12,16,19,23,26,29,31,33,34,35,35,34,33,31,29,26,23,19,16,12, 9, 6, 4, 2, 1, 0,
- 0, 1, 2, 3, 5, 8,10,13,16,19,21,24,26,27,28,29,29,28,27,26,24,21,19,16,13,10, 8, 5, 3, 2, 1, 0,
- 0, 1, 1, 2, 4, 6, 8,10,12,15,17,19,20,21,22,23,23,22,21,20,19,17,15,12,10, 8, 6, 4, 2, 1, 1, 0,
- 0, 0, 1, 2, 3, 5, 6, 8, 9,11,12,14,15,16,17,17,17,17,16,15,14,12,11, 9, 8, 6, 5, 3, 2, 1, 0, 0,
- 0, 0, 1, 1, 2, 3, 4, 5, 6, 7, 9,10,11,11,12,12,12,12,11,11,10, 9, 7, 6, 5, 4, 3, 2, 1, 1, 0, 0,
- 0, 0, 1, 1, 1, 2, 2, 3, 4, 5, 5, 6, 7, 7, 7, 7, 7, 7, 7, 7, 6, 5, 5, 4, 3, 2, 2, 1, 1, 1, 0, 0,
- 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 2, 2, 1, 1, 1, 1, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 4, 4, 4, 4, 4, 4, 4, 4, 8, 4, 4, 8, 4, 4, 4, 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 4, 4, 4, 4, 8, 8, 12, 12, 12, 12, 16, 16, 16, 16, 16, 16, 12, 12, 12, 12, 8, 8, 4, 4, 4, 4, 0, 0, 0,
+ 0, 0, 4, 4, 4, 8, 8, 12, 16, 20, 20, 24, 28, 28, 28, 28, 28, 28, 28, 28, 24, 20, 20, 16, 12, 8, 8, 4, 4, 4, 0, 0,
+ 0, 0, 4, 4, 8, 12, 16, 20, 24, 28, 36, 40, 44, 44, 48, 48, 48, 48, 44, 44, 40, 36, 28, 24, 20, 16, 12, 8, 4, 4, 0, 0,
+ 0, 0, 4, 8, 12, 20, 24, 32, 36, 44, 48, 56, 60, 64, 68, 68, 68, 68, 64, 60, 56, 48, 44, 36, 32, 24, 20, 12, 8, 4, 0, 0,
+ 0, 4, 4, 8, 16, 24, 32, 40, 48, 60, 68, 76, 80, 84, 88, 92, 92, 88, 84, 80, 76, 68, 60, 48, 40, 32, 24, 16, 8, 4, 4, 0,
+ 0, 4, 8, 12, 20, 32, 40, 52, 64, 76, 84, 96,104,108,112,116,116,112,108,104, 96, 84, 76, 64, 52, 40, 32, 20, 12, 8, 4, 0,
+ 0, 4, 8, 16, 24, 36, 48, 64, 76, 92,104,116,124,132,136,140,140,136,132,124,116,104, 92, 76, 64, 48, 36, 24, 16, 8, 4, 0,
+ 0, 4, 12, 20, 28, 44, 60, 76, 92,104,120,136,148,156,160,164,164,160,156,148,136,120,104, 92, 76, 60, 44, 28, 20, 12, 4, 0,
+ 0, 4, 12, 20, 36, 48, 68, 84,104,120,140,152,168,176,184,188,188,184,176,168,152,140,120,104, 84, 68, 48, 36, 20, 12, 4, 0,
+ 0, 4, 12, 24, 36, 56, 76, 96,116,136,152,172,184,196,204,208,208,204,196,184,172,152,136,116, 96, 76, 56, 36, 24, 12, 4, 0,
+ 0, 4, 12, 24, 44, 60, 80,104,124,148,168,184,200,212,224,228,228,224,212,200,184,168,148,124,104, 80, 60, 44, 24, 12, 4, 0,
+ 0, 4, 12, 28, 44, 64, 84,108,132,156,176,196,212,228,236,240,240,236,228,212,196,176,156,132,108, 84, 64, 44, 28, 12, 4, 0,
+ 0, 4, 16, 28, 48, 68, 88,112,136,160,184,204,224,236,244,252,252,244,236,224,204,184,160,136,112, 88, 68, 48, 28, 16, 4, 0,
+ 1, 4, 16, 28, 48, 68, 92,116,140,164,188,208,228,240,252,255,255,252,240,228,208,188,164,140,116, 92, 68, 48, 28, 16, 4, 1,
+ 1, 4, 16, 28, 48, 68, 92,116,140,164,188,208,228,240,252,255,255,252,240,228,208,188,164,140,116, 92, 68, 48, 28, 16, 4, 1,
+ 0, 4, 16, 28, 48, 68, 88,112,136,160,184,204,224,236,244,252,252,244,236,224,204,184,160,136,112, 88, 68, 48, 28, 16, 4, 0,
+ 0, 4, 12, 28, 44, 64, 84,108,132,156,176,196,212,228,236,240,240,236,228,212,196,176,156,132,108, 84, 64, 44, 28, 12, 4, 0,
+ 0, 4, 12, 24, 44, 60, 80,104,124,148,168,184,200,212,224,228,228,224,212,200,184,168,148,124,104, 80, 60, 44, 24, 12, 4, 0,
+ 0, 4, 12, 24, 36, 56, 76, 96,116,136,152,172,184,196,204,208,208,204,196,184,172,152,136,116, 96, 76, 56, 36, 24, 12, 4, 0,
+ 0, 4, 12, 20, 36, 48, 68, 84,104,120,140,152,168,176,184,188,188,184,176,168,152,140,120,104, 84, 68, 48, 36, 20, 12, 4, 0,
+ 0, 4, 12, 20, 28, 44, 60, 76, 92,104,120,136,148,156,160,164,164,160,156,148,136,120,104, 92, 76, 60, 44, 28, 20, 12, 4, 0,
+ 0, 4, 8, 16, 24, 36, 48, 64, 76, 92,104,116,124,132,136,140,140,136,132,124,116,104, 92, 76, 64, 48, 36, 24, 16, 8, 4, 0,
+ 0, 4, 8, 12, 20, 32, 40, 52, 64, 76, 84, 96,104,108,112,116,116,112,108,104, 96, 84, 76, 64, 52, 40, 32, 20, 12, 8, 4, 0,
+ 0, 4, 4, 8, 16, 24, 32, 40, 48, 60, 68, 76, 80, 84, 88, 92, 92, 88, 84, 80, 76, 68, 60, 48, 40, 32, 24, 16, 8, 4, 4, 0,
+ 0, 0, 4, 8, 12, 20, 24, 32, 36, 44, 48, 56, 60, 64, 68, 68, 68, 68, 64, 60, 56, 48, 44, 36, 32, 24, 20, 12, 8, 4, 0, 0,
+ 0, 0, 4, 4, 8, 12, 16, 20, 24, 28, 36, 40, 44, 44, 48, 48, 48, 48, 44, 44, 40, 36, 28, 24, 20, 16, 12, 8, 4, 4, 0, 0,
+ 0, 0, 4, 4, 4, 8, 8, 12, 16, 20, 20, 24, 28, 28, 28, 28, 28, 28, 28, 28, 24, 20, 20, 16, 12, 8, 8, 4, 4, 4, 0, 0,
+ 0, 0, 0, 4, 4, 4, 4, 8, 8, 12, 12, 12, 12, 16, 16, 16, 16, 16, 16, 12, 12, 12, 12, 8, 8, 4, 4, 4, 4, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 4, 4, 4, 4, 4, 4, 4, 4, 8, 4, 4, 8, 4, 4, 4, 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
//error:0.000022
};
static const uint8_t obmc16[256]={
- 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,
- 0, 0, 1, 2, 3, 4, 5, 5, 5, 5, 4, 3, 2, 1, 0, 0,
- 0, 1, 3, 6, 8,11,13,14,14,13,11, 8, 6, 3, 1, 0,
- 0, 2, 6,10,15,20,24,26,26,24,20,15,10, 6, 2, 0,
- 0, 3, 8,16,23,30,35,38,38,35,30,23,16, 8, 3, 0,
- 1, 4,11,20,30,39,46,49,49,46,39,30,20,11, 4, 1,
- 1, 5,13,24,35,46,54,58,58,54,46,35,24,13, 5, 1,
- 0, 5,14,26,38,49,58,63,63,58,49,38,26,14, 5, 0,
- 0, 5,14,26,38,49,58,63,63,58,49,38,26,14, 5, 0,
- 1, 5,13,24,35,46,54,58,58,54,46,35,24,13, 5, 1,
- 1, 4,11,20,30,39,46,49,49,46,39,30,20,11, 4, 1,
- 0, 3, 8,16,23,30,35,38,38,35,30,23,16, 8, 3, 0,
- 0, 2, 6,10,15,20,24,26,26,24,20,15,10, 6, 2, 0,
- 0, 1, 3, 6, 8,11,13,14,14,13,11, 8, 6, 3, 1, 0,
- 0, 0, 1, 2, 3, 4, 5, 5, 5, 5, 4, 3, 2, 1, 0, 0,
- 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 4, 4, 4, 4, 4, 4, 0, 0, 0, 0, 0,
+ 0, 0, 4, 8, 12, 16, 20, 20, 20, 20, 16, 12, 8, 4, 0, 0,
+ 0, 4, 12, 24, 32, 44, 52, 56, 56, 52, 44, 32, 24, 12, 4, 0,
+ 0, 8, 24, 40, 60, 80, 96,104,104, 96, 80, 60, 40, 24, 8, 0,
+ 0, 12, 32, 64, 92,120,140,152,152,140,120, 92, 64, 32, 12, 0,
+ 4, 16, 44, 80,120,156,184,196,196,184,156,120, 80, 44, 16, 4,
+ 4, 20, 52, 96,140,184,216,232,232,216,184,140, 96, 52, 20, 4,
+ 0, 20, 56,104,152,196,232,252,252,232,196,152,104, 56, 20, 0,
+ 0, 20, 56,104,152,196,232,252,252,232,196,152,104, 56, 20, 0,
+ 4, 20, 52, 96,140,184,216,232,232,216,184,140, 96, 52, 20, 4,
+ 4, 16, 44, 80,120,156,184,196,196,184,156,120, 80, 44, 16, 4,
+ 0, 12, 32, 64, 92,120,140,152,152,140,120, 92, 64, 32, 12, 0,
+ 0, 8, 24, 40, 60, 80, 96,104,104, 96, 80, 60, 40, 24, 8, 0,
+ 0, 4, 12, 24, 32, 44, 52, 56, 56, 52, 44, 32, 24, 12, 4, 0,
+ 0, 0, 4, 8, 12, 16, 20, 20, 20, 20, 16, 12, 8, 4, 0, 0,
+ 0, 0, 0, 0, 0, 4, 4, 4, 4, 4, 4, 0, 0, 0, 0, 0,
//error:0.000022
};
#endif
#define LOG2_MB_SIZE 4
#define MB_SIZE (1<<LOG2_MB_SIZE)
+#define ENCODER_EXTRA_BITS 4
+#define HTAPS_MAX 8
typedef struct x_and_coeff{
int16_t x;
int height;
int qlog; ///< log(qscale)/log[2^(1/6)]
DWTELEM *buf;
+ IDWTELEM *ibuf;
int buf_x_offset;
int buf_y_offset;
int stride_line; ///< Stride measured in lines, not pixels.
int width;
int height;
SubBand band[MAX_DECOMPOSITIONS][4];
+
+ int htaps;
+ int8_t hcoeff[HTAPS_MAX/2];
+ int diag_mc;
+ int fast_mc;
+
+ int last_htaps;
+ int8_t last_hcoeff[HTAPS_MAX/2];
+ int last_diag_mc;
}Plane;
typedef struct SnowContext{
AVFrame input_picture; ///< new_picture with the internal linesizes
AVFrame current_picture;
AVFrame last_picture[MAX_REF_FRAMES];
+ uint8_t *halfpel_plane[MAX_REF_FRAMES][4][4];
AVFrame mconly_picture;
// uint8_t q_context[16];
uint8_t header_state[32];
int last_spatial_decomposition_type;
int temporal_decomposition_type;
int spatial_decomposition_count;
+ int last_spatial_decomposition_count;
int temporal_decomposition_count;
int max_ref_frames;
int ref_frames;
int16_t (*ref_mvs[MAX_REF_FRAMES])[2];
uint32_t *ref_scores[MAX_REF_FRAMES];
DWTELEM *spatial_dwt_buffer;
+ IDWTELEM *spatial_idwt_buffer;
int colorspace_type;
int chroma_h_shift;
int chroma_v_shift;
}SnowContext;
typedef struct {
- DWTELEM *b0;
- DWTELEM *b1;
- DWTELEM *b2;
- DWTELEM *b3;
+ IDWTELEM *b0;
+ IDWTELEM *b1;
+ IDWTELEM *b2;
+ IDWTELEM *b3;
int y;
} dwt_compose_t;
static void iterative_me(SnowContext *s);
-static void slice_buffer_init(slice_buffer * buf, int line_count, int max_allocated_lines, int line_width, DWTELEM * base_buffer)
+static void slice_buffer_init(slice_buffer * buf, int line_count, int max_allocated_lines, int line_width, IDWTELEM * base_buffer)
{
int i;
buf->line_count = line_count;
buf->line_width = line_width;
buf->data_count = max_allocated_lines;
- buf->line = (DWTELEM * *) av_mallocz (sizeof(DWTELEM *) * line_count);
- buf->data_stack = (DWTELEM * *) av_malloc (sizeof(DWTELEM *) * max_allocated_lines);
+ buf->line = av_mallocz (sizeof(IDWTELEM *) * line_count);
+ buf->data_stack = av_malloc (sizeof(IDWTELEM *) * max_allocated_lines);
for (i = 0; i < max_allocated_lines; i++)
{
- buf->data_stack[i] = (DWTELEM *) av_malloc (sizeof(DWTELEM) * line_width);
+ buf->data_stack[i] = av_malloc (sizeof(IDWTELEM) * line_width);
}
buf->data_stack_top = max_allocated_lines - 1;
}
-static DWTELEM * slice_buffer_load_line(slice_buffer * buf, int line)
+static IDWTELEM * slice_buffer_load_line(slice_buffer * buf, int line)
{
int offset;
- DWTELEM * buffer;
+ IDWTELEM * buffer;
// av_log(NULL, AV_LOG_DEBUG, "Cache hit: %d\n", line);
static void slice_buffer_release(slice_buffer * buf, int line)
{
int offset;
- DWTELEM * buffer;
+ IDWTELEM * buffer;
assert(line >= 0 && line < buf->line_count);
assert(buf->line[line]);
for (i = buf->data_count - 1; i >= 0; i--)
{
- assert(buf->data_stack[i]);
av_freep(&buf->data_stack[i]);
}
- assert(buf->data_stack);
av_freep(&buf->data_stack);
- assert(buf->line);
av_freep(&buf->line);
}
return v;
}
-static av_always_inline void lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
+static av_always_inline void
+lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
+ int dst_step, int src_step, int ref_step,
+ int width, int mul, int add, int shift,
+ int highpass, int inverse){
const int mirror_left= !highpass;
const int mirror_right= (width&1) ^ highpass;
const int w= (width>>1) - 1 + (highpass & width);
}
for(i=0; i<w; i++){
- dst[i*dst_step] = LIFT(src[i*src_step], ((mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add)>>shift), inverse);
+ dst[i*dst_step] =
+ LIFT(src[i*src_step],
+ ((mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add)>>shift),
+ inverse);
}
if(mirror_right){
- dst[w*dst_step] = LIFT(src[w*src_step], ((mul*2*ref[w*ref_step]+add)>>shift), inverse);
+ dst[w*dst_step] =
+ LIFT(src[w*src_step],
+ ((mul*2*ref[w*ref_step]+add)>>shift),
+ inverse);
}
}
-#ifndef lift5
-static av_always_inline void lift5(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
+static av_always_inline void
+inv_lift(IDWTELEM *dst, IDWTELEM *src, IDWTELEM *ref,
+ int dst_step, int src_step, int ref_step,
+ int width, int mul, int add, int shift,
+ int highpass, int inverse){
const int mirror_left= !highpass;
const int mirror_right= (width&1) ^ highpass;
const int w= (width>>1) - 1 + (highpass & width);
int i;
+#define LIFT(src, ref, inv) ((src) + ((inv) ? - (ref) : + (ref)))
if(mirror_left){
- int r= 3*2*ref[0];
- r += r>>4;
- r += r>>8;
- dst[0] = LIFT(src[0], ((r+add)>>shift), inverse);
+ dst[0] = LIFT(src[0], ((mul*2*ref[0]+add)>>shift), inverse);
dst += dst_step;
src += src_step;
}
for(i=0; i<w; i++){
- int r= 3*(ref[i*ref_step] + ref[(i+1)*ref_step]);
- r += r>>4;
- r += r>>8;
- dst[i*dst_step] = LIFT(src[i*src_step], ((r+add)>>shift), inverse);
+ dst[i*dst_step] =
+ LIFT(src[i*src_step],
+ ((mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add)>>shift),
+ inverse);
}
if(mirror_right){
- int r= 3*2*ref[w*ref_step];
- r += r>>4;
- r += r>>8;
- dst[w*dst_step] = LIFT(src[w*src_step], ((r+add)>>shift), inverse);
+ dst[w*dst_step] =
+ LIFT(src[w*src_step],
+ ((mul*2*ref[w*ref_step]+add)>>shift),
+ inverse);
}
}
-#endif
#ifndef liftS
-static av_always_inline void liftS(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
+static av_always_inline void
+liftS(DWTELEM *dst, DWTELEM *src, DWTELEM *ref,
+ int dst_step, int src_step, int ref_step,
+ int width, int mul, int add, int shift,
+ int highpass, int inverse){
const int mirror_left= !highpass;
const int mirror_right= (width&1) ^ highpass;
const int w= (width>>1) - 1 + (highpass & width);
int i;
assert(shift == 4);
-#define LIFTS(src, ref, inv) ((inv) ? (src) + (((ref) + 4*(src))>>shift): -((-16*4*(src) + 4*(ref) + add + 5 + (5<<27))/(5*16) - (1<<23)))
+#define LIFTS(src, ref, inv) \
+ ((inv) ? \
+ (src) + (((ref) + 4*(src))>>shift): \
+ -((-16*(src) + (ref) + add/4 + 1 + (5<<25))/(5*4) - (1<<23)))
if(mirror_left){
dst[0] = LIFTS(src[0], mul*2*ref[0]+add, inverse);
dst += dst_step;
}
for(i=0; i<w; i++){
- dst[i*dst_step] = LIFTS(src[i*src_step], mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add, inverse);
+ dst[i*dst_step] =
+ LIFTS(src[i*src_step],
+ mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add,
+ inverse);
}
if(mirror_right){
- dst[w*dst_step] = LIFTS(src[w*src_step], mul*2*ref[w*ref_step]+add, inverse);
- }
-}
-#endif
-
-
-static void inplace_lift(DWTELEM *dst, int width, int *coeffs, int n, int shift, int start, int inverse){
- int x, i;
-
- for(x=start; x<width; x+=2){
- int64_t sum=0;
-
- for(i=0; i<n; i++){
- int x2= x + 2*i - n + 1;
- if (x2< 0) x2= -x2;
- else if(x2>=width) x2= 2*width-x2-2;
- sum += coeffs[i]*(int64_t)dst[x2];
- }
- if(inverse) dst[x] -= (sum + (1<<shift)/2)>>shift;
- else dst[x] += (sum + (1<<shift)/2)>>shift;
- }
-}
-
-static void inplace_liftV(DWTELEM *dst, int width, int height, int stride, int *coeffs, int n, int shift, int start, int inverse){
- int x, y, i;
- for(y=start; y<height; y+=2){
- for(x=0; x<width; x++){
- int64_t sum=0;
-
- for(i=0; i<n; i++){
- int y2= y + 2*i - n + 1;
- if (y2< 0) y2= -y2;
- else if(y2>=height) y2= 2*height-y2-2;
- sum += coeffs[i]*(int64_t)dst[x + y2*stride];
- }
- if(inverse) dst[x + y*stride] -= (sum + (1<<shift)/2)>>shift;
- else dst[x + y*stride] += (sum + (1<<shift)/2)>>shift;
- }
- }
-}
-
-#define SCALEX 1
-#define LX0 0
-#define LX1 1
-
-#if 0 // more accurate 9/7
-#define N1 2
-#define SHIFT1 14
-#define COEFFS1 (int[]){-25987,-25987}
-#define N2 2
-#define SHIFT2 19
-#define COEFFS2 (int[]){-27777,-27777}
-#define N3 2
-#define SHIFT3 15
-#define COEFFS3 (int[]){28931,28931}
-#define N4 2
-#define SHIFT4 15
-#define COEFFS4 (int[]){14533,14533}
-#elif 1 // 13/7 CRF
-#define N1 4
-#define SHIFT1 4
-#define COEFFS1 (int[]){1,-9,-9,1}
-#define N2 4
-#define SHIFT2 4
-#define COEFFS2 (int[]){-1,5,5,-1}
-#define N3 0
-#define SHIFT3 1
-#define COEFFS3 NULL
-#define N4 0
-#define SHIFT4 1
-#define COEFFS4 NULL
-#elif 1 // 3/5
-#define LX0 1
-#define LX1 0
-#define SCALEX 0.5
-#define N1 2
-#define SHIFT1 1
-#define COEFFS1 (int[]){1,1}
-#define N2 2
-#define SHIFT2 2
-#define COEFFS2 (int[]){-1,-1}
-#define N3 0
-#define SHIFT3 0
-#define COEFFS3 NULL
-#define N4 0
-#define SHIFT4 0
-#define COEFFS4 NULL
-#elif 1 // 11/5
-#define N1 0
-#define SHIFT1 1
-#define COEFFS1 NULL
-#define N2 2
-#define SHIFT2 2
-#define COEFFS2 (int[]){-1,-1}
-#define N3 2
-#define SHIFT3 0
-#define COEFFS3 (int[]){-1,-1}
-#define N4 4
-#define SHIFT4 7
-#define COEFFS4 (int[]){-5,29,29,-5}
-#define SCALEX 4
-#elif 1 // 9/7 CDF
-#define N1 2
-#define SHIFT1 7
-#define COEFFS1 (int[]){-203,-203}
-#define N2 2
-#define SHIFT2 12
-#define COEFFS2 (int[]){-217,-217}
-#define N3 2
-#define SHIFT3 7
-#define COEFFS3 (int[]){113,113}
-#define N4 2
-#define SHIFT4 9
-#define COEFFS4 (int[]){227,227}
-#define SCALEX 1
-#elif 1 // 7/5 CDF
-#define N1 0
-#define SHIFT1 1
-#define COEFFS1 NULL
-#define N2 2
-#define SHIFT2 2
-#define COEFFS2 (int[]){-1,-1}
-#define N3 2
-#define SHIFT3 0
-#define COEFFS3 (int[]){-1,-1}
-#define N4 2
-#define SHIFT4 4
-#define COEFFS4 (int[]){3,3}
-#elif 1 // 9/7 MN
-#define N1 4
-#define SHIFT1 4
-#define COEFFS1 (int[]){1,-9,-9,1}
-#define N2 2
-#define SHIFT2 2
-#define COEFFS2 (int[]){1,1}
-#define N3 0
-#define SHIFT3 1
-#define COEFFS3 NULL
-#define N4 0
-#define SHIFT4 1
-#define COEFFS4 NULL
-#else // 13/7 CRF
-#define N1 4
-#define SHIFT1 4
-#define COEFFS1 (int[]){1,-9,-9,1}
-#define N2 4
-#define SHIFT2 4
-#define COEFFS2 (int[]){-1,5,5,-1}
-#define N3 0
-#define SHIFT3 1
-#define COEFFS3 NULL
-#define N4 0
-#define SHIFT4 1
-#define COEFFS4 NULL
-#endif
-static void horizontal_decomposeX(DWTELEM *b, int width){
- DWTELEM temp[width];
- const int width2= width>>1;
- const int w2= (width+1)>>1;
- int x;
-
- inplace_lift(b, width, COEFFS1, N1, SHIFT1, LX1, 0);
- inplace_lift(b, width, COEFFS2, N2, SHIFT2, LX0, 0);
- inplace_lift(b, width, COEFFS3, N3, SHIFT3, LX1, 0);
- inplace_lift(b, width, COEFFS4, N4, SHIFT4, LX0, 0);
-
- for(x=0; x<width2; x++){
- temp[x ]= b[2*x ];
- temp[x+w2]= b[2*x + 1];
+ dst[w*dst_step] =
+ LIFTS(src[w*src_step], mul*2*ref[w*ref_step]+add, inverse);
}
- if(width&1)
- temp[x ]= b[2*x ];
- memcpy(b, temp, width*sizeof(int));
}
+static av_always_inline void
+inv_liftS(IDWTELEM *dst, IDWTELEM *src, IDWTELEM *ref,
+ int dst_step, int src_step, int ref_step,
+ int width, int mul, int add, int shift,
+ int highpass, int inverse){
+ const int mirror_left= !highpass;
+ const int mirror_right= (width&1) ^ highpass;
+ const int w= (width>>1) - 1 + (highpass & width);
+ int i;
-static void horizontal_composeX(DWTELEM *b, int width){
- DWTELEM temp[width];
- const int width2= width>>1;
- int x;
- const int w2= (width+1)>>1;
-
- memcpy(temp, b, width*sizeof(int));
- for(x=0; x<width2; x++){
- b[2*x ]= temp[x ];
- b[2*x + 1]= temp[x+w2];
- }
- if(width&1)
- b[2*x ]= temp[x ];
-
- inplace_lift(b, width, COEFFS4, N4, SHIFT4, LX0, 1);
- inplace_lift(b, width, COEFFS3, N3, SHIFT3, LX1, 1);
- inplace_lift(b, width, COEFFS2, N2, SHIFT2, LX0, 1);
- inplace_lift(b, width, COEFFS1, N1, SHIFT1, LX1, 1);
-}
-
-static void spatial_decomposeX(DWTELEM *buffer, int width, int height, int stride){
- int x, y;
-
- for(y=0; y<height; y++){
- for(x=0; x<width; x++){
- buffer[y*stride + x] *= SCALEX;
- }
- }
-
- for(y=0; y<height; y++){
- horizontal_decomposeX(buffer + y*stride, width);
+ assert(shift == 4);
+#define LIFTS(src, ref, inv) \
+ ((inv) ? \
+ (src) + (((ref) + 4*(src))>>shift): \
+ -((-16*(src) + (ref) + add/4 + 1 + (5<<25))/(5*4) - (1<<23)))
+ if(mirror_left){
+ dst[0] = LIFTS(src[0], mul*2*ref[0]+add, inverse);
+ dst += dst_step;
+ src += src_step;
}
- inplace_liftV(buffer, width, height, stride, COEFFS1, N1, SHIFT1, LX1, 0);
- inplace_liftV(buffer, width, height, stride, COEFFS2, N2, SHIFT2, LX0, 0);
- inplace_liftV(buffer, width, height, stride, COEFFS3, N3, SHIFT3, LX1, 0);
- inplace_liftV(buffer, width, height, stride, COEFFS4, N4, SHIFT4, LX0, 0);
-}
-
-static void spatial_composeX(DWTELEM *buffer, int width, int height, int stride){
- int x, y;
-
- inplace_liftV(buffer, width, height, stride, COEFFS4, N4, SHIFT4, LX0, 1);
- inplace_liftV(buffer, width, height, stride, COEFFS3, N3, SHIFT3, LX1, 1);
- inplace_liftV(buffer, width, height, stride, COEFFS2, N2, SHIFT2, LX0, 1);
- inplace_liftV(buffer, width, height, stride, COEFFS1, N1, SHIFT1, LX1, 1);
-
- for(y=0; y<height; y++){
- horizontal_composeX(buffer + y*stride, width);
+ for(i=0; i<w; i++){
+ dst[i*dst_step] =
+ LIFTS(src[i*src_step],
+ mul*(ref[i*ref_step] + ref[(i+1)*ref_step])+add,
+ inverse);
}
- for(y=0; y<height; y++){
- for(x=0; x<width; x++){
- buffer[y*stride + x] /= SCALEX;
- }
+ if(mirror_right){
+ dst[w*dst_step] =
+ LIFTS(src[w*src_step], mul*2*ref[w*ref_step]+add, inverse);
}
}
+#endif
static void horizontal_decompose53i(DWTELEM *b, int width){
DWTELEM temp[width];
lift (temp+w2, b +1, b , 1, 2, 2, width, W_AM, W_AO, W_AS, 1, 1);
liftS(temp , b , temp+w2, 1, 2, 1, width, W_BM, W_BO, W_BS, 0, 0);
- lift5(b +w2, temp+w2, temp , 1, 1, 1, width, W_CM, W_CO, W_CS, 1, 0);
+ lift (b +w2, temp+w2, temp , 1, 1, 1, width, W_CM, W_CO, W_CS, 1, 0);
lift (b , temp , b +w2, 1, 1, 1, width, W_DM, W_DO, W_DS, 0, 0);
}
int i;
for(i=0; i<width; i++){
-#ifdef lift5
b1[i] += (W_CM*(b0[i] + b2[i])+W_CO)>>W_CS;
-#else
- int r= 3*(b0[i] + b2[i]);
- r+= r>>4;
- r+= r>>8;
- b1[i] += (r+W_CO)>>W_CS;
-#endif
}
}
switch(type){
case DWT_97: spatial_decompose97i(buffer, width>>level, height>>level, stride<<level); break;
case DWT_53: spatial_decompose53i(buffer, width>>level, height>>level, stride<<level); break;
- case DWT_X: spatial_decomposeX (buffer, width>>level, height>>level, stride<<level); break;
}
}
}
-static void horizontal_compose53i(DWTELEM *b, int width){
- DWTELEM temp[width];
+static void horizontal_compose53i(IDWTELEM *b, int width){
+ IDWTELEM temp[width];
const int width2= width>>1;
const int w2= (width+1)>>1;
int x;
b[width -1] = A3;
b[width2-1] = A2;
#else
- lift(temp , b , b+w2, 1, 1, 1, width, 1, 2, 2, 0, 1);
- lift(temp+w2, b+w2, temp, 1, 1, 1, width, -1, 0, 1, 1, 1);
+ inv_lift(temp , b , b+w2, 1, 1, 1, width, 1, 2, 2, 0, 1);
+ inv_lift(temp+w2, b+w2, temp, 1, 1, 1, width, -1, 0, 1, 1, 1);
#endif
for(x=0; x<width2; x++){
b[2*x ]= temp[x ];
b[2*x ]= temp[x ];
}
-static void vertical_compose53iH0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
+static void vertical_compose53iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
int i;
for(i=0; i<width; i++){
}
}
-static void vertical_compose53iL0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
+static void vertical_compose53iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
int i;
for(i=0; i<width; i++){
cs->y = -1;
}
-static void spatial_compose53i_init(dwt_compose_t *cs, DWTELEM *buffer, int height, int stride){
+static void spatial_compose53i_init(dwt_compose_t *cs, IDWTELEM *buffer, int height, int stride){
cs->b0 = buffer + mirror(-1-1, height-1)*stride;
cs->b1 = buffer + mirror(-1 , height-1)*stride;
cs->y = -1;
static void spatial_compose53i_dy_buffered(dwt_compose_t *cs, slice_buffer * sb, int width, int height, int stride_line){
int y= cs->y;
- DWTELEM *b0= cs->b0;
- DWTELEM *b1= cs->b1;
- DWTELEM *b2= slice_buffer_get_line(sb, mirror(y+1, height-1) * stride_line);
- DWTELEM *b3= slice_buffer_get_line(sb, mirror(y+2, height-1) * stride_line);
+ IDWTELEM *b0= cs->b0;
+ IDWTELEM *b1= cs->b1;
+ IDWTELEM *b2= slice_buffer_get_line(sb, mirror(y+1, height-1) * stride_line);
+ IDWTELEM *b3= slice_buffer_get_line(sb, mirror(y+2, height-1) * stride_line);
{START_TIMER
if(y+1<(unsigned)height) vertical_compose53iL0(b1, b2, b3, width);
cs->y += 2;
}
-static void spatial_compose53i_dy(dwt_compose_t *cs, DWTELEM *buffer, int width, int height, int stride){
+static void spatial_compose53i_dy(dwt_compose_t *cs, IDWTELEM *buffer, int width, int height, int stride){
int y= cs->y;
- DWTELEM *b0= cs->b0;
- DWTELEM *b1= cs->b1;
- DWTELEM *b2= buffer + mirror(y+1, height-1)*stride;
- DWTELEM *b3= buffer + mirror(y+2, height-1)*stride;
+ IDWTELEM *b0= cs->b0;
+ IDWTELEM *b1= cs->b1;
+ IDWTELEM *b2= buffer + mirror(y+1, height-1)*stride;
+ IDWTELEM *b3= buffer + mirror(y+2, height-1)*stride;
{START_TIMER
if(y+1<(unsigned)height) vertical_compose53iL0(b1, b2, b3, width);
cs->y += 2;
}
-static void spatial_compose53i(DWTELEM *buffer, int width, int height, int stride){
+static void spatial_compose53i(IDWTELEM *buffer, int width, int height, int stride){
dwt_compose_t cs;
spatial_compose53i_init(&cs, buffer, height, stride);
while(cs.y <= height)
}
-void ff_snow_horizontal_compose97i(DWTELEM *b, int width){
- DWTELEM temp[width];
+void ff_snow_horizontal_compose97i(IDWTELEM *b, int width){
+ IDWTELEM temp[width];
const int w2= (width+1)>>1;
- lift (temp , b , b +w2, 1, 1, 1, width, W_DM, W_DO, W_DS, 0, 1);
- lift5(temp+w2, b +w2, temp , 1, 1, 1, width, W_CM, W_CO, W_CS, 1, 1);
- liftS(b , temp , temp+w2, 2, 1, 1, width, W_BM, W_BO, W_BS, 0, 1);
- lift (b+1 , temp+w2, b , 2, 1, 2, width, W_AM, W_AO, W_AS, 1, 0);
+ inv_lift (temp , b , b +w2, 1, 1, 1, width, W_DM, W_DO, W_DS, 0, 1);
+ inv_lift (temp+w2, b +w2, temp , 1, 1, 1, width, W_CM, W_CO, W_CS, 1, 1);
+ inv_liftS(b , temp , temp+w2, 2, 1, 1, width, W_BM, W_BO, W_BS, 0, 1);
+ inv_lift (b+1 , temp+w2, b , 2, 1, 2, width, W_AM, W_AO, W_AS, 1, 0);
}
-static void vertical_compose97iH0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
+static void vertical_compose97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
int i;
for(i=0; i<width; i++){
}
}
-static void vertical_compose97iH1(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
+static void vertical_compose97iH1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
int i;
for(i=0; i<width; i++){
-#ifdef lift5
b1[i] -= (W_CM*(b0[i] + b2[i])+W_CO)>>W_CS;
-#else
- int r= 3*(b0[i] + b2[i]);
- r+= r>>4;
- r+= r>>8;
- b1[i] -= (r+W_CO)>>W_CS;
-#endif
}
}
-static void vertical_compose97iL0(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
+static void vertical_compose97iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
int i;
for(i=0; i<width; i++){
}
}
-static void vertical_compose97iL1(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, int width){
+static void vertical_compose97iL1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width){
int i;
for(i=0; i<width; i++){
}
}
-void ff_snow_vertical_compose97i(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, DWTELEM *b3, DWTELEM *b4, DWTELEM *b5, int width){
+void ff_snow_vertical_compose97i(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width){
int i;
for(i=0; i<width; i++){
-#ifndef lift5
- int r;
-#endif
b4[i] -= (W_DM*(b3[i] + b5[i])+W_DO)>>W_DS;
-#ifdef lift5
b3[i] -= (W_CM*(b2[i] + b4[i])+W_CO)>>W_CS;
-#else
- r= 3*(b2[i] + b4[i]);
- r+= r>>4;
- r+= r>>8;
- b3[i] -= (r+W_CO)>>W_CS;
-#endif
#ifdef liftS
b2[i] += (W_BM*(b1[i] + b3[i])+W_BO)>>W_BS;
#else
cs->y = -3;
}
-static void spatial_compose97i_init(dwt_compose_t *cs, DWTELEM *buffer, int height, int stride){
+static void spatial_compose97i_init(dwt_compose_t *cs, IDWTELEM *buffer, int height, int stride){
cs->b0 = buffer + mirror(-3-1, height-1)*stride;
cs->b1 = buffer + mirror(-3 , height-1)*stride;
cs->b2 = buffer + mirror(-3+1, height-1)*stride;
static void spatial_compose97i_dy_buffered(DSPContext *dsp, dwt_compose_t *cs, slice_buffer * sb, int width, int height, int stride_line){
int y = cs->y;
- DWTELEM *b0= cs->b0;
- DWTELEM *b1= cs->b1;
- DWTELEM *b2= cs->b2;
- DWTELEM *b3= cs->b3;
- DWTELEM *b4= slice_buffer_get_line(sb, mirror(y + 3, height - 1) * stride_line);
- DWTELEM *b5= slice_buffer_get_line(sb, mirror(y + 4, height - 1) * stride_line);
+ IDWTELEM *b0= cs->b0;
+ IDWTELEM *b1= cs->b1;
+ IDWTELEM *b2= cs->b2;
+ IDWTELEM *b3= cs->b3;
+ IDWTELEM *b4= slice_buffer_get_line(sb, mirror(y + 3, height - 1) * stride_line);
+ IDWTELEM *b5= slice_buffer_get_line(sb, mirror(y + 4, height - 1) * stride_line);
{START_TIMER
if(y>0 && y+4<height){
cs->y += 2;
}
-static void spatial_compose97i_dy(dwt_compose_t *cs, DWTELEM *buffer, int width, int height, int stride){
+static void spatial_compose97i_dy(dwt_compose_t *cs, IDWTELEM *buffer, int width, int height, int stride){
int y = cs->y;
- DWTELEM *b0= cs->b0;
- DWTELEM *b1= cs->b1;
- DWTELEM *b2= cs->b2;
- DWTELEM *b3= cs->b3;
- DWTELEM *b4= buffer + mirror(y+3, height-1)*stride;
- DWTELEM *b5= buffer + mirror(y+4, height-1)*stride;
+ IDWTELEM *b0= cs->b0;
+ IDWTELEM *b1= cs->b1;
+ IDWTELEM *b2= cs->b2;
+ IDWTELEM *b3= cs->b3;
+ IDWTELEM *b4= buffer + mirror(y+3, height-1)*stride;
+ IDWTELEM *b5= buffer + mirror(y+4, height-1)*stride;
{START_TIMER
if(y+3<(unsigned)height) vertical_compose97iL1(b3, b4, b5, width);
cs->y += 2;
}
-static void spatial_compose97i(DWTELEM *buffer, int width, int height, int stride){
+static void spatial_compose97i(IDWTELEM *buffer, int width, int height, int stride){
dwt_compose_t cs;
spatial_compose97i_init(&cs, buffer, height, stride);
while(cs.y <= height)
switch(type){
case DWT_97: spatial_compose97i_buffered_init(cs+level, sb, height>>level, stride_line<<level); break;
case DWT_53: spatial_compose53i_buffered_init(cs+level, sb, height>>level, stride_line<<level); break;
- /* not slicified yet */
- case DWT_X: /*spatial_composeX(buffer, width>>level, height>>level, stride<<level); break;*/
- av_log(NULL, AV_LOG_ERROR, "spatial_composeX neither buffered nor slicified yet.\n"); break;
}
}
}
-static void ff_spatial_idwt_init(dwt_compose_t *cs, DWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count){
+static void ff_spatial_idwt_init(dwt_compose_t *cs, IDWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count){
int level;
for(level=decomposition_count-1; level>=0; level--){
switch(type){
case DWT_97: spatial_compose97i_init(cs+level, buffer, height>>level, stride<<level); break;
case DWT_53: spatial_compose53i_init(cs+level, buffer, height>>level, stride<<level); break;
- /* not slicified yet */
- case DWT_X: spatial_composeX(buffer, width>>level, height>>level, stride<<level); break;
}
}
}
-static void ff_spatial_idwt_slice(dwt_compose_t *cs, DWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count, int y){
+static void ff_spatial_idwt_slice(dwt_compose_t *cs, IDWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count, int y){
const int support = type==1 ? 3 : 5;
int level;
if(type==2) return;
break;
case DWT_53: spatial_compose53i_dy(cs+level, buffer, width>>level, height>>level, stride<<level);
break;
- case DWT_X: break;
}
}
}
break;
case DWT_53: spatial_compose53i_dy_buffered(cs+level, slice_buf, width>>level, height>>level, stride_line<<level);
break;
- case DWT_X: break;
}
}
}
}
-static void ff_spatial_idwt(DWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count){
- if(type==2){
- int level;
- for(level=decomposition_count-1; level>=0; level--)
- spatial_composeX (buffer, width>>level, height>>level, stride<<level);
- }else{
+static void ff_spatial_idwt(IDWTELEM *buffer, int width, int height, int stride, int type, int decomposition_count){
dwt_compose_t cs[MAX_DECOMPOSITIONS];
int y;
ff_spatial_idwt_init(cs, buffer, width, height, stride, type, decomposition_count);
for(y=0; y<height; y+=4)
ff_spatial_idwt_slice(cs, buffer, width, height, stride, type, decomposition_count, y);
- }
}
-static int encode_subband_c0run(SnowContext *s, SubBand *b, DWTELEM *src, DWTELEM *parent, int stride, int orientation){
+static int encode_subband_c0run(SnowContext *s, SubBand *b, IDWTELEM *src, IDWTELEM *parent, int stride, int orientation){
const int w= b->width;
const int h= b->height;
int x, y;
return 0;
}
-static int encode_subband(SnowContext *s, SubBand *b, DWTELEM *src, DWTELEM *parent, int stride, int orientation){
+static int encode_subband(SnowContext *s, SubBand *b, IDWTELEM *src, IDWTELEM *parent, int stride, int orientation){
// encode_subband_qtree(s, b, src, parent, stride, orientation);
// encode_subband_z0run(s, b, src, parent, stride, orientation);
return encode_subband_c0run(s, b, src, parent, stride, orientation);
START_TIMER
- if(b->buf == s->spatial_dwt_buffer || s->qlog == LOSSLESS_QLOG){
+ if(b->ibuf == s->spatial_idwt_buffer || s->qlog == LOSSLESS_QLOG){
qadd= 0;
qmul= 1<<QEXPSHIFT;
}
for(y=start_y; y<h; y++){
int x = 0;
int v;
- DWTELEM * line = slice_buffer_get_line(sb, y * b->stride_line + b->buf_y_offset) + b->buf_x_offset;
- memset(line, 0, b->width*sizeof(DWTELEM));
+ IDWTELEM * line = slice_buffer_get_line(sb, y * b->stride_line + b->buf_y_offset) + b->buf_x_offset;
+ memset(line, 0, b->width*sizeof(IDWTELEM));
v = b->x_coeff[new_index].coeff;
x = b->x_coeff[new_index++].x;
while(x < w)
int plane_index, level, orientation;
for(plane_index=0; plane_index<3; plane_index++){
- for(level=0; level<s->spatial_decomposition_count; level++){
+ for(level=0; level<MAX_DECOMPOSITIONS; level++){
for(orientation=level ? 1:0; orientation<4; orientation++){
memset(s->plane[plane_index].band[level][orientation].state, MID_STATE, sizeof(s->plane[plane_index].band[level][orientation].state));
}
}
}
-static void mc_block(uint8_t *dst, const uint8_t *src, uint8_t *tmp, int stride, int b_w, int b_h, int dx, int dy){
- int x, y;
+static void mc_block(Plane *p, uint8_t *dst, const uint8_t *src, uint8_t *tmp, int stride, int b_w, int b_h, int dx, int dy){
+ const static uint8_t weight[64]={
+ 8,7,6,5,4,3,2,1,
+ 7,7,0,0,0,0,0,1,
+ 6,0,6,0,0,0,2,0,
+ 5,0,0,5,0,3,0,0,
+ 4,0,0,0,4,0,0,0,
+ 3,0,0,5,0,3,0,0,
+ 2,0,6,0,0,0,2,0,
+ 1,7,0,0,0,0,0,1,
+ };
+
+ const static uint8_t brane[256]={
+ 0x00,0x01,0x01,0x01,0x01,0x01,0x01,0x01,0x11,0x12,0x12,0x12,0x12,0x12,0x12,0x12,
+ 0x04,0x05,0xcc,0xcc,0xcc,0xcc,0xcc,0x41,0x15,0x16,0xcc,0xcc,0xcc,0xcc,0xcc,0x52,
+ 0x04,0xcc,0x05,0xcc,0xcc,0xcc,0x41,0xcc,0x15,0xcc,0x16,0xcc,0xcc,0xcc,0x52,0xcc,
+ 0x04,0xcc,0xcc,0x05,0xcc,0x41,0xcc,0xcc,0x15,0xcc,0xcc,0x16,0xcc,0x52,0xcc,0xcc,
+ 0x04,0xcc,0xcc,0xcc,0x41,0xcc,0xcc,0xcc,0x15,0xcc,0xcc,0xcc,0x16,0xcc,0xcc,0xcc,
+ 0x04,0xcc,0xcc,0x41,0xcc,0x05,0xcc,0xcc,0x15,0xcc,0xcc,0x52,0xcc,0x16,0xcc,0xcc,
+ 0x04,0xcc,0x41,0xcc,0xcc,0xcc,0x05,0xcc,0x15,0xcc,0x52,0xcc,0xcc,0xcc,0x16,0xcc,
+ 0x04,0x41,0xcc,0xcc,0xcc,0xcc,0xcc,0x05,0x15,0x52,0xcc,0xcc,0xcc,0xcc,0xcc,0x16,
+ 0x44,0x45,0x45,0x45,0x45,0x45,0x45,0x45,0x55,0x56,0x56,0x56,0x56,0x56,0x56,0x56,
+ 0x48,0x49,0xcc,0xcc,0xcc,0xcc,0xcc,0x85,0x59,0x5A,0xcc,0xcc,0xcc,0xcc,0xcc,0x96,
+ 0x48,0xcc,0x49,0xcc,0xcc,0xcc,0x85,0xcc,0x59,0xcc,0x5A,0xcc,0xcc,0xcc,0x96,0xcc,
+ 0x48,0xcc,0xcc,0x49,0xcc,0x85,0xcc,0xcc,0x59,0xcc,0xcc,0x5A,0xcc,0x96,0xcc,0xcc,
+ 0x48,0xcc,0xcc,0xcc,0x49,0xcc,0xcc,0xcc,0x59,0xcc,0xcc,0xcc,0x96,0xcc,0xcc,0xcc,
+ 0x48,0xcc,0xcc,0x85,0xcc,0x49,0xcc,0xcc,0x59,0xcc,0xcc,0x96,0xcc,0x5A,0xcc,0xcc,
+ 0x48,0xcc,0x85,0xcc,0xcc,0xcc,0x49,0xcc,0x59,0xcc,0x96,0xcc,0xcc,0xcc,0x5A,0xcc,
+ 0x48,0x85,0xcc,0xcc,0xcc,0xcc,0xcc,0x49,0x59,0x96,0xcc,0xcc,0xcc,0xcc,0xcc,0x5A,
+ };
+
+ const static uint8_t needs[16]={
+ 0,1,0,0,
+ 2,4,2,0,
+ 0,1,0,0,
+ 15
+ };
+
+ int x, y, b, r, l;
+ int16_t tmpIt [64*(32+HTAPS_MAX)];
+ uint8_t tmp2t[3][stride*(32+HTAPS_MAX)];
+ int16_t *tmpI= tmpIt;
+ uint8_t *tmp2= tmp2t[0];
+ uint8_t *hpel[11];
START_TIMER
- for(y=0; y < b_h+5; y++){
- for(x=0; x < b_w; x++){
- int a0= src[x ];
- int a1= src[x + 1];
- int a2= src[x + 2];
- int a3= src[x + 3];
- int a4= src[x + 4];
- int a5= src[x + 5];
-// int am= 9*(a1+a2) - (a0+a3);
- int am= 20*(a2+a3) - 5*(a1+a4) + (a0+a5);
-// int am= 18*(a2+a3) - 2*(a1+a4);
-// int aL= (-7*a0 + 105*a1 + 35*a2 - 5*a3)>>3;
-// int aR= (-7*a3 + 105*a2 + 35*a1 - 5*a0)>>3;
-
-// if(b_w==16) am= 8*(a1+a2);
-
- if(dx<8) am = (32*a2*( 8-dx) + am* dx + 128)>>8;
- else am = ( am*(16-dx) + 32*a3*(dx-8) + 128)>>8;
-
- /* FIXME Try increasing tmp buffer to 16 bits and not clipping here. Should give marginally better results. - Robert*/
- if(am&(~255)) am= ~(am>>31);
-
- tmp[x] = am;
-
-/* if (dx< 4) tmp[x + y*stride]= (16*a1*( 4-dx) + aL* dx + 32)>>6;
- else if(dx< 8) tmp[x + y*stride]= ( aL*( 8-dx) + am*(dx- 4) + 32)>>6;
- else if(dx<12) tmp[x + y*stride]= ( am*(12-dx) + aR*(dx- 8) + 32)>>6;
- else tmp[x + y*stride]= ( aR*(16-dx) + 16*a2*(dx-12) + 32)>>6;*/
+ assert(dx<16 && dy<16);
+ r= brane[dx + 16*dy]&15;
+ l= brane[dx + 16*dy]>>4;
+
+ b= needs[l] | needs[r];
+ if(p && !p->diag_mc)
+ b= 15;
+
+ if(b&5){
+ for(y=0; y < b_h+HTAPS_MAX-1; y++){
+ for(x=0; x < b_w; x++){
+ int a_1=src[x + HTAPS_MAX/2-4];
+ int a0= src[x + HTAPS_MAX/2-3];
+ int a1= src[x + HTAPS_MAX/2-2];
+ int a2= src[x + HTAPS_MAX/2-1];
+ int a3= src[x + HTAPS_MAX/2+0];
+ int a4= src[x + HTAPS_MAX/2+1];
+ int a5= src[x + HTAPS_MAX/2+2];
+ int a6= src[x + HTAPS_MAX/2+3];
+ int am=0;
+ if(!p || p->fast_mc){
+ am= 20*(a2+a3) - 5*(a1+a4) + (a0+a5);
+ tmpI[x]= am;
+ am= (am+16)>>5;
+ }else{
+ am= p->hcoeff[0]*(a2+a3) + p->hcoeff[1]*(a1+a4) + p->hcoeff[2]*(a0+a5) + p->hcoeff[3]*(a_1+a6);
+ tmpI[x]= am;
+ am= (am+32)>>6;
+ }
+
+ if(am&(~255)) am= ~(am>>31);
+ tmp2[x]= am;
+ }
+ tmpI+= 64;
+ tmp2+= stride;
+ src += stride;
+ }
+ src -= stride*y;
+ }
+ src += HTAPS_MAX/2 - 1;
+ tmp2= tmp2t[1];
+
+ if(b&2){
+ for(y=0; y < b_h; y++){
+ for(x=0; x < b_w+1; x++){
+ int a_1=src[x + (HTAPS_MAX/2-4)*stride];
+ int a0= src[x + (HTAPS_MAX/2-3)*stride];
+ int a1= src[x + (HTAPS_MAX/2-2)*stride];
+ int a2= src[x + (HTAPS_MAX/2-1)*stride];
+ int a3= src[x + (HTAPS_MAX/2+0)*stride];
+ int a4= src[x + (HTAPS_MAX/2+1)*stride];
+ int a5= src[x + (HTAPS_MAX/2+2)*stride];
+ int a6= src[x + (HTAPS_MAX/2+3)*stride];
+ int am=0;
+ if(!p || p->fast_mc)
+ am= (20*(a2+a3) - 5*(a1+a4) + (a0+a5) + 16)>>5;
+ else
+ am= (p->hcoeff[0]*(a2+a3) + p->hcoeff[1]*(a1+a4) + p->hcoeff[2]*(a0+a5) + p->hcoeff[3]*(a_1+a6) + 32)>>6;
+
+ if(am&(~255)) am= ~(am>>31);
+ tmp2[x]= am;
+ }
+ src += stride;
+ tmp2+= stride;
+ }
+ src -= stride*y;
+ }
+ src += stride*(HTAPS_MAX/2 - 1);
+ tmp2= tmp2t[2];
+ tmpI= tmpIt;
+ if(b&4){
+ for(y=0; y < b_h; y++){
+ for(x=0; x < b_w; x++){
+ int a_1=tmpI[x + (HTAPS_MAX/2-4)*64];
+ int a0= tmpI[x + (HTAPS_MAX/2-3)*64];
+ int a1= tmpI[x + (HTAPS_MAX/2-2)*64];
+ int a2= tmpI[x + (HTAPS_MAX/2-1)*64];
+ int a3= tmpI[x + (HTAPS_MAX/2+0)*64];
+ int a4= tmpI[x + (HTAPS_MAX/2+1)*64];
+ int a5= tmpI[x + (HTAPS_MAX/2+2)*64];
+ int a6= tmpI[x + (HTAPS_MAX/2+3)*64];
+ int am=0;
+ if(!p || p->fast_mc)
+ am= (20*(a2+a3) - 5*(a1+a4) + (a0+a5) + 512)>>10;
+ else
+ am= (p->hcoeff[0]*(a2+a3) + p->hcoeff[1]*(a1+a4) + p->hcoeff[2]*(a0+a5) + p->hcoeff[3]*(a_1+a6) + 2048)>>12;
+ if(am&(~255)) am= ~(am>>31);
+ tmp2[x]= am;
+ }
+ tmpI+= 64;
+ tmp2+= stride;
+ }
+ }
+
+ hpel[ 0]= src;
+ hpel[ 1]= tmp2t[0] + stride*(HTAPS_MAX/2-1);
+ hpel[ 2]= src + 1;
+
+ hpel[ 4]= tmp2t[1];
+ hpel[ 5]= tmp2t[2];
+ hpel[ 6]= tmp2t[1] + 1;
+
+ hpel[ 8]= src + stride;
+ hpel[ 9]= hpel[1] + stride;
+ hpel[10]= hpel[8] + 1;
+
+ if(b==15){
+ uint8_t *src1= hpel[dx/8 + dy/8*4 ];
+ uint8_t *src2= hpel[dx/8 + dy/8*4+1];
+ uint8_t *src3= hpel[dx/8 + dy/8*4+4];
+ uint8_t *src4= hpel[dx/8 + dy/8*4+5];
+ dx&=7;
+ dy&=7;
+ for(y=0; y < b_h; y++){
+ for(x=0; x < b_w; x++){
+ dst[x]= ((8-dx)*(8-dy)*src1[x] + dx*(8-dy)*src2[x]+
+ (8-dx)* dy *src3[x] + dx* dy *src4[x]+32)>>6;
+ }
+ src1+=stride;
+ src2+=stride;
+ src3+=stride;
+ src4+=stride;
+ dst +=stride;
}
- tmp += stride;
- src += stride;
- }
- tmp -= (b_h+5)*stride;
-
- for(y=0; y < b_h; y++){
- for(x=0; x < b_w; x++){
- int a0= tmp[x + 0*stride];
- int a1= tmp[x + 1*stride];
- int a2= tmp[x + 2*stride];
- int a3= tmp[x + 3*stride];
- int a4= tmp[x + 4*stride];
- int a5= tmp[x + 5*stride];
- int am= 20*(a2+a3) - 5*(a1+a4) + (a0+a5);
-// int am= 18*(a2+a3) - 2*(a1+a4);
-/* int aL= (-7*a0 + 105*a1 + 35*a2 - 5*a3)>>3;
- int aR= (-7*a3 + 105*a2 + 35*a1 - 5*a0)>>3;*/
-
-// if(b_w==16) am= 8*(a1+a2);
-
- if(dy<8) am = (32*a2*( 8-dy) + am* dy + 128)>>8;
- else am = ( am*(16-dy) + 32*a3*(dy-8) + 128)>>8;
-
- if(am&(~255)) am= ~(am>>31);
-
- dst[x] = am;
-/* if (dy< 4) tmp[x + y*stride]= (16*a1*( 4-dy) + aL* dy + 32)>>6;
- else if(dy< 8) tmp[x + y*stride]= ( aL*( 8-dy) + am*(dy- 4) + 32)>>6;
- else if(dy<12) tmp[x + y*stride]= ( am*(12-dy) + aR*(dy- 8) + 32)>>6;
- else tmp[x + y*stride]= ( aR*(16-dy) + 16*a2*(dy-12) + 32)>>6;*/
+ }else{
+ uint8_t *src1= hpel[l];
+ uint8_t *src2= hpel[r];
+ int a= weight[((dx&7) + (8*(dy&7)))];
+ int b= 8-a;
+ for(y=0; y < b_h; y++){
+ for(x=0; x < b_w; x++){
+ dst[x]= (a*src1[x] + b*src2[x] + 4)>>3;
+ }
+ src1+=stride;
+ src2+=stride;
+ dst +=stride;
}
- dst += stride;
- tmp += stride;
}
STOP_TIMER("mc_block")
}
#define mca(dx,dy,b_w)\
static void mc_block_hpel ## dx ## dy ## b_w(uint8_t *dst, const uint8_t *src, int stride, int h){\
- uint8_t tmp[stride*(b_w+5)];\
+ uint8_t tmp[stride*(b_w+HTAPS_MAX-1)];\
assert(h==b_w);\
- mc_block(dst, src-2-2*stride, tmp, stride, b_w, b_w, dx, dy);\
+ mc_block(NULL, dst, src-(HTAPS_MAX/2-1)-(HTAPS_MAX/2-1)*stride, tmp, stride, b_w, b_w, dx, dy);\
}
mca( 0, 0,16)
const int dx= mx&15;
const int dy= my&15;
const int tab_index= 3 - (b_w>>2) + (b_w>>4);
- sx += (mx>>4) - 2;
- sy += (my>>4) - 2;
+ sx += (mx>>4) - (HTAPS_MAX/2-1);
+ sy += (my>>4) - (HTAPS_MAX/2-1);
src += sx + sy*stride;
- if( (unsigned)sx >= w - b_w - 4
- || (unsigned)sy >= h - b_h - 4){
- ff_emulated_edge_mc(tmp + MB_SIZE, src, stride, b_w+5, b_h+5, sx, sy, w, h);
+ if( (unsigned)sx >= w - b_w - (HTAPS_MAX-2)
+ || (unsigned)sy >= h - b_h - (HTAPS_MAX-2)){
+ ff_emulated_edge_mc(tmp + MB_SIZE, src, stride, b_w+HTAPS_MAX-1, b_h+HTAPS_MAX-1, sx, sy, w, h);
src= tmp + MB_SIZE;
}
// assert(b_w == b_h || 2*b_w == b_h || b_w == 2*b_h);
// assert(!(b_w&(b_w-1)));
assert(b_w>1 && b_h>1);
assert(tab_index>=0 && tab_index<4 || b_w==32);
- if((dx&3) || (dy&3) || !(b_w == b_h || 2*b_w == b_h || b_w == 2*b_h) || (b_w&(b_w-1)))
- mc_block(dst, src, tmp, stride, b_w, b_h, dx, dy);
+ if((dx&3) || (dy&3) || !(b_w == b_h || 2*b_w == b_h || b_w == 2*b_h) || (b_w&(b_w-1)) || !s->plane[plane_index].fast_mc )
+ mc_block(&s->plane[plane_index], dst, src, tmp, stride, b_w, b_h, dx, dy);
else if(b_w==32){
int y;
for(y=0; y<b_h; y+=16){
- s->dsp.put_h264_qpel_pixels_tab[0][dy+(dx>>2)](dst + y*stride, src + 2 + (y+2)*stride,stride);
- s->dsp.put_h264_qpel_pixels_tab[0][dy+(dx>>2)](dst + 16 + y*stride, src + 18 + (y+2)*stride,stride);
+ s->dsp.put_h264_qpel_pixels_tab[0][dy+(dx>>2)](dst + y*stride, src + 3 + (y+3)*stride,stride);
+ s->dsp.put_h264_qpel_pixels_tab[0][dy+(dx>>2)](dst + 16 + y*stride, src + 19 + (y+3)*stride,stride);
}
}else if(b_w==b_h)
- s->dsp.put_h264_qpel_pixels_tab[tab_index ][dy+(dx>>2)](dst,src + 2 + 2*stride,stride);
+ s->dsp.put_h264_qpel_pixels_tab[tab_index ][dy+(dx>>2)](dst,src + 3 + 3*stride,stride);
else if(b_w==2*b_h){
- s->dsp.put_h264_qpel_pixels_tab[tab_index+1][dy+(dx>>2)](dst ,src + 2 + 2*stride,stride);
- s->dsp.put_h264_qpel_pixels_tab[tab_index+1][dy+(dx>>2)](dst+b_h,src + 2 + b_h + 2*stride,stride);
+ s->dsp.put_h264_qpel_pixels_tab[tab_index+1][dy+(dx>>2)](dst ,src + 3 + 3*stride,stride);
+ s->dsp.put_h264_qpel_pixels_tab[tab_index+1][dy+(dx>>2)](dst+b_h,src + 3 + b_h + 3*stride,stride);
}else{
assert(2*b_w==b_h);
- s->dsp.put_h264_qpel_pixels_tab[tab_index ][dy+(dx>>2)](dst ,src + 2 + 2*stride ,stride);
- s->dsp.put_h264_qpel_pixels_tab[tab_index ][dy+(dx>>2)](dst+b_w*stride,src + 2 + 2*stride+b_w*stride,stride);
+ s->dsp.put_h264_qpel_pixels_tab[tab_index ][dy+(dx>>2)](dst ,src + 3 + 3*stride ,stride);
+ s->dsp.put_h264_qpel_pixels_tab[tab_index ][dy+(dx>>2)](dst+b_w*stride,src + 3 + 3*stride+b_w*stride,stride);
}
}
}
void ff_snow_inner_add_yblock(const uint8_t *obmc, const int obmc_stride, uint8_t * * block, int b_w, int b_h,
int src_x, int src_y, int src_stride, slice_buffer * sb, int add, uint8_t * dst8){
int y, x;
- DWTELEM * dst;
+ IDWTELEM * dst;
for(y=0; y<b_h; y++){
//FIXME ugly misuse of obmc_stride
const uint8_t *obmc1= obmc + y*obmc_stride;
v <<= 8 - LOG2_OBMC_MAX;
if(FRAC_BITS != 8){
- v += 1<<(7 - FRAC_BITS);
v >>= 8 - FRAC_BITS;
}
if(add){
}
//FIXME name clenup (b_w, block_w, b_width stuff)
-static av_always_inline void add_yblock(SnowContext *s, int sliced, slice_buffer *sb, DWTELEM *dst, uint8_t *dst8, const uint8_t *obmc, int src_x, int src_y, int b_w, int b_h, int w, int h, int dst_stride, int src_stride, int obmc_stride, int b_x, int b_y, int add, int offset_dst, int plane_index){
+static av_always_inline void add_yblock(SnowContext *s, int sliced, slice_buffer *sb, IDWTELEM *dst, uint8_t *dst8, const uint8_t *obmc, int src_x, int src_y, int b_w, int b_h, int w, int h, int dst_stride, int src_stride, int obmc_stride, int b_x, int b_y, int add, int offset_dst, int plane_index){
const int b_width = s->b_width << s->block_max_depth;
const int b_height= s->b_height << s->block_max_depth;
const int b_stride= b_width;
v <<= 8 - LOG2_OBMC_MAX;
if(FRAC_BITS != 8){
- v += 1<<(7 - FRAC_BITS);
v >>= 8 - FRAC_BITS;
}
if(add){
#endif
}
-static av_always_inline void predict_slice_buffered(SnowContext *s, slice_buffer * sb, DWTELEM * old_buffer, int plane_index, int add, int mb_y){
+static av_always_inline void predict_slice_buffered(SnowContext *s, slice_buffer * sb, IDWTELEM * old_buffer, int plane_index, int add, int mb_y){
Plane *p= &s->plane[plane_index];
const int mb_w= s->b_width << s->block_max_depth;
const int mb_h= s->b_height << s->block_max_depth;
for(y=block_w*mb_y; y<FFMIN(h,block_w*(mb_y+1)); y++)
{
// DWTELEM * line = slice_buffer_get_line(sb, y);
- DWTELEM * line = sb->line[y];
+ IDWTELEM * line = sb->line[y];
for(x=0; x<w; x++)
{
// int v= buf[x + y*w] + (128<<FRAC_BITS) + (1<<(FRAC_BITS-1));
for(y=block_w*mb_y; y<FFMIN(h,block_w*(mb_y+1)); y++)
{
// DWTELEM * line = slice_buffer_get_line(sb, y);
- DWTELEM * line = sb->line[y];
+ IDWTELEM * line = sb->line[y];
for(x=0; x<w; x++)
{
line[x] -= 128 << FRAC_BITS;
STOP_TIMER("predict_slice")
}
-static av_always_inline void predict_slice(SnowContext *s, DWTELEM *buf, int plane_index, int add, int mb_y){
+static av_always_inline void predict_slice(SnowContext *s, IDWTELEM *buf, int plane_index, int add, int mb_y){
Plane *p= &s->plane[plane_index];
const int mb_w= s->b_width << s->block_max_depth;
const int mb_h= s->b_height << s->block_max_depth;
STOP_TIMER("predict_slice")
}
-static av_always_inline void predict_plane(SnowContext *s, DWTELEM *buf, int plane_index, int add){
+static av_always_inline void predict_plane(SnowContext *s, IDWTELEM *buf, int plane_index, int add){
const int mb_h= s->b_height << s->block_max_depth;
int mb_y;
for(mb_y=0; mb_y<=mb_h; mb_y++)
const int obmc_stride= plane_index ? block_size : 2*block_size;
const int ref_stride= s->current_picture.linesize[plane_index];
uint8_t *src= s-> input_picture.data[plane_index];
- DWTELEM *dst= (DWTELEM*)s->m.obmc_scratchpad + plane_index*block_size*block_size*4;
+ IDWTELEM *dst= (IDWTELEM*)s->m.obmc_scratchpad + plane_index*block_size*block_size*4; //FIXME change to unsigned
const int b_stride = s->b_width << s->block_max_depth;
const int w= p->width;
const int h= p->height;
b->type|= BLOCK_INTRA;
b->color[plane_index]= 0;
- memset(dst, 0, obmc_stride*obmc_stride*sizeof(DWTELEM));
+ memset(dst, 0, obmc_stride*obmc_stride*sizeof(IDWTELEM));
for(i=0; i<4; i++){
int mb_x2= mb_x + (i &1) - 1;
const int ref_stride= s->current_picture.linesize[plane_index];
uint8_t *dst= s->current_picture.data[plane_index];
uint8_t *src= s-> input_picture.data[plane_index];
- DWTELEM *pred= (DWTELEM*)s->m.obmc_scratchpad + plane_index*block_size*block_size*4;
+ IDWTELEM *pred= (IDWTELEM*)s->m.obmc_scratchpad + plane_index*block_size*block_size*4;
uint8_t cur[ref_stride*2*MB_SIZE]; //FIXME alignment
- uint8_t tmp[ref_stride*(2*MB_SIZE+5)];
+ uint8_t tmp[ref_stride*(2*MB_SIZE+HTAPS_MAX-1)];
const int b_stride = s->b_width << s->block_max_depth;
const int b_height = s->b_height<< s->block_max_depth;
const int w= p->width;
for(y=y0; y<y1; y++){
const uint8_t *obmc1= obmc_edged + y*obmc_stride;
- const DWTELEM *pred1 = pred + y*obmc_stride;
+ const IDWTELEM *pred1 = pred + y*obmc_stride;
uint8_t *cur1 = cur + y*ref_stride;
uint8_t *dst1 = dst + sx + (sy+y)*ref_stride;
for(x=x0; x<x1; x++){
+#if FRAC_BITS >= LOG2_OBMC_MAX
int v = (cur1[x] * obmc1[x]) << (FRAC_BITS - LOG2_OBMC_MAX);
+#else
+ int v = (cur1[x] * obmc1[x] + (1<<(LOG2_OBMC_MAX - FRAC_BITS-1))) >> (LOG2_OBMC_MAX - FRAC_BITS);
+#endif
v = (v + pred1[x]) >> FRAC_BITS;
if(v&(~255)) v= ~(v>>31);
dst1[x] = v;
const int ref_stride= s->current_picture.linesize[plane_index];
uint8_t *dst= s->current_picture.data[plane_index];
uint8_t *src= s-> input_picture.data[plane_index];
- static const DWTELEM zero_dst[4096]; //FIXME
+ static const IDWTELEM zero_dst[4096]; //FIXME
const int b_stride = s->b_width << s->block_max_depth;
const int w= p->width;
const int h= p->height;
}
}
-static void quantize(SnowContext *s, SubBand *b, DWTELEM *src, int stride, int bias){
+static void quantize(SnowContext *s, SubBand *b, IDWTELEM *dst, DWTELEM *src, int stride, int bias){
const int level= b->level;
const int w= b->width;
const int h= b->height;
const int qlog= av_clip(s->qlog + b->qlog, 0, QROOT*16);
- const int qmul= qexp[qlog&(QROOT-1)]<<(qlog>>QSHIFT);
+ const int qmul= qexp[qlog&(QROOT-1)]<<((qlog>>QSHIFT) + ENCODER_EXTRA_BITS);
int x,y, thres1, thres2;
// START_TIMER
- if(s->qlog == LOSSLESS_QLOG) return;
+ if(s->qlog == LOSSLESS_QLOG){
+ for(y=0; y<h; y++)
+ for(x=0; x<w; x++)
+ dst[x + y*stride]= src[x + y*stride];
+ return;
+ }
bias= bias ? 0 : (3*qmul)>>3;
thres1= ((qmul - bias)>>QEXPSHIFT) - 1;
if(i>=0){
i<<= QEXPSHIFT;
i/= qmul; //FIXME optimize
- src[x + y*stride]= i;
+ dst[x + y*stride]= i;
}else{
i= -i;
i<<= QEXPSHIFT;
i/= qmul; //FIXME optimize
- src[x + y*stride]= -i;
+ dst[x + y*stride]= -i;
}
}else
- src[x + y*stride]= 0;
+ dst[x + y*stride]= 0;
}
}
}else{
if(i>=0){
i<<= QEXPSHIFT;
i= (i + bias) / qmul; //FIXME optimize
- src[x + y*stride]= i;
+ dst[x + y*stride]= i;
}else{
i= -i;
i<<= QEXPSHIFT;
i= (i + bias) / qmul; //FIXME optimize
- src[x + y*stride]= -i;
+ dst[x + y*stride]= -i;
}
}else
- src[x + y*stride]= 0;
+ dst[x + y*stride]= 0;
}
}
}
}
}
-static void dequantize_slice_buffered(SnowContext *s, slice_buffer * sb, SubBand *b, DWTELEM *src, int stride, int start_y, int end_y){
+static void dequantize_slice_buffered(SnowContext *s, slice_buffer * sb, SubBand *b, IDWTELEM *src, int stride, int start_y, int end_y){
const int w= b->width;
const int qlog= av_clip(s->qlog + b->qlog, 0, QROOT*16);
const int qmul= qexp[qlog&(QROOT-1)]<<(qlog>>QSHIFT);
for(y=start_y; y<end_y; y++){
// DWTELEM * line = slice_buffer_get_line_from_address(sb, src + (y * stride));
- DWTELEM * line = slice_buffer_get_line(sb, (y * b->stride_line) + b->buf_y_offset) + b->buf_x_offset;
+ IDWTELEM * line = slice_buffer_get_line(sb, (y * b->stride_line) + b->buf_y_offset) + b->buf_x_offset;
for(x=0; x<w; x++){
int i= line[x];
if(i<0){
}
}
-static void dequantize(SnowContext *s, SubBand *b, DWTELEM *src, int stride){
+static void dequantize(SnowContext *s, SubBand *b, IDWTELEM *src, int stride){
const int w= b->width;
const int h= b->height;
const int qlog= av_clip(s->qlog + b->qlog, 0, QROOT*16);
}
}
-static void decorrelate(SnowContext *s, SubBand *b, DWTELEM *src, int stride, int inverse, int use_median){
+static void decorrelate(SnowContext *s, SubBand *b, IDWTELEM *src, int stride, int inverse, int use_median){
const int w= b->width;
const int h= b->height;
int x,y;
}
}
-static void correlate_slice_buffered(SnowContext *s, slice_buffer * sb, SubBand *b, DWTELEM *src, int stride, int inverse, int use_median, int start_y, int end_y){
+static void correlate_slice_buffered(SnowContext *s, slice_buffer * sb, SubBand *b, IDWTELEM *src, int stride, int inverse, int use_median, int start_y, int end_y){
const int w= b->width;
int x,y;
// START_TIMER
- DWTELEM * line=0; // silence silly "could be used without having been initialized" warning
- DWTELEM * prev;
+ IDWTELEM * line=0; // silence silly "could be used without having been initialized" warning
+ IDWTELEM * prev;
if (start_y != 0)
line = slice_buffer_get_line(sb, ((start_y - 1) * b->stride_line) + b->buf_y_offset) + b->buf_x_offset;
// STOP_TIMER("correlate")
}
-static void correlate(SnowContext *s, SubBand *b, DWTELEM *src, int stride, int inverse, int use_median){
+static void correlate(SnowContext *s, SubBand *b, IDWTELEM *src, int stride, int inverse, int use_median){
const int w= b->width;
const int h= b->height;
int x,y;
}
}
-static void encode_header(SnowContext *s){
+static void encode_qlogs(SnowContext *s){
int plane_index, level, orientation;
+
+ for(plane_index=0; plane_index<2; plane_index++){
+ for(level=0; level<s->spatial_decomposition_count; level++){
+ for(orientation=level ? 1:0; orientation<4; orientation++){
+ if(orientation==2) continue;
+ put_symbol(&s->c, s->header_state, s->plane[plane_index].band[level][orientation].qlog, 1);
+ }
+ }
+ }
+}
+
+static void encode_header(SnowContext *s){
+ int plane_index, i;
uint8_t kstate[32];
memset(kstate, MID_STATE, sizeof(kstate));
s->last_qbias=
s->last_mv_scale=
s->last_block_max_depth= 0;
+ for(plane_index=0; plane_index<2; plane_index++){
+ Plane *p= &s->plane[plane_index];
+ p->last_htaps=0;
+ p->last_diag_mc=0;
+ memset(p->last_hcoeff, 0, sizeof(p->last_hcoeff));
+ }
}
if(s->keyframe){
put_symbol(&s->c, s->header_state, s->version, 0);
// put_rac(&s->c, s->header_state, s->rate_scalability);
put_symbol(&s->c, s->header_state, s->max_ref_frames-1, 0);
+ encode_qlogs(s);
+ }
+
+ if(!s->keyframe){
+ int update_mc=0;
for(plane_index=0; plane_index<2; plane_index++){
- for(level=0; level<s->spatial_decomposition_count; level++){
- for(orientation=level ? 1:0; orientation<4; orientation++){
- if(orientation==2) continue;
- put_symbol(&s->c, s->header_state, s->plane[plane_index].band[level][orientation].qlog, 1);
- }
+ Plane *p= &s->plane[plane_index];
+ update_mc |= p->last_htaps != p->htaps;
+ update_mc |= p->last_diag_mc != p->diag_mc;
+ update_mc |= !!memcmp(p->last_hcoeff, p->hcoeff, sizeof(p->hcoeff));
+ }
+ put_rac(&s->c, s->header_state, update_mc);
+ if(update_mc){
+ for(plane_index=0; plane_index<2; plane_index++){
+ Plane *p= &s->plane[plane_index];
+ put_rac(&s->c, s->header_state, p->diag_mc);
+ put_symbol(&s->c, s->header_state, p->htaps/2-1, 0);
+ for(i= p->htaps/2; i; i--)
+ put_symbol(&s->c, s->header_state, FFABS(p->hcoeff[i]), 0);
+
+ p->last_diag_mc= p->diag_mc;
+ p->last_htaps= p->htaps;
+ memcpy(p->last_hcoeff, p->hcoeff, sizeof(p->hcoeff));
}
}
+ if(s->last_spatial_decomposition_count != s->spatial_decomposition_count){
+ put_rac(&s->c, s->header_state, 1);
+ put_symbol(&s->c, s->header_state, s->spatial_decomposition_count, 0);
+ encode_qlogs(s);
+ }else
+ put_rac(&s->c, s->header_state, 0);
}
+
put_symbol(&s->c, s->header_state, s->spatial_decomposition_type - s->last_spatial_decomposition_type, 1);
put_symbol(&s->c, s->header_state, s->qlog - s->last_qlog , 1);
put_symbol(&s->c, s->header_state, s->mv_scale - s->last_mv_scale, 1);
s->last_qbias = s->qbias;
s->last_mv_scale = s->mv_scale;
s->last_block_max_depth = s->block_max_depth;
+ s->last_spatial_decomposition_count= s->spatial_decomposition_count;
}
-static int decode_header(SnowContext *s){
+static void decode_qlogs(SnowContext *s){
int plane_index, level, orientation;
+
+ for(plane_index=0; plane_index<3; plane_index++){
+ for(level=0; level<s->spatial_decomposition_count; level++){
+ for(orientation=level ? 1:0; orientation<4; orientation++){
+ int q;
+ if (plane_index==2) q= s->plane[1].band[level][orientation].qlog;
+ else if(orientation==2) q= s->plane[plane_index].band[level][1].qlog;
+ else q= get_symbol(&s->c, s->header_state, 1);
+ s->plane[plane_index].band[level][orientation].qlog= q;
+ }
+ }
+ }
+}
+
+static int decode_header(SnowContext *s){
+ int plane_index;
uint8_t kstate[32];
memset(kstate, MID_STATE, sizeof(kstate));
// s->rate_scalability= get_rac(&s->c, s->header_state);
s->max_ref_frames= get_symbol(&s->c, s->header_state, 0)+1;
- for(plane_index=0; plane_index<3; plane_index++){
- for(level=0; level<s->spatial_decomposition_count; level++){
- for(orientation=level ? 1:0; orientation<4; orientation++){
- int q;
- if (plane_index==2) q= s->plane[1].band[level][orientation].qlog;
- else if(orientation==2) q= s->plane[plane_index].band[level][1].qlog;
- else q= get_symbol(&s->c, s->header_state, 1);
- s->plane[plane_index].band[level][orientation].qlog= q;
+ decode_qlogs(s);
+ }
+
+ if(!s->keyframe){
+ if(get_rac(&s->c, s->header_state)){
+ for(plane_index=0; plane_index<2; plane_index++){
+ int htaps, i, sum=0, absum=0;
+ Plane *p= &s->plane[plane_index];
+ p->diag_mc= get_rac(&s->c, s->header_state);
+ htaps= get_symbol(&s->c, s->header_state, 0)*2 + 2;
+ if((unsigned)htaps > HTAPS_MAX || htaps==0)
+ return -1;
+ p->htaps= htaps;
+ for(i= htaps/2; i; i--){
+ p->hcoeff[i]= get_symbol(&s->c, s->header_state, 0) * (1-2*(i&1));
+ sum += p->hcoeff[i];
}
+ p->hcoeff[0]= 32-sum;
}
+ s->plane[2].diag_mc= s->plane[1].diag_mc;
+ s->plane[2].htaps = s->plane[1].htaps;
+ memcpy(s->plane[2].hcoeff, s->plane[1].hcoeff, sizeof(s->plane[1].hcoeff));
+ }
+ if(get_rac(&s->c, s->header_state)){
+ s->spatial_decomposition_count= get_symbol(&s->c, s->header_state, 0);
+ decode_qlogs(s);
}
}
s->spatial_decomposition_type+= get_symbol(&s->c, s->header_state, 1);
- if(s->spatial_decomposition_type > 2){
+ if(s->spatial_decomposition_type > 1){
av_log(s->avctx, AV_LOG_ERROR, "spatial_decomposition_type %d not supported", s->spatial_decomposition_type);
return -1;
}
static int common_init(AVCodecContext *avctx){
SnowContext *s = avctx->priv_data;
int width, height;
- int level, orientation, plane_index, dec;
int i, j;
s->avctx= avctx;
if(!qexp[0])
init_qexp();
- dec= s->spatial_decomposition_count= 5;
- s->spatial_decomposition_type= avctx->prediction_method; //FIXME add decorrelator type r transform_type
-
- s->chroma_h_shift= 1; //FIXME XXX
- s->chroma_v_shift= 1;
-
// dec += FFMAX(s->chroma_h_shift, s->chroma_v_shift);
width= s->avctx->width;
height= s->avctx->height;
- s->spatial_dwt_buffer= av_mallocz(width*height*sizeof(DWTELEM));
+ s->spatial_idwt_buffer= av_mallocz(width*height*sizeof(IDWTELEM));
+ s->spatial_dwt_buffer= av_mallocz(width*height*sizeof(DWTELEM)); //FIXME this doesnt belong here
+
+ for(i=0; i<MAX_REF_FRAMES; i++)
+ for(j=0; j<MAX_REF_FRAMES; j++)
+ scale_mv_ref[i][j] = 256*(i+1)/(j+1);
+
+ s->avctx->get_buffer(s->avctx, &s->mconly_picture);
+
+ return 0;
+}
- s->mv_scale= (s->avctx->flags & CODEC_FLAG_QPEL) ? 2 : 4;
- s->block_max_depth= (s->avctx->flags & CODEC_FLAG_4MV) ? 1 : 0;
+static int common_init_after_header(AVCodecContext *avctx){
+ SnowContext *s = avctx->priv_data;
+ int plane_index, level, orientation;
for(plane_index=0; plane_index<3; plane_index++){
int w= s->avctx->width;
}
s->plane[plane_index].width = w;
s->plane[plane_index].height= h;
+
//av_log(NULL, AV_LOG_DEBUG, "%d %d\n", w, h);
for(level=s->spatial_decomposition_count-1; level>=0; level--){
for(orientation=level ? 1 : 0; orientation<4; orientation++){
b->buf += b->stride>>1;
b->buf_y_offset = b->stride_line >> 1;
}
+ b->ibuf= s->spatial_idwt_buffer + (b->buf - s->spatial_dwt_buffer);
if(level)
b->parent= &s->plane[plane_index].band[level-1][orientation];
+ //FIXME avoid this realloc
+ av_freep(&b->x_coeff);
b->x_coeff=av_mallocz(((b->width+1) * b->height+1)*sizeof(x_and_coeff));
}
w= (w+1)>>1;
}
}
- for(i=0; i<MAX_REF_FRAMES; i++)
- for(j=0; j<MAX_REF_FRAMES; j++)
- scale_mv_ref[i][j] = 256*(i+1)/(j+1);
-
- reset_contexts(s);
-/*
- width= s->width= avctx->width;
- height= s->height= avctx->height;
-
- assert(width && height);
-*/
- s->avctx->get_buffer(s->avctx, &s->mconly_picture);
-
return 0;
}
for(level=0; level<s->spatial_decomposition_count; level++){
for(orientation=level ? 1 : 0; orientation<4; orientation++){
SubBand *b= &s->plane[0].band[level][orientation];
- DWTELEM *buf= b->buf;
+ IDWTELEM *buf= b->ibuf;
const int w= b->width;
const int h= b->height;
const int stride= b->stride;
const int qmul= qexp[qlog&(QROOT-1)]<<(qlog>>QSHIFT);
const int qdiv= (1<<16)/qmul;
int x, y;
+ //FIXME this is ugly
+ for(y=0; y<h; y++)
+ for(x=0; x<w; x++)
+ buf[x+y*stride]= b->buf[x+y*stride];
if(orientation==0)
decorrelate(s, b, buf, stride, 1, 0);
for(y=0; y<h; y++)
for(x=0; x<w; x++)
coef_sum+= abs(buf[x+y*stride]) * qdiv >> 16;
- if(orientation==0)
- correlate(s, b, buf, stride, 1, 0);
}
}
for(level=0; level<s->spatial_decomposition_count; level++){
for(orientation=level ? 1 : 0; orientation<4; orientation++){
SubBand *b= &p->band[level][orientation];
- DWTELEM *buf= b->buf;
+ IDWTELEM *ibuf= b->ibuf;
int64_t error=0;
- memset(s->spatial_dwt_buffer, 0, sizeof(int)*width*height);
- buf[b->width/2 + b->height/2*b->stride]= 256*256;
- ff_spatial_idwt(s->spatial_dwt_buffer, width, height, width, s->spatial_decomposition_type, s->spatial_decomposition_count);
+ memset(s->spatial_idwt_buffer, 0, sizeof(*s->spatial_idwt_buffer)*width*height);
+ ibuf[b->width/2 + b->height/2*b->stride]= 256*16;
+ ff_spatial_idwt(s->spatial_idwt_buffer, width, height, width, s->spatial_decomposition_type, s->spatial_decomposition_count);
for(y=0; y<height; y++){
for(x=0; x<width; x++){
- int64_t d= s->spatial_dwt_buffer[x + y*width];
+ int64_t d= s->spatial_idwt_buffer[x + y*width]*16;
error += d*d;
}
}
return -1;
}
+ s->spatial_decomposition_type= avctx->prediction_method; //FIXME add decorrelator type r transform_type
+
+ s->chroma_h_shift= 1; //FIXME XXX
+ s->chroma_v_shift= 1;
+
+ s->mv_scale = (avctx->flags & CODEC_FLAG_QPEL) ? 2 : 4;
+ s->block_max_depth= (avctx->flags & CODEC_FLAG_4MV ) ? 1 : 0;
+
+ for(plane_index=0; plane_index<3; plane_index++){
+ s->plane[plane_index].diag_mc= 1;
+ s->plane[plane_index].htaps= 6;
+ s->plane[plane_index].hcoeff[0]= 40;
+ s->plane[plane_index].hcoeff[1]= -10;
+ s->plane[plane_index].hcoeff[2]= 2;
+ s->plane[plane_index].fast_mc= 1;
+ }
+
common_init(avctx);
alloc_blocks(s);
}
s->pass1_rc= !(avctx->flags & (CODEC_FLAG_QSCALE|CODEC_FLAG_PASS2));
- for(plane_index=0; plane_index<3; plane_index++){
- calculate_vissual_weight(s, &s->plane[plane_index]);
- }
-
-
avctx->coded_frame= &s->current_picture;
switch(avctx->pix_fmt){
// case PIX_FMT_YUV444P:
return 0;
}
+static void halfpel_interpol(SnowContext *s, uint8_t *halfpel[4][4], AVFrame *frame){
+ int p,x,y;
+
+ assert(!(s->avctx->flags & CODEC_FLAG_EMU_EDGE));
+
+ for(p=0; p<3; p++){
+ int is_chroma= !!p;
+ int w= s->avctx->width >>is_chroma;
+ int h= s->avctx->height >>is_chroma;
+ int ls= frame->linesize[p];
+ uint8_t *src= frame->data[p];
+
+ halfpel[1][p]= (uint8_t*)av_malloc(ls * (h+2*EDGE_WIDTH)) + EDGE_WIDTH*(1+ls);
+ halfpel[2][p]= (uint8_t*)av_malloc(ls * (h+2*EDGE_WIDTH)) + EDGE_WIDTH*(1+ls);
+ halfpel[3][p]= (uint8_t*)av_malloc(ls * (h+2*EDGE_WIDTH)) + EDGE_WIDTH*(1+ls);
+
+ halfpel[0][p]= src;
+ for(y=0; y<h; y++){
+ for(x=0; x<w; x++){
+ int i= y*ls + x;
+
+ halfpel[1][p][i]= (20*(src[i] + src[i+1]) - 5*(src[i-1] + src[i+2]) + (src[i-2] + src[i+3]) + 16 )>>5;
+ }
+ }
+ for(y=0; y<h; y++){
+ for(x=0; x<w; x++){
+ int i= y*ls + x;
+
+ halfpel[2][p][i]= (20*(src[i] + src[i+ls]) - 5*(src[i-ls] + src[i+2*ls]) + (src[i-2*ls] + src[i+3*ls]) + 16 )>>5;
+ }
+ }
+ src= halfpel[1][p];
+ for(y=0; y<h; y++){
+ for(x=0; x<w; x++){
+ int i= y*ls + x;
+
+ halfpel[3][p][i]= (20*(src[i] + src[i+ls]) - 5*(src[i-ls] + src[i+2*ls]) + (src[i-2*ls] + src[i+3*ls]) + 16 )>>5;
+ }
+ }
+
+//FIXME border!
+ }
+}
+
static int frame_start(SnowContext *s){
AVFrame tmp;
int w= s->avctx->width; //FIXME round up to x16 ?
tmp= s->last_picture[s->max_ref_frames-1];
memmove(s->last_picture+1, s->last_picture, (s->max_ref_frames-1)*sizeof(AVFrame));
+ memmove(s->halfpel_plane+1, s->halfpel_plane, (s->max_ref_frames-1)*sizeof(void*)*4*4);
+#ifdef USE_HALFPEL_PLANE
+ if(s->current_picture.data[0])
+ halfpel_interpol(s, s->halfpel_plane[0], &s->current_picture);
+#endif
s->last_picture[0]= s->current_picture;
s->current_picture= tmp;
redo_frame:
+ if(pict->pict_type == I_TYPE)
+ s->spatial_decomposition_count= 5;
+ else
+ s->spatial_decomposition_count= 5;
+
s->m.pict_type = pict->pict_type;
s->qbias= pict->pict_type == P_TYPE ? 2 : 0;
+ common_init_after_header(avctx);
+
+ if(s->last_spatial_decomposition_count != s->spatial_decomposition_count){
+ for(plane_index=0; plane_index<3; plane_index++){
+ calculate_vissual_weight(s, &s->plane[plane_index]);
+ }
+ }
+
encode_header(s);
s->m.misc_bits = 8*(s->c.bytestream - s->c.bytestream_start);
encode_blocks(s, 1);
if(pict->data[plane_index]) //FIXME gray hack
for(y=0; y<h; y++){
for(x=0; x<w; x++){
- s->spatial_dwt_buffer[y*w + x]= pict->data[plane_index][y*pict->linesize[plane_index] + x]<<FRAC_BITS;
+ s->spatial_idwt_buffer[y*w + x]= pict->data[plane_index][y*pict->linesize[plane_index] + x]<<FRAC_BITS;
}
}
- predict_plane(s, s->spatial_dwt_buffer, plane_index, 0);
+ predict_plane(s, s->spatial_idwt_buffer, plane_index, 0);
if( plane_index==0
&& pict->pict_type == P_TYPE
if(s->qlog == LOSSLESS_QLOG){
for(y=0; y<h; y++){
for(x=0; x<w; x++){
- s->spatial_dwt_buffer[y*w + x]= (s->spatial_dwt_buffer[y*w + x] + (1<<(FRAC_BITS-1))-1)>>FRAC_BITS;
+ s->spatial_dwt_buffer[y*w + x]= (s->spatial_idwt_buffer[y*w + x] + (1<<(FRAC_BITS-1))-1)>>FRAC_BITS;
+ }
+ }
+ }else{
+ for(y=0; y<h; y++){
+ for(x=0; x<w; x++){
+ s->spatial_dwt_buffer[y*w + x]=s->spatial_idwt_buffer[y*w + x]<<ENCODER_EXTRA_BITS;
}
}
}
for(orientation=level ? 1 : 0; orientation<4; orientation++){
SubBand *b= &p->band[level][orientation];
- quantize(s, b, b->buf, b->stride, s->qbias);
+ quantize(s, b, b->ibuf, b->buf, b->stride, s->qbias);
if(orientation==0)
- decorrelate(s, b, b->buf, b->stride, pict->pict_type == P_TYPE, 0);
- encode_subband(s, b, b->buf, b->parent ? b->parent->buf : NULL, b->stride, orientation);
+ decorrelate(s, b, b->ibuf, b->stride, pict->pict_type == P_TYPE, 0);
+ encode_subband(s, b, b->ibuf, b->parent ? b->parent->ibuf : NULL, b->stride, orientation);
assert(b->parent==NULL || b->parent->stride == b->stride*2);
if(orientation==0)
- correlate(s, b, b->buf, b->stride, 1, 0);
+ correlate(s, b, b->ibuf, b->stride, 1, 0);
}
}
// av_log(NULL, AV_LOG_DEBUG, "plane:%d bits:%d\n", plane_index, put_bits_count(&s->c.pb) - bits);
for(orientation=level ? 1 : 0; orientation<4; orientation++){
SubBand *b= &p->band[level][orientation];
- dequantize(s, b, b->buf, b->stride);
+ dequantize(s, b, b->ibuf, b->stride);
}
}
- ff_spatial_idwt(s->spatial_dwt_buffer, w, h, w, s->spatial_decomposition_type, s->spatial_decomposition_count);
+ ff_spatial_idwt(s->spatial_idwt_buffer, w, h, w, s->spatial_decomposition_type, s->spatial_decomposition_count);
if(s->qlog == LOSSLESS_QLOG){
for(y=0; y<h; y++){
for(x=0; x<w; x++){
- s->spatial_dwt_buffer[y*w + x]<<=FRAC_BITS;
+ s->spatial_idwt_buffer[y*w + x]<<=FRAC_BITS;
}
}
}
{START_TIMER
- predict_plane(s, s->spatial_dwt_buffer, plane_index, 1);
+ predict_plane(s, s->spatial_idwt_buffer, plane_index, 1);
STOP_TIMER("pred-conv")}
}else{
//ME/MC only
}
}
}else{
- memset(s->spatial_dwt_buffer, 0, sizeof(DWTELEM)*w*h);
- predict_plane(s, s->spatial_dwt_buffer, plane_index, 1);
+ memset(s->spatial_idwt_buffer, 0, sizeof(IDWTELEM)*w*h);
+ predict_plane(s, s->spatial_idwt_buffer, plane_index, 1);
}
}
if(s->avctx->flags&CODEC_FLAG_PSNR){
}
}
- if(s->last_picture[s->max_ref_frames-1].data[0])
+ if(s->last_picture[s->max_ref_frames-1].data[0]){
avctx->release_buffer(avctx, &s->last_picture[s->max_ref_frames-1]);
+ for(i=0; i<9; i++)
+ if(s->halfpel_plane[s->max_ref_frames-1][1+i/3][i%3])
+ av_free(s->halfpel_plane[s->max_ref_frames-1][1+i/3][i%3] - EDGE_WIDTH*(1+s->current_picture.linesize[i%3]));
+ }
s->current_picture.coded_picture_number = avctx->frame_number;
s->current_picture.pict_type = pict->pict_type;
int plane_index, level, orientation, i;
av_freep(&s->spatial_dwt_buffer);
+ av_freep(&s->spatial_idwt_buffer);
av_freep(&s->m.me.scratchpad);
av_freep(&s->m.me.map);
static int decode_init(AVCodecContext *avctx)
{
SnowContext *s = avctx->priv_data;
- int block_size;
avctx->pix_fmt= PIX_FMT_YUV420P;
common_init(avctx);
- block_size = MB_SIZE >> s->block_max_depth;
- slice_buffer_init(&s->sb, s->plane[0].height, (block_size) + (s->spatial_decomposition_count * (s->spatial_decomposition_count + 3)) + 1, s->plane[0].width, s->spatial_dwt_buffer);
-
return 0;
}
RangeCoder * const c= &s->c;
int bytes_read;
AVFrame *picture = data;
- int level, orientation, plane_index;
+ int level, orientation, plane_index, i;
ff_init_range_decoder(c, buf, buf_size);
ff_build_rac_states(c, 0.05*(1LL<<32), 256-8);
s->current_picture.pict_type= FF_I_TYPE; //FIXME I vs. P
- decode_header(s);
+ if(decode_header(s)<0)
+ return -1;
+ common_init_after_header(avctx);
+
+ // realloc slice buffer for the case that spatial_decomposition_count changed
+ slice_buffer_destroy(&s->sb);
+ slice_buffer_init(&s->sb, s->plane[0].height, (MB_SIZE >> s->block_max_depth) + s->spatial_decomposition_count * 8 + 1, s->plane[0].width, s->spatial_idwt_buffer);
+
+ for(plane_index=0; plane_index<3; plane_index++){
+ Plane *p= &s->plane[plane_index];
+ p->fast_mc= p->diag_mc && p->htaps==6 && p->hcoeff[0]==40
+ && p->hcoeff[1]==-10
+ && p->hcoeff[2]==2;
+ }
+
if(!s->block) alloc_blocks(s);
frame_start(s);
if(s->avctx->debug&2048){
memset(s->spatial_dwt_buffer, 0, sizeof(DWTELEM)*w*h);
- predict_plane(s, s->spatial_dwt_buffer, plane_index, 1);
+ predict_plane(s, s->spatial_idwt_buffer, plane_index, 1);
for(y=0; y<h; y++){
for(x=0; x<w; x++){
int correlate_end_y = FFMIN(b->height, end_y + 1);
int correlate_start_y = FFMIN(b->height, (start_y ? start_y + 1 : 0));
decode_subband_slice_buffered(s, correlate_band, &s->sb, correlate_start_y, correlate_end_y, decode_state[0][0]);
- correlate_slice_buffered(s, &s->sb, correlate_band, correlate_band->buf, correlate_band->stride, 1, 0, correlate_start_y, correlate_end_y);
- dequantize_slice_buffered(s, &s->sb, correlate_band, correlate_band->buf, correlate_band->stride, start_y, end_y);
+ correlate_slice_buffered(s, &s->sb, correlate_band, correlate_band->ibuf, correlate_band->stride, 1, 0, correlate_start_y, correlate_end_y);
+ dequantize_slice_buffered(s, &s->sb, correlate_band, correlate_band->ibuf, correlate_band->stride, start_y, end_y);
}
else
decode_subband_slice_buffered(s, b, &s->sb, start_y, end_y, decode_state[level][orientation]);
if(s->qlog == LOSSLESS_QLOG){
for(; yq<slice_h && yq<h; yq++){
- DWTELEM * line = slice_buffer_get_line(&s->sb, yq);
+ IDWTELEM * line = slice_buffer_get_line(&s->sb, yq);
for(x=0; x<w; x++){
line[x] <<= FRAC_BITS;
}
}
}
- predict_slice_buffered(s, &s->sb, s->spatial_dwt_buffer, plane_index, 1, mb_y);
+ predict_slice_buffered(s, &s->sb, s->spatial_idwt_buffer, plane_index, 1, mb_y);
y = FFMIN(p->height, slice_starty);
end_y = FFMIN(p->height, slice_h);
emms_c();
- if(s->last_picture[s->max_ref_frames-1].data[0])
+ if(s->last_picture[s->max_ref_frames-1].data[0]){
avctx->release_buffer(avctx, &s->last_picture[s->max_ref_frames-1]);
+ for(i=0; i<9; i++)
+ if(s->halfpel_plane[s->max_ref_frames-1][1+i/3][i%3])
+ av_free(s->halfpel_plane[s->max_ref_frames-1][1+i/3][i%3] - EDGE_WIDTH*(1+s->current_picture.linesize[i%3]));
+ }
if(!(s->avctx->debug&2048))
*picture= s->current_picture;