43 for (i = 0; i < 8; i++) {
48 d1 = (a - d + 3 +
rnd) >> 3;
49 d2 = (a - d + b - c + 4 -
rnd) >> 3;
67 for (i = 0; i < 8; i++) {
72 d1 = (a - d + 3 +
rnd) >> 3;
73 d2 = (a - d + b - c + 4 -
rnd) >> 3;
89 int rnd1 = 4, rnd2 = 3;
90 for (i = 0; i < 8; i++) {
98 top[48] = ((a * 8) - d1 + rnd1) >> 3;
99 top[56] = ((b * 8) - d2 + rnd2) >> 3;
100 bottom[0] = ((c * 8) + d2 + rnd1) >> 3;
101 bottom[8] = ((d * 8) + d1 + rnd2) >> 3;
115 int rnd1 = flags & 2 ? 3 : 4;
117 for (i = 0; i < 8; i++) {
125 left[6] = ((a * 8) - d1 + rnd1) >> 3;
126 left[7] = ((b * 8) - d2 + rnd2) >> 3;
127 right[0] = ((c * 8) + d2 + rnd1) >> 3;
128 right[1] = ((d * 8) + d1 + rnd2) >> 3;
130 right += right_stride;
150 5 * (src[-1 * stride] - src[0 * stride]) + 4) >> 3;
151 int a0_sign = a0 >> 31;
153 a0 = (a0 ^ a0_sign) - a0_sign;
155 int a1 =
FFABS((2 * (src[-4 * stride] - src[-1 * stride]) -
156 5 * (src[-3 * stride] - src[-2 * stride]) + 4) >> 3);
157 int a2 =
FFABS((2 * (src[ 0 * stride] - src[ 3 * stride]) -
158 5 * (src[ 1 * stride] - src[ 2 * stride]) + 4) >> 3);
159 if (a1 < a0 || a2 < a0) {
161 int clip_sign = clip >> 31;
163 clip = ((clip ^ clip_sign) - clip_sign) >> 1;
166 int d = 5 * (a3 -
a0);
167 int d_sign = (d >> 31);
169 d = ((d ^ d_sign) - d_sign) >> 3;
172 if (d_sign ^ clip_sign)
176 d = (d ^ d_sign) - d_sign;
202 for (i = 0; i <
len; i += 4) {
249 dc = (3 * dc + 1) >> 1;
250 dc = (3 * dc + 16) >> 5;
252 for (i = 0; i < 8; i++) {
273 for (i = 0; i < 8; i++) {
274 t1 = 12 * (src[ 0] + src[32]) + 4;
275 t2 = 12 * (src[ 0] - src[32]) + 4;
276 t3 = 16 * src[16] + 6 * src[48];
277 t4 = 6 * src[16] - 16 * src[48];
284 t1 = 16 * src[ 8] + 15 * src[24] + 9 * src[40] + 4 * src[56];
285 t2 = 15 * src[ 8] - 4 * src[24] - 16 * src[40] - 9 * src[56];
286 t3 = 9 * src[ 8] - 16 * src[24] + 4 * src[40] + 15 * src[56];
287 t4 = 4 * src[ 8] - 9 * src[24] + 15 * src[40] - 16 * src[56];
289 dst[0] = (t5 +
t1) >> 3;
290 dst[1] = (t6 +
t2) >> 3;
291 dst[2] = (t7 +
t3) >> 3;
292 dst[3] = (t8 +
t4) >> 3;
293 dst[4] = (t8 -
t4) >> 3;
294 dst[5] = (t7 -
t3) >> 3;
295 dst[6] = (t6 -
t2) >> 3;
296 dst[7] = (t5 -
t1) >> 3;
304 for (i = 0; i < 8; i++) {
305 t1 = 12 * (src[ 0] + src[32]) + 64;
306 t2 = 12 * (src[ 0] - src[32]) + 64;
307 t3 = 16 * src[16] + 6 * src[48];
308 t4 = 6 * src[16] - 16 * src[48];
315 t1 = 16 * src[ 8] + 15 * src[24] + 9 * src[40] + 4 * src[56];
316 t2 = 15 * src[ 8] - 4 * src[24] - 16 * src[40] - 9 * src[56];
317 t3 = 9 * src[ 8] - 16 * src[24] + 4 * src[40] + 15 * src[56];
318 t4 = 4 * src[ 8] - 9 * src[24] + 15 * src[40] - 16 * src[56];
320 dst[ 0] = (t5 +
t1) >> 7;
321 dst[ 8] = (t6 +
t2) >> 7;
322 dst[16] = (t7 +
t3) >> 7;
323 dst[24] = (t8 +
t4) >> 7;
324 dst[32] = (t8 - t4 + 1) >> 7;
325 dst[40] = (t7 - t3 + 1) >> 7;
326 dst[48] = (t6 - t2 + 1) >> 7;
327 dst[56] = (t5 - t1 + 1) >> 7;
340 dc = (3 * dc + 1) >> 1;
341 dc = (17 * dc + 64) >> 7;
343 for (i = 0; i < 4; i++) {
365 for (i = 0; i < 4; i++) {
366 t1 = 12 * (src[0] + src[4]) + 4;
367 t2 = 12 * (src[0] - src[4]) + 4;
368 t3 = 16 * src[2] + 6 * src[6];
369 t4 = 6 * src[2] - 16 * src[6];
376 t1 = 16 * src[1] + 15 * src[3] + 9 * src[5] + 4 * src[7];
377 t2 = 15 * src[1] - 4 * src[3] - 16 * src[5] - 9 * src[7];
378 t3 = 9 * src[1] - 16 * src[3] + 4 * src[5] + 15 * src[7];
379 t4 = 4 * src[1] - 9 * src[3] + 15 * src[5] - 16 * src[7];
381 dst[0] = (t5 +
t1) >> 3;
382 dst[1] = (t6 +
t2) >> 3;
383 dst[2] = (t7 +
t3) >> 3;
384 dst[3] = (t8 +
t4) >> 3;
385 dst[4] = (t8 -
t4) >> 3;
386 dst[5] = (t7 -
t3) >> 3;
387 dst[6] = (t6 -
t2) >> 3;
388 dst[7] = (t5 -
t1) >> 3;
395 for (i = 0; i < 8; i++) {
396 t1 = 17 * (src[ 0] + src[16]) + 64;
397 t2 = 17 * (src[ 0] - src[16]) + 64;
398 t3 = 22 * src[ 8] + 10 * src[24];
399 t4 = 22 * src[24] - 10 * src[ 8];
417 dc = (17 * dc + 4) >> 3;
418 dc = (12 * dc + 64) >> 7;
420 for (i = 0; i < 8; i++) {
438 for (i = 0; i < 8; i++) {
439 t1 = 17 * (src[0] + src[2]) + 4;
440 t2 = 17 * (src[0] - src[2]) + 4;
441 t3 = 22 * src[1] + 10 * src[3];
442 t4 = 22 * src[3] - 10 * src[1];
444 dst[0] = (t1 +
t3) >> 3;
445 dst[1] = (t2 -
t4) >> 3;
446 dst[2] = (t2 +
t4) >> 3;
447 dst[3] = (t1 -
t3) >> 3;
454 for (i = 0; i < 4; i++) {
455 t1 = 12 * (src[ 0] + src[32]) + 64;
456 t2 = 12 * (src[ 0] - src[32]) + 64;
457 t3 = 16 * src[16] + 6 * src[48];
458 t4 = 6 * src[16] - 16 * src[48];
465 t1 = 16 * src[ 8] + 15 * src[24] + 9 * src[40] + 4 * src[56];
466 t2 = 15 * src[ 8] - 4 * src[24] - 16 * src[40] - 9 * src[56];
467 t3 = 9 * src[ 8] - 16 * src[24] + 4 * src[40] + 15 * src[56];
468 t4 = 4 * src[ 8] - 9 * src[24] + 15 * src[40] - 16 * src[56];
490 dc = (17 * dc + 4) >> 3;
491 dc = (17 * dc + 64) >> 7;
493 for (i = 0; i < 4; i++) {
510 for (i = 0; i < 4; i++) {
511 t1 = 17 * (src[0] + src[2]) + 4;
512 t2 = 17 * (src[0] - src[2]) + 4;
513 t3 = 22 * src[1] + 10 * src[3];
514 t4 = 22 * src[3] - 10 * src[1];
516 dst[0] = (t1 +
t3) >> 3;
517 dst[1] = (t2 -
t4) >> 3;
518 dst[2] = (t2 +
t4) >> 3;
519 dst[3] = (t1 -
t3) >> 3;
526 for (i = 0; i < 4; i++) {
527 t1 = 17 * (src[0] + src[16]) + 64;
528 t2 = 17 * (src[0] - src[16]) + 64;
529 t3 = 22 * src[8] + 10 * src[24];
530 t4 = 22 * src[24] - 10 * src[8];
545 #define VC1_MSPEL_FILTER_16B(DIR, TYPE) \ 546 static av_always_inline int vc1_mspel_ ## DIR ## _filter_16bits(const TYPE *src, \ 554 return -4 * src[-stride] + 53 * src[0] + \ 555 18 * src[stride] - 3 * src[stride * 2]; \ 557 return -1 * src[-stride] + 9 * src[0] + \ 558 9 * src[stride] - 1 * src[stride * 2]; \ 560 return -3 * src[-stride] + 18 * src[0] + \ 561 53 * src[stride] - 4 * src[stride * 2]; \ 577 return (-4 * src[-stride] + 53 * src[0] +
578 18 * src[stride] - 3 * src[stride * 2] + 32 - r) >> 6;
580 return (-1 * src[-stride] + 9 * src[0] +
581 9 * src[stride] - 1 * src[stride * 2] + 8 - r) >> 4;
583 return (-3 * src[-stride] + 18 * src[0] +
584 53 * src[stride] - 4 * src[stride * 2] + 32 - r) >> 6;
590 #define VC1_MSPEL_MC(OP, OP4, OPNAME) \ 591 static av_always_inline void OPNAME ## vc1_mspel_mc(uint8_t *dst, \ 592 const uint8_t *src, \ 604 static const int shift_value[] = { 0, 5, 1, 5 }; \ 605 int shift = (shift_value[hmode] + shift_value[vmode]) >> 1; \ 606 int16_t tmp[11 * 8], *tptr = tmp; \ 608 r = (1 << (shift - 1)) + rnd - 1; \ 611 for (j = 0; j < 8; j++) { \ 612 for (i = 0; i < 11; i++) \ 613 tptr[i] = (vc1_mspel_ver_filter_16bits(src + i, stride, vmode) + r) >> shift; \ 620 for (j = 0; j < 8; j++) { \ 621 for (i = 0; i < 8; i++) \ 622 OP(dst[i], (vc1_mspel_hor_filter_16bits(tptr + i, 1, hmode) + r) >> 7); \ 631 for (j = 0; j < 8; j++) { \ 632 for (i = 0; i < 8; i++) \ 633 OP(dst[i], vc1_mspel_filter(src + i, stride, vmode, r)); \ 642 for (j = 0; j < 8; j++) { \ 643 for (i = 0; i < 8; i++) \ 644 OP(dst[i], vc1_mspel_filter(src + i, 1, hmode, rnd)); \ 649 static av_always_inline void OPNAME ## vc1_mspel_mc_16(uint8_t *dst, \ 650 const uint8_t *src, \ 662 static const int shift_value[] = { 0, 5, 1, 5 }; \ 663 int shift = (shift_value[hmode] + shift_value[vmode]) >> 1; \ 664 int16_t tmp[19 * 16], *tptr = tmp; \ 666 r = (1 << (shift - 1)) + rnd - 1; \ 669 for (j = 0; j < 16; j++) { \ 670 for (i = 0; i < 19; i++) \ 671 tptr[i] = (vc1_mspel_ver_filter_16bits(src + i, stride, vmode) + r) >> shift; \ 678 for (j = 0; j < 16; j++) { \ 679 for (i = 0; i < 16; i++) \ 680 OP(dst[i], (vc1_mspel_hor_filter_16bits(tptr + i, 1, hmode) + r) >> 7); \ 689 for (j = 0; j < 16; j++) { \ 690 for (i = 0; i < 16; i++) \ 691 OP(dst[i], vc1_mspel_filter(src + i, stride, vmode, r)); \ 700 for (j = 0; j < 16; j++) { \ 701 for (i = 0; i < 16; i++) \ 702 OP(dst[i], vc1_mspel_filter(src + i, 1, hmode, rnd)); \ 707 static void OPNAME ## pixels8x8_c(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int rnd){\ 710 OP4(*(uint32_t*)(block ), AV_RN32(pixels ));\ 711 OP4(*(uint32_t*)(block+4), AV_RN32(pixels+4));\ 716 static void OPNAME ## pixels16x16_c(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int rnd){\ 718 for(i=0; i<16; i++){\ 719 OP4(*(uint32_t*)(block ), AV_RN32(pixels ));\ 720 OP4(*(uint32_t*)(block+ 4), AV_RN32(pixels+ 4));\ 721 OP4(*(uint32_t*)(block+ 8), AV_RN32(pixels+ 8));\ 722 OP4(*(uint32_t*)(block+12), AV_RN32(pixels+12));\ 728 #define op_put(a, b) (a) = av_clip_uint8(b) 729 #define op_avg(a, b) (a) = ((a) + av_clip_uint8(b) + 1) >> 1 730 #define op4_avg(a, b) (a) = rnd_avg32(a, b) 731 #define op4_put(a, b) (a) = (b) 738 #define PUT_VC1_MSPEL(a, b) \ 739 static void put_vc1_mspel_mc ## a ## b ## _c(uint8_t *dst, \ 740 const uint8_t *src, \ 741 ptrdiff_t stride, int rnd) \ 743 put_vc1_mspel_mc(dst, src, stride, a, b, rnd); \ 745 static void avg_vc1_mspel_mc ## a ## b ## _c(uint8_t *dst, \ 746 const uint8_t *src, \ 747 ptrdiff_t stride, int rnd) \ 749 avg_vc1_mspel_mc(dst, src, stride, a, b, rnd); \ 751 static void put_vc1_mspel_mc ## a ## b ## _16_c(uint8_t *dst, \ 752 const uint8_t *src, \ 753 ptrdiff_t stride, int rnd) \ 755 put_vc1_mspel_mc_16(dst, src, stride, a, b, rnd); \ 757 static void avg_vc1_mspel_mc ## a ## b ## _16_c(uint8_t *dst, \ 758 const uint8_t *src, \ 759 ptrdiff_t stride, int rnd) \ 761 avg_vc1_mspel_mc_16(dst, src, stride, a, b, rnd); \ 783 #define chroma_mc(a) \ 784 ((A * src[a] + B * src[a + 1] + \ 785 C * src[stride + a] + D * src[stride + a + 1] + 32 - 4) >> 6) 788 ptrdiff_t
stride,
int h,
int x,
int y)
790 const int A = (8 - x) * (8 - y);
791 const int B = (x) * (8 - y);
792 const int C = (8 - x) * (y);
793 const int D = (x) * (y);
796 av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
798 for (i = 0; i <
h; i++) {
813 ptrdiff_t
stride,
int h,
int x,
int y)
815 const int A = (8 - x) * (8 - y);
816 const int B = (x) * (8 - y);
817 const int C = (8 - x) * (y);
818 const int D = (x) * (y);
821 av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
823 for (i = 0; i <
h; i++) {
833 #define avg2(a, b) (((a) + (b) + 1) >> 1) 836 ptrdiff_t
stride,
int h,
int x,
int y)
838 const int A = (8 - x) * (8 - y);
839 const int B = (x) * (8 - y);
840 const int C = (8 - x) * (y);
841 const int D = (x) * (y);
844 av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
846 for (i = 0; i <
h; i++) {
862 ptrdiff_t
stride,
int h,
int x,
int y)
864 const int A = (8 - x) * (8 - y);
865 const int B = ( x) * (8 - y);
866 const int C = (8 - x) * ( y);
867 const int D = ( x) * ( y);
870 av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
872 for (i = 0; i <
h; i++) {
882 #if CONFIG_WMV3IMAGE_DECODER || CONFIG_VC1IMAGE_DECODER 885 int advance,
int count)
888 int a = src[(offset >> 16)];
889 int b = src[(offset >> 16) + 1];
890 *dst++ = a + ((b -
a) * (offset & 0xFFFF) >> 16);
903 int alpha,
int scaled,
911 a1 = a1 + ((b1 -
a1) * offset1 >> 16);
917 a2 = a2 + ((b2 -
a2) * offset2 >> 16);
919 a1 = a1 + ((a2 -
a1) * alpha >> 16);
927 int offset,
int width)
929 sprite_v_template(dst, src1a, src1b, offset, 0,
NULL,
NULL, 0, 0, 1, width);
932 static void sprite_v_double_noscale_c(
uint8_t *dst,
const uint8_t *src1a,
934 int alpha,
int width)
936 sprite_v_template(dst, src1a,
NULL, 0, 1, src2a,
NULL, 0, alpha, 0, width);
939 static void sprite_v_double_onescale_c(
uint8_t *dst,
944 int alpha,
int width)
946 sprite_v_template(dst, src1a, src1b, offset1, 1, src2a,
NULL, 0, alpha, 1,
950 static void sprite_v_double_twoscale_c(
uint8_t *dst,
960 sprite_v_template(dst, src1a, src1b, offset1, 1, src2a, src2b, offset2,
965 #define FN_ASSIGN(X, Y) \ 966 dsp->put_vc1_mspel_pixels_tab[1][X+4*Y] = put_vc1_mspel_mc##X##Y##_c; \ 967 dsp->put_vc1_mspel_pixels_tab[0][X+4*Y] = put_vc1_mspel_mc##X##Y##_16_c; \ 968 dsp->avg_vc1_mspel_pixels_tab[1][X+4*Y] = avg_vc1_mspel_mc##X##Y##_c; \ 969 dsp->avg_vc1_mspel_pixels_tab[0][X+4*Y] = avg_vc1_mspel_mc##X##Y##_16_c 1022 #if CONFIG_WMV3IMAGE_DECODER || CONFIG_VC1IMAGE_DECODER
static void vc1_inv_trans_4x4_c(uint8_t *dest, ptrdiff_t stride, int16_t *block)
void(* sprite_v_double_onescale)(uint8_t *dst, const uint8_t *src1a, const uint8_t *src1b, int offset1, const uint8_t *src2a, int alpha, int width)
void(* vc1_inv_trans_8x4)(uint8_t *dest, ptrdiff_t stride, int16_t *block)
static av_always_inline int vc1_filter_line(uint8_t *src, int stride, int pq)
VC-1 in-loop deblocking filter for one line.
void(* vc1_v_loop_filter16)(uint8_t *src, int stride, int pq)
static void vc1_v_loop_filter16_c(uint8_t *src, int stride, int pq)
static void vc1_v_loop_filter4_c(uint8_t *src, int stride, int pq)
void(* vc1_v_overlap)(uint8_t *src, int stride)
void(* sprite_v_single)(uint8_t *dst, const uint8_t *src1a, const uint8_t *src1b, int offset, int width)
void(* vc1_h_loop_filter16)(uint8_t *src, int stride, int pq)
The reader does not expect b to be semantically here and if the code is changed by maybe adding a a division or other the signedness will almost certainly be mistaken To avoid this confusion a new type was SUINT is the C unsigned type but it holds a signed int to use the same example SUINT a
static void vc1_inv_trans_8x4_c(uint8_t *dest, ptrdiff_t stride, int16_t *block)
void(* vc1_h_s_overlap)(int16_t *left, int16_t *right, int left_stride, int right_stride, int flags)
static void put_no_rnd_vc1_chroma_mc8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride, int h, int x, int y)
int(* startcode_find_candidate)(const uint8_t *buf, int size)
Search buf from the start for up to size bytes.
void(* vc1_v_loop_filter4)(uint8_t *src, int stride, int pq)
vc1op_pixels_func put_vc1_mspel_pixels_tab[2][16]
#define VC1_MSPEL_FILTER_16B(DIR, TYPE)
The exact code depends on how similar the blocks are and how related they are to the block
#define av_assert2(cond)
assert() equivalent, that does lie in speed critical code.
void(* vc1_inv_trans_4x8_dc)(uint8_t *dest, ptrdiff_t stride, int16_t *block)
static void vc1_h_loop_filter16_c(uint8_t *src, int stride, int pq)
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf offset
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
void(* vc1_inv_trans_8x8)(int16_t *b)
av_cold void ff_vc1dsp_init_mips(VC1DSPContext *dsp)
static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, ptrdiff_t stride, int16_t *block)
void(* vc1_h_loop_filter8)(uint8_t *src, int stride, int pq)
void(* vc1_inv_trans_4x4)(uint8_t *dest, ptrdiff_t stride, int16_t *block)
static void vc1_h_overlap_c(uint8_t *src, int stride)
simple assert() macros that are a bit more flexible than ISO C assert().
static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, ptrdiff_t stride, int16_t *block)
void(* vc1_v_loop_filter8)(uint8_t *src, int stride, int pq)
static void avg_no_rnd_vc1_chroma_mc4_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride, int h, int x, int y)
Accelerated start code search function for start codes common to MPEG-1/2/4 video, VC-1, H.264/5.
static void vc1_loop_filter(uint8_t *src, int step, int stride, int len, int pq)
VC-1 in-loop deblocking filter.
static void vc1_inv_trans_8x8_c(int16_t block[64])
void ff_vc1dsp_init_x86(VC1DSPContext *dsp)
void(* vc1_h_overlap)(uint8_t *src, int stride)
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
static void vc1_v_s_overlap_c(int16_t *top, int16_t *bottom)
void(* vc1_v_s_overlap)(int16_t *top, int16_t *bottom)
static void vc1_v_loop_filter8_c(uint8_t *src, int stride, int pq)
av_cold void ff_vc1dsp_init_aarch64(VC1DSPContext *dsp)
void(* vc1_inv_trans_8x8_dc)(uint8_t *dest, ptrdiff_t stride, int16_t *block)
static void vc1_inv_trans_4x8_c(uint8_t *dest, ptrdiff_t stride, int16_t *block)
s EdgeDetect Foobar g libavfilter vf_edgedetect c libavfilter vf_foobar c edit libavfilter and add an entry for foobar following the pattern of the other filters edit libavfilter allfilters and add an entry for foobar following the pattern of the other filters configure make j< whatever > ffmpeg ffmpeg i you should get a foobar png with Lena edge detected That s your new playground is ready Some little details about what s going which in turn will define variables for the build system and the C
h264_chroma_mc_func avg_no_rnd_vc1_chroma_pixels_tab[3]
static double b1(void *priv, double x, double y)
int ff_startcode_find_candidate_c(const uint8_t *buf, int size)
#define PUT_VC1_MSPEL(a, b)
void(* sprite_v_double_noscale)(uint8_t *dst, const uint8_t *src1a, const uint8_t *src2a, int alpha, int width)
static const int16_t alpha[]
Tag MUST be and< 10hcoeff half pel interpolation filter coefficients, hcoeff[0] are the 2 middle coefficients[1] are the next outer ones and so on, resulting in a filter like:...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2]...the sign of the coefficients is not explicitly stored but alternates after each coeff and coeff[0] is positive, so...,+,-,+,-,+,+,-,+,-,+,...hcoeff[0] is not explicitly stored but found by subtracting the sum of all stored coefficients with signs from 32 hcoeff[0]=32-hcoeff[1]-hcoeff[2]-...a good choice for hcoeff and htaps is htaps=6 hcoeff={40,-10, 2}an alternative which requires more computations at both encoder and decoder side and may or may not be better is htaps=8 hcoeff={42,-14, 6,-2}ref_frames minimum of the number of available reference frames and max_ref_frames for example the first frame after a key frame always has ref_frames=1spatial_decomposition_type wavelet type 0 is a 9/7 symmetric compact integer wavelet 1 is a 5/3 symmetric compact integer wavelet others are reserved stored as delta from last, last is reset to 0 if always_reset||keyframeqlog quality(logarithmic quantizer scale) stored as delta from last, last is reset to 0 if always_reset||keyframemv_scale stored as delta from last, last is reset to 0 if always_reset||keyframe FIXME check that everything works fine if this changes between framesqbias dequantization bias stored as delta from last, last is reset to 0 if always_reset||keyframeblock_max_depth maximum depth of the block tree stored as delta from last, last is reset to 0 if always_reset||keyframequant_table quantization tableHighlevel bitstream structure:==============================--------------------------------------------|Header|--------------------------------------------|------------------------------------|||Block0||||split?||||yes no||||.........intra?||||:Block01:yes no||||:Block02:.................||||:Block03::y DC::ref index:||||:Block04::cb DC::motion x:||||.........:cr DC::motion y:||||.................|||------------------------------------||------------------------------------|||Block1|||...|--------------------------------------------|------------------------------------|||Y subbands||Cb subbands||Cr subbands||||------||------||------|||||LL0||HL0||||LL0||HL0||||LL0||HL0|||||------||------||------||||------||------||------|||||LH0||HH0||||LH0||HH0||||LH0||HH0|||||------||------||------||||------||------||------|||||HL1||LH1||||HL1||LH1||||HL1||LH1|||||------||------||------||||------||------||------|||||HH1||HL2||||HH1||HL2||||HH1||HL2|||||...||...||...|||------------------------------------|--------------------------------------------Decoding process:=================------------|||Subbands|------------||||------------|Intra DC||||LL0 subband prediction------------|\Dequantization-------------------\||Reference frames|\IDWT|--------------|Motion\|||Frame 0||Frame 1||Compensation.OBMC v-------|--------------|--------------.\------> Frame n output Frame Frame<----------------------------------/|...|-------------------Range Coder:============Binary Range Coder:-------------------The implemented range coder is an adapted version based upon"Range encoding: an algorithm for removing redundancy from a digitised message."by G.N.N.Martin.The symbols encoded by the Snow range coder are bits(0|1).The associated probabilities are not fix but change depending on the symbol mix seen so far.bit seen|new state---------+-----------------------------------------------0|256-state_transition_table[256-old_state];1|state_transition_table[old_state];state_transition_table={0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194, 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209, 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225, 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};FIXME Range Coding of integers:-------------------------FIXME Neighboring Blocks:===================left and top are set to the respective blocks unless they are outside of the image in which case they are set to the Null block top-left is set to the top left block unless it is outside of the image in which case it is set to the left block if this block has no larger parent block or it is at the left side of its parent block and the top right block is not outside of the image then the top right block is used for top-right else the top-left block is used Null block y, cb, cr are 128 level, ref, mx and my are 0 Motion Vector Prediction:=========================1.the motion vectors of all the neighboring blocks are scaled to compensate for the difference of reference frames scaled_mv=(mv *(256 *(current_reference+1)/(mv.reference+1))+128)> the median of the scaled top and top right vectors is used as motion vector prediction the used motion vector is the sum of the predictor and(mvx_diff, mvy_diff)*mv_scale Intra DC Prediction block[y][x] dc[1]
Tag MUST be and< 10hcoeff half pel interpolation filter coefficients, hcoeff[0] are the 2 middle coefficients[1] are the next outer ones and so on, resulting in a filter like:...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2]...the sign of the coefficients is not explicitly stored but alternates after each coeff and coeff[0] is positive, so...,+,-,+,-,+,+,-,+,-,+,...hcoeff[0] is not explicitly stored but found by subtracting the sum of all stored coefficients with signs from 32 hcoeff[0]=32-hcoeff[1]-hcoeff[2]-...a good choice for hcoeff and htaps is htaps=6 hcoeff={40,-10, 2}an alternative which requires more computations at both encoder and decoder side and may or may not be better is htaps=8 hcoeff={42,-14, 6,-2}ref_frames minimum of the number of available reference frames and max_ref_frames for example the first frame after a key frame always has ref_frames=1spatial_decomposition_type wavelet type 0 is a 9/7 symmetric compact integer wavelet 1 is a 5/3 symmetric compact integer wavelet others are reserved stored as delta from last, last is reset to 0 if always_reset||keyframeqlog quality(logarithmic quantizer scale) stored as delta from last, last is reset to 0 if always_reset||keyframemv_scale stored as delta from last, last is reset to 0 if always_reset||keyframe FIXME check that everything works fine if this changes between framesqbias dequantization bias stored as delta from last, last is reset to 0 if always_reset||keyframeblock_max_depth maximum depth of the block tree stored as delta from last, last is reset to 0 if always_reset||keyframequant_table quantization tableHighlevel bitstream structure:==============================--------------------------------------------|Header|--------------------------------------------|------------------------------------|||Block0||||split?||||yes no||||.........intra?||||:Block01:yes no||||:Block02:.................||||:Block03::y DC::ref index:||||:Block04::cb DC::motion x:||||.........:cr DC::motion y:||||.................|||------------------------------------||------------------------------------|||Block1|||...|--------------------------------------------|------------------------------------|||Y subbands||Cb subbands||Cr subbands||||------||------||------|||||LL0||HL0||||LL0||HL0||||LL0||HL0|||||------||------||------||||------||------||------|||||LH0||HH0||||LH0||HH0||||LH0||HH0|||||------||------||------||||------||------||------|||||HL1||LH1||||HL1||LH1||||HL1||LH1|||||------||------||------||||------||------||------|||||HH1||HL2||||HH1||HL2||||HH1||HL2|||||...||...||...|||------------------------------------|--------------------------------------------Decoding process:=================------------|||Subbands|------------||||------------|Intra DC||||LL0 subband prediction------------|\Dequantization-------------------\||Reference frames|\IDWT|--------------|Motion\|||Frame 0||Frame 1||Compensation.OBMC v-------|--------------|--------------.\------> Frame n output Frame Frame<----------------------------------/|...|-------------------Range Coder:============Binary Range Coder:-------------------The implemented range coder is an adapted version based upon"Range encoding: an algorithm for removing redundancy from a digitised message."by G.N.N.Martin.The symbols encoded by the Snow range coder are bits(0|1).The associated probabilities are not fix but change depending on the symbol mix seen so far.bit seen|new state---------+-----------------------------------------------0|256-state_transition_table[256-old_state];1|state_transition_table[old_state];state_transition_table={0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194, 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209, 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225, 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};FIXME Range Coding of integers:-------------------------FIXME Neighboring Blocks:===================left and top are set to the respective blocks unless they are outside of the image in which case they are set to the Null block top-left is set to the top left block unless it is outside of the image in which case it is set to the left block if this block has no larger parent block or it is at the left side of its parent block and the top right block is not outside of the image then the top right block is used for top-right else the top-left block is used Null block y, cb, cr are 128 level, ref, mx and my are 0 Motion Vector Prediction:=========================1.the motion vectors of all the neighboring blocks are scaled to compensate for the difference of reference frames scaled_mv=(mv *(256 *(current_reference+1)/(mv.reference+1))+128)> the median of the scaled left
void(* vc1_inv_trans_4x8)(uint8_t *dest, ptrdiff_t stride, int16_t *block)
vc1op_pixels_func avg_vc1_mspel_pixels_tab[2][16]
void(* vc1_inv_trans_8x4_dc)(uint8_t *dest, ptrdiff_t stride, int16_t *block)
void(* sprite_v_double_twoscale)(uint8_t *dst, const uint8_t *src1a, const uint8_t *src1b, int offset1, const uint8_t *src2a, const uint8_t *src2b, int offset2, int alpha, int width)
static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, ptrdiff_t stride, int16_t *block)
static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, ptrdiff_t stride, int16_t *block)
static void put_no_rnd_vc1_chroma_mc4_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride, int h, int x, int y)
static void vc1_v_overlap_c(uint8_t *src, int stride)
h264_chroma_mc_func put_no_rnd_vc1_chroma_pixels_tab[3]
void(* vc1_inv_trans_4x4_dc)(uint8_t *dest, ptrdiff_t stride, int16_t *block)
#define flags(name, subs,...)
GLint GLenum GLboolean GLsizei stride
static void vc1_h_loop_filter8_c(uint8_t *src, int stride, int pq)
common internal and external API header
static double clip(void *opaque, double val)
Clip value val in the minval - maxval range.
static void vc1_h_loop_filter4_c(uint8_t *src, int stride, int pq)
#define VC1_MSPEL_MC(OP, OP4, OPNAME)
static av_always_inline int vc1_mspel_filter(const uint8_t *src, int stride, int mode, int r)
av_cold void ff_vc1dsp_init_arm(VC1DSPContext *dsp)
static void vc1_h_s_overlap_c(int16_t *left, int16_t *right, int left_stride, int right_stride, int flags)
void(* vc1_h_loop_filter4)(uint8_t *src, int stride, int pq)
av_cold void ff_vc1dsp_init_ppc(VC1DSPContext *dsp)
void(* sprite_h)(uint8_t *dst, const uint8_t *src, int offset, int advance, int count)
av_cold void ff_vc1dsp_init(VC1DSPContext *dsp)
mode
Use these values in ebur128_init (or'ed).
static void avg_no_rnd_vc1_chroma_mc8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride, int h, int x, int y)
trying all byte sequences megabyte in length and selecting the best looking sequence will yield cases to try But a word about which is also called distortion Distortion can be quantified by almost any quality measurement one chooses the sum of squared differences is used but more complex methods that consider psychovisual effects can be used as well It makes no difference in this discussion First step
static double b2(void *priv, double x, double y)