summaryrefslogtreecommitdiff
path: root/src/libffmpeg/libavcodec/dsputil.c
diff options
context:
space:
mode:
Diffstat (limited to 'src/libffmpeg/libavcodec/dsputil.c')
-rw-r--r--src/libffmpeg/libavcodec/dsputil.c364
1 files changed, 286 insertions, 78 deletions
diff --git a/src/libffmpeg/libavcodec/dsputil.c b/src/libffmpeg/libavcodec/dsputil.c
index 270424706..114d67b50 100644
--- a/src/libffmpeg/libavcodec/dsputil.c
+++ b/src/libffmpeg/libavcodec/dsputil.c
@@ -1,6 +1,7 @@
/*
* DSP utils
* Copyright (c) 2000, 2001 Fabrice Bellard.
+ * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
@@ -28,7 +29,7 @@
#include "dsputil.h"
#include "mpegvideo.h"
#include "simple_idct.h"
-
+#include "faandct.h"
uint8_t cropTbl[256 + 2 * MAX_NEG_CROP];
uint32_t squareTbl[512];
@@ -44,6 +45,19 @@ const uint8_t ff_zigzag_direct[64] = {
53, 60, 61, 54, 47, 55, 62, 63
};
+/* Specific zigzag scan for 248 idct. NOTE that unlike the
+ specification, we interleave the fields */
+const uint8_t ff_zigzag248_direct[64] = {
+ 0, 8, 1, 9, 16, 24, 2, 10,
+ 17, 25, 32, 40, 48, 56, 33, 41,
+ 18, 26, 3, 11, 4, 12, 19, 27,
+ 34, 42, 49, 57, 50, 58, 35, 43,
+ 20, 28, 5, 13, 6, 14, 21, 29,
+ 36, 44, 51, 59, 52, 60, 37, 45,
+ 22, 30, 7, 15, 23, 31, 38, 46,
+ 53, 61, 54, 62, 39, 47, 55, 63,
+};
+
/* not permutated inverse zigzag_direct + 1 for MMX quantizer */
uint16_t __align8 inv_zigzag_direct16[64];
@@ -205,13 +219,13 @@ static void bswap_buf(uint32_t *dst, uint32_t *src, int w){
}
}
-static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size)
+static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h)
{
int s, i;
uint32_t *sq = squareTbl + 256;
s = 0;
- for (i = 0; i < 8; i++) {
+ for (i = 0; i < h; i++) {
s += sq[pix1[0] - pix2[0]];
s += sq[pix1[1] - pix2[1]];
s += sq[pix1[2] - pix2[2]];
@@ -226,13 +240,13 @@ static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size)
return s;
}
-static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size)
+static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
uint32_t *sq = squareTbl + 256;
s = 0;
- for (i = 0; i < 16; i++) {
+ for (i = 0; i < h; i++) {
s += sq[pix1[ 0] - pix2[ 0]];
s += sq[pix1[ 1] - pix2[ 1]];
s += sq[pix1[ 2] - pix2[ 2]];
@@ -2248,13 +2262,82 @@ static void put_mspel8_mc22_c(uint8_t *dst, uint8_t *src, int stride){
wmv2_mspel8_v_lowpass(dst, halfH+8, stride, 8, 8);
}
+static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
+ int x;
+ const int strength= ff_h263_loop_filter_strength[qscale];
+
+ for(x=0; x<8; x++){
+ int d1, d2, ad1;
+ int p0= src[x-2*stride];
+ int p1= src[x-1*stride];
+ int p2= src[x+0*stride];
+ int p3= src[x+1*stride];
+ int d = (p0 - p3 + 4*(p2 - p1)) / 8;
+
+ if (d<-2*strength) d1= 0;
+ else if(d<- strength) d1=-2*strength - d;
+ else if(d< strength) d1= d;
+ else if(d< 2*strength) d1= 2*strength - d;
+ else d1= 0;
+
+ p1 += d1;
+ p2 -= d1;
+ if(p1&256) p1= ~(p1>>31);
+ if(p2&256) p2= ~(p2>>31);
+
+ src[x-1*stride] = p1;
+ src[x+0*stride] = p2;
+
+ ad1= ABS(d1)>>1;
+
+ d2= clip((p0-p3)/4, -ad1, ad1);
+
+ src[x-2*stride] = p0 - d2;
+ src[x+ stride] = p3 + d2;
+ }
+}
+
+static void h263_h_loop_filter_c(uint8_t *src, int stride, int qscale){
+ int y;
+ const int strength= ff_h263_loop_filter_strength[qscale];
+
+ for(y=0; y<8; y++){
+ int d1, d2, ad1;
+ int p0= src[y*stride-2];
+ int p1= src[y*stride-1];
+ int p2= src[y*stride+0];
+ int p3= src[y*stride+1];
+ int d = (p0 - p3 + 4*(p2 - p1)) / 8;
+
+ if (d<-2*strength) d1= 0;
+ else if(d<- strength) d1=-2*strength - d;
+ else if(d< strength) d1= d;
+ else if(d< 2*strength) d1= 2*strength - d;
+ else d1= 0;
+
+ p1 += d1;
+ p2 -= d1;
+ if(p1&256) p1= ~(p1>>31);
+ if(p2&256) p2= ~(p2>>31);
+
+ src[y*stride-1] = p1;
+ src[y*stride+0] = p2;
-static inline int pix_abs16x16_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+ ad1= ABS(d1)>>1;
+
+ d2= clip((p0-p3)/4, -ad1, ad1);
+
+ src[y*stride-2] = p0 - d2;
+ src[y*stride+1] = p3 + d2;
+ }
+}
+
+static inline int pix_abs16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - pix2[0]);
s += abs(pix1[1] - pix2[1]);
s += abs(pix1[2] - pix2[2]);
@@ -2277,12 +2360,12 @@ static inline int pix_abs16x16_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static int pix_abs16x16_x2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+static int pix_abs16_x2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
@@ -2305,13 +2388,13 @@ static int pix_abs16x16_x2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static int pix_abs16x16_y2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+static int pix_abs16_y2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
@@ -2335,13 +2418,13 @@ static int pix_abs16x16_y2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static int pix_abs16x16_xy2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+static int pix_abs16_xy2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<16;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
@@ -2365,12 +2448,12 @@ static int pix_abs16x16_xy2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static inline int pix_abs8x8_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+static inline int pix_abs8_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - pix2[0]);
s += abs(pix1[1] - pix2[1]);
s += abs(pix1[2] - pix2[2]);
@@ -2385,12 +2468,12 @@ static inline int pix_abs8x8_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static int pix_abs8x8_x2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+static int pix_abs8_x2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix2[1]));
s += abs(pix1[1] - avg2(pix2[1], pix2[2]));
s += abs(pix1[2] - avg2(pix2[2], pix2[3]));
@@ -2405,13 +2488,13 @@ static int pix_abs8x8_x2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static int pix_abs8x8_y2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+static int pix_abs8_y2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg2(pix2[0], pix3[0]));
s += abs(pix1[1] - avg2(pix2[1], pix3[1]));
s += abs(pix1[2] - avg2(pix2[2], pix3[2]));
@@ -2427,13 +2510,13 @@ static int pix_abs8x8_y2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static int pix_abs8x8_xy2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
+static int pix_abs8_xy2_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
{
int s, i;
uint8_t *pix3 = pix2 + line_size;
s = 0;
- for(i=0;i<8;i++) {
+ for(i=0;i<h;i++) {
s += abs(pix1[0] - avg4(pix2[0], pix2[1], pix3[0], pix3[1]));
s += abs(pix1[1] - avg4(pix2[1], pix2[2], pix3[1], pix3[2]));
s += abs(pix1[2] - avg4(pix2[2], pix2[3], pix3[2], pix3[3]));
@@ -2449,14 +2532,6 @@ static int pix_abs8x8_xy2_c(uint8_t *pix1, uint8_t *pix2, int line_size)
return s;
}
-static int sad16x16_c(void *s, uint8_t *a, uint8_t *b, int stride){
- return pix_abs16x16_c(a,b,stride);
-}
-
-static int sad8x8_c(void *s, uint8_t *a, uint8_t *b, int stride){
- return pix_abs8x8_c(a,b,stride);
-}
-
/**
* permutes an 8x8 block.
* @param block the block which will be permuted according to the given permutation vector
@@ -2486,6 +2561,53 @@ void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scant
}
}
+static int zero_cmp(void *s, uint8_t *a, uint8_t *b, int stride, int h){
+ return 0;
+}
+
+void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){
+ int i;
+
+ memset(cmp, 0, sizeof(void*)*5);
+
+ for(i=0; i<5; i++){
+ switch(type&0xFF){
+ case FF_CMP_SAD:
+ cmp[i]= c->sad[i];
+ break;
+ case FF_CMP_SATD:
+ cmp[i]= c->hadamard8_diff[i];
+ break;
+ case FF_CMP_SSE:
+ cmp[i]= c->sse[i];
+ break;
+ case FF_CMP_DCT:
+ cmp[i]= c->dct_sad[i];
+ break;
+ case FF_CMP_PSNR:
+ cmp[i]= c->quant_psnr[i];
+ break;
+ case FF_CMP_BIT:
+ cmp[i]= c->bit[i];
+ break;
+ case FF_CMP_RD:
+ cmp[i]= c->rd[i];
+ break;
+ case FF_CMP_VSAD:
+ cmp[i]= c->vsad[i];
+ break;
+ case FF_CMP_VSSE:
+ cmp[i]= c->vsse[i];
+ break;
+ case FF_CMP_ZERO:
+ cmp[i]= zero_cmp;
+ break;
+ default:
+ av_log(NULL, AV_LOG_ERROR,"internal error in cmp function selection\n");
+ }
+ }
+}
+
/**
* memset(blocks, 0, sizeof(DCTELEM)*6*64)
*/
@@ -2559,10 +2681,12 @@ o2= (i1)-(i2);
#define BUTTERFLYA(x,y) (ABS((x)+(y)) + ABS((x)-(y)))
-static int hadamard8_diff_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride){
+static int hadamard8_diff8x8_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
int i;
int temp[64];
int sum=0;
+
+ assert(h==8);
for(i=0; i<8; i++){
//FIXME try pointer walks
@@ -2609,17 +2733,19 @@ if(sum>maxi){
return sum;
}
-static int hadamard8_abs_c(uint8_t *src, int stride, int mean){
+static int hadamard8_intra8x8_c(/*MpegEncContext*/ void *s, uint8_t *src, uint8_t *dummy, int stride, int h){
int i;
int temp[64];
int sum=0;
-//FIXME OOOPS ignore 0 term instead of mean mess
+
+ assert(h==8);
+
for(i=0; i<8; i++){
//FIXME try pointer walks
- BUTTERFLY2(temp[8*i+0], temp[8*i+1], src[stride*i+0]-mean,src[stride*i+1]-mean);
- BUTTERFLY2(temp[8*i+2], temp[8*i+3], src[stride*i+2]-mean,src[stride*i+3]-mean);
- BUTTERFLY2(temp[8*i+4], temp[8*i+5], src[stride*i+4]-mean,src[stride*i+5]-mean);
- BUTTERFLY2(temp[8*i+6], temp[8*i+7], src[stride*i+6]-mean,src[stride*i+7]-mean);
+ BUTTERFLY2(temp[8*i+0], temp[8*i+1], src[stride*i+0],src[stride*i+1]);
+ BUTTERFLY2(temp[8*i+2], temp[8*i+3], src[stride*i+2],src[stride*i+3]);
+ BUTTERFLY2(temp[8*i+4], temp[8*i+5], src[stride*i+4],src[stride*i+5]);
+ BUTTERFLY2(temp[8*i+6], temp[8*i+7], src[stride*i+6],src[stride*i+7]);
BUTTERFLY1(temp[8*i+0], temp[8*i+2]);
BUTTERFLY1(temp[8*i+1], temp[8*i+3]);
@@ -2650,14 +2776,18 @@ static int hadamard8_abs_c(uint8_t *src, int stride, int mean){
+BUTTERFLYA(temp[8*3+i], temp[8*7+i]);
}
+ sum -= ABS(temp[8*0] + temp[8*4]); // -mean
+
return sum;
}
-static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
+static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
MpegEncContext * const s= (MpegEncContext *)c;
uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
DCTELEM * const temp= (DCTELEM*)aligned_temp;
int sum=0, i;
+
+ assert(h==8);
s->dsp.diff_pixels(temp, src1, src2, stride);
s->dsp.fdct(temp);
@@ -2670,13 +2800,14 @@ static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2
void simple_idct(DCTELEM *block); //FIXME
-static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
+static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
MpegEncContext * const s= (MpegEncContext *)c;
uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64*2/8];
DCTELEM * const temp= (DCTELEM*)aligned_temp;
DCTELEM * const bak = ((DCTELEM*)aligned_temp)+64;
int sum=0, i;
+ assert(h==8);
s->mb_intra=0;
s->dsp.diff_pixels(temp, src1, src2, stride);
@@ -2684,7 +2815,7 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
memcpy(bak, temp, 64*sizeof(DCTELEM));
s->block_last_index[0/*FIXME*/]= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
- s->dct_unquantize(s, temp, 0, s->qscale);
+ s->dct_unquantize_inter(s, temp, 0, s->qscale);
simple_idct(temp); //FIXME
for(i=0; i<64; i++)
@@ -2693,7 +2824,7 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
return sum;
}
-static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
+static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
MpegEncContext * const s= (MpegEncContext *)c;
const uint8_t *scantable= s->intra_scantable.permutated;
uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
@@ -2705,6 +2836,8 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
uint8_t * length;
uint8_t * last_length;
+ assert(h==8);
+
for(i=0; i<8; i++){
((uint32_t*)(bak + i*stride))[0]= ((uint32_t*)(src2 + i*stride))[0];
((uint32_t*)(bak + i*stride))[1]= ((uint32_t*)(src2 + i*stride))[1];
@@ -2757,17 +2890,20 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
}
if(last>=0){
- s->dct_unquantize(s, temp, 0, s->qscale);
+ if(s->mb_intra)
+ s->dct_unquantize_intra(s, temp, 0, s->qscale);
+ else
+ s->dct_unquantize_inter(s, temp, 0, s->qscale);
}
s->dsp.idct_add(bak, stride, temp);
- distoration= s->dsp.sse[1](NULL, bak, src1, stride);
+ distoration= s->dsp.sse[1](NULL, bak, src1, stride, 8);
return distoration + ((bits*s->qscale*s->qscale*109 + 64)>>7);
}
-static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride){
+static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
MpegEncContext * const s= (MpegEncContext *)c;
const uint8_t *scantable= s->intra_scantable.permutated;
uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
@@ -2776,6 +2912,8 @@ static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, in
const int esc_length= s->ac_esc_length;
uint8_t * length;
uint8_t * last_length;
+
+ assert(h==8);
s->dsp.diff_pixels(temp, src1, src2, stride);
@@ -2825,12 +2963,73 @@ static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, in
return bits;
}
+static int vsad_intra16_c(/*MpegEncContext*/ void *c, uint8_t *s, uint8_t *dummy, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x+=4){
+ score+= ABS(s[x ] - s[x +stride]) + ABS(s[x+1] - s[x+1+stride])
+ +ABS(s[x+2] - s[x+2+stride]) + ABS(s[x+3] - s[x+3+stride]);
+ }
+ s+= stride;
+ }
+
+ return score;
+}
-WARPER88_1616(hadamard8_diff_c, hadamard8_diff16_c)
-WARPER88_1616(dct_sad8x8_c, dct_sad16x16_c)
-WARPER88_1616(quant_psnr8x8_c, quant_psnr16x16_c)
-WARPER88_1616(rd8x8_c, rd16x16_c)
-WARPER88_1616(bit8x8_c, bit16x16_c)
+static int vsad16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x++){
+ score+= ABS(s1[x ] - s2[x ] - s1[x +stride] + s2[x +stride]);
+ }
+ s1+= stride;
+ s2+= stride;
+ }
+
+ return score;
+}
+
+#define SQ(a) ((a)*(a))
+static int vsse_intra16_c(/*MpegEncContext*/ void *c, uint8_t *s, uint8_t *dummy, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x+=4){
+ score+= SQ(s[x ] - s[x +stride]) + SQ(s[x+1] - s[x+1+stride])
+ +SQ(s[x+2] - s[x+2+stride]) + SQ(s[x+3] - s[x+3+stride]);
+ }
+ s+= stride;
+ }
+
+ return score;
+}
+
+static int vsse16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int stride, int h){
+ int score=0;
+ int x,y;
+
+ for(y=1; y<h; y++){
+ for(x=0; x<16; x++){
+ score+= SQ(s1[x ] - s2[x ] - s1[x +stride] + s2[x +stride]);
+ }
+ s1+= stride;
+ s2+= stride;
+ }
+
+ return score;
+}
+
+WARPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
+WARPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
+WARPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
+WARPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
+WARPER8_16_SQ(rd8x8_c, rd16_c)
+WARPER8_16_SQ(bit8x8_c, bit16_c)
/* XXX: those functions should be suppressed ASAP when all IDCTs are
converted */
@@ -2869,10 +3068,18 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
int i;
#ifdef CONFIG_ENCODERS
- if(avctx->dct_algo==FF_DCT_FASTINT)
+ if(avctx->dct_algo==FF_DCT_FASTINT) {
c->fdct = fdct_ifast;
- else
+ c->fdct248 = fdct_ifast248;
+ }
+ else if(avctx->dct_algo==FF_DCT_FAAN) {
+ c->fdct = ff_faandct;
+ c->fdct248 = ff_faandct248;
+ }
+ else {
c->fdct = ff_jpeg_fdct_islow; //slow/accurate/default
+ c->fdct248 = ff_fdct248_islow;
+ }
#endif //CONFIG_ENCODERS
if(avctx->idct_algo==FF_IDCT_INT){
@@ -2896,18 +3103,16 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
c->clear_blocks = clear_blocks_c;
c->pix_sum = pix_sum_c;
c->pix_norm1 = pix_norm1_c;
- c->sse[0]= sse16_c;
- c->sse[1]= sse8_c;
/* TODO [0] 16 [1] 8 */
- c->pix_abs16x16 = pix_abs16x16_c;
- c->pix_abs16x16_x2 = pix_abs16x16_x2_c;
- c->pix_abs16x16_y2 = pix_abs16x16_y2_c;
- c->pix_abs16x16_xy2 = pix_abs16x16_xy2_c;
- c->pix_abs8x8 = pix_abs8x8_c;
- c->pix_abs8x8_x2 = pix_abs8x8_x2_c;
- c->pix_abs8x8_y2 = pix_abs8x8_y2_c;
- c->pix_abs8x8_xy2 = pix_abs8x8_xy2_c;
+ c->pix_abs[0][0] = pix_abs16_c;
+ c->pix_abs[0][1] = pix_abs16_x2_c;
+ c->pix_abs[0][2] = pix_abs16_y2_c;
+ c->pix_abs[0][3] = pix_abs16_xy2_c;
+ c->pix_abs[1][0] = pix_abs8_c;
+ c->pix_abs[1][1] = pix_abs8_x2_c;
+ c->pix_abs[1][2] = pix_abs8_y2_c;
+ c->pix_abs[1][3] = pix_abs8_xy2_c;
#define dspfunc(PFX, IDX, NUM) \
c->PFX ## _pixels_tab[IDX][0] = PFX ## _pixels ## NUM ## _c; \
@@ -3004,29 +3209,32 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
c->put_mspel_pixels_tab[6]= put_mspel8_mc22_c;
c->put_mspel_pixels_tab[7]= put_mspel8_mc32_c;
- c->hadamard8_diff[0]= hadamard8_diff16_c;
- c->hadamard8_diff[1]= hadamard8_diff_c;
- c->hadamard8_abs = hadamard8_abs_c;
-
- c->dct_sad[0]= dct_sad16x16_c;
- c->dct_sad[1]= dct_sad8x8_c;
+#define SET_CMP_FUNC(name) \
+ c->name[0]= name ## 16_c;\
+ c->name[1]= name ## 8x8_c;
- c->sad[0]= sad16x16_c;
- c->sad[1]= sad8x8_c;
-
- c->quant_psnr[0]= quant_psnr16x16_c;
- c->quant_psnr[1]= quant_psnr8x8_c;
-
- c->rd[0]= rd16x16_c;
- c->rd[1]= rd8x8_c;
-
- c->bit[0]= bit16x16_c;
- c->bit[1]= bit8x8_c;
+ SET_CMP_FUNC(hadamard8_diff)
+ c->hadamard8_diff[4]= hadamard8_intra16_c;
+ SET_CMP_FUNC(dct_sad)
+ c->sad[0]= pix_abs16_c;
+ c->sad[1]= pix_abs8_c;
+ c->sse[0]= sse16_c;
+ c->sse[1]= sse8_c;
+ SET_CMP_FUNC(quant_psnr)
+ SET_CMP_FUNC(rd)
+ SET_CMP_FUNC(bit)
+ c->vsad[0]= vsad16_c;
+ c->vsad[4]= vsad_intra16_c;
+ c->vsse[0]= vsse16_c;
+ c->vsse[4]= vsse_intra16_c;
c->add_bytes= add_bytes_c;
c->diff_bytes= diff_bytes_c;
c->sub_hfyu_median_prediction= sub_hfyu_median_prediction_c;
c->bswap_buf= bswap_buf;
+
+ c->h263_h_loop_filter= h263_h_loop_filter_c;
+ c->h263_v_loop_filter= h263_v_loop_filter_c;
#ifdef HAVE_MMX
dsputil_init_mmx(c, avctx);
@@ -3068,7 +3276,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
c->idct_permutation[i]= ((i&7)<<3) | (i>>3);
break;
default:
- fprintf(stderr, "Internal error, IDCT permutation not set\n");
+ av_log(avctx, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n");
}
}
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-27 20:12:22 +0000