comparison dsputil.c @ 4001:34fdffe98bd0 libavcodec

Rename ABS macro to FFABS.

4000:eeab02251056

            int sy= (ori&2) ? stride>>1 : 0;

            for(i=0; i<size; i++){
                for(j=0; j<size; j++){
                    int v= tmp[sx + sy + i*stride + j] * scale[type][dec_count-3][level][ori];
                    s += ABS(v);
                }
            }
        }
    }
    assert(s>=0);

        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;

        if(p2&256) p2= ~(p2>>31);

        src[y*stride-1] = p1;
        src[y*stride+0] = p2;

        ad1= ABS(d1)>>1;

        d2= clip((p0-p3)/4, -ad1, ad1);

        src[y*stride-2] = p0 - d2;
        src[y*stride+1] = p3 + d2;

            const int p2 = pix[-3*xstride];
            const int q0 = pix[0];
            const int q1 = pix[1*xstride];
            const int q2 = pix[2*xstride];

            if( ABS( p0 - q0 ) < alpha &&
                ABS( p1 - p0 ) < beta &&
                ABS( q1 - q0 ) < beta ) {

                int tc = tc0[i];
                int i_delta;

                if( ABS( p2 - p0 ) < beta ) {
                    pix[-2*xstride] = p1 + clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
                    tc++;
                }
                if( ABS( q2 - q0 ) < beta ) {
                    pix[   xstride] = q1 + clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
                    tc++;
                }

                i_delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

            const int p0 = pix[-1*xstride];
            const int p1 = pix[-2*xstride];
            const int q0 = pix[0];
            const int q1 = pix[1*xstride];

            if( ABS( p0 - q0 ) < alpha &&
                ABS( p1 - p0 ) < beta &&
                ABS( q1 - q0 ) < beta ) {

                int delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

                pix[-xstride] = clip_uint8( p0 + delta );    /* p0' */
                pix[0]        = clip_uint8( q0 - delta );    /* q0' */

        const int p0 = pix[-1*xstride];
        const int p1 = pix[-2*xstride];
        const int q0 = pix[0];
        const int q1 = pix[1*xstride];

        if( ABS( p0 - q0 ) < alpha &&
            ABS( p1 - p0 ) < beta &&
            ABS( q1 - q0 ) < beta ) {

            pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */
            pix[0]        = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */
        }
        pix += ystride;

        for(x=0; x<16; x++){
            score1+= (s1[x  ] - s2[x ])*(s1[x  ] - s2[x ]);
        }
        if(y+1<h){
            for(x=0; x<15; x++){
                score2+= ABS(  s1[x  ] - s1[x  +stride]
                             - s1[x+1] + s1[x+1+stride])
                        -ABS(  s2[x  ] - s2[x  +stride]
                             - s2[x+1] + s2[x+1+stride]);
            }
        }
        s1+= stride;
        s2+= stride;
    }

    if(c) return score1 + ABS(score2)*c->avctx->nsse_weight;
    else  return score1 + ABS(score2)*8;
}

static int nsse8_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
    MpegEncContext *c = v;
    int score1=0;

        for(x=0; x<8; x++){
            score1+= (s1[x  ] - s2[x ])*(s1[x  ] - s2[x ]);
        }
        if(y+1<h){
            for(x=0; x<7; x++){
                score2+= ABS(  s1[x  ] - s1[x  +stride]
                             - s1[x+1] + s1[x+1+stride])
                        -ABS(  s2[x  ] - s2[x  +stride]
                             - s2[x+1] + s2[x+1+stride]);
            }
        }
        s1+= stride;
        s2+= stride;
    }

    if(c) return score1 + ABS(score2)*c->avctx->nsse_weight;
    else  return score1 + ABS(score2)*8;
}

static int try_8x8basis_c(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale){
    int i;
    unsigned int sum=0;

    b= y;\
    x= a+b;\
    y= a-b;\
}

#define BUTTERFLYA(x,y) (ABS((x)+(y)) + ABS((x)-(y)))

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;

            +BUTTERFLYA(temp[8*1+i], temp[8*5+i])
            +BUTTERFLYA(temp[8*2+i], temp[8*6+i])
            +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, int h){

    s->dsp.diff_pixels(temp, src1, src2, stride);
    s->dsp.fdct(temp);

    for(i=0; i<64; i++)
        sum+= ABS(temp[i]);

    return sum;
}

#ifdef CONFIG_GPL

        DCT8_1D
#undef SRC
#undef DST

#define SRC(x) dct[x][i]
#define DST(x,v) sum += ABS(v)
    for( i = 0; i < 8; i++ )
        DCT8_1D
#undef SRC
#undef DST
    return sum;

    s->dsp.diff_pixels(temp, src1, src2, stride);
    s->dsp.fdct(temp);

    for(i=0; i<64; i++)
        sum= FFMAX(sum, ABS(temp[i]));

    return sum;
}

static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, 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;

    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;
    }

4001:34fdffe98bd0

            int sy= (ori&2) ? stride>>1 : 0;

            for(i=0; i<size; i++){
                for(j=0; j<size; j++){
                    int v= tmp[sx + sy + i*stride + j] * scale[type][dec_count-3][level][ori];
                    s += FFABS(v);
                }
            }
        }
    }
    assert(s>=0);

        if(p2&256) p2= ~(p2>>31);

        src[x-1*stride] = p1;
        src[x+0*stride] = p2;

        ad1= FFABS(d1)>>1;

        d2= clip((p0-p3)/4, -ad1, ad1);

        src[x-2*stride] = p0 - d2;
        src[x+  stride] = p3 + d2;

        if(p2&256) p2= ~(p2>>31);

        src[y*stride-1] = p1;
        src[y*stride+0] = p2;

        ad1= FFABS(d1)>>1;

        d2= clip((p0-p3)/4, -ad1, ad1);

        src[y*stride-2] = p0 - d2;
        src[y*stride+1] = p3 + d2;

            const int p2 = pix[-3*xstride];
            const int q0 = pix[0];
            const int q1 = pix[1*xstride];
            const int q2 = pix[2*xstride];

            if( FFABS( p0 - q0 ) < alpha &&
                FFABS( p1 - p0 ) < beta &&
                FFABS( q1 - q0 ) < beta ) {

                int tc = tc0[i];
                int i_delta;

                if( FFABS( p2 - p0 ) < beta ) {
                    pix[-2*xstride] = p1 + clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
                    tc++;
                }
                if( FFABS( q2 - q0 ) < beta ) {
                    pix[   xstride] = q1 + clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
                    tc++;
                }

                i_delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

            const int p0 = pix[-1*xstride];
            const int p1 = pix[-2*xstride];
            const int q0 = pix[0];
            const int q1 = pix[1*xstride];

            if( FFABS( p0 - q0 ) < alpha &&
                FFABS( p1 - p0 ) < beta &&
                FFABS( q1 - q0 ) < beta ) {

                int delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

                pix[-xstride] = clip_uint8( p0 + delta );    /* p0' */
                pix[0]        = clip_uint8( q0 - delta );    /* q0' */

        const int p0 = pix[-1*xstride];
        const int p1 = pix[-2*xstride];
        const int q0 = pix[0];
        const int q1 = pix[1*xstride];

        if( FFABS( p0 - q0 ) < alpha &&
            FFABS( p1 - p0 ) < beta &&
            FFABS( q1 - q0 ) < beta ) {

            pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */
            pix[0]        = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */
        }
        pix += ystride;

        for(x=0; x<16; x++){
            score1+= (s1[x  ] - s2[x ])*(s1[x  ] - s2[x ]);
        }
        if(y+1<h){
            for(x=0; x<15; x++){
                score2+= FFABS(  s1[x  ] - s1[x  +stride]
                             - s1[x+1] + s1[x+1+stride])
                        -FFABS(  s2[x  ] - s2[x  +stride]
                             - s2[x+1] + s2[x+1+stride]);
            }
        }
        s1+= stride;
        s2+= stride;
    }

    if(c) return score1 + FFABS(score2)*c->avctx->nsse_weight;
    else  return score1 + FFABS(score2)*8;
}

static int nsse8_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
    MpegEncContext *c = v;
    int score1=0;

        for(x=0; x<8; x++){
            score1+= (s1[x  ] - s2[x ])*(s1[x  ] - s2[x ]);
        }
        if(y+1<h){
            for(x=0; x<7; x++){
                score2+= FFABS(  s1[x  ] - s1[x  +stride]
                             - s1[x+1] + s1[x+1+stride])
                        -FFABS(  s2[x  ] - s2[x  +stride]
                             - s2[x+1] + s2[x+1+stride]);
            }
        }
        s1+= stride;
        s2+= stride;
    }

    if(c) return score1 + FFABS(score2)*c->avctx->nsse_weight;
    else  return score1 + FFABS(score2)*8;
}

static int try_8x8basis_c(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale){
    int i;
    unsigned int sum=0;

    b= y;\
    x= a+b;\
    y= a-b;\
}

#define BUTTERFLYA(x,y) (FFABS((x)+(y)) + FFABS((x)-(y)))

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;

            +BUTTERFLYA(temp[8*1+i], temp[8*5+i])
            +BUTTERFLYA(temp[8*2+i], temp[8*6+i])
            +BUTTERFLYA(temp[8*3+i], temp[8*7+i]);
    }

    sum -= FFABS(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, int h){

    s->dsp.diff_pixels(temp, src1, src2, stride);
    s->dsp.fdct(temp);

    for(i=0; i<64; i++)
        sum+= FFABS(temp[i]);

    return sum;
}

#ifdef CONFIG_GPL

        DCT8_1D
#undef SRC
#undef DST

#define SRC(x) dct[x][i]
#define DST(x,v) sum += FFABS(v)
    for( i = 0; i < 8; i++ )
        DCT8_1D
#undef SRC
#undef DST
    return sum;

    s->dsp.diff_pixels(temp, src1, src2, stride);
    s->dsp.fdct(temp);

    for(i=0; i<64; i++)
        sum= FFMAX(sum, FFABS(temp[i]));

    return sum;
}

static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){

    int score=0;
    int x,y;

    for(y=1; y<h; y++){
        for(x=0; x<16; x+=4){
            score+= FFABS(s[x  ] - s[x  +stride]) + FFABS(s[x+1] - s[x+1+stride])
                   +FFABS(s[x+2] - s[x+2+stride]) + FFABS(s[x+3] - s[x+3+stride]);
        }
        s+= stride;
    }

    return score;

    int score=0;
    int x,y;

    for(y=1; y<h; y++){
        for(x=0; x<16; x++){
            score+= FFABS(s1[x  ] - s2[x ] - s1[x  +stride] + s2[x +stride]);
        }
        s1+= stride;
        s2+= stride;
    }