/***************************************************************************** * predict.c: h264 encoder ***************************************************************************** * Copyright (C) 2003 Laurent Aimar * $Id: predict.c,v 1.1 2004/06/03 19:27:07 fenrir Exp $ * * Authors: Laurent Aimar <fenrir@via.ecp.fr> *          Loren Merritt <lorenm@u.washington.edu> * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111, USA. *****************************************************************************//* predict4x4 are inspired from ffmpeg h264 decoder */#include "common.h"#include "clip1.h"#ifdef _MSC_VER#undef HAVE_MMXEXT  /* not finished now */#endif#ifdef HAVE_MMXEXT#   include "i386/predict.h"#endif/**************************************************************************** * 16x16 prediction for intra luma block ****************************************************************************/#define PREDICT_16x16_DC(v) \    for( i = 0; i < 16; i++ )\    {\        uint32_t *p = (uint32_t*)src;\        *p++ = v;\        *p++ = v;\        *p++ = v;\        *p++ = v;\        src += FDEC_STRIDE;\    }static void predict_16x16_dc( uint8_t *src ){    uint32_t dc = 0;    int i;    for( i = 0; i < 16; i++ )    {        dc += src[-1 + i * FDEC_STRIDE];        dc += src[i - FDEC_STRIDE];    }    dc = (( dc + 16 ) >> 5) * 0x01010101;    PREDICT_16x16_DC(dc);}static void predict_16x16_dc_left( uint8_t *src ){    uint32_t dc = 0;    int i;    for( i = 0; i < 16; i++ )    {        dc += src[-1 + i * FDEC_STRIDE];    }    dc = (( dc + 8 ) >> 4) * 0x01010101;    PREDICT_16x16_DC(dc);}static void predict_16x16_dc_top( uint8_t *src ){    uint32_t dc = 0;    int i;    for( i = 0; i < 16; i++ )    {        dc += src[i - FDEC_STRIDE];    }    dc = (( dc + 8 ) >> 4) * 0x01010101;    PREDICT_16x16_DC(dc);}static void predict_16x16_dc_128( uint8_t *src ){    int i;    PREDICT_16x16_DC(0x80808080);}static void predict_16x16_h( uint8_t *src ){    int i;    for( i = 0; i < 16; i++ )    {        const uint32_t v = 0x01010101 * src[-1];        uint32_t *p = (uint32_t*)src;        *p++ = v;        *p++ = v;        *p++ = v;        *p++ = v;        src += FDEC_STRIDE;    }}static void predict_16x16_v( uint8_t *src ){    uint32_t v0 = *(uint32_t*)&src[ 0-FDEC_STRIDE];    uint32_t v1 = *(uint32_t*)&src[ 4-FDEC_STRIDE];    uint32_t v2 = *(uint32_t*)&src[ 8-FDEC_STRIDE];    uint32_t v3 = *(uint32_t*)&src[12-FDEC_STRIDE];    int i;    for( i = 0; i < 16; i++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = v0;        *p++ = v1;        *p++ = v2;        *p++ = v3;        src += FDEC_STRIDE;    }}static void predict_16x16_p( uint8_t *src ){    int x, y, i;    int a, b, c;    int H = 0;    int V = 0;    int i00;    /* calcule H and V */    for( i = 0; i <= 7; i++ )    {        H += ( i + 1 ) * ( src[ 8 + i - FDEC_STRIDE ] - src[6 -i -FDEC_STRIDE] );        V += ( i + 1 ) * ( src[-1 + (8+i)*FDEC_STRIDE] - src[-1 + (6-i)*FDEC_STRIDE] );    }    a = 16 * ( src[-1 + 15*FDEC_STRIDE] + src[15 - FDEC_STRIDE] );    b = ( 5 * H + 32 ) >> 6;    c = ( 5 * V + 32 ) >> 6;    i00 = a - b * 7 - c * 7 + 16;    for( y = 0; y < 16; y++ )    {        int pix = i00;        for( x = 0; x < 16; x++ )        {            src[x] = x264_clip_uint8( pix>>5 );            pix += b;        }        src += FDEC_STRIDE;        i00 += c;    }}/**************************************************************************** * 8x8 prediction for intra chroma block ****************************************************************************/static void predict_8x8c_dc_128( uint8_t *src ){    int y;    for( y = 0; y < 8; y++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = 0x80808080;        *p++ = 0x80808080;        src += FDEC_STRIDE;    }}static void predict_8x8c_dc_left( uint8_t *src ){    int y;    uint32_t dc0 = 0, dc1 = 0;    for( y = 0; y < 4; y++ )    {        dc0 += src[y * FDEC_STRIDE     - 1];        dc1 += src[(y+4) * FDEC_STRIDE - 1];    }    dc0 = (( dc0 + 2 ) >> 2)*0x01010101;    dc1 = (( dc1 + 2 ) >> 2)*0x01010101;    for( y = 0; y < 4; y++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = dc0;        *p++ = dc0;        src += FDEC_STRIDE;    }    for( y = 0; y < 4; y++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = dc1;        *p++ = dc1;        src += FDEC_STRIDE;    }}static void predict_8x8c_dc_top( uint8_t *src ){    int y, x;    uint32_t dc0 = 0, dc1 = 0;    for( x = 0; x < 4; x++ )    {        dc0 += src[x     - FDEC_STRIDE];        dc1 += src[x + 4 - FDEC_STRIDE];    }    dc0 = (( dc0 + 2 ) >> 2)*0x01010101;    dc1 = (( dc1 + 2 ) >> 2)*0x01010101;    for( y = 0; y < 8; y++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = dc0;        *p++ = dc1;        src += FDEC_STRIDE;    }}static void predict_8x8c_dc( uint8_t *src ){    int y;    int s0 = 0, s1 = 0, s2 = 0, s3 = 0;    uint32_t dc0, dc1, dc2, dc3;    int i;    /*          s0 s1       s2       s3    */    for( i = 0; i < 4; i++ )    {        s0 += src[i - FDEC_STRIDE];        s1 += src[i + 4 - FDEC_STRIDE];        s2 += src[-1 + i * FDEC_STRIDE];        s3 += src[-1 + (i+4)*FDEC_STRIDE];    }    /*       dc0 dc1       dc2 dc3     */    dc0 = (( s0 + s2 + 4 ) >> 3)*0x01010101;    dc1 = (( s1 + 2 ) >> 2)*0x01010101;    dc2 = (( s3 + 2 ) >> 2)*0x01010101;    dc3 = (( s1 + s3 + 4 ) >> 3)*0x01010101;    for( y = 0; y < 4; y++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = dc0;        *p++ = dc1;        src += FDEC_STRIDE;    }    for( y = 0; y < 4; y++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = dc2;        *p++ = dc3;        src += FDEC_STRIDE;    }}static void predict_8x8c_h( uint8_t *src ){    int i;    for( i = 0; i < 8; i++ )    {        uint32_t v = 0x01010101 * src[-1];        uint32_t *p = (uint32_t*)src;        *p++ = v;        *p++ = v;        src += FDEC_STRIDE;    }}static void predict_8x8c_v( uint8_t *src ){    uint32_t v0 = *(uint32_t*)&src[0-FDEC_STRIDE];    uint32_t v1 = *(uint32_t*)&src[4-FDEC_STRIDE];    int i;    for( i = 0; i < 8; i++ )    {        uint32_t *p = (uint32_t*)src;        *p++ = v0;        *p++ = v1;        src += FDEC_STRIDE;    }}static void predict_8x8c_p( uint8_t *src ){    int i;    int x,y;    int a, b, c;    int H = 0;    int V = 0;    int i00;    for( i = 0; i < 4; i++ )    {        H += ( i + 1 ) * ( src[4+i - FDEC_STRIDE] - src[2 - i -FDEC_STRIDE] );        V += ( i + 1 ) * ( src[-1 +(i+4)*FDEC_STRIDE] - src[-1+(2-i)*FDEC_STRIDE] );    }    a = 16 * ( src[-1+7*FDEC_STRIDE] + src[7 - FDEC_STRIDE] );    b = ( 17 * H + 16 ) >> 5;    c = ( 17 * V + 16 ) >> 5;    i00 = a -3*b -3*c + 16;    for( y = 0; y < 8; y++ )    {        int pix = i00;        for( x = 0; x < 8; x++ )        {            src[x] = x264_clip_uint8( pix>>5 );            pix += b;        }        src += FDEC_STRIDE;        i00 += c;    }}/**************************************************************************** * 4x4 prediction for intra luma block ****************************************************************************/#define PREDICT_4x4_DC(v) \{\    *(uint32_t*)&src[0*FDEC_STRIDE] =\    *(uint32_t*)&src[1*FDEC_STRIDE] =\    *(uint32_t*)&src[2*FDEC_STRIDE] =\    *(uint32_t*)&src[3*FDEC_STRIDE] = v;\}static void predict_4x4_dc_128( uint8_t *src ){    PREDICT_4x4_DC(0x80808080);}static void predict_4x4_dc_left( uint8_t *src ){    uint32_t dc = (( src[-1+0*FDEC_STRIDE] + src[-1+FDEC_STRIDE]+                     src[-1+2*FDEC_STRIDE] + src[-1+3*FDEC_STRIDE] + 2 ) >> 2)*0x01010101;    PREDICT_4x4_DC(dc);}static void predict_4x4_dc_top( uint8_t *src ){    uint32_t dc = (( src[0 - FDEC_STRIDE] + src[1 - FDEC_STRIDE] +                     src[2 - FDEC_STRIDE] + src[3 - FDEC_STRIDE] + 2 ) >> 2)*0x01010101;    PREDICT_4x4_DC(dc);}static void predict_4x4_dc( uint8_t *src ){    uint32_t dc = (( src[-1+0*FDEC_STRIDE] + src[-1+FDEC_STRIDE] +                     src[-1+2*FDEC_STRIDE] + src[-1+3*FDEC_STRIDE] +                     src[0 - FDEC_STRIDE]  + src[1 - FDEC_STRIDE] +                     src[2 - FDEC_STRIDE]  + src[3 - FDEC_STRIDE] + 4 ) >> 3)*0x01010101;    PREDICT_4x4_DC(dc);}static void predict_4x4_h( uint8_t *src ){    *(uint32_t*)&src[0*FDEC_STRIDE] = src[0*FDEC_STRIDE-1] * 0x01010101;    *(uint32_t*)&src[1*FDEC_STRIDE] = src[1*FDEC_STRIDE-1] * 0x01010101;    *(uint32_t*)&src[2*FDEC_STRIDE] = src[2*FDEC_STRIDE-1] * 0x01010101;    *(uint32_t*)&src[3*FDEC_STRIDE] = src[3*FDEC_STRIDE-1] * 0x01010101;}static void predict_4x4_v( uint8_t *src ){    uint32_t top = *((uint32_t*)&src[-FDEC_STRIDE]);    PREDICT_4x4_DC(top);}#define PREDICT_4x4_LOAD_LEFT \    const int l0 = src[-1+0*FDEC_STRIDE];   \    const int l1 = src[-1+1*FDEC_STRIDE];   \    const int l2 = src[-1+2*FDEC_STRIDE];   \    UNUSED const int l3 = src[-1+3*FDEC_STRIDE];#define PREDICT_4x4_LOAD_TOP \    const int t0 = src[0-1*FDEC_STRIDE];   \    const int t1 = src[1-1*FDEC_STRIDE];   \    const int t2 = src[2-1*FDEC_STRIDE];   \    UNUSED const int t3 = src[3-1*FDEC_STRIDE];#define PREDICT_4x4_LOAD_TOP_RIGHT \    const int t4 = src[4-1*FDEC_STRIDE];   \    const int t5 = src[5-1*FDEC_STRIDE];   \    const int t6 = src[6-1*FDEC_STRIDE];   \    UNUSED const int t7 = src[7-1*FDEC_STRIDE];static void predict_4x4_ddl( uint8_t *src ){    PREDICT_4x4_LOAD_TOP    PREDICT_4x4_LOAD_TOP_RIGHT    src[0*FDEC_STRIDE+0] = ( t0 + 2*t1 + t2 + 2 ) >> 2;    src[0*FDEC_STRIDE+1] =    src[1*FDEC_STRIDE+0] = ( t1 + 2*t2 + t3 + 2 ) >> 2;    src[0*FDEC_STRIDE+2] =    src[1*FDEC_STRIDE+1] =    src[2*FDEC_STRIDE+0] = ( t2 + 2*t3 + t4 + 2 ) >> 2;    src[0*FDEC_STRIDE+3] =    src[1*FDEC_STRIDE+2] =    src[2*FDEC_STRIDE+1] =    src[3*FDEC_STRIDE+0] = ( t3 + 2*t4 + t5 + 2 ) >> 2;    src[1*FDEC_STRIDE+3] =    src[2*FDEC_STRIDE+2] =    src[3*FDEC_STRIDE+1] = ( t4 + 2*t5 + t6 + 2 ) >> 2;    src[2*FDEC_STRIDE+3] =    src[3*FDEC_STRIDE+2] = ( t5 + 2*t6 + t7 + 2 ) >> 2;    src[3*FDEC_STRIDE+3] = ( t6 + 3*t7 + 2 ) >> 2;}static void predict_4x4_ddr( uint8_t *src ){    const int lt = src[-1-FDEC_STRIDE];    PREDICT_4x4_LOAD_LEFT    PREDICT_4x4_LOAD_TOP    src[0*FDEC_STRIDE+0] =    src[1*FDEC_STRIDE+1] =    src[2*FDEC_STRIDE+2] =    src[3*FDEC_STRIDE+3] = ( t0 + 2 * lt + l0 + 2 ) >> 2;    src[0*FDEC_STRIDE+1] =    src[1*FDEC_STRIDE+2] =    src[2*FDEC_STRIDE+3] = ( lt + 2 * t0 + t1 + 2 ) >> 2;    src[0*FDEC_STRIDE+2] =    src[1*FDEC_STRIDE+3] = ( t0 + 2 * t1 + t2 + 2 ) >> 2;    src[0*FDEC_STRIDE+3] = ( t1 + 2 * t2 + t3 + 2 ) >> 2;    src[1*FDEC_STRIDE+0] =    src[2*FDEC_STRIDE+1] =    src[3*FDEC_STRIDE+2] = ( lt + 2 * l0 + l1 + 2 ) >> 2;    src[2*FDEC_STRIDE+0] =    src[3*FDEC_STRIDE+1] = ( l0 + 2 * l1 + l2 + 2 ) >> 2;    src[3*FDEC_STRIDE+0] = ( l1 + 2 * l2 + l3 + 2 ) >> 2;}