macroblock.c.svn-base

此段代码是h.264在linux下的编码源程序,在linux下编译可以得到可执行程序

/*****************************************************************************
 * macroblock.c: h264 encoder library
 *****************************************************************************
 * Copyright (C) 2003 Laurent Aimar
 * $Id: macroblock.c,v 1.1 2004/06/03 19:27:08 fenrir Exp $
 *
 * Authors: Laurent Aimar <[e-mail]>
 *
 * 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.
 *****************************************************************************/

#include <stdio.h>
#include <string.h>

#include "common/common.h"
#include "macroblock.h"


/* def_quant4_mf only for probe_skip; actual encoding uses matrices from set.c */
/* FIXME this seems to make better decisions with cqm=jvt, but could screw up
 * with general custom matrices. */
static const DECLARE_ALIGNED ( int, def_quant4_mf[6][4][4], 16) =
{
    { { 13107, 8066, 13107, 8066 }, { 8066, 5243, 8066, 5243 },
      { 13107, 8066, 13107, 8066 }, { 8066, 5243, 8066, 5243 } },
    { { 11916, 7490, 11916, 7490 }, { 7490, 4660, 7490, 4660 },
      { 11916, 7490, 11916, 7490 }, { 7490, 4660, 7490, 4660 } },
    { { 10082, 6554, 10082, 6554 }, { 6554, 4194, 6554, 4194 },
      { 10082, 6554, 10082, 6554 }, { 6554, 4194, 6554, 4194 } },
    { {  9362, 5825,  9362, 5825 }, { 5825, 3647, 5825, 3647 },
      {  9362, 5825,  9362, 5825 }, { 5825, 3647, 5825, 3647 } },
    { {  8192, 5243,  8192, 5243 }, { 5243, 3355, 5243, 3355 },
      {  8192, 5243,  8192, 5243 }, { 5243, 3355, 5243, 3355 } },
    { {  7282, 4559,  7282, 4559 }, { 4559, 2893, 4559, 2893 },
      {  7282, 4559,  7282, 4559 }, { 4559, 2893, 4559, 2893 } }
};

#define ZIG(i,y,x) level[i] = dct[x][y];
static inline void zigzag_scan_2x2_dc( int level[4], int16_t dct[2][2] )
{
    ZIG(0,0,0)
    ZIG(1,0,1)
    ZIG(2,1,0)
    ZIG(3,1,1)
}
#undef ZIG

static void quant_8x8( x264_t *h, int16_t dct[8][8], int quant_mf[6][8][8], int i_qscale, int b_intra )
{
    const int i_qbits = 16 + i_qscale / 6;
    const int i_mf = i_qscale % 6;
    const int f = h->mb.i_luma_deadzone[b_intra] << (i_qbits-6);
    h->quantf.quant_8x8_core( dct, quant_mf[i_mf], i_qbits, f );
}
static void quant_4x4( x264_t *h, int16_t dct[4][4], int quant_mf[6][4][4], int i_qscale, int b_intra )
{
    const int i_qbits = 15 + i_qscale / 6;
    const int i_mf = i_qscale % 6;
    const int f = h->mb.i_luma_deadzone[b_intra] << (i_qbits-6);
    h->quantf.quant_4x4_core( dct, quant_mf[i_mf], i_qbits, f );
}
static void quant_4x4_chroma( x264_t *h, int16_t dct[4][4], int quant_mf[6][4][4], int i_qscale, int b_intra )
{
    const int i_qbits = 15 + i_qscale / 6;
    const int i_mf = i_qscale % 6;
    const int f = ( 1 << (i_qbits + b_intra) ) / 6;
    h->quantf.quant_4x4_core( dct, quant_mf[i_mf], i_qbits, f );
}
static void quant_4x4_dc( x264_t *h, int16_t dct[4][4], int quant_mf[6][4][4], int i_qscale )
{
    const int i_qbits = 16 + i_qscale / 6;
    const int i_mf = i_qscale % 6;
    const int f = h->mb.i_luma_deadzone[1] << (i_qbits-6);
    h->quantf.quant_4x4_dc_core( dct, quant_mf[i_mf][0][0], i_qbits, f );
}
static void quant_2x2_dc( x264_t *h, int16_t dct[2][2], int quant_mf[6][4][4], int i_qscale, int b_intra )
{
    const int i_qbits = 16 + i_qscale / 6;
    const int i_mf = i_qscale % 6;
    const int f = ( 1 << (i_qbits + b_intra) ) / 6;
    h->quantf.quant_2x2_dc_core( dct, quant_mf[i_mf][0][0], i_qbits, f );
}

/* (ref: JVT-B118)
 * x264_mb_decimate_score: given dct coeffs it returns a score to see if we could empty this dct coeffs
 * to 0 (low score means set it to null)
 * Used in inter macroblock (luma and chroma)
 *  luma: for a 8x8 block: if score < 4 -> null
 *        for the complete mb: if score < 6 -> null
 *  chroma: for the complete mb: if score < 7 -> null
 */
static int x264_mb_decimate_score( int *dct, int i_max )
{
    static const int i_ds_table4[16] = {
        3,2,2,1,1,1,0,0,0,0,0,0,0,0,0,0 };
    static const int i_ds_table8[64] = {
        3,3,3,3,2,2,2,2,2,2,2,2,1,1,1,1,
        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 };

    const int *ds_table = (i_max == 64) ? i_ds_table8 : i_ds_table4;
    int i_score = 0;
    int idx = i_max - 1;

    while( idx >= 0 && dct[idx] == 0 )
        idx--;

    while( idx >= 0 )
    {
        int i_run;

        if( abs( dct[idx--] ) > 1 )
            return 9;

        i_run = 0;
        while( idx >= 0 && dct[idx] == 0 )
        {
            idx--;
            i_run++;
        }
        i_score += ds_table[i_run];
    }

    return i_score;
}

void x264_mb_encode_i4x4( x264_t *h, int idx, int i_qscale )
{
    int x = 4 * block_idx_x[idx];
    int y = 4 * block_idx_y[idx];
    uint8_t *p_src = &h->mb.pic.p_fenc[0][x+y*FENC_STRIDE];
    uint8_t *p_dst = &h->mb.pic.p_fdec[0][x+y*FDEC_STRIDE];
    DECLARE_ALIGNED( int16_t, dct4x4[4][4], 16 );

    if( h->mb.b_lossless )
    {
        h->zigzagf.sub_4x4( h->dct.block[idx].luma4x4, p_src, p_dst );
        return;
    }

    h->dctf.sub4x4_dct( dct4x4, p_src, p_dst );

    if( h->mb.b_trellis )
        x264_quant_4x4_trellis( h, dct4x4, CQM_4IY, i_qscale, DCT_LUMA_4x4, 1 );
    else
        quant_4x4( h, dct4x4, h->quant4_mf[CQM_4IY], i_qscale, 1 );

    h->zigzagf.scan_4x4( h->dct.block[idx].luma4x4, dct4x4 );
    h->quantf.dequant_4x4( dct4x4, h->dequant4_mf[CQM_4IY], i_qscale );

    /* output samples to fdec */
    h->dctf.add4x4_idct( p_dst, dct4x4 );
}

void x264_mb_encode_i8x8( x264_t *h, int idx, int i_qscale )
{
    int x = 8 * (idx&1);
    int y = 8 * (idx>>1);
    uint8_t *p_src = &h->mb.pic.p_fenc[0][x+y*FENC_STRIDE];
    uint8_t *p_dst = &h->mb.pic.p_fdec[0][x+y*FDEC_STRIDE];
    DECLARE_ALIGNED( int16_t, dct8x8[8][8], 16 );

    h->dctf.sub8x8_dct8( dct8x8, p_src, p_dst );

    if( h->mb.b_trellis )
        x264_quant_8x8_trellis( h, dct8x8, CQM_8IY, i_qscale, 1 );
    else 
        quant_8x8( h, dct8x8, h->quant8_mf[CQM_8IY], i_qscale, 1 );

    h->zigzagf.scan_8x8( h->dct.luma8x8[idx], dct8x8 );
    h->quantf.dequant_8x8( dct8x8, h->dequant8_mf[CQM_8IY], i_qscale );
    h->dctf.add8x8_idct8( p_dst, dct8x8 );
}

static void x264_mb_encode_i16x16( x264_t *h, int i_qscale )
{
    uint8_t  *p_src = h->mb.pic.p_fenc[0];
    uint8_t  *p_dst = h->mb.pic.p_fdec[0];

    DECLARE_ALIGNED( int16_t, dct4x4[16+1][4][4], 16 );

    int i;

    if( h->mb.b_lossless )
    {
        for( i = 0; i < 16; i++ )
        {
            int oe = block_idx_x[i]*4 + block_idx_y[i]*4*FENC_STRIDE;
            int od = block_idx_x[i]*4 + block_idx_y[i]*4*FDEC_STRIDE;
            h->zigzagf.sub_4x4ac( h->dct.block[i].residual_ac, p_src+oe, p_dst+od );
            dct4x4[0][block_idx_x[i]][block_idx_y[i]] = p_src[oe] - p_dst[od];
            p_dst[od] = p_src[oe];
        }
        h->zigzagf.scan_4x4( h->dct.luma16x16_dc, dct4x4[0] );
        return;
    }

    h->dctf.sub16x16_dct( &dct4x4[1], p_src, p_dst );
    for( i = 0; i < 16; i++ )
    {
        /* copy dc coeff */
        dct4x4[0][block_idx_y[i]][block_idx_x[i]] = dct4x4[1+i][0][0];

        /* quant/scan/dequant */
        if( h->mb.b_trellis )
            x264_quant_4x4_trellis( h, dct4x4[1+i], CQM_4IY, i_qscale, DCT_LUMA_AC, 1 );
        else
            quant_4x4( h, dct4x4[1+i], h->quant4_mf[CQM_4IY], i_qscale, 1 );

        h->zigzagf.scan_4x4ac( h->dct.block[i].residual_ac, dct4x4[1+i] );
        h->quantf.dequant_4x4( dct4x4[1+i], h->dequant4_mf[CQM_4IY], i_qscale );
    }

    h->dctf.dct4x4dc( dct4x4[0] );
    quant_4x4_dc( h, dct4x4[0], h->quant4_mf[CQM_4IY], i_qscale );
    h->zigzagf.scan_4x4( h->dct.luma16x16_dc, dct4x4[0] );

    /* output samples to fdec */
    h->dctf.idct4x4dc( dct4x4[0] );
    x264_mb_dequant_4x4_dc( dct4x4[0], h->dequant4_mf[CQM_4IY], i_qscale );  /* XXX not inversed */

    /* calculate dct coeffs */
    for( i = 0; i < 16; i++ )
    {
        /* copy dc coeff */
        dct4x4[1+i][0][0] = dct4x4[0][block_idx_y[i]][block_idx_x[i]];
    }
    /* put pixels to fdec */
    h->dctf.add16x16_idct( p_dst, &dct4x4[1] );
}

void x264_mb_encode_8x8_chroma( x264_t *h, int b_inter, int i_qscale )
{
    int i, ch;
    int b_decimate = b_inter && (h->sh.i_type == SLICE_TYPE_B || h->param.analyse.b_dct_decimate);

    for( ch = 0; ch < 2; ch++ )
    {
        uint8_t  *p_src = h->mb.pic.p_fenc[1+ch];
        uint8_t  *p_dst = h->mb.pic.p_fdec[1+ch];
        int i_decimate_score = 0;

        DECLARE_ALIGNED( int16_t, dct2x2[2][2] , 16 );
        DECLARE_ALIGNED( int16_t, dct4x4[4][4][4], 16 );

        if( h->mb.b_lossless )
        {
            for( i = 0; i < 4; i++ )
            {
                int oe = block_idx_x[i]*4 + block_idx_y[i]*4*FENC_STRIDE;
                int od = block_idx_x[i]*4 + block_idx_y[i]*4*FDEC_STRIDE;
                h->zigzagf.sub_4x4ac( h->dct.block[16+i+ch*4].residual_ac, p_src+oe, p_dst+od );
                h->dct.chroma_dc[ch][i] = p_src[oe] - p_dst[od];
                p_dst[od] = p_src[oe];
            }
            continue;
        }
            
        h->dctf.sub8x8_dct( dct4x4, p_src, p_dst );
        /* calculate dct coeffs */
        for( i = 0; i < 4; i++ )
        {
            /* copy dc coeff */
            dct2x2[block_idx_y[i]][block_idx_x[i]] = dct4x4[i][0][0];

            /* no trellis; it doesn't seem to help chroma noticeably */
            quant_4x4_chroma( h, dct4x4[i], h->quant4_mf[CQM_4IC + b_inter], i_qscale, !b_inter );
            h->zigzagf.scan_4x4ac( h->dct.block[16+i+ch*4].residual_ac, dct4x4[i] );

            if( b_decimate )
            {
                i_decimate_score += x264_mb_decimate_score( h->dct.block[16+i+ch*4].residual_ac, 15 );
            }
        }

        h->dctf.dct2x2dc( dct2x2 );
        quant_2x2_dc( h, dct2x2, h->quant4_mf[CQM_4IC + b_inter], i_qscale, !b_inter );
        zigzag_scan_2x2_dc( h->dct.chroma_dc[ch], dct2x2 );

        /* output samples to fdec */
        h->dctf.idct2x2dc( dct2x2 );
        x264_mb_dequant_2x2_dc( dct2x2, h->dequant4_mf[CQM_4IC + b_inter], i_qscale );  /* XXX not inversed */

        if( b_decimate && i_decimate_score < 7 )
        {
            /* Near null chroma 8x8 block so make it null (bits saving) */
            memset( &h->dct.block[16+ch*4], 0, 4 * sizeof( *h->dct.block ) );
            if( !array_non_zero( (int*)dct2x2, sizeof(dct2x2)/sizeof(int) ) )