slicetype_decision.c

x264/h264 編碼, you can put any dir, No error, No warnning.In vc6 build.

/*****************************************************************************
 * slicetype_decision.c: h264 encoder library
 *****************************************************************************
 * Copyright (C) 2005 Loren Merritt
 *
 * Authors: Loren Merritt <[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 <string.h>
#include <math.h>
#include <limits.h>

#include "common/common.h"
#include "enc_macroblock.h"
#include "me.h"
#include "ratecontrol.h"

typedef struct
{
    /* 16x16 */
    int i_ref;
    int       i_rd16x16;
    x264_me_t me16x16;

    /* 8x8 */
    int       i_cost8x8;
    int       mvc[16][5][2]; /* [ref][0] is 16x16 mv,
                                [ref][1..4] are 8x8 mv from partition [0..3] */
    x264_me_t me8x8[4];

    /* Sub 4x4 */
    int       i_cost4x4[4]; /* cost per 8x8 partition */
    x264_me_t me4x4[4][4];

    /* Sub 8x4 */
    int       i_cost8x4[4]; /* cost per 8x8 partition */
    x264_me_t me8x4[4][2];

    /* Sub 4x8 */
    int       i_cost4x8[4]; /* cost per 8x8 partition */
    x264_me_t me4x8[4][4];

    /* 16x8 */
    int       i_cost16x8;
    x264_me_t me16x8[2];

    /* 8x16 */
    int       i_cost8x16;
    x264_me_t me8x16[2];

} x264_mb_analysis_list_t;

typedef struct
{
    /* conduct the analysis using this lamda and QP */
    int i_lambda;
    int i_lambda2;
    int i_qp;
    int16_t *p_cost_mv;
    int b_mbrd;


    /* I: Intra part */
    /* Take some shortcuts in intra search if intra is deemed unlikely */
    int b_fast_intra;
    int b_try_pskip;

    /* Luma part */
    int i_satd_i16x16;
    int i_satd_i16x16_dir[7];
    int i_predict16x16;

    int i_satd_i8x8;
    int i_satd_i8x8_dir[12][4];
    int i_predict8x8[4];

    int i_satd_i4x4;
    int i_predict4x4[16];

    /* Chroma part */
    int i_satd_i8x8chroma;
    int i_predict8x8chroma;

    /* II: Inter part P/B frame */
    x264_mb_analysis_list_t l0;
    x264_mb_analysis_list_t l1;

    int i_cost16x16bi; /* used the same ref and mv as l0 and l1 (at least for now) */
    int i_cost16x16direct;
    int i_cost8x8bi;
    int i_cost8x8direct[4];
    int i_cost16x8bi;
    int i_cost8x16bi;
    int i_rd16x16bi;
    int i_rd16x16direct;
    int i_rd16x8bi;
    int i_rd8x16bi;
    int i_rd8x8bi;

    int i_mb_partition16x8[2]; /* mb_partition_e */
    int i_mb_partition8x16[2];
    int i_mb_type16x8; /* mb_class_e */
    int i_mb_type8x16;

    int b_direct_available;

} x264_mb_analysis_t;

/* lambda = pow(2,qp/6-2) */
static const int i_qp0_cost_table[52] = {
   1, 1, 1, 1, 1, 1, 1, 1,  /*  0-7 */
   1, 1, 1, 1,              /*  8-11 */
   1, 1, 1, 1, 2, 2, 2, 2,  /* 12-19 */
   3, 3, 3, 4, 4, 4, 5, 6,  /* 20-27 */
   6, 7, 8, 9,10,11,13,14,  /* 28-35 */
  16,18,20,23,25,29,32,36,  /* 36-43 */
  40,45,51,57,64,72,81,91   /* 44-51 */
};

/* pow(lambda,2) * .9 */
static const int i_qp0_cost2_table[52] = {
   1,   1,   1,   1,   1,   1, /*  0-5  */
   1,   1,   1,   1,   1,   1, /*  6-11 */
   1,   1,   1,   2,   2,   3, /* 12-17 */
   4,   5,   6,   7,   9,  11, /* 18-23 */
  14,  18,  23,  29,  36,  46, /* 24-29 */
  58,  73,  91, 115, 145, 183, /* 30-35 */
 230, 290, 366, 461, 581, 731, /* 36-41 */
 922,1161,1463,1843,2322,2926, /* 42-47 */
3686,4645,5852,7373
};

extern void x264_mb_analyse_load_costs( x264_t *h, x264_mb_analysis_t *a );
extern void x264_cpu_restore( uint32_t cpu );

static void x264_lowres_context_init( x264_t *h, x264_mb_analysis_t *a )
{
    a->i_qp = 12; // arbitrary, but low because SATD scores are 1/4 normal
    a->i_lambda = i_qp0_cost_table[ a->i_qp ];
    x264_mb_analyse_load_costs( h, a );
    h->mb.i_me_method = X264_MIN( X264_ME_HEX, h->param.analyse.i_me_method ); // maybe dia?
    h->mb.i_subpel_refine = 4; // 3 should be enough, but not tweaking for speed now
    h->mb.b_chroma_me = 0;
}

int x264_slicetype_mb_cost( x264_t *h, x264_mb_analysis_t *a,
                            x264_frame_t **frames, int p0, int p1, int b,
                            int dist_scale_factor )
{
    x264_frame_t *fref0 = frames[p0];
    x264_frame_t *fref1 = frames[p1];
    x264_frame_t *fenc  = frames[b];
    const int b_bidir = (b < p1);
    const int i_mb_x = h->mb.i_mb_x;
    const int i_mb_y = h->mb.i_mb_y;
    const int i_mb_stride = h->sps->i_mb_width;
    const int i_mb_xy = i_mb_x + i_mb_y * i_mb_stride;
    const int i_stride = fenc->i_stride_lowres;
    const int i_pel_offset = 8 * ( i_mb_x + i_mb_y * i_stride );

    uint8_t pix1[9*9], pix2[8*8];
    x264_me_t m[2];
    int i_bcost = COST_MAX;
    int i_cost_bak;
    int l, i;

    h->mb.pic.p_fenc[0] = h->mb.pic.fenc_buf;
    h->mc.copy[PIXEL_8x8]( h->mb.pic.p_fenc[0], FENC_STRIDE, &fenc->lowres[0][i_pel_offset], i_stride, 8 );

    if( !p0 && !p1 && !b )
        goto lowres_intra_mb;

    // no need for h->mb.mv_min[]
    h->mb.mv_min_fpel[0] = -8*h->mb.i_mb_x - 4;
    h->mb.mv_max_fpel[0] = 8*( h->sps->i_mb_width - h->mb.i_mb_x - 1 ) + 4;
    h->mb.mv_min_spel[0] = 4*( h->mb.mv_min_fpel[0] - 8 );
    h->mb.mv_max_spel[0] = 4*( h->mb.mv_max_fpel[0] + 8 );
    if( h->mb.i_mb_x <= 1 )
    {
        h->mb.mv_min_fpel[1] = -8*h->mb.i_mb_y - 4;
        h->mb.mv_max_fpel[1] = 8*( h->sps->i_mb_height - h->mb.i_mb_y - 1 ) + 4;
        h->mb.mv_min_spel[1] = 4*( h->mb.mv_min_fpel[1] - 8 );
        h->mb.mv_max_spel[1] = 4*( h->mb.mv_max_fpel[1] + 8 );
    }

#define LOAD_HPELS_LUMA(dst, src) \
    { \
        (dst)[0] = &(src)[0][i_pel_offset]; \
        (dst)[1] = &(src)[1][i_pel_offset]; \
        (dst)[2] = &(src)[2][i_pel_offset]; \
        (dst)[3] = &(src)[3][i_pel_offset]; \
    }
#define SAVE_MVS( mv0, mv1 ) \
    { \
        fenc->mv[0][i_mb_xy][0] = mv0[0]; \
        fenc->mv[0][i_mb_xy][1] = mv0[1]; \
        if( b_bidir ) \
        { \
            fenc->mv[1][i_mb_xy][0] = mv1[0]; \
            fenc->mv[1][i_mb_xy][1] = mv1[1]; \
        } \
    }
#define CLIP_MV( mv ) \
    { \
        mv[0] = x264_clip3( mv[0], h->mb.mv_min_spel[0], h->mb.mv_max_spel[0] ); \
        mv[1] = x264_clip3( mv[1], h->mb.mv_min_spel[1], h->mb.mv_max_spel[1] ); \
    }
#define TRY_BIDIR( mv0, mv1, penalty ) \
    { \
        int stride2 = 8; \
        uint8_t *src2; \
        int i_cost; \
        h->mc.mc_luma( m[0].p_fref, m[0].i_stride[0], pix1, 8, \
                       (mv0)[0], (mv0)[1], 8, 8 ); \
        src2 = h->mc.get_ref( m[1].p_fref, m[1].i_stride[0], pix2, &stride2, \
                       (mv1)[0], (mv1)[1], 8, 8 ); \
        h->mc.avg[PIXEL_8x8]( pix1, 8, src2, stride2 ); \
        i_cost = penalty + h->pixf.mbcmp[PIXEL_8x8]( \
                           m[0].p_fenc[0], FENC_STRIDE, pix1, 8 ); \
        if( i_bcost > i_cost ) \
        { \
            i_bcost = i_cost; \
            SAVE_MVS( mv0, mv1 ); \
        } \
    }

    m[0].i_pixel = PIXEL_8x8;
    m[0].p_cost_mv = a->p_cost_mv;
    m[0].i_stride[0] = i_stride;
    m[0].p_fenc[0] = h->mb.pic.p_fenc[0];
    LOAD_HPELS_LUMA( m[0].p_fref, fref0->lowres );

    if( b_bidir )
    {
        int16_t *mvr = fref1->mv[0][i_mb_xy];
        int dmv[2][2];
        int mv0[2] = {0,0};

        m[1] = m[0];
        LOAD_HPELS_LUMA( m[1].p_fref, fref1->lowres );

        dmv[0][0] = ( mvr[0] * dist_scale_factor + 128 ) >> 8;
        dmv[0][1] = ( mvr[1] * dist_scale_factor + 128 ) >> 8;
        dmv[1][0] = dmv[0][0] - mvr[0];
        dmv[1][1] = dmv[0][1] - mvr[1];
        CLIP_MV( dmv[0] );
        CLIP_MV( dmv[1] );

        TRY_BIDIR( dmv[0], dmv[1], 0 );
        if( dmv[0][0] || dmv[0][1] || dmv[1][0] || dmv[1][1] )
           TRY_BIDIR( mv0, mv0, 0 );
//      if( i_bcost < 60 ) // arbitrary threshold
//          return i_bcost;
    }

    i_cost_bak = i_bcost;
    for( l = 0; l < 1 + b_bidir; l++ )
    {
        int mvc[4][2] = {{0}}, i_mvc;
        int16_t (*fenc_mv)[2] = &fenc->mv[l][i_mb_xy];
        i_mvc = 0;
        if( i_mb_x > 0 )
        {
            mvc[i_mvc][0] = fenc_mv[-1][0];
            mvc[i_mvc][1] = fenc_mv[-1][1];
            i_mvc++;
        }
        if( i_mb_y > 0 )
        {
            mvc[i_mvc][0] = fenc_mv[-i_mb_stride][0];
            mvc[i_mvc][1] = fenc_mv[-i_mb_stride][1];
            i_mvc++;
            if( i_mb_x < h->sps->i_mb_width - 1 )
            {
                mvc[i_mvc][0] = fenc_mv[-i_mb_stride+1][0];
                mvc[i_mvc][1] = fenc_mv[-i_mb_stride+1][1];
                i_mvc++;
            }