h264pred.c

播放H264文件的播放器

    src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
}

static void pred16x16_vertical_c(uint8_t *src, int stride){
    int i;
    const uint32_t a= ((uint32_t*)(src-stride))[0];
    const uint32_t b= ((uint32_t*)(src-stride))[1];
    const uint32_t c= ((uint32_t*)(src-stride))[2];
    const uint32_t d= ((uint32_t*)(src-stride))[3];

    for(i=0; i<16; i++){
        ((uint32_t*)(src+i*stride))[0]= a;
        ((uint32_t*)(src+i*stride))[1]= b;
        ((uint32_t*)(src+i*stride))[2]= c;
        ((uint32_t*)(src+i*stride))[3]= d;
    }
}

static void pred16x16_horizontal_c(uint8_t *src, int stride){
    int i;

    for(i=0; i<16; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]=
        ((uint32_t*)(src+i*stride))[2]=
        ((uint32_t*)(src+i*stride))[3]= src[-1+i*stride]*0x01010101;
    }
}

static void pred16x16_dc_c(uint8_t *src, int stride){
    int i, dc=0;

    for(i=0;i<16; i++){
        dc+= src[-1+i*stride];
    }

    for(i=0;i<16; i++){
        dc+= src[i-stride];
    }

    dc= 0x01010101*((dc + 16)>>5);

    for(i=0; i<16; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]=
        ((uint32_t*)(src+i*stride))[2]=
        ((uint32_t*)(src+i*stride))[3]= dc;
    }
}

static void pred16x16_left_dc_c(uint8_t *src, int stride){
    int i, dc=0;

    for(i=0;i<16; i++){
        dc+= src[-1+i*stride];
    }

    dc= 0x01010101*((dc + 8)>>4);

    for(i=0; i<16; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]=
        ((uint32_t*)(src+i*stride))[2]=
        ((uint32_t*)(src+i*stride))[3]= dc;
    }
}

static void pred16x16_top_dc_c(uint8_t *src, int stride){
    int i, dc=0;

    for(i=0;i<16; i++){
        dc+= src[i-stride];
    }
    dc= 0x01010101*((dc + 8)>>4);

    for(i=0; i<16; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]=
        ((uint32_t*)(src+i*stride))[2]=
        ((uint32_t*)(src+i*stride))[3]= dc;
    }
}

static void pred16x16_128_dc_c(uint8_t *src, int stride){
    int i;

    for(i=0; i<16; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]=
        ((uint32_t*)(src+i*stride))[2]=
        ((uint32_t*)(src+i*stride))[3]= 0x01010101U*128U;
    }
}

static _inline void pred16x16_plane_compat_c(uint8_t *src, int stride, const int svq3, const int rv40){
  int i, j, k;
  int a;
  uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
  const uint8_t * const src0 = src+7-stride;
  const uint8_t *src1 = src+8*stride-1;
  const uint8_t *src2 = src1-2*stride;      // == src+6*stride-1;
  int H = src0[1] - src0[-1];
  int V = src1[0] - src2[ 0];
  for(k=2; k<=8; ++k) {
    src1 += stride; src2 -= stride;
    H += k*(src0[k] - src0[-k]);
    V += k*(src1[0] - src2[ 0]);
  }
  if(svq3){
    H = ( 5*(H/4) ) / 16;
    V = ( 5*(V/4) ) / 16;

    /* required for 100% accuracy */
    i = H; H = V; V = i;
  }else if(rv40){
    H = ( H + (H>>2) ) >> 4;
    V = ( V + (V>>2) ) >> 4;
  }else{
    H = ( 5*H+32 ) >> 6;
    V = ( 5*V+32 ) >> 6;
  }

  a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);
  for(j=16; j>0; --j) {
    int b = a;
    a += V;
    for(i=-16; i<0; i+=4) {
      src[16+i] = cm[ (b    ) >> 5 ];
      src[17+i] = cm[ (b+  H) >> 5 ];
      src[18+i] = cm[ (b+2*H) >> 5 ];
      src[19+i] = cm[ (b+3*H) >> 5 ];
      b += 4*H;
    }
    src += stride;
  }
}

static void pred16x16_plane_c(uint8_t *src, int stride){
    pred16x16_plane_compat_c(src, stride, 0, 0);
}

static void pred16x16_plane_svq3_c(uint8_t *src, int stride){
    pred16x16_plane_compat_c(src, stride, 1, 0);
}

static void pred16x16_plane_rv40_c(uint8_t *src, int stride){
    pred16x16_plane_compat_c(src, stride, 0, 1);
}

static void pred8x8_vertical_c(uint8_t *src, int stride){
    int i;
    const uint32_t a= ((uint32_t*)(src-stride))[0];
    const uint32_t b= ((uint32_t*)(src-stride))[1];

    for(i=0; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]= a;
        ((uint32_t*)(src+i*stride))[1]= b;
    }
}

static void pred8x8_horizontal_c(uint8_t *src, int stride){
    int i;

    for(i=0; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]= src[-1+i*stride]*0x01010101;
    }
}

static void pred8x8_128_dc_c(uint8_t *src, int stride){
    int i;

    for(i=0; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]= 0x01010101U*128U;
    }
}

static void pred8x8_left_dc_c(uint8_t *src, int stride){
    int i;
    int dc0, dc2;

    dc0=dc2=0;
    for(i=0;i<4; i++){
        dc0+= src[-1+i*stride];
        dc2+= src[-1+(i+4)*stride];
    }
    dc0= 0x01010101*((dc0 + 2)>>2);
    dc2= 0x01010101*((dc2 + 2)>>2);

    for(i=0; i<4; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]= dc0;
    }
    for(i=4; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]= dc2;
    }
}

static void pred8x8_left_dc_rv40_c(uint8_t *src, int stride){
    int i;
    int dc0;

    dc0=0;
    for(i=0;i<8; i++)
        dc0+= src[-1+i*stride];
    dc0= 0x01010101*((dc0 + 4)>>3);

    for(i=0; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]= dc0;
    }
}

static void pred8x8_top_dc_c(uint8_t *src, int stride){
    int i;
    int dc0, dc1;

    dc0=dc1=0;
    for(i=0;i<4; i++){
        dc0+= src[i-stride];
        dc1+= src[4+i-stride];
    }
    dc0= 0x01010101*((dc0 + 2)>>2);
    dc1= 0x01010101*((dc1 + 2)>>2);

    for(i=0; i<4; i++){
        ((uint32_t*)(src+i*stride))[0]= dc0;
        ((uint32_t*)(src+i*stride))[1]= dc1;
    }
    for(i=4; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]= dc0;
        ((uint32_t*)(src+i*stride))[1]= dc1;
    }
}

static void pred8x8_top_dc_rv40_c(uint8_t *src, int stride){
    int i;
    int dc0;

    dc0=0;
    for(i=0;i<8; i++)
        dc0+= src[i-stride];
    dc0= 0x01010101*((dc0 + 4)>>3);

    for(i=0; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]=
        ((uint32_t*)(src+i*stride))[1]= dc0;
    }
}


static void pred8x8_dc_c(uint8_t *src, int stride){
    int i;
    int dc0, dc1, dc2, dc3;

    dc0=dc1=dc2=0;
    for(i=0;i<4; i++){
        dc0+= src[-1+i*stride] + src[i-stride];
        dc1+= src[4+i-stride];
        dc2+= src[-1+(i+4)*stride];
    }
    dc3= 0x01010101*((dc1 + dc2 + 4)>>3);
    dc0= 0x01010101*((dc0 + 4)>>3);
    dc1= 0x01010101*((dc1 + 2)>>2);
    dc2= 0x01010101*((dc2 + 2)>>2);

    for(i=0; i<4; i++){
        ((uint32_t*)(src+i*stride))[0]= dc0;
        ((uint32_t*)(src+i*stride))[1]= dc1;
    }
    for(i=4; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]= dc2;
        ((uint32_t*)(src+i*stride))[1]= dc3;
    }
}

//the following 4 function should not be optimized!
static void pred8x8_mad_cow_dc_l0t(uint8_t *src, int stride){
    pred8x8_top_dc_c(src, stride);
    pred4x4_dc_c(src, NULL, stride);
}

static void pred8x8_mad_cow_dc_0lt(uint8_t *src, int stride){
    pred8x8_dc_c(src, stride);
    pred4x4_top_dc_c(src, NULL, stride);
}

static void pred8x8_mad_cow_dc_l00(uint8_t *src, int stride){
    pred8x8_left_dc_c(src, stride);
    pred4x4_128_dc_c(src + 4*stride    , NULL, stride);
    pred4x4_128_dc_c(src + 4*stride + 4, NULL, stride);
}

static void pred8x8_mad_cow_dc_0l0(uint8_t *src, int stride){
    pred8x8_left_dc_c(src, stride);
    pred4x4_128_dc_c(src    , NULL, stride);
    pred4x4_128_dc_c(src + 4, NULL, stride);
}

static void pred8x8_dc_rv40_c(uint8_t *src, int stride){
    int i;
    int dc0=0;

    for(i=0;i<4; i++){
        dc0+= src[-1+i*stride] + src[i-stride];
        dc0+= src[4+i-stride];
        dc0+= src[-1+(i+4)*stride];
    }
    dc0= 0x01010101*((dc0 + 8)>>4);

    for(i=0; i<4; i++){
        ((uint32_t*)(src+i*stride))[0]= dc0;
        ((uint32_t*)(src+i*stride))[1]= dc0;
    }
    for(i=4; i<8; i++){
        ((uint32_t*)(src+i*stride))[0]= dc0;
        ((uint32_t*)(src+i*stride))[1]= dc0;
    }
}

static void pred8x8_plane_c(uint8_t *src, int stride){
  int j, k;
  int a;
  uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
  const uint8_t * const src0 = src+3-stride;
  const uint8_t *src1 = src+4*stride-1;
  const uint8_t *src2 = src1-2*stride;      // == src+2*stride-1;
  int H = src0[1] - src0[-1];
  int V = src1[0] - src2[ 0];
  for(k=2; k<=4; ++k) {
    src1 += stride; src2 -= stride;
    H += k*(src0[k] - src0[-k]);
    V += k*(src1[0] - src2[ 0]);
  }
  H = ( 17*H+16 ) >> 5;
  V = ( 17*V+16 ) >> 5;

  a = 16*(src1[0] + src2[8]+1) - 3*(V+H);
  for(j=8; j>0; --j) {
    int b = a;
    a += V;
    src[0] = cm[ (b    ) >> 5 ];
    src[1] = cm[ (b+  H) >> 5 ];
    src[2] = cm[ (b+2*H) >> 5 ];
    src[3] = cm[ (b+3*H) >> 5 ];
    src[4] = cm[ (b+4*H) >> 5 ];
    src[5] = cm[ (b+5*H) >> 5 ];
    src[6] = cm[ (b+6*H) >> 5 ];
    src[7] = cm[ (b+7*H) >> 5 ];
    src += stride;
  }
}

#define SRC(x,y) src[(x)+(y)*stride]
#define PL(y) \
    const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;
#define PREDICT_8x8_LOAD_LEFT \
    const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \
                     + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \
    PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \
    const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2

#define PT(x) \
    const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
#define PREDICT_8x8_LOAD_TOP \
    const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \
                     + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \
    PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \
    const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
                     + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2

#define PTR(x) \
    t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
#define PREDICT_8x8_LOAD_TOPRIGHT \
    int t8, t9, t10, t11, t12, t13, t14, t15; \
    if(has_topright) { \
        PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \
        t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \
    } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);

#define PREDICT_8x8_LOAD_TOPLEFT \
    const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2

#define PREDICT_8x8_DC(v) \
    int y; \
    for( y = 0; y < 8; y++ ) { \
        ((uint32_t*)src)[0] = \
        ((uint32_t*)src)[1] = v; \
        src += stride; \
    }