/*****************************************************************************
* macroblock.h: h264 encoder library
*****************************************************************************
- * Copyright (C) 2003 Laurent Aimar
- * $Id: macroblock.h,v 1.1 2004/06/03 19:27:07 fenrir Exp $
+ * Copyright (C) 2005-2008 x264 project
*
- * Authors: Laurent Aimar <fenrir@via.ecp.fr>
+ * Authors: Loren Merritt <lorenm@u.washington.edu>
+ * Laurent Aimar <fenrir@via.ecp.fr>
+ * Fiona Glaser <fiona@x264.com>
*
* 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
*
* 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.
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
*****************************************************************************/
#ifndef X264_MACROBLOCK_H
/* XXX mb_type isn't the one written in the bitstream -> only internal usage */
-#define IS_INTRA(type) ( (type) == I_4x4 || (type) == I_8x8 || (type) == I_16x16 )
+#define IS_INTRA(type) ( (type) == I_4x4 || (type) == I_8x8 || (type) == I_16x16 || (type) == I_PCM )
#define IS_SKIP(type) ( (type) == P_SKIP || (type) == B_SKIP )
#define IS_DIRECT(type) ( (type) == B_DIRECT )
enum mb_class_e
B_DIRECT, B_L0_L0, B_L0_L1, B_L0_BI, B_L1_L0, B_L1_L1,
B_L1_BI, B_BI_L0, B_BI_L1, B_BI_BI, B_8x8, B_SKIP
};
-static const uint8_t x264_mb_type_list0_table[X264_MBTYPE_MAX][2] =
-{
- {0,0}, {0,0}, {0,0}, {0,0}, /* INTRA */
- {1,1}, /* P_L0 */
- {0,0}, /* P_8x8 */
- {1,1}, /* P_SKIP */
- {0,0}, /* B_DIRECT */
- {1,1}, {1,0}, {1,1}, /* B_L0_* */
- {0,1}, {0,0}, {0,1}, /* B_L1_* */
- {1,1}, {1,0}, {1,1}, /* B_BI_* */
- {0,0}, /* B_8x8 */
- {0,0} /* B_SKIP */
-};
-static const uint8_t x264_mb_type_list1_table[X264_MBTYPE_MAX][2] =
-{
- {0,0}, {0,0}, {0,0}, {0,0}, /* INTRA */
- {0,0}, /* P_L0 */
- {0,0}, /* P_8x8 */
- {0,0}, /* P_SKIP */
- {0,0}, /* B_DIRECT */
- {0,0}, {0,1}, {0,1}, /* B_L0_* */
- {1,0}, {1,1}, {1,1}, /* B_L1_* */
- {1,0}, {1,1}, {1,1}, /* B_BI_* */
- {0,0}, /* B_8x8 */
- {0,0} /* B_SKIP */
+static const uint8_t x264_mb_type_list_table[X264_MBTYPE_MAX][2][2] =
+{
+ {{0,0},{0,0}}, {{0,0},{0,0}}, {{0,0},{0,0}}, {{0,0},{0,0}}, /* INTRA */
+ {{1,1},{0,0}}, /* P_L0 */
+ {{0,0},{0,0}}, /* P_8x8 */
+ {{1,1},{0,0}}, /* P_SKIP */
+ {{0,0},{0,0}}, /* B_DIRECT */
+ {{1,1},{0,0}}, {{1,0},{0,1}}, {{1,1},{0,1}}, /* B_L0_* */
+ {{0,1},{1,0}}, {{0,0},{1,1}}, {{0,1},{1,1}}, /* B_L1_* */
+ {{1,1},{1,0}}, {{1,0},{1,1}}, {{1,1},{1,1}}, /* B_BI_* */
+ {{0,0},{0,0}}, /* B_8x8 */
+ {{0,0},{0,0}} /* B_SKIP */
};
#define IS_SUB4x4(type) ( (type ==D_L0_4x4)||(type ==D_L1_4x4)||(type ==D_BI_4x4))
enum mb_partition_e
{
/* sub partition type for P_8x8 and B_8x8 */
- D_L0_4x4 = 0,
- D_L0_8x4 = 1,
- D_L0_4x8 = 2,
- D_L0_8x8 = 3,
+ D_L0_4x4 = 0,
+ D_L0_8x4 = 1,
+ D_L0_4x8 = 2,
+ D_L0_8x8 = 3,
/* sub partition type for B_8x8 only */
- D_L1_4x4 = 4,
- D_L1_8x4 = 5,
- D_L1_4x8 = 6,
- D_L1_8x8 = 7,
+ D_L1_4x4 = 4,
+ D_L1_8x4 = 5,
+ D_L1_4x8 = 6,
+ D_L1_8x8 = 7,
- D_BI_4x4 = 8,
- D_BI_8x4 = 9,
- D_BI_4x8 = 10,
- D_BI_8x8 = 11,
- D_DIRECT_8x8 = 12,
+ D_BI_4x4 = 8,
+ D_BI_8x4 = 9,
+ D_BI_4x8 = 10,
+ D_BI_8x8 = 11,
+ D_DIRECT_8x8 = 12,
/* partition */
- D_8x8 = 13,
- D_16x8 = 14,
- D_8x16 = 15,
- D_16x16 = 16,
+ D_8x8 = 13,
+ D_16x8 = 14,
+ D_8x16 = 15,
+ D_16x16 = 16,
+ X264_PARTTYPE_MAX = 17,
};
static const uint8_t x264_mb_partition_listX_table[2][17] =
{ 4, 6, 12, 14 },
{ 5, 7, 13, 15 }
};
+static const uint8_t block_idx_xy_1d[16] =
+{
+ 0, 1, 4, 5, 2, 3, 6, 7, 8, 9, 12, 13, 10, 11, 14, 15
+};
+static const uint8_t block_idx_yx_1d[16] =
+{
+ 0, 4, 1, 5, 8, 12, 9, 13, 2, 6, 3, 7, 10, 14, 11, 15
+};
+static const uint8_t block_idx_xy_fenc[16] =
+{
+ 0*4 + 0*4*FENC_STRIDE, 1*4 + 0*4*FENC_STRIDE,
+ 0*4 + 1*4*FENC_STRIDE, 1*4 + 1*4*FENC_STRIDE,
+ 2*4 + 0*4*FENC_STRIDE, 3*4 + 0*4*FENC_STRIDE,
+ 2*4 + 1*4*FENC_STRIDE, 3*4 + 1*4*FENC_STRIDE,
+ 0*4 + 2*4*FENC_STRIDE, 1*4 + 2*4*FENC_STRIDE,
+ 0*4 + 3*4*FENC_STRIDE, 1*4 + 3*4*FENC_STRIDE,
+ 2*4 + 2*4*FENC_STRIDE, 3*4 + 2*4*FENC_STRIDE,
+ 2*4 + 3*4*FENC_STRIDE, 3*4 + 3*4*FENC_STRIDE
+};
+static const uint16_t block_idx_xy_fdec[16] =
+{
+ 0*4 + 0*4*FDEC_STRIDE, 1*4 + 0*4*FDEC_STRIDE,
+ 0*4 + 1*4*FDEC_STRIDE, 1*4 + 1*4*FDEC_STRIDE,
+ 2*4 + 0*4*FDEC_STRIDE, 3*4 + 0*4*FDEC_STRIDE,
+ 2*4 + 1*4*FDEC_STRIDE, 3*4 + 1*4*FDEC_STRIDE,
+ 0*4 + 2*4*FDEC_STRIDE, 1*4 + 2*4*FDEC_STRIDE,
+ 0*4 + 3*4*FDEC_STRIDE, 1*4 + 3*4*FDEC_STRIDE,
+ 2*4 + 2*4*FDEC_STRIDE, 3*4 + 2*4*FDEC_STRIDE,
+ 2*4 + 3*4*FDEC_STRIDE, 3*4 + 3*4*FDEC_STRIDE
+};
-static const uint8_t i_chroma_qp_table[52] =
+static const uint8_t i_chroma_qp_table[52+12*2] =
{
+ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 1, 2, 3, 4, 5, 6, 7, 8, 9,
10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
20, 21, 22, 23, 24, 25, 26, 27, 28, 29,
29, 30, 31, 32, 32, 33, 34, 34, 35, 35,
36, 36, 37, 37, 37, 38, 38, 38, 39, 39,
- 39, 39
+ 39, 39,
+ 39, 39, 39, 39, 39, 39, 39, 39, 39, 39, 39, 39,
};
enum cabac_ctx_block_cat_e
* h->mb. need only valid values from other blocks */
void x264_mb_predict_mv_ref16x16( x264_t *h, int i_list, int i_ref, int16_t mvc[8][2], int *i_mvc );
-
-int x264_mb_predict_intra4x4_mode( x264_t *h, int idx );
-int x264_mb_predict_non_zero_code( x264_t *h, int idx );
-
-/* x264_mb_transform_8x8_allowed:
- * check whether any partition is smaller than 8x8 (or at least
- * might be, according to just partition type.)
- * doesn't check for cbp */
-int x264_mb_transform_8x8_allowed( x264_t *h );
-
void x264_mb_mc( x264_t *h );
void x264_mb_mc_8x8( x264_t *h, int i8 );
+static ALWAYS_INLINE uint32_t pack16to32( int a, int b )
+{
+#ifdef WORDS_BIGENDIAN
+ return b + (a<<16);
+#else
+ return a + (b<<16);
+#endif
+}
+static ALWAYS_INLINE uint32_t pack8to16( int a, int b )
+{
+#ifdef WORDS_BIGENDIAN
+ return b + (a<<8);
+#else
+ return a + (b<<8);
+#endif
+}
+static ALWAYS_INLINE uint32_t pack8to32( int a, int b, int c, int d )
+{
+#ifdef WORDS_BIGENDIAN
+ return d + (c<<8) + (b<<16) + (a<<24);
+#else
+ return a + (b<<8) + (c<<16) + (d<<24);
+#endif
+}
static ALWAYS_INLINE uint32_t pack16to32_mask( int a, int b )
{
#ifdef WORDS_BIGENDIAN
}
static ALWAYS_INLINE void x264_macroblock_cache_rect1( void *dst, int width, int height, uint8_t val )
{
- int dy;
if( width == 4 )
{
uint32_t val2 = val * 0x01010101;
- for( dy = 0; dy < height; dy++ )
- ((uint32_t*)dst)[2*dy] = val2;
+ ((uint32_t*)dst)[0] = val2;
+ if( height >= 2 ) ((uint32_t*)dst)[2] = val2;
+ if( height == 4 ) ((uint32_t*)dst)[4] = val2;
+ if( height == 4 ) ((uint32_t*)dst)[6] = val2;
}
else // 2
{
uint32_t val2 = val * 0x0101;
- for( dy = 0; dy < height; dy++ )
- ((uint16_t*)dst)[4*dy] = val2;
+ ((uint16_t*)dst)[ 0] = val2;
+ if( height >= 2 ) ((uint16_t*)dst)[ 4] = val2;
+ if( height == 4 ) ((uint16_t*)dst)[ 8] = val2;
+ if( height == 4 ) ((uint16_t*)dst)[12] = val2;
}
}
static ALWAYS_INLINE void x264_macroblock_cache_rect4( void *dst, int width, int height, uint32_t val )
{
- int dy, dx;
+ int dy;
if( width == 1 || WORD_SIZE < 8 )
{
for( dy = 0; dy < height; dy++ )
- for( dx = 0; dx < width; dx++ )
- ((uint32_t*)dst)[dx+8*dy] = val;
+ {
+ ((uint32_t*)dst)[8*dy+0] = val;
+ if( width >= 2 ) ((uint32_t*)dst)[8*dy+1] = val;
+ if( width == 4 ) ((uint32_t*)dst)[8*dy+2] = val;
+ if( width == 4 ) ((uint32_t*)dst)[8*dy+3] = val;
+ }
}
else
{
uint64_t val64 = val + ((uint64_t)val<<32);
for( dy = 0; dy < height; dy++ )
- for( dx = 0; dx < width/2; dx++ )
- ((uint64_t*)dst)[dx+4*dy] = val64;
+ {
+ ((uint64_t*)dst)[4*dy+0] = val64;
+ if( width == 4 ) ((uint64_t*)dst)[4*dy+1] = val64;
+ }
}
}
#define x264_macroblock_cache_mv_ptr(a,x,y,w,h,l,mv) x264_macroblock_cache_mv(a,x,y,w,h,l,*(uint32_t*)mv)
#define array_non_zero_int array_non_zero_int_c
static ALWAYS_INLINE int array_non_zero_int_c( void *v, int i_count )
{
- uint64_t *x = v;
+ union {uint16_t s[4]; uint64_t l;} *x = v;
if(i_count == 8)
- return !!x[0];
+ return !!x[0].l;
else if(i_count == 16)
- return !!(x[0]|x[1]);
+ return !!(x[0].l|x[1].l);
else if(i_count == 32)
- return !!(x[0]|x[1]|x[2]|x[3]);
+ return !!(x[0].l|x[1].l|x[2].l|x[3].l);
else
{
int i;
i_count /= sizeof(uint64_t);
for( i = 0; i < i_count; i++ )
- if( x[i] ) return 1;
+ if( x[i].l ) return 1;
return 0;
}
}
-/* This function and its MMX version only work on arrays of size 16 */
-static ALWAYS_INLINE int array_non_zero_count( int16_t *v )
+static inline int x264_mb_predict_intra4x4_mode( x264_t *h, int idx )
+{
+ const int ma = h->mb.cache.intra4x4_pred_mode[x264_scan8[idx] - 1];
+ const int mb = h->mb.cache.intra4x4_pred_mode[x264_scan8[idx] - 8];
+ const int m = X264_MIN( x264_mb_pred_mode4x4_fix(ma),
+ x264_mb_pred_mode4x4_fix(mb) );
+
+ if( m < 0 )
+ return I_PRED_4x4_DC;
+
+ return m;
+}
+static inline int x264_mb_predict_non_zero_code( x264_t *h, int idx )
{
- int i;
- int i_nz;
+ const int za = h->mb.cache.non_zero_count[x264_scan8[idx] - 1];
+ const int zb = h->mb.cache.non_zero_count[x264_scan8[idx] - 8];
- for( i = 0, i_nz = 0; i < 16; i++ )
- if( v[i] )
- i_nz++;
+ int i_ret = za + zb;
- return i_nz;
+ if( i_ret < 0x80 )
+ {
+ i_ret = ( i_ret + 1 ) >> 1;
+ }
+ return i_ret & 0x7f;
+}
+/* x264_mb_transform_8x8_allowed:
+ * check whether any partition is smaller than 8x8 (or at least
+ * might be, according to just partition type.)
+ * doesn't check for cbp */
+static inline int x264_mb_transform_8x8_allowed( x264_t *h )
+{
+ // intra and skip are disallowed
+ // large partitions are allowed
+ // direct and 8x8 are conditional
+ static const uint8_t partition_tab[X264_MBTYPE_MAX] = {
+ 0,0,0,0,1,2,0,1,1,1,1,1,1,1,1,1,1,1,0,
+ };
+
+ if( !h->pps->b_transform_8x8_mode )
+ return 0;
+ if( h->mb.i_type != P_8x8 )
+ return partition_tab[h->mb.i_type];
+ return *(uint32_t*)h->mb.i_sub_partition == D_L0_8x8*0x01010101;
}
#endif
projects / x264 / blobdiff