1 /*****************************************************************************
2 * deblock.c: h264 encoder library
3 *****************************************************************************
4 * Copyright (C) 2003-2008 x264 project
6 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
7 * Loren Merritt <lorenm@u.washington.edu>
8 * Fiona Glaser <fiona@x264.com>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
23 *****************************************************************************/
27 /* Deblocking filter */
28 static const uint8_t i_alpha_table[52+12*3] =
30 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
31 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
32 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
33 0, 0, 0, 0, 0, 0, 4, 4, 5, 6,
34 7, 8, 9, 10, 12, 13, 15, 17, 20, 22,
35 25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
36 80, 90,101,113,127,144,162,182,203,226,
38 255,255,255,255,255,255,255,255,255,255,255,255,
40 static const uint8_t i_beta_table[52+12*3] =
42 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
43 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
44 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
45 0, 0, 0, 0, 0, 0, 2, 2, 2, 3,
46 3, 3, 3, 4, 4, 4, 6, 6, 7, 7,
47 8, 8, 9, 9, 10, 10, 11, 11, 12, 12,
48 13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
50 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
52 static const int8_t i_tc0_table[52+12*3][4] =
54 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
55 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
56 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
57 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
58 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
59 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
60 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 1 },
61 {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 1, 1 }, {-1, 0, 1, 1 }, {-1, 1, 1, 1 },
62 {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 },
63 {-1, 1, 1, 2 }, {-1, 1, 2, 3 }, {-1, 1, 2, 3 }, {-1, 2, 2, 3 }, {-1, 2, 2, 4 }, {-1, 2, 3, 4 },
64 {-1, 2, 3, 4 }, {-1, 3, 3, 5 }, {-1, 3, 4, 6 }, {-1, 3, 4, 6 }, {-1, 4, 5, 7 }, {-1, 4, 5, 8 },
65 {-1, 4, 6, 9 }, {-1, 5, 7,10 }, {-1, 6, 8,11 }, {-1, 6, 8,13 }, {-1, 7,10,14 }, {-1, 8,11,16 },
66 {-1, 9,12,18 }, {-1,10,13,20 }, {-1,11,15,23 }, {-1,13,17,25 },
67 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
68 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
70 #define alpha_table(x) i_alpha_table[(x)+24]
71 #define beta_table(x) i_beta_table[(x)+24]
72 #define tc0_table(x) i_tc0_table[(x)+24]
75 static inline void deblock_luma_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
77 for( int i = 0; i < 4; i++ )
84 for( int d = 0; d < 4; d++ )
86 int p2 = pix[-3*xstride];
87 int p1 = pix[-2*xstride];
88 int p0 = pix[-1*xstride];
89 int q0 = pix[ 0*xstride];
90 int q1 = pix[ 1*xstride];
91 int q2 = pix[ 2*xstride];
93 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
97 if( abs( p2 - p0 ) < beta )
100 pix[-2*xstride] = p1 + x264_clip3( (( p2 + ((p0 + q0 + 1) >> 1)) >> 1) - p1, -tc0[i], tc0[i] );
103 if( abs( q2 - q0 ) < beta )
106 pix[ 1*xstride] = q1 + x264_clip3( (( q2 + ((p0 + q0 + 1) >> 1)) >> 1) - q1, -tc0[i], tc0[i] );
110 delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
111 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
112 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
118 static void deblock_v_luma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
120 deblock_luma_c( pix, stride, 1, alpha, beta, tc0 );
122 static void deblock_h_luma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
124 deblock_luma_c( pix, 1, stride, alpha, beta, tc0 );
127 static inline void deblock_chroma_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
129 for( int i = 0; i < 4; i++ )
137 for( int d = 0; d < 2; d++, pix += ystride-2 )
138 for( int e = 0; e < 2; e++, pix++ )
140 int p1 = pix[-2*xstride];
141 int p0 = pix[-1*xstride];
142 int q0 = pix[ 0*xstride];
143 int q1 = pix[ 1*xstride];
145 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
147 int delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
148 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
149 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
154 static void deblock_v_chroma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
156 deblock_chroma_c( pix, stride, 2, alpha, beta, tc0 );
158 static void deblock_h_chroma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
160 deblock_chroma_c( pix, 2, stride, alpha, beta, tc0 );
163 static inline void deblock_luma_intra_c( pixel *pix, int xstride, int ystride, int alpha, int beta )
165 for( int d = 0; d < 16; d++ )
167 int p2 = pix[-3*xstride];
168 int p1 = pix[-2*xstride];
169 int p0 = pix[-1*xstride];
170 int q0 = pix[ 0*xstride];
171 int q1 = pix[ 1*xstride];
172 int q2 = pix[ 2*xstride];
174 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
176 if(abs( p0 - q0 ) < ((alpha >> 2) + 2) )
178 if( abs( p2 - p0 ) < beta ) /* p0', p1', p2' */
180 const int p3 = pix[-4*xstride];
181 pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
182 pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
183 pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
186 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
187 if( abs( q2 - q0 ) < beta ) /* q0', q1', q2' */
189 const int q3 = pix[3*xstride];
190 pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
191 pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
192 pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
195 pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
199 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
200 pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
206 static void deblock_v_luma_intra_c( pixel *pix, int stride, int alpha, int beta )
208 deblock_luma_intra_c( pix, stride, 1, alpha, beta );
210 static void deblock_h_luma_intra_c( pixel *pix, int stride, int alpha, int beta )
212 deblock_luma_intra_c( pix, 1, stride, alpha, beta );
215 static inline void deblock_chroma_intra_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int dir )
217 for( int d = 0; d < (dir?16:8); d++, pix += ystride-2 )
218 for( int e = 0; e < (dir?1:2); e++, pix++ )
220 int p1 = pix[-2*xstride];
221 int p0 = pix[-1*xstride];
222 int q0 = pix[ 0*xstride];
223 int q1 = pix[ 1*xstride];
225 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
227 pix[-1*xstride] = (2*p1 + p0 + q1 + 2) >> 2; /* p0' */
228 pix[ 0*xstride] = (2*q1 + q0 + p1 + 2) >> 2; /* q0' */
232 static void deblock_v_chroma_intra_c( pixel *pix, int stride, int alpha, int beta )
234 deblock_chroma_intra_c( pix, stride, 2, alpha, beta, 1 );
236 static void deblock_h_chroma_intra_c( pixel *pix, int stride, int alpha, int beta )
238 deblock_chroma_intra_c( pix, 2, stride, alpha, beta, 0 );
241 static void deblock_strength_c( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
242 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4], int mvy_limit,
245 for( int dir = 0; dir < 2; dir++ )
247 int s1 = dir ? 1 : 8;
248 int s2 = dir ? 8 : 1;
249 for( int edge = 0; edge < 4; edge++ )
250 for( int i = 0, loc = X264_SCAN8_0+edge*s2; i < 4; i++, loc += s1 )
253 if( nnz[loc] || nnz[locn] )
254 bs[dir][edge][i] = 2;
255 else if( ref[0][loc] != ref[0][locn] ||
256 abs( mv[0][loc][0] - mv[0][locn][0] ) >= 4 ||
257 abs( mv[0][loc][1] - mv[0][locn][1] ) >= mvy_limit ||
258 (bframe && (ref[1][loc] != ref[1][locn] ||
259 abs( mv[1][loc][0] - mv[1][locn][0] ) >= 4 ||
260 abs( mv[1][loc][1] - mv[1][locn][1] ) >= mvy_limit )))
262 bs[dir][edge][i] = 1;
265 bs[dir][edge][i] = 0;
270 static inline void deblock_edge( x264_t *h, pixel *pix, int i_stride, uint8_t bS[4], int i_qp, int b_chroma, x264_deblock_inter_t pf_inter )
272 int index_a = i_qp-QP_BD_OFFSET + h->sh.i_alpha_c0_offset;
273 int index_b = i_qp-QP_BD_OFFSET + h->sh.i_beta_offset;
274 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
275 int beta = beta_table(index_b) << (BIT_DEPTH-8);
278 if( !M32(bS) || !alpha || !beta )
281 tc[0] = (tc0_table(index_a)[bS[0]] << (BIT_DEPTH-8)) + b_chroma;
282 tc[1] = (tc0_table(index_a)[bS[1]] << (BIT_DEPTH-8)) + b_chroma;
283 tc[2] = (tc0_table(index_a)[bS[2]] << (BIT_DEPTH-8)) + b_chroma;
284 tc[3] = (tc0_table(index_a)[bS[3]] << (BIT_DEPTH-8)) + b_chroma;
286 pf_inter( pix, i_stride, alpha, beta, tc );
289 static inline void deblock_edge_intra( x264_t *h, pixel *pix, int i_stride, uint8_t bS[4], int i_qp, int b_chroma, x264_deblock_intra_t pf_intra )
291 int index_a = i_qp-QP_BD_OFFSET + h->sh.i_alpha_c0_offset;
292 int index_b = i_qp-QP_BD_OFFSET + h->sh.i_beta_offset;
293 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
294 int beta = beta_table(index_b) << (BIT_DEPTH-8);
296 if( !alpha || !beta )
299 pf_intra( pix, i_stride, alpha, beta );
302 void x264_frame_deblock_row( x264_t *h, int mb_y )
304 int b_interlaced = h->sh.b_mbaff;
305 int qp_thresh = 15 - X264_MIN( h->sh.i_alpha_c0_offset, h->sh.i_beta_offset ) - X264_MAX( 0, h->param.analyse.i_chroma_qp_offset );
306 int stridey = h->fdec->i_stride[0];
307 int stride2y = stridey << b_interlaced;
308 int strideuv = h->fdec->i_stride[1];
309 int stride2uv = strideuv << b_interlaced;
311 for( int mb_x = 0; mb_x < h->mb.i_mb_width; mb_x += (~b_interlaced | mb_y)&1, mb_y ^= b_interlaced )
313 x264_prefetch_fenc( h, h->fdec, mb_x, mb_y );
314 x264_macroblock_cache_load_neighbours_deblock( h, mb_x, mb_y );
316 int mb_xy = h->mb.i_mb_xy;
317 int transform_8x8 = h->mb.mb_transform_size[h->mb.i_mb_xy];
318 int intra_cur = IS_INTRA( h->mb.type[mb_xy] );
319 uint8_t (*bs)[4][4] = h->deblock_strength[mb_y&1][mb_x];
321 pixel *pixy = h->fdec->plane[0] + 16*mb_y*stridey + 16*mb_x;
322 pixel *pixuv = h->fdec->plane[1] + 8*mb_y*strideuv + 16*mb_x;
323 if( mb_y & b_interlaced )
329 int qp = h->mb.qp[mb_xy];
330 int qpc = h->chroma_qp_table[qp];
331 int first_edge_only = h->mb.type[mb_xy] == P_SKIP || qp <= qp_thresh;
333 #define FILTER( intra, dir, edge, qp, chroma_qp )\
336 deblock_edge##intra( h, pixy + 4*edge*(dir?stride2y:1),\
337 stride2y, bs[dir][edge], qp, 0,\
338 h->loopf.deblock_luma##intra[dir] );\
340 deblock_edge##intra( h, pixuv + 2*edge*(dir?stride2uv:2),\
341 stride2uv, bs[dir][edge], chroma_qp, 1,\
342 h->loopf.deblock_chroma##intra[dir] );\
345 if( h->mb.i_neighbour & MB_LEFT )
347 int qpl = h->mb.qp[h->mb.i_mb_left_xy];
348 int qp_left = (qp + qpl + 1) >> 1;
349 int qpc_left = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpl] + 1) >> 1;
350 int intra_left = IS_INTRA( h->mb.type[h->mb.i_mb_left_xy] );
351 if( intra_cur || intra_left )
352 FILTER( _intra, 0, 0, qp_left, qpc_left );
354 FILTER( , 0, 0, qp_left, qpc_left );
357 if( !first_edge_only )
359 if( !transform_8x8 ) FILTER( , 0, 1, qp, qpc );
360 FILTER( , 0, 2, qp, qpc );
361 if( !transform_8x8 ) FILTER( , 0, 3, qp, qpc );
364 if( h->mb.i_neighbour & MB_TOP )
366 int qpt = h->mb.qp[h->mb.i_mb_top_xy];
367 int qp_top = (qp + qpt + 1) >> 1;
368 int qpc_top = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpt] + 1) >> 1;
369 int intra_top = IS_INTRA( h->mb.type[h->mb.i_mb_top_xy] );
370 if( ~b_interlaced & (intra_cur | intra_top) )
371 FILTER( _intra, 1, 0, qp_top, qpc_top );
375 M32( bs[1][0] ) = 0x03030303;
376 FILTER( , 1, 0, qp_top, qpc_top );
380 if( !first_edge_only )
382 if( !transform_8x8 ) FILTER( , 1, 1, qp, qpc );
383 FILTER( , 1, 2, qp, qpc );
384 if( !transform_8x8 ) FILTER( , 1, 3, qp, qpc );
391 /* For deblock-aware RD.
393 * deblock macroblock edges
394 * support analysis partitions smaller than 16x16
396 * handle duplicate refs correctly
397 * handle cavlc+8x8dct correctly
399 void x264_macroblock_deblock( x264_t *h )
401 int qp_thresh = 15 - X264_MIN( h->sh.i_alpha_c0_offset, h->sh.i_beta_offset ) - X264_MAX( 0, h->param.analyse.i_chroma_qp_offset );
403 if( qp <= qp_thresh || h->mb.i_type == P_SKIP )
406 uint8_t (*bs)[4][4] = h->deblock_strength[h->mb.i_mb_y&1][h->mb.i_mb_x];
407 if( IS_INTRA( h->mb.i_type ) )
408 memset( bs, 3, 2*4*4*sizeof(uint8_t) );
410 h->loopf.deblock_strength( h->mb.cache.non_zero_count, h->mb.cache.ref, h->mb.cache.mv,
411 bs, 4 >> h->sh.b_mbaff, h->sh.i_type == SLICE_TYPE_B );
413 int transform_8x8 = h->mb.b_transform_8x8;
414 pixel *fdec = h->mb.pic.p_fdec[0];
416 #define FILTER( dir, edge )\
419 deblock_edge( h, fdec + 4*edge*(dir?FDEC_STRIDE:1),\
420 FDEC_STRIDE, bs[dir][edge], qp, 0,\
421 h->loopf.deblock_luma[dir] );\
424 if( !transform_8x8 ) FILTER( 0, 1 );
426 if( !transform_8x8 ) FILTER( 0, 3 );
428 if( !transform_8x8 ) FILTER( 1, 1 );
430 if( !transform_8x8 ) FILTER( 1, 3 );
436 void x264_deblock_v_luma_sse2( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
437 void x264_deblock_h_luma_sse2( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
438 void x264_deblock_v_chroma_sse2( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
439 void x264_deblock_h_chroma_sse2( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
440 void x264_deblock_v_luma_intra_sse2( uint8_t *pix, int stride, int alpha, int beta );
441 void x264_deblock_h_luma_intra_sse2( uint8_t *pix, int stride, int alpha, int beta );
442 void x264_deblock_v_chroma_intra_sse2( uint8_t *pix, int stride, int alpha, int beta );
443 void x264_deblock_h_chroma_intra_sse2( uint8_t *pix, int stride, int alpha, int beta );
444 void x264_deblock_strength_mmxext( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
445 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
446 int mvy_limit, int bframe );
447 void x264_deblock_strength_sse2 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
448 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
449 int mvy_limit, int bframe );
450 void x264_deblock_strength_ssse3 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
451 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
452 int mvy_limit, int bframe );
454 void x264_deblock_h_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
455 void x264_deblock_v8_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
456 void x264_deblock_v_chroma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
457 void x264_deblock_h_chroma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
458 void x264_deblock_h_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
459 void x264_deblock_v8_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
460 void x264_deblock_v_chroma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
461 void x264_deblock_h_chroma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
463 // FIXME this wrapper has a significant cpu cost
464 static void x264_deblock_v_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
466 x264_deblock_v8_luma_mmxext( pix, stride, alpha, beta, tc0 );
467 x264_deblock_v8_luma_mmxext( pix+8, stride, alpha, beta, tc0+2 );
469 static void x264_deblock_v_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta )
471 x264_deblock_v8_luma_intra_mmxext( pix, stride, alpha, beta );
472 x264_deblock_v8_luma_intra_mmxext( pix+8, stride, alpha, beta );
478 void x264_deblock_v_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
479 void x264_deblock_h_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
483 void x264_deblock_v_luma_neon( uint8_t *, int, int, int, int8_t * );
484 void x264_deblock_h_luma_neon( uint8_t *, int, int, int, int8_t * );
485 void x264_deblock_v_chroma_neon( uint8_t *, int, int, int, int8_t * );
486 void x264_deblock_h_chroma_neon( uint8_t *, int, int, int, int8_t * );
489 void x264_deblock_init( int cpu, x264_deblock_function_t *pf )
491 pf->deblock_luma[1] = deblock_v_luma_c;
492 pf->deblock_luma[0] = deblock_h_luma_c;
493 pf->deblock_chroma[1] = deblock_v_chroma_c;
494 pf->deblock_chroma[0] = deblock_h_chroma_c;
495 pf->deblock_luma_intra[1] = deblock_v_luma_intra_c;
496 pf->deblock_luma_intra[0] = deblock_h_luma_intra_c;
497 pf->deblock_chroma_intra[1] = deblock_v_chroma_intra_c;
498 pf->deblock_chroma_intra[0] = deblock_h_chroma_intra_c;
499 pf->deblock_strength = deblock_strength_c;
502 if( cpu&X264_CPU_MMXEXT )
504 #if !X264_HIGH_BIT_DEPTH
506 pf->deblock_luma[1] = x264_deblock_v_luma_mmxext;
507 pf->deblock_luma[0] = x264_deblock_h_luma_mmxext;
508 pf->deblock_chroma[1] = x264_deblock_v_chroma_mmxext;
509 pf->deblock_chroma[0] = x264_deblock_h_chroma_mmxext;
510 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_mmxext;
511 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_mmxext;
512 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_mmxext;
513 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_mmxext;
515 #endif // !X264_HIGH_BIT_DEPTH
516 pf->deblock_strength = x264_deblock_strength_mmxext;
517 if( cpu&X264_CPU_SSE2 )
519 pf->deblock_strength = x264_deblock_strength_sse2;
520 #if !X264_HIGH_BIT_DEPTH
521 if( !(cpu&X264_CPU_STACK_MOD4) )
523 pf->deblock_luma[1] = x264_deblock_v_luma_sse2;
524 pf->deblock_luma[0] = x264_deblock_h_luma_sse2;
525 pf->deblock_chroma[1] = x264_deblock_v_chroma_sse2;
526 pf->deblock_chroma[0] = x264_deblock_h_chroma_sse2;
527 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_sse2;
528 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_sse2;
529 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_sse2;
530 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_sse2;
532 #endif // !X264_HIGH_BIT_DEPTH
534 if( cpu&X264_CPU_SSSE3 )
535 pf->deblock_strength = x264_deblock_strength_ssse3;
539 #if !X264_HIGH_BIT_DEPTH
541 if( cpu&X264_CPU_ALTIVEC )
543 pf->deblock_luma[1] = x264_deblock_v_luma_altivec;
544 pf->deblock_luma[0] = x264_deblock_h_luma_altivec;
546 #endif // HAVE_ALTIVEC
549 if( cpu&X264_CPU_NEON )
551 pf->deblock_luma[1] = x264_deblock_v_luma_neon;
552 pf->deblock_luma[0] = x264_deblock_h_luma_neon;
553 // pf->deblock_chroma[1] = x264_deblock_v_chroma_neon;
554 // pf->deblock_chroma[0] = x264_deblock_h_chroma_neon;
557 #endif // !X264_HIGH_BIT_DEPTH