1 /*****************************************************************************
2 * deblock.c: deblocking
3 *****************************************************************************
4 * Copyright (C) 2003-2011 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.
24 * This program is also available under a commercial proprietary license.
25 * For more information, contact us at licensing@x264.com.
26 *****************************************************************************/
30 /* Deblocking filter */
31 static const uint8_t i_alpha_table[52+12*3] =
33 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
34 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
35 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
36 0, 0, 0, 0, 0, 0, 4, 4, 5, 6,
37 7, 8, 9, 10, 12, 13, 15, 17, 20, 22,
38 25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
39 80, 90,101,113,127,144,162,182,203,226,
41 255,255,255,255,255,255,255,255,255,255,255,255,
43 static const uint8_t i_beta_table[52+12*3] =
45 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
46 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
47 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
48 0, 0, 0, 0, 0, 0, 2, 2, 2, 3,
49 3, 3, 3, 4, 4, 4, 6, 6, 7, 7,
50 8, 8, 9, 9, 10, 10, 11, 11, 12, 12,
51 13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
53 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
55 static const int8_t i_tc0_table[52+12*3][4] =
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, 0 },
61 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
62 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
63 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 1 },
64 {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 1, 1 }, {-1, 0, 1, 1 }, {-1, 1, 1, 1 },
65 {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 },
66 {-1, 1, 1, 2 }, {-1, 1, 2, 3 }, {-1, 1, 2, 3 }, {-1, 2, 2, 3 }, {-1, 2, 2, 4 }, {-1, 2, 3, 4 },
67 {-1, 2, 3, 4 }, {-1, 3, 3, 5 }, {-1, 3, 4, 6 }, {-1, 3, 4, 6 }, {-1, 4, 5, 7 }, {-1, 4, 5, 8 },
68 {-1, 4, 6, 9 }, {-1, 5, 7,10 }, {-1, 6, 8,11 }, {-1, 6, 8,13 }, {-1, 7,10,14 }, {-1, 8,11,16 },
69 {-1, 9,12,18 }, {-1,10,13,20 }, {-1,11,15,23 }, {-1,13,17,25 },
70 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
71 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
73 #define alpha_table(x) i_alpha_table[(x)+24]
74 #define beta_table(x) i_beta_table[(x)+24]
75 #define tc0_table(x) i_tc0_table[(x)+24]
78 static ALWAYS_INLINE void deblock_edge_luma_c( pixel *pix, int xstride, int alpha, int beta, int8_t tc0 )
80 int p2 = pix[-3*xstride];
81 int p1 = pix[-2*xstride];
82 int p0 = pix[-1*xstride];
83 int q0 = pix[ 0*xstride];
84 int q1 = pix[ 1*xstride];
85 int q2 = pix[ 2*xstride];
87 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
91 if( abs( p2 - p0 ) < beta )
94 pix[-2*xstride] = p1 + x264_clip3( (( p2 + ((p0 + q0 + 1) >> 1)) >> 1) - p1, -tc0, tc0 );
97 if( abs( q2 - q0 ) < beta )
100 pix[ 1*xstride] = q1 + x264_clip3( (( q2 + ((p0 + q0 + 1) >> 1)) >> 1) - q1, -tc0, tc0 );
104 delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
105 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
106 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
109 static inline void deblock_luma_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
111 for( int i = 0; i < 4; i++ )
118 for( int d = 0; d < 4; d++, pix += ystride )
119 deblock_edge_luma_c( pix, xstride, alpha, beta, tc0[i] );
122 static void deblock_v_luma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
124 deblock_luma_c( pix, stride, 1, alpha, beta, tc0 );
126 static void deblock_h_luma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
128 deblock_luma_c( pix, 1, stride, alpha, beta, tc0 );
131 static ALWAYS_INLINE void deblock_edge_chroma_c( pixel *pix, int xstride, int alpha, int beta, int8_t tc )
133 int p1 = pix[-2*xstride];
134 int p0 = pix[-1*xstride];
135 int q0 = pix[ 0*xstride];
136 int q1 = pix[ 1*xstride];
138 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
140 int delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
141 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
142 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
145 static inline void deblock_chroma_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
147 for( int i = 0; i < 4; i++ )
155 for( int d = 0; d < 2; d++, pix += ystride-2 )
156 for( int e = 0; e < 2; e++, pix++ )
157 deblock_edge_chroma_c( pix, xstride, alpha, beta, tc0[i] );
160 static void deblock_v_chroma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
162 deblock_chroma_c( pix, stride, 2, alpha, beta, tc0 );
164 static void deblock_h_chroma_c( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 )
166 deblock_chroma_c( pix, 2, stride, alpha, beta, tc0 );
169 static ALWAYS_INLINE void deblock_edge_luma_intra_c( pixel *pix, int xstride, int alpha, int beta )
171 int p2 = pix[-3*xstride];
172 int p1 = pix[-2*xstride];
173 int p0 = pix[-1*xstride];
174 int q0 = pix[ 0*xstride];
175 int q1 = pix[ 1*xstride];
176 int q2 = pix[ 2*xstride];
178 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
180 if( abs( p0 - q0 ) < ((alpha >> 2) + 2) )
182 if( abs( p2 - p0 ) < beta ) /* p0', p1', p2' */
184 const int p3 = pix[-4*xstride];
185 pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
186 pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
187 pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
190 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
191 if( abs( q2 - q0 ) < beta ) /* q0', q1', q2' */
193 const int q3 = pix[3*xstride];
194 pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
195 pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
196 pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
199 pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
203 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
204 pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
208 static inline void deblock_luma_intra_c( pixel *pix, int xstride, int ystride, int alpha, int beta )
210 for( int d = 0; d < 16; d++, pix += ystride )
211 deblock_edge_luma_intra_c( pix, xstride, alpha, beta );
213 static void deblock_v_luma_intra_c( pixel *pix, int stride, int alpha, int beta )
215 deblock_luma_intra_c( pix, stride, 1, alpha, beta );
217 static void deblock_h_luma_intra_c( pixel *pix, int stride, int alpha, int beta )
219 deblock_luma_intra_c( pix, 1, stride, alpha, beta );
222 static ALWAYS_INLINE void deblock_edge_chroma_intra_c( pixel *pix, int xstride, int alpha, int beta )
224 int p1 = pix[-2*xstride];
225 int p0 = pix[-1*xstride];
226 int q0 = pix[ 0*xstride];
227 int q1 = pix[ 1*xstride];
229 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
231 pix[-1*xstride] = (2*p1 + p0 + q1 + 2) >> 2; /* p0' */
232 pix[ 0*xstride] = (2*q1 + q0 + p1 + 2) >> 2; /* q0' */
235 static inline void deblock_chroma_intra_c( pixel *pix, int xstride, int ystride, int alpha, int beta, int dir )
237 for( int d = 0; d < (dir?16:8); d++, pix += ystride-2 )
238 for( int e = 0; e < (dir?1:2); e++, pix++ )
239 deblock_edge_chroma_intra_c( pix, xstride, alpha, beta );
241 static void deblock_v_chroma_intra_c( pixel *pix, int stride, int alpha, int beta )
243 deblock_chroma_intra_c( pix, stride, 2, alpha, beta, 1 );
245 static void deblock_h_chroma_intra_c( pixel *pix, int stride, int alpha, int beta )
247 deblock_chroma_intra_c( pix, 2, stride, alpha, beta, 0 );
250 static void deblock_strength_c( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
251 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4], int mvy_limit,
252 int bframe, x264_t *h )
254 for( int dir = 0; dir < 2; dir++ )
256 int s1 = dir ? 1 : 8;
257 int s2 = dir ? 8 : 1;
258 for( int edge = 0; edge < 4; edge++ )
259 for( int i = 0, loc = X264_SCAN8_0+edge*s2; i < 4; i++, loc += s1 )
262 if( nnz[loc] || nnz[locn] )
263 bs[dir][edge][i] = 2;
264 else if( ref[0][loc] != ref[0][locn] ||
265 abs( mv[0][loc][0] - mv[0][locn][0] ) >= 4 ||
266 abs( mv[0][loc][1] - mv[0][locn][1] ) >= mvy_limit ||
267 (bframe && (ref[1][loc] != ref[1][locn] ||
268 abs( mv[1][loc][0] - mv[1][locn][0] ) >= 4 ||
269 abs( mv[1][loc][1] - mv[1][locn][1] ) >= mvy_limit )))
271 bs[dir][edge][i] = 1;
274 bs[dir][edge][i] = 0;
278 void deblock_strength_mbaff_c( uint8_t nnz_cache[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
279 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
280 int mvy_limit, int bframe, x264_t *h )
282 int neighbour_field[2];
283 neighbour_field[0] = h->mb.i_mb_left_xy[0] >= 0 && h->mb.field[h->mb.i_mb_left_xy[0]];
284 neighbour_field[1] = h->mb.i_mb_top_xy >= 0 && h->mb.field[h->mb.i_mb_top_xy];
285 int intra_cur = IS_INTRA( h->mb.i_type );
289 for( int dir = 0; dir < 2; dir++ )
291 int edge_stride = dir ? 8 : 1;
292 int part_stride = dir ? 1 : 8;
293 for( int edge = 0; edge < 4; edge++ )
295 for( int i = 0, q = X264_SCAN8_0+edge*edge_stride; i < 4; i++, q += part_stride )
297 int p = q - edge_stride;
298 if( nnz_cache[q] || nnz_cache[p] )
300 bs[dir][edge][i] = 2;
302 else if( (edge == 0 && h->mb.b_interlaced != neighbour_field[dir]) ||
303 ref[0][q] != ref[0][p] ||
304 abs( mv[0][q][0] - mv[0][p][0] ) >= 4 ||
305 abs( mv[0][q][1] - mv[0][p][1] ) >= mvy_limit ||
306 (bframe && (ref[1][q] != ref[1][p] ||
307 abs( mv[1][q][0] - mv[1][p][0] ) >= 4 ||
308 abs( mv[1][q][1] - mv[1][p][1] ) >= mvy_limit )) )
310 bs[dir][edge][i] = 1;
313 bs[dir][edge][i] = 0;
320 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 )
322 int index_a = i_qp-QP_BD_OFFSET + h->sh.i_alpha_c0_offset;
323 int index_b = i_qp-QP_BD_OFFSET + h->sh.i_beta_offset;
324 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
325 int beta = beta_table(index_b) << (BIT_DEPTH-8);
328 if( !M32(bS) || !alpha || !beta )
331 tc[0] = (tc0_table(index_a)[bS[0]] << (BIT_DEPTH-8)) + b_chroma;
332 tc[1] = (tc0_table(index_a)[bS[1]] << (BIT_DEPTH-8)) + b_chroma;
333 tc[2] = (tc0_table(index_a)[bS[2]] << (BIT_DEPTH-8)) + b_chroma;
334 tc[3] = (tc0_table(index_a)[bS[3]] << (BIT_DEPTH-8)) + b_chroma;
336 pf_inter( pix, i_stride, alpha, beta, tc );
339 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 )
341 int index_a = i_qp-QP_BD_OFFSET + h->sh.i_alpha_c0_offset;
342 int index_b = i_qp-QP_BD_OFFSET + h->sh.i_beta_offset;
343 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
344 int beta = beta_table(index_b) << (BIT_DEPTH-8);
346 if( !alpha || !beta )
349 pf_intra( pix, i_stride, alpha, beta );
352 void x264_frame_deblock_row( x264_t *h, int mb_y )
354 int b_interlaced = h->sh.b_mbaff;
355 int qp_thresh = 15 - X264_MIN( h->sh.i_alpha_c0_offset, h->sh.i_beta_offset ) - X264_MAX( 0, h->pps->i_chroma_qp_index_offset );
356 int stridey = h->fdec->i_stride[0];
357 int strideuv = h->fdec->i_stride[1];
359 for( int mb_x = 0; mb_x < h->mb.i_mb_width; mb_x += (~b_interlaced | mb_y)&1, mb_y ^= b_interlaced )
361 x264_prefetch_fenc( h, h->fdec, mb_x, mb_y );
362 x264_macroblock_cache_load_neighbours_deblock( h, mb_x, mb_y );
364 int mb_xy = h->mb.i_mb_xy;
365 int transform_8x8 = h->mb.mb_transform_size[h->mb.i_mb_xy];
366 int intra_cur = IS_INTRA( h->mb.type[mb_xy] );
367 uint8_t (*bs)[4][4] = h->deblock_strength[mb_y&1][mb_x];
369 pixel *pixy = h->fdec->plane[0] + 16*mb_y*stridey + 16*mb_x;
370 pixel *pixuv = h->fdec->plane[1] + 8*mb_y*strideuv + 16*mb_x;
371 if( mb_y & h->mb.b_interlaced )
377 int stride2y = stridey << h->mb.b_interlaced;
378 int stride2uv = strideuv << h->mb.b_interlaced;
379 int qp = h->mb.qp[mb_xy];
380 int qpc = h->chroma_qp_table[qp];
381 int first_edge_only = h->mb.type[mb_xy] == P_SKIP || qp <= qp_thresh;
383 #define FILTER( intra, dir, edge, qp, chroma_qp )\
386 deblock_edge##intra( h, pixy + 4*edge*(dir?stride2y:1),\
387 stride2y, bs[dir][edge], qp, 0,\
388 h->loopf.deblock_luma##intra[dir] );\
390 deblock_edge##intra( h, pixuv + 2*edge*(dir?stride2uv:2),\
391 stride2uv, bs[dir][edge], chroma_qp, 1,\
392 h->loopf.deblock_chroma##intra[dir] );\
395 if( h->mb.i_neighbour & MB_LEFT )
397 int qpl = h->mb.qp[h->mb.i_mb_left_xy[0]];
398 int qp_left = (qp + qpl + 1) >> 1;
399 int qpc_left = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpl] + 1) >> 1;
400 int intra_left = IS_INTRA( h->mb.type[h->mb.i_mb_left_xy[0]] );
401 if( intra_cur || intra_left )
402 FILTER( _intra, 0, 0, qp_left, qpc_left );
404 FILTER( , 0, 0, qp_left, qpc_left );
407 if( !first_edge_only )
409 if( !transform_8x8 ) FILTER( , 0, 1, qp, qpc );
410 FILTER( , 0, 2, qp, qpc );
411 if( !transform_8x8 ) FILTER( , 0, 3, qp, qpc );
414 if( h->mb.i_neighbour & MB_TOP )
416 int qpt = h->mb.qp[h->mb.i_mb_top_xy];
417 int qp_top = (qp + qpt + 1) >> 1;
418 int qpc_top = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpt] + 1) >> 1;
419 int intra_top = IS_INTRA( h->mb.type[h->mb.i_mb_top_xy] );
420 if( ~b_interlaced & (intra_cur | intra_top) )
421 FILTER( _intra, 1, 0, qp_top, qpc_top );
425 M32( bs[1][0] ) = 0x03030303;
426 FILTER( , 1, 0, qp_top, qpc_top );
430 if( !first_edge_only )
432 if( !transform_8x8 ) FILTER( , 1, 1, qp, qpc );
433 FILTER( , 1, 2, qp, qpc );
434 if( !transform_8x8 ) FILTER( , 1, 3, qp, qpc );
441 /* For deblock-aware RD.
443 * deblock macroblock edges
444 * support analysis partitions smaller than 16x16
446 * handle duplicate refs correctly
447 * handle cavlc+8x8dct correctly
449 void x264_macroblock_deblock( x264_t *h )
451 int qp_thresh = 15 - X264_MIN( h->sh.i_alpha_c0_offset, h->sh.i_beta_offset ) - X264_MAX( 0, h->pps->i_chroma_qp_index_offset );
453 if( qp <= qp_thresh || h->mb.i_type == P_SKIP )
456 uint8_t (*bs)[4][4] = h->deblock_strength[h->mb.i_mb_y&1][h->mb.i_mb_x];
457 if( IS_INTRA( h->mb.i_type ) )
458 memset( bs, 3, 2*4*4*sizeof(uint8_t) );
460 h->loopf.deblock_strength( h->mb.cache.non_zero_count, h->mb.cache.ref, h->mb.cache.mv,
461 bs, 4 >> h->sh.b_mbaff, h->sh.i_type == SLICE_TYPE_B, h );
463 int transform_8x8 = h->mb.b_transform_8x8;
464 pixel *fdec = h->mb.pic.p_fdec[0];
466 #define FILTER( dir, edge )\
469 deblock_edge( h, fdec + 4*edge*(dir?FDEC_STRIDE:1),\
470 FDEC_STRIDE, bs[dir][edge], qp, 0,\
471 h->loopf.deblock_luma[dir] );\
474 if( !transform_8x8 ) FILTER( 0, 1 );
476 if( !transform_8x8 ) FILTER( 0, 3 );
478 if( !transform_8x8 ) FILTER( 1, 1 );
480 if( !transform_8x8 ) FILTER( 1, 3 );
486 void x264_deblock_v_luma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
487 void x264_deblock_v_luma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
488 void x264_deblock_h_luma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
489 void x264_deblock_h_luma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
490 void x264_deblock_v_chroma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
491 void x264_deblock_v_chroma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
492 void x264_deblock_h_chroma_sse2( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
493 void x264_deblock_h_chroma_avx ( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
494 void x264_deblock_v_luma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
495 void x264_deblock_v_luma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
496 void x264_deblock_h_luma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
497 void x264_deblock_h_luma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
498 void x264_deblock_v_chroma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
499 void x264_deblock_v_chroma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
500 void x264_deblock_h_chroma_intra_sse2( pixel *pix, int stride, int alpha, int beta );
501 void x264_deblock_h_chroma_intra_avx ( pixel *pix, int stride, int alpha, int beta );
502 void x264_deblock_strength_mmxext( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
503 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
504 int mvy_limit, int bframe, x264_t *h );
505 void x264_deblock_strength_sse2 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
506 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
507 int mvy_limit, int bframe, x264_t *h );
508 void x264_deblock_strength_ssse3 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
509 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
510 int mvy_limit, int bframe, x264_t *h );
511 void x264_deblock_strength_avx ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
512 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
513 int mvy_limit, int bframe, x264_t *h );
515 void x264_deblock_h_luma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
516 void x264_deblock_v8_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
517 void x264_deblock_v_chroma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
518 void x264_deblock_h_chroma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
519 void x264_deblock_h_luma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
520 void x264_deblock_v8_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
521 void x264_deblock_v_chroma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
522 void x264_deblock_h_chroma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
525 void x264_deblock_v_luma_mmxext( pixel *pix, int stride, int alpha, int beta, int8_t *tc0 );
526 void x264_deblock_v_luma_intra_mmxext( pixel *pix, int stride, int alpha, int beta );
528 // FIXME this wrapper has a significant cpu cost
529 static void x264_deblock_v_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
531 x264_deblock_v8_luma_mmxext( pix, stride, alpha, beta, tc0 );
532 x264_deblock_v8_luma_mmxext( pix+8, stride, alpha, beta, tc0+2 );
534 static void x264_deblock_v_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta )
536 x264_deblock_v8_luma_intra_mmxext( pix, stride, alpha, beta );
537 x264_deblock_v8_luma_intra_mmxext( pix+8, stride, alpha, beta );
539 #endif // HIGH_BIT_DEPTH
544 void x264_deblock_v_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
545 void x264_deblock_h_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
549 void x264_deblock_v_luma_neon( uint8_t *, int, int, int, int8_t * );
550 void x264_deblock_h_luma_neon( uint8_t *, int, int, int, int8_t * );
551 void x264_deblock_v_chroma_neon( uint8_t *, int, int, int, int8_t * );
552 void x264_deblock_h_chroma_neon( uint8_t *, int, int, int, int8_t * );
555 void x264_deblock_init( int cpu, x264_deblock_function_t *pf, int b_mbaff )
557 pf->deblock_luma[1] = deblock_v_luma_c;
558 pf->deblock_luma[0] = deblock_h_luma_c;
559 pf->deblock_chroma[1] = deblock_v_chroma_c;
560 pf->deblock_chroma[0] = deblock_h_chroma_c;
561 pf->deblock_luma_intra[1] = deblock_v_luma_intra_c;
562 pf->deblock_luma_intra[0] = deblock_h_luma_intra_c;
563 pf->deblock_chroma_intra[1] = deblock_v_chroma_intra_c;
564 pf->deblock_chroma_intra[0] = deblock_h_chroma_intra_c;
565 pf->deblock_strength = deblock_strength_c;
568 if( cpu&X264_CPU_MMXEXT )
571 pf->deblock_luma[1] = x264_deblock_v_luma_mmxext;
572 pf->deblock_luma[0] = x264_deblock_h_luma_mmxext;
573 pf->deblock_chroma[1] = x264_deblock_v_chroma_mmxext;
574 pf->deblock_chroma[0] = x264_deblock_h_chroma_mmxext;
575 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_mmxext;
576 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_mmxext;
577 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_mmxext;
578 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_mmxext;
580 pf->deblock_strength = x264_deblock_strength_mmxext;
581 if( cpu&X264_CPU_SSE2 )
583 pf->deblock_strength = x264_deblock_strength_sse2;
584 if( !(cpu&X264_CPU_STACK_MOD4) )
586 pf->deblock_luma[1] = x264_deblock_v_luma_sse2;
587 pf->deblock_luma[0] = x264_deblock_h_luma_sse2;
588 pf->deblock_chroma[1] = x264_deblock_v_chroma_sse2;
589 pf->deblock_chroma[0] = x264_deblock_h_chroma_sse2;
590 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_sse2;
591 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_sse2;
592 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_sse2;
593 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_sse2;
596 if( cpu&X264_CPU_SSSE3 )
597 pf->deblock_strength = x264_deblock_strength_ssse3;
598 if( cpu&X264_CPU_AVX )
600 pf->deblock_strength = x264_deblock_strength_avx;
601 if( !(cpu&X264_CPU_STACK_MOD4) )
603 pf->deblock_luma[1] = x264_deblock_v_luma_avx;
604 pf->deblock_luma[0] = x264_deblock_h_luma_avx;
605 pf->deblock_chroma[1] = x264_deblock_v_chroma_avx;
606 pf->deblock_chroma[0] = x264_deblock_h_chroma_avx;
607 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_avx;
608 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_avx;
609 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_avx;
610 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_avx;
618 if( cpu&X264_CPU_ALTIVEC )
620 pf->deblock_luma[1] = x264_deblock_v_luma_altivec;
621 pf->deblock_luma[0] = x264_deblock_h_luma_altivec;
623 #endif // HAVE_ALTIVEC
626 if( cpu&X264_CPU_NEON )
628 pf->deblock_luma[1] = x264_deblock_v_luma_neon;
629 pf->deblock_luma[0] = x264_deblock_h_luma_neon;
630 // pf->deblock_chroma[1] = x264_deblock_v_chroma_neon;
631 // pf->deblock_chroma[0] = x264_deblock_h_chroma_neon;
634 #endif // !HIGH_BIT_DEPTH
636 if( b_mbaff ) pf->deblock_strength = deblock_strength_mbaff_c;