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 /* cavlc + 8x8 transform stores nnz per 16 coeffs for the purpose of
28 * entropy coding, but per 64 coeffs for the purpose of deblocking */
29 static void munge_cavlc_nnz_row( x264_t *h, int mb_y, uint8_t (*buf)[16] )
31 uint32_t (*src)[6] = (uint32_t(*)[6])h->mb.non_zero_count + mb_y * h->sps->i_mb_width;
32 int8_t *transform = h->mb.mb_transform_size + mb_y * h->sps->i_mb_width;
33 for( int x = 0; x<h->sps->i_mb_width; x++ )
35 memcpy( buf+x, src+x, 16 );
38 int nnz = src[x][0] | src[x][1];
39 src[x][0] = src[x][1] = ((uint16_t)nnz ? 0x0101 : 0) + (nnz>>16 ? 0x01010000 : 0);
40 nnz = src[x][2] | src[x][3];
41 src[x][2] = src[x][3] = ((uint16_t)nnz ? 0x0101 : 0) + (nnz>>16 ? 0x01010000 : 0);
46 static void restore_cavlc_nnz_row( x264_t *h, int mb_y, uint8_t (*buf)[16] )
48 uint8_t (*dst)[24] = h->mb.non_zero_count + mb_y * h->sps->i_mb_width;
49 for( int x = 0; x < h->sps->i_mb_width; x++ )
50 memcpy( dst+x, buf+x, 16 );
53 static void munge_cavlc_nnz( x264_t *h, int mb_y, uint8_t (*buf)[16], void (*func)(x264_t*, int, uint8_t (*)[16]) )
57 func( h, mb_y-1, buf + h->sps->i_mb_width );
60 func( h, mb_y+1, buf + h->sps->i_mb_width * 2 );
62 func( h, mb_y-2, buf + h->sps->i_mb_width * 3 );
67 /* Deblocking filter */
68 static const uint8_t i_alpha_table[52+12*2] =
70 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
71 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
72 0, 0, 0, 0, 0, 0, 4, 4, 5, 6,
73 7, 8, 9, 10, 12, 13, 15, 17, 20, 22,
74 25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
75 80, 90,101,113,127,144,162,182,203,226,
77 255,255,255,255,255,255,255,255,255,255,255,255,
79 static const uint8_t i_beta_table[52+12*2] =
81 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
82 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
83 0, 0, 0, 0, 0, 0, 2, 2, 2, 3,
84 3, 3, 3, 4, 4, 4, 6, 6, 7, 7,
85 8, 8, 9, 9, 10, 10, 11, 11, 12, 12,
86 13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
88 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
90 static const int8_t i_tc0_table[52+12*2][4] =
92 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
93 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
94 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
95 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 },
96 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 1 },
97 {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 1, 1 }, {-1, 0, 1, 1 }, {-1, 1, 1, 1 },
98 {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 },
99 {-1, 1, 1, 2 }, {-1, 1, 2, 3 }, {-1, 1, 2, 3 }, {-1, 2, 2, 3 }, {-1, 2, 2, 4 }, {-1, 2, 3, 4 },
100 {-1, 2, 3, 4 }, {-1, 3, 3, 5 }, {-1, 3, 4, 6 }, {-1, 3, 4, 6 }, {-1, 4, 5, 7 }, {-1, 4, 5, 8 },
101 {-1, 4, 6, 9 }, {-1, 5, 7,10 }, {-1, 6, 8,11 }, {-1, 6, 8,13 }, {-1, 7,10,14 }, {-1, 8,11,16 },
102 {-1, 9,12,18 }, {-1,10,13,20 }, {-1,11,15,23 }, {-1,13,17,25 },
103 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
104 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
106 #define alpha_table(x) i_alpha_table[(x)+12]
107 #define beta_table(x) i_beta_table[(x)+12]
108 #define tc0_table(x) i_tc0_table[(x)+12]
111 static inline void deblock_luma_c( uint8_t *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
113 for( int i = 0; i < 4; i++ )
120 for( int d = 0; d < 4; d++ )
122 int p2 = pix[-3*xstride];
123 int p1 = pix[-2*xstride];
124 int p0 = pix[-1*xstride];
125 int q0 = pix[ 0*xstride];
126 int q1 = pix[ 1*xstride];
127 int q2 = pix[ 2*xstride];
129 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
133 if( abs( p2 - p0 ) < beta )
136 pix[-2*xstride] = p1 + x264_clip3( (( p2 + ((p0 + q0 + 1) >> 1)) >> 1) - p1, -tc0[i], tc0[i] );
139 if( abs( q2 - q0 ) < beta )
142 pix[ 1*xstride] = q1 + x264_clip3( (( q2 + ((p0 + q0 + 1) >> 1)) >> 1) - q1, -tc0[i], tc0[i] );
146 delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
147 pix[-1*xstride] = x264_clip_uint8( p0 + delta ); /* p0' */
148 pix[ 0*xstride] = x264_clip_uint8( q0 - delta ); /* q0' */
154 static void deblock_v_luma_c( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
156 deblock_luma_c( pix, stride, 1, alpha, beta, tc0 );
158 static void deblock_h_luma_c( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
160 deblock_luma_c( pix, 1, stride, alpha, beta, tc0 );
163 static inline void deblock_chroma_c( uint8_t *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0 )
165 for( int i = 0; i < 4; i++ )
173 for( int d = 0; d < 2; d++ )
175 int p1 = pix[-2*xstride];
176 int p0 = pix[-1*xstride];
177 int q0 = pix[ 0*xstride];
178 int q1 = pix[ 1*xstride];
180 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
182 int delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
183 pix[-1*xstride] = x264_clip_uint8( p0 + delta ); /* p0' */
184 pix[ 0*xstride] = x264_clip_uint8( q0 - delta ); /* q0' */
190 static void deblock_v_chroma_c( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
192 deblock_chroma_c( pix, stride, 1, alpha, beta, tc0 );
194 static void deblock_h_chroma_c( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
196 deblock_chroma_c( pix, 1, stride, alpha, beta, tc0 );
199 static inline void deblock_luma_intra_c( uint8_t *pix, int xstride, int ystride, int alpha, int beta )
201 for( int d = 0; d < 16; d++ )
203 int p2 = pix[-3*xstride];
204 int p1 = pix[-2*xstride];
205 int p0 = pix[-1*xstride];
206 int q0 = pix[ 0*xstride];
207 int q1 = pix[ 1*xstride];
208 int q2 = pix[ 2*xstride];
210 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
212 if(abs( p0 - q0 ) < ((alpha >> 2) + 2) )
214 if( abs( p2 - p0 ) < beta ) /* p0', p1', p2' */
216 const int p3 = pix[-4*xstride];
217 pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
218 pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
219 pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
222 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
223 if( abs( q2 - q0 ) < beta ) /* q0', q1', q2' */
225 const int q3 = pix[3*xstride];
226 pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
227 pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
228 pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
231 pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
235 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
236 pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
242 static void deblock_v_luma_intra_c( uint8_t *pix, int stride, int alpha, int beta )
244 deblock_luma_intra_c( pix, stride, 1, alpha, beta );
246 static void deblock_h_luma_intra_c( uint8_t *pix, int stride, int alpha, int beta )
248 deblock_luma_intra_c( pix, 1, stride, alpha, beta );
251 static inline void deblock_chroma_intra_c( uint8_t *pix, int xstride, int ystride, int alpha, int beta )
253 for( int d = 0; d < 8; d++ )
255 int p1 = pix[-2*xstride];
256 int p0 = pix[-1*xstride];
257 int q0 = pix[ 0*xstride];
258 int q1 = pix[ 1*xstride];
260 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
262 pix[-1*xstride] = (2*p1 + p0 + q1 + 2) >> 2; /* p0' */
263 pix[ 0*xstride] = (2*q1 + q0 + p1 + 2) >> 2; /* q0' */
268 static void deblock_v_chroma_intra_c( uint8_t *pix, int stride, int alpha, int beta )
270 deblock_chroma_intra_c( pix, stride, 1, alpha, beta );
272 static void deblock_h_chroma_intra_c( uint8_t *pix, int stride, int alpha, int beta )
274 deblock_chroma_intra_c( pix, 1, stride, alpha, beta );
277 static void deblock_strength_c( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE], int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4], int mvy_limit, int bframe, int step, int first_edge_only )
279 for( int dir = 0; dir < 2; dir++ )
281 int s1 = dir ? 1 : 8;
282 int s2 = dir ? 8 : 1;
283 for( int edge = 0; edge < (first_edge_only ? 1 : 4); edge += step )
284 for( int i = 0, loc = X264_SCAN8_0+edge*s2; i < 4; i++, loc += s1 )
287 if( nnz[loc] || nnz[locn] )
288 bs[dir][edge][i] = 2;
289 else if( ref[0][loc] != ref[0][locn] ||
290 abs( mv[0][loc][0] - mv[0][locn][0] ) >= 4 ||
291 abs( mv[0][loc][1] - mv[0][locn][1] ) >= mvy_limit ||
292 (bframe && (ref[1][loc] != ref[1][locn] ||
293 abs( mv[1][loc][0] - mv[1][locn][0] ) >= 4 ||
294 abs( mv[1][loc][1] - mv[1][locn][1] ) >= mvy_limit )))
296 bs[dir][edge][i] = 1;
299 bs[dir][edge][i] = 0;
304 static inline void deblock_edge( x264_t *h, uint8_t *pix1, uint8_t *pix2, int i_stride, uint8_t bS[4], int i_qp, int b_chroma, x264_deblock_inter_t pf_inter )
306 int index_a = i_qp + h->sh.i_alpha_c0_offset;
307 int alpha = alpha_table(index_a);
308 int beta = beta_table(i_qp + h->sh.i_beta_offset);
311 if( !M32(bS) || !alpha || !beta )
314 tc[0] = tc0_table(index_a)[bS[0]] + b_chroma;
315 tc[1] = tc0_table(index_a)[bS[1]] + b_chroma;
316 tc[2] = tc0_table(index_a)[bS[2]] + b_chroma;
317 tc[3] = tc0_table(index_a)[bS[3]] + b_chroma;
319 pf_inter( pix1, i_stride, alpha, beta, tc );
321 pf_inter( pix2, i_stride, alpha, beta, tc );
324 static inline void deblock_edge_intra( x264_t *h, uint8_t *pix1, uint8_t *pix2, int i_stride, uint8_t bS[4], int i_qp, int b_chroma, x264_deblock_intra_t pf_intra )
326 int alpha = alpha_table(i_qp + h->sh.i_alpha_c0_offset);
327 int beta = beta_table(i_qp + h->sh.i_beta_offset);
329 if( !alpha || !beta )
332 pf_intra( pix1, i_stride, alpha, beta );
334 pf_intra( pix2, i_stride, alpha, beta );
337 void x264_frame_deblock_row( x264_t *h, int mb_y )
339 int b_interlaced = h->sh.b_mbaff;
340 int mvy_limit = 4 >> b_interlaced;
341 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);
342 int stridey = h->fdec->i_stride[0];
343 int stride2y = stridey << b_interlaced;
344 int strideuv = h->fdec->i_stride[1];
345 int stride2uv = strideuv << b_interlaced;
346 int deblock_ref_table[2][32+2];
347 uint8_t (*nnz_backup)[16] = h->scratch_buffer;
349 for( int l = 0; l < 2; l++ )
351 int refs = (l ? h->i_ref1 : h->i_ref0) << h->sh.b_mbaff;
352 x264_frame_t **fref = l ? h->fref1 : h->fref0;
353 deblock_ref_table(l,-2) = -2;
354 deblock_ref_table(l,-1) = -1;
355 for( int i = 0; i < refs; i++ )
357 /* Mask off high bits to avoid frame num collisions with -1/-2.
358 * frame num values don't actually have to be correct, just unique.
359 * frame num values can't cover a range of more than 32. */
360 if( !h->mb.b_interlaced )
361 deblock_ref_table(l,i) = fref[i]->i_frame_num&63;
363 deblock_ref_table(l,i) = ((fref[i>>1]->i_frame_num&63)<<1) + (i&1);
367 if( !h->pps->b_cabac && h->pps->b_transform_8x8_mode )
368 munge_cavlc_nnz( h, mb_y, nnz_backup, munge_cavlc_nnz_row );
370 for( int mb_x = 0; mb_x < h->sps->i_mb_width; mb_x += (~b_interlaced | mb_y)&1, mb_y ^= b_interlaced )
372 ALIGNED_ARRAY_16( uint8_t, bs, [2][4][4] );
374 x264_prefetch_fenc( h, h->fdec, mb_x, mb_y );
375 x264_macroblock_cache_load_deblock( h, mb_x, mb_y, deblock_ref_table );
377 int mb_xy = h->mb.i_mb_xy;
378 int transform_8x8 = h->mb.mb_transform_size[mb_xy];
379 int intra_cur = IS_INTRA( h->mb.type[mb_xy] );
381 uint8_t *pixy = h->fdec->plane[0] + 16*mb_y*stridey + 16*mb_x;
382 uint8_t *pixu = h->fdec->plane[1] + 8*mb_y*strideuv + 8*mb_x;
383 uint8_t *pixv = h->fdec->plane[2] + 8*mb_y*strideuv + 8*mb_x;
384 if( b_interlaced && (mb_y&1) )
391 int qp = h->mb.qp[mb_xy];
392 int qpc = h->chroma_qp_table[qp];
393 int first_edge_only = h->mb.type[mb_xy] == P_SKIP || qp <= qp_thresh;
395 #define FILTER( intra, dir, edge, qp, chroma_qp )\
398 deblock_edge##intra( h, pixy + 4*edge*(dir?stride2y:1), NULL,\
399 stride2y, bs[dir][edge], qp, 0,\
400 h->loopf.deblock_luma##intra[dir] );\
402 deblock_edge##intra( h, pixu + 2*edge*(dir?stride2uv:1), pixv + 2*edge*(dir?stride2uv:1),\
403 stride2uv, bs[dir][edge], chroma_qp, 1,\
404 h->loopf.deblock_chroma##intra[dir] );\
408 memset( bs, 3, sizeof(bs) );
410 h->loopf.deblock_strength( h->mb.cache.non_zero_count, h->mb.cache.ref, h->mb.cache.mv, bs, mvy_limit, h->sh.i_type == SLICE_TYPE_B, transform_8x8 + 1, first_edge_only );
412 if( h->mb.i_neighbour & MB_LEFT )
414 int qpl = h->mb.qp[h->mb.i_mb_left_xy];
415 int qp_left = (qp + qpl + 1) >> 1;
416 int qpc_left = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpl] + 1) >> 1;
417 int intra_left = IS_INTRA( h->mb.type[h->mb.i_mb_left_xy] );
418 if( intra_cur || intra_left )
419 FILTER( _intra, 0, 0, qp_left, qpc_left );
421 FILTER( , 0, 0, qp_left, qpc_left );
424 if( !first_edge_only )
426 if( !transform_8x8 ) FILTER( , 0, 1, qp, qpc );
427 FILTER( , 0, 2, qp, qpc );
428 if( !transform_8x8 ) FILTER( , 0, 3, qp, qpc );
431 if( h->mb.i_neighbour & MB_TOP )
433 int qpt = h->mb.qp[h->mb.i_mb_top_xy];
434 int qp_top = (qp + qpt + 1) >> 1;
435 int qpc_top = (h->chroma_qp_table[qp] + h->chroma_qp_table[qpt] + 1) >> 1;
436 int intra_top = IS_INTRA( h->mb.type[h->mb.i_mb_top_xy] );
437 if( !b_interlaced && (intra_cur || intra_top) )
438 FILTER( _intra, 1, 0, qp_top, qpc_top );
442 memset( bs[1][0], 3, sizeof(bs[1][0]) );
443 FILTER( , 1, 0, qp_top, qpc_top );
447 if( !first_edge_only )
449 if( !transform_8x8 ) FILTER( , 1, 1, qp, qpc );
450 FILTER( , 1, 2, qp, qpc );
451 if( !transform_8x8 ) FILTER( , 1, 3, qp, qpc );
455 if( !h->pps->b_cabac && h->pps->b_transform_8x8_mode )
456 munge_cavlc_nnz( h, mb_y, nnz_backup, restore_cavlc_nnz_row );
460 void x264_deblock_v_chroma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
461 void x264_deblock_h_chroma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
462 void x264_deblock_v_chroma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
463 void x264_deblock_h_chroma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
465 void x264_deblock_v_luma_sse2( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
466 void x264_deblock_h_luma_sse2( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
467 void x264_deblock_v_luma_intra_sse2( uint8_t *pix, int stride, int alpha, int beta );
468 void x264_deblock_h_luma_intra_sse2( uint8_t *pix, int stride, int alpha, int beta );
469 void x264_deblock_strength_mmxext( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
470 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
471 int mvy_limit, int bframe, int step, int first_edge_only );
472 void x264_deblock_strength_sse2 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
473 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
474 int mvy_limit, int bframe, int step, int first_edge_only );
475 void x264_deblock_strength_ssse3 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
476 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][4][4],
477 int mvy_limit, int bframe, int step, int first_edge_only );
479 void x264_deblock_h_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
480 void x264_deblock_v8_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
481 void x264_deblock_h_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
482 void x264_deblock_v8_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta );
484 static void x264_deblock_v_luma_mmxext( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
486 x264_deblock_v8_luma_mmxext( pix, stride, alpha, beta, tc0 );
487 x264_deblock_v8_luma_mmxext( pix+8, stride, alpha, beta, tc0+2 );
489 static void x264_deblock_v_luma_intra_mmxext( uint8_t *pix, int stride, int alpha, int beta )
491 x264_deblock_v8_luma_intra_mmxext( pix, stride, alpha, beta );
492 x264_deblock_v8_luma_intra_mmxext( pix+8, stride, alpha, beta );
498 void x264_deblock_v_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
499 void x264_deblock_h_luma_altivec( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 );
503 void x264_deblock_v_luma_neon( uint8_t *, int, int, int, int8_t * );
504 void x264_deblock_h_luma_neon( uint8_t *, int, int, int, int8_t * );
505 void x264_deblock_v_chroma_neon( uint8_t *, int, int, int, int8_t * );
506 void x264_deblock_h_chroma_neon( uint8_t *, int, int, int, int8_t * );
509 void x264_deblock_init( int cpu, x264_deblock_function_t *pf )
511 pf->deblock_luma[1] = deblock_v_luma_c;
512 pf->deblock_luma[0] = deblock_h_luma_c;
513 pf->deblock_chroma[1] = deblock_v_chroma_c;
514 pf->deblock_chroma[0] = deblock_h_chroma_c;
515 pf->deblock_luma_intra[1] = deblock_v_luma_intra_c;
516 pf->deblock_luma_intra[0] = deblock_h_luma_intra_c;
517 pf->deblock_chroma_intra[1] = deblock_v_chroma_intra_c;
518 pf->deblock_chroma_intra[0] = deblock_h_chroma_intra_c;
519 pf->deblock_strength = deblock_strength_c;
522 if( cpu&X264_CPU_MMXEXT )
524 pf->deblock_chroma[1] = x264_deblock_v_chroma_mmxext;
525 pf->deblock_chroma[0] = x264_deblock_h_chroma_mmxext;
526 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_mmxext;
527 pf->deblock_chroma_intra[0] = x264_deblock_h_chroma_intra_mmxext;
529 pf->deblock_luma[1] = x264_deblock_v_luma_mmxext;
530 pf->deblock_luma[0] = x264_deblock_h_luma_mmxext;
531 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_mmxext;
532 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_mmxext;
534 pf->deblock_strength = x264_deblock_strength_mmxext;
535 if( cpu&X264_CPU_SSE2 )
537 pf->deblock_strength = x264_deblock_strength_sse2;
538 if( !(cpu&X264_CPU_STACK_MOD4) )
540 pf->deblock_luma[1] = x264_deblock_v_luma_sse2;
541 pf->deblock_luma[0] = x264_deblock_h_luma_sse2;
542 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_sse2;
543 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_sse2;
546 if( cpu&X264_CPU_SSSE3 )
547 pf->deblock_strength = x264_deblock_strength_ssse3;
552 if( cpu&X264_CPU_ALTIVEC )
554 pf->deblock_luma[1] = x264_deblock_v_luma_altivec;
555 pf->deblock_luma[0] = x264_deblock_h_luma_altivec;
557 #endif // HAVE_ALTIVEC
560 if( cpu&X264_CPU_NEON )
562 pf->deblock_luma[1] = x264_deblock_v_luma_neon;
563 pf->deblock_luma[0] = x264_deblock_h_luma_neon;
564 pf->deblock_chroma[1] = x264_deblock_v_chroma_neon;
565 pf->deblock_chroma[0] = x264_deblock_h_chroma_neon;