1 /*****************************************************************************
2 * deblock.c: deblocking
3 *****************************************************************************
4 * Copyright (C) 2003-2015 x264 project
6 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
7 * Loren Merritt <lorenm@u.washington.edu>
8 * Fiona Glaser <fiona@x264.com>
9 * Henrik Gramner <henrik@gramner.com>
11 * This program is free software; you can redistribute it and/or modify
12 * it under the terms of the GNU General Public License as published by
13 * the Free Software Foundation; either version 2 of the License, or
14 * (at your option) any later version.
16 * This program is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 * GNU General Public License for more details.
21 * You should have received a copy of the GNU General Public License
22 * along with this program; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
25 * This program is also available under a commercial proprietary license.
26 * For more information, contact us at licensing@x264.com.
27 *****************************************************************************/
31 /* Deblocking filter */
32 static const uint8_t i_alpha_table[52+12*3] =
34 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
35 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
36 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
37 0, 0, 0, 0, 0, 0, 4, 4, 5, 6,
38 7, 8, 9, 10, 12, 13, 15, 17, 20, 22,
39 25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
40 80, 90,101,113,127,144,162,182,203,226,
42 255,255,255,255,255,255,255,255,255,255,255,255,
44 static const uint8_t i_beta_table[52+12*3] =
46 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
47 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
48 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
49 0, 0, 0, 0, 0, 0, 2, 2, 2, 3,
50 3, 3, 3, 4, 4, 4, 6, 6, 7, 7,
51 8, 8, 9, 9, 10, 10, 11, 11, 12, 12,
52 13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
54 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18,
56 static const int8_t i_tc0_table[52+12*3][4] =
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, 0 },
64 {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 0 }, {-1, 0, 0, 1 },
65 {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 0, 1 }, {-1, 0, 1, 1 }, {-1, 0, 1, 1 }, {-1, 1, 1, 1 },
66 {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 1 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 }, {-1, 1, 1, 2 },
67 {-1, 1, 1, 2 }, {-1, 1, 2, 3 }, {-1, 1, 2, 3 }, {-1, 2, 2, 3 }, {-1, 2, 2, 4 }, {-1, 2, 3, 4 },
68 {-1, 2, 3, 4 }, {-1, 3, 3, 5 }, {-1, 3, 4, 6 }, {-1, 3, 4, 6 }, {-1, 4, 5, 7 }, {-1, 4, 5, 8 },
69 {-1, 4, 6, 9 }, {-1, 5, 7,10 }, {-1, 6, 8,11 }, {-1, 6, 8,13 }, {-1, 7,10,14 }, {-1, 8,11,16 },
70 {-1, 9,12,18 }, {-1,10,13,20 }, {-1,11,15,23 }, {-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 },
72 {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 }, {-1,13,17,25 },
74 #define alpha_table(x) i_alpha_table[(x)+24]
75 #define beta_table(x) i_beta_table[(x)+24]
76 #define tc0_table(x) i_tc0_table[(x)+24]
79 static ALWAYS_INLINE void deblock_edge_luma_c( pixel *pix, intptr_t xstride, int alpha, int beta, int8_t tc0 )
81 int p2 = pix[-3*xstride];
82 int p1 = pix[-2*xstride];
83 int p0 = pix[-1*xstride];
84 int q0 = pix[ 0*xstride];
85 int q1 = pix[ 1*xstride];
86 int q2 = pix[ 2*xstride];
88 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
92 if( abs( p2 - p0 ) < beta )
95 pix[-2*xstride] = p1 + x264_clip3( (( p2 + ((p0 + q0 + 1) >> 1)) >> 1) - p1, -tc0, tc0 );
98 if( abs( q2 - q0 ) < beta )
101 pix[ 1*xstride] = q1 + x264_clip3( (( q2 + ((p0 + q0 + 1) >> 1)) >> 1) - q1, -tc0, tc0 );
105 delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
106 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
107 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
110 static inline void deblock_luma_c( pixel *pix, intptr_t xstride, intptr_t ystride, int alpha, int beta, int8_t *tc0 )
112 for( int i = 0; i < 4; i++ )
119 for( int d = 0; d < 4; d++, pix += ystride )
120 deblock_edge_luma_c( pix, xstride, alpha, beta, tc0[i] );
123 static void deblock_h_luma_mbaff_c( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
125 for( int d = 0; d < 8; d++, pix += stride )
126 deblock_edge_luma_c( pix, 1, alpha, beta, tc0[d>>1] );
128 static void deblock_v_luma_c( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
130 deblock_luma_c( pix, stride, 1, alpha, beta, tc0 );
132 static void deblock_h_luma_c( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
134 deblock_luma_c( pix, 1, stride, alpha, beta, tc0 );
137 static ALWAYS_INLINE void deblock_edge_chroma_c( pixel *pix, intptr_t xstride, int alpha, int beta, int8_t tc )
139 int p1 = pix[-2*xstride];
140 int p0 = pix[-1*xstride];
141 int q0 = pix[ 0*xstride];
142 int q1 = pix[ 1*xstride];
144 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
146 int delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
147 pix[-1*xstride] = x264_clip_pixel( p0 + delta ); /* p0' */
148 pix[ 0*xstride] = x264_clip_pixel( q0 - delta ); /* q0' */
151 static ALWAYS_INLINE void deblock_chroma_c( pixel *pix, int height, intptr_t xstride, intptr_t ystride, int alpha, int beta, int8_t *tc0 )
153 for( int i = 0; i < 4; i++ )
158 pix += height*ystride;
161 for( int d = 0; d < height; d++, pix += ystride-2 )
162 for( int e = 0; e < 2; e++, pix++ )
163 deblock_edge_chroma_c( pix, xstride, alpha, beta, tc0[i] );
166 static void deblock_h_chroma_mbaff_c( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
168 deblock_chroma_c( pix, 1, 2, stride, alpha, beta, tc0 );
170 static void deblock_v_chroma_c( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
172 deblock_chroma_c( pix, 2, stride, 2, alpha, beta, tc0 );
174 static void deblock_h_chroma_c( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
176 deblock_chroma_c( pix, 2, 2, stride, alpha, beta, tc0 );
178 static void deblock_h_chroma_422_c( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
180 deblock_chroma_c( pix, 4, 2, stride, alpha, beta, tc0 );
183 static ALWAYS_INLINE void deblock_edge_luma_intra_c( pixel *pix, intptr_t xstride, int alpha, int beta )
185 int p2 = pix[-3*xstride];
186 int p1 = pix[-2*xstride];
187 int p0 = pix[-1*xstride];
188 int q0 = pix[ 0*xstride];
189 int q1 = pix[ 1*xstride];
190 int q2 = pix[ 2*xstride];
192 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
194 if( abs( p0 - q0 ) < ((alpha >> 2) + 2) )
196 if( abs( p2 - p0 ) < beta ) /* p0', p1', p2' */
198 const int p3 = pix[-4*xstride];
199 pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
200 pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
201 pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
204 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
205 if( abs( q2 - q0 ) < beta ) /* q0', q1', q2' */
207 const int q3 = pix[3*xstride];
208 pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
209 pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
210 pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
213 pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
217 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
218 pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
222 static inline void deblock_luma_intra_c( pixel *pix, intptr_t xstride, intptr_t ystride, int alpha, int beta )
224 for( int d = 0; d < 16; d++, pix += ystride )
225 deblock_edge_luma_intra_c( pix, xstride, alpha, beta );
227 static void deblock_h_luma_intra_mbaff_c( pixel *pix, intptr_t ystride, int alpha, int beta )
229 for( int d = 0; d < 8; d++, pix += ystride )
230 deblock_edge_luma_intra_c( pix, 1, alpha, beta );
232 static void deblock_v_luma_intra_c( pixel *pix, intptr_t stride, int alpha, int beta )
234 deblock_luma_intra_c( pix, stride, 1, alpha, beta );
236 static void deblock_h_luma_intra_c( pixel *pix, intptr_t stride, int alpha, int beta )
238 deblock_luma_intra_c( pix, 1, stride, alpha, beta );
241 static ALWAYS_INLINE void deblock_edge_chroma_intra_c( pixel *pix, intptr_t xstride, int alpha, int beta )
243 int p1 = pix[-2*xstride];
244 int p0 = pix[-1*xstride];
245 int q0 = pix[ 0*xstride];
246 int q1 = pix[ 1*xstride];
248 if( abs( p0 - q0 ) < alpha && abs( p1 - p0 ) < beta && abs( q1 - q0 ) < beta )
250 pix[-1*xstride] = (2*p1 + p0 + q1 + 2) >> 2; /* p0' */
251 pix[ 0*xstride] = (2*q1 + q0 + p1 + 2) >> 2; /* q0' */
254 static ALWAYS_INLINE void deblock_chroma_intra_c( pixel *pix, int width, int height, intptr_t xstride, intptr_t ystride, int alpha, int beta )
256 for( int d = 0; d < height; d++, pix += ystride-2 )
257 for( int e = 0; e < width; e++, pix++ )
258 deblock_edge_chroma_intra_c( pix, xstride, alpha, beta );
260 static void deblock_h_chroma_intra_mbaff_c( pixel *pix, intptr_t stride, int alpha, int beta )
262 deblock_chroma_intra_c( pix, 2, 4, 2, stride, alpha, beta );
264 static void deblock_v_chroma_intra_c( pixel *pix, intptr_t stride, int alpha, int beta )
266 deblock_chroma_intra_c( pix, 1, 16, stride, 2, alpha, beta );
268 static void deblock_h_chroma_intra_c( pixel *pix, intptr_t stride, int alpha, int beta )
270 deblock_chroma_intra_c( pix, 2, 8, 2, stride, alpha, beta );
272 static void deblock_h_chroma_422_intra_c( pixel *pix, intptr_t stride, int alpha, int beta )
274 deblock_chroma_intra_c( pix, 2, 16, 2, stride, alpha, beta );
277 static void deblock_strength_c( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
278 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4], int mvy_limit,
281 for( int dir = 0; dir < 2; dir++ )
283 int s1 = dir ? 1 : 8;
284 int s2 = dir ? 8 : 1;
285 for( int edge = 0; edge < 4; edge++ )
286 for( int i = 0, loc = X264_SCAN8_0+edge*s2; i < 4; i++, loc += s1 )
289 if( nnz[loc] || nnz[locn] )
290 bs[dir][edge][i] = 2;
291 else if( ref[0][loc] != ref[0][locn] ||
292 abs( mv[0][loc][0] - mv[0][locn][0] ) >= 4 ||
293 abs( mv[0][loc][1] - mv[0][locn][1] ) >= mvy_limit ||
294 (bframe && (ref[1][loc] != ref[1][locn] ||
295 abs( mv[1][loc][0] - mv[1][locn][0] ) >= 4 ||
296 abs( mv[1][loc][1] - mv[1][locn][1] ) >= mvy_limit )))
298 bs[dir][edge][i] = 1;
301 bs[dir][edge][i] = 0;
306 static ALWAYS_INLINE void deblock_edge( x264_t *h, pixel *pix, intptr_t i_stride, uint8_t bS[4], int i_qp,
307 int a, int b, int b_chroma, x264_deblock_inter_t pf_inter )
309 int index_a = i_qp + a;
310 int index_b = i_qp + b;
311 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
312 int beta = beta_table(index_b) << (BIT_DEPTH-8);
315 if( !M32(bS) || !alpha || !beta )
318 tc[0] = (tc0_table(index_a)[bS[0]] << (BIT_DEPTH-8)) + b_chroma;
319 tc[1] = (tc0_table(index_a)[bS[1]] << (BIT_DEPTH-8)) + b_chroma;
320 tc[2] = (tc0_table(index_a)[bS[2]] << (BIT_DEPTH-8)) + b_chroma;
321 tc[3] = (tc0_table(index_a)[bS[3]] << (BIT_DEPTH-8)) + b_chroma;
323 pf_inter( pix, i_stride, alpha, beta, tc );
326 static ALWAYS_INLINE void deblock_edge_intra( x264_t *h, pixel *pix, intptr_t i_stride, uint8_t bS[4], int i_qp,
327 int a, int b, int b_chroma, x264_deblock_intra_t pf_intra )
329 int index_a = i_qp + a;
330 int index_b = i_qp + b;
331 int alpha = alpha_table(index_a) << (BIT_DEPTH-8);
332 int beta = beta_table(index_b) << (BIT_DEPTH-8);
334 if( !alpha || !beta )
337 pf_intra( pix, i_stride, alpha, beta );
340 static ALWAYS_INLINE void x264_macroblock_cache_load_neighbours_deblock( x264_t *h, int mb_x, int mb_y )
342 int deblock_on_slice_edges = h->sh.i_disable_deblocking_filter_idc != 2;
344 h->mb.i_neighbour = 0;
345 h->mb.i_mb_xy = mb_y * h->mb.i_mb_stride + mb_x;
346 h->mb.b_interlaced = PARAM_INTERLACED && h->mb.field[h->mb.i_mb_xy];
347 h->mb.i_mb_top_y = mb_y - (1 << MB_INTERLACED);
348 h->mb.i_mb_top_xy = mb_x + h->mb.i_mb_stride*h->mb.i_mb_top_y;
349 h->mb.i_mb_left_xy[1] =
350 h->mb.i_mb_left_xy[0] = h->mb.i_mb_xy - 1;
355 if( mb_x && h->mb.field[h->mb.i_mb_xy - 1] != MB_INTERLACED )
356 h->mb.i_mb_left_xy[0] -= h->mb.i_mb_stride;
360 if( h->mb.i_mb_top_xy >= 0 && MB_INTERLACED && !h->mb.field[h->mb.i_mb_top_xy] )
362 h->mb.i_mb_top_xy += h->mb.i_mb_stride;
365 if( mb_x && h->mb.field[h->mb.i_mb_xy - 1] != MB_INTERLACED )
366 h->mb.i_mb_left_xy[1] += h->mb.i_mb_stride;
370 if( mb_x > 0 && (deblock_on_slice_edges ||
371 h->mb.slice_table[h->mb.i_mb_left_xy[0]] == h->mb.slice_table[h->mb.i_mb_xy]) )
372 h->mb.i_neighbour |= MB_LEFT;
373 if( mb_y > MB_INTERLACED && (deblock_on_slice_edges
374 || h->mb.slice_table[h->mb.i_mb_top_xy] == h->mb.slice_table[h->mb.i_mb_xy]) )
375 h->mb.i_neighbour |= MB_TOP;
378 void x264_frame_deblock_row( x264_t *h, int mb_y )
380 int b_interlaced = SLICE_MBAFF;
381 int a = h->sh.i_alpha_c0_offset - QP_BD_OFFSET;
382 int b = h->sh.i_beta_offset - QP_BD_OFFSET;
383 int qp_thresh = 15 - X264_MIN( a, b ) - X264_MAX( 0, h->pps->i_chroma_qp_index_offset );
384 int stridey = h->fdec->i_stride[0];
385 int strideuv = h->fdec->i_stride[1];
386 int chroma444 = CHROMA444;
387 int chroma_height = 16 >> CHROMA_V_SHIFT;
388 intptr_t uvdiff = chroma444 ? h->fdec->plane[2] - h->fdec->plane[1] : 1;
390 for( int mb_x = 0; mb_x < h->mb.i_mb_width; mb_x += (~b_interlaced | mb_y)&1, mb_y ^= b_interlaced )
392 x264_prefetch_fenc( h, h->fdec, mb_x, mb_y );
393 x264_macroblock_cache_load_neighbours_deblock( h, mb_x, mb_y );
395 int mb_xy = h->mb.i_mb_xy;
396 int transform_8x8 = h->mb.mb_transform_size[mb_xy];
397 int intra_cur = IS_INTRA( h->mb.type[mb_xy] );
398 uint8_t (*bs)[8][4] = h->deblock_strength[mb_y&1][h->param.b_sliced_threads?mb_xy:mb_x];
400 pixel *pixy = h->fdec->plane[0] + 16*mb_y*stridey + 16*mb_x;
401 pixel *pixuv = h->fdec->plane[1] + chroma_height*mb_y*strideuv + 16*mb_x;
403 if( mb_y & MB_INTERLACED )
406 pixuv -= (chroma_height-1)*strideuv;
409 int stride2y = stridey << MB_INTERLACED;
410 int stride2uv = strideuv << MB_INTERLACED;
411 int qp = h->mb.qp[mb_xy];
412 int qpc = h->chroma_qp_table[qp];
413 int first_edge_only = (h->mb.partition[mb_xy] == D_16x16 && !h->mb.cbp[mb_xy] && !intra_cur) || qp <= qp_thresh;
415 #define FILTER( intra, dir, edge, qp, chroma_qp )\
418 if( !(edge & 1) || !transform_8x8 )\
420 deblock_edge##intra( h, pixy + 4*edge*(dir?stride2y:1),\
421 stride2y, bs[dir][edge], qp, a, b, 0,\
422 h->loopf.deblock_luma##intra[dir] );\
423 if( CHROMA_FORMAT == CHROMA_444 )\
425 deblock_edge##intra( h, pixuv + 4*edge*(dir?stride2uv:1),\
426 stride2uv, bs[dir][edge], chroma_qp, a, b, 0,\
427 h->loopf.deblock_luma##intra[dir] );\
428 deblock_edge##intra( h, pixuv + uvdiff + 4*edge*(dir?stride2uv:1),\
429 stride2uv, bs[dir][edge], chroma_qp, a, b, 0,\
430 h->loopf.deblock_luma##intra[dir] );\
432 else if( CHROMA_FORMAT == CHROMA_420 && !(edge & 1) )\
434 deblock_edge##intra( h, pixuv + edge*(dir?2*stride2uv:4),\
435 stride2uv, bs[dir][edge], chroma_qp, a, b, 1,\
436 h->loopf.deblock_chroma##intra[dir] );\
439 if( CHROMA_FORMAT == CHROMA_422 && (dir || !(edge & 1)) )\
441 deblock_edge##intra( h, pixuv + edge*(dir?4*stride2uv:4),\
442 stride2uv, bs[dir][edge], chroma_qp, a, b, 1,\
443 h->loopf.deblock_chroma##intra[dir] );\
447 if( h->mb.i_neighbour & MB_LEFT )
449 if( b_interlaced && h->mb.field[h->mb.i_mb_left_xy[0]] != MB_INTERLACED )
454 x264_deblock_inter_t luma_deblock = h->loopf.deblock_luma_mbaff;
455 x264_deblock_inter_t chroma_deblock = h->loopf.deblock_chroma_mbaff;
456 x264_deblock_intra_t luma_intra_deblock = h->loopf.deblock_luma_intra_mbaff;
457 x264_deblock_intra_t chroma_intra_deblock = h->loopf.deblock_chroma_intra_mbaff;
458 int c = chroma444 ? 0 : 1;
460 left_qp[0] = h->mb.qp[h->mb.i_mb_left_xy[0]];
461 luma_qp[0] = (qp + left_qp[0] + 1) >> 1;
462 chroma_qp[0] = (qpc + h->chroma_qp_table[left_qp[0]] + 1) >> 1;
463 if( intra_cur || IS_INTRA( h->mb.type[h->mb.i_mb_left_xy[0]] ) )
465 deblock_edge_intra( h, pixy, 2*stridey, bs[0][0], luma_qp[0], a, b, 0, luma_intra_deblock );
466 deblock_edge_intra( h, pixuv, 2*strideuv, bs[0][0], chroma_qp[0], a, b, c, chroma_intra_deblock );
468 deblock_edge_intra( h, pixuv + uvdiff, 2*strideuv, bs[0][0], chroma_qp[0], a, b, c, chroma_intra_deblock );
472 deblock_edge( h, pixy, 2*stridey, bs[0][0], luma_qp[0], a, b, 0, luma_deblock );
473 deblock_edge( h, pixuv, 2*strideuv, bs[0][0], chroma_qp[0], a, b, c, chroma_deblock );
475 deblock_edge( h, pixuv + uvdiff, 2*strideuv, bs[0][0], chroma_qp[0], a, b, c, chroma_deblock );
478 int offy = MB_INTERLACED ? 4 : 0;
479 int offuv = MB_INTERLACED ? 4-CHROMA_V_SHIFT : 0;
480 left_qp[1] = h->mb.qp[h->mb.i_mb_left_xy[1]];
481 luma_qp[1] = (qp + left_qp[1] + 1) >> 1;
482 chroma_qp[1] = (qpc + h->chroma_qp_table[left_qp[1]] + 1) >> 1;
483 if( intra_cur || IS_INTRA( h->mb.type[h->mb.i_mb_left_xy[1]] ) )
485 deblock_edge_intra( h, pixy + (stridey<<offy), 2*stridey, bs[0][4], luma_qp[1], a, b, 0, luma_intra_deblock );
486 deblock_edge_intra( h, pixuv + (strideuv<<offuv), 2*strideuv, bs[0][4], chroma_qp[1], a, b, c, chroma_intra_deblock );
488 deblock_edge_intra( h, pixuv + uvdiff + (strideuv<<offuv), 2*strideuv, bs[0][4], chroma_qp[1], a, b, c, chroma_intra_deblock );
492 deblock_edge( h, pixy + (stridey<<offy), 2*stridey, bs[0][4], luma_qp[1], a, b, 0, luma_deblock );
493 deblock_edge( h, pixuv + (strideuv<<offuv), 2*strideuv, bs[0][4], chroma_qp[1], a, b, c, chroma_deblock );
495 deblock_edge( h, pixuv + uvdiff + (strideuv<<offuv), 2*strideuv, bs[0][4], chroma_qp[1], a, b, c, chroma_deblock );
500 int qpl = h->mb.qp[h->mb.i_mb_xy-1];
501 int qp_left = (qp + qpl + 1) >> 1;
502 int qpc_left = (qpc + h->chroma_qp_table[qpl] + 1) >> 1;
503 int intra_left = IS_INTRA( h->mb.type[h->mb.i_mb_xy-1] );
504 int intra_deblock = intra_cur || intra_left;
506 /* Any MB that was coded, or that analysis decided to skip, has quality commensurate with its QP.
507 * But if deblocking affects neighboring MBs that were force-skipped, blur might accumulate there.
508 * So reset their effective QP to max, to indicate that lack of guarantee. */
509 if( h->fdec->mb_info && M32( bs[0][0] ) )
511 #define RESET_EFFECTIVE_QP(xy) h->fdec->effective_qp[xy] |= 0xff * !!(h->fdec->mb_info[xy] & X264_MBINFO_CONSTANT);
512 RESET_EFFECTIVE_QP(mb_xy);
513 RESET_EFFECTIVE_QP(h->mb.i_mb_left_xy[0]);
517 FILTER( _intra, 0, 0, qp_left, qpc_left );
519 FILTER( , 0, 0, qp_left, qpc_left );
522 if( !first_edge_only )
524 FILTER( , 0, 1, qp, qpc );
525 FILTER( , 0, 2, qp, qpc );
526 FILTER( , 0, 3, qp, qpc );
529 if( h->mb.i_neighbour & MB_TOP )
531 if( b_interlaced && !(mb_y&1) && !MB_INTERLACED && h->mb.field[h->mb.i_mb_top_xy] )
533 int mbn_xy = mb_xy - 2 * h->mb.i_mb_stride;
535 for( int j = 0; j < 2; j++, mbn_xy += h->mb.i_mb_stride )
537 int qpt = h->mb.qp[mbn_xy];
538 int qp_top = (qp + qpt + 1) >> 1;
539 int qpc_top = (qpc + h->chroma_qp_table[qpt] + 1) >> 1;
540 int intra_top = IS_INTRA( h->mb.type[mbn_xy] );
541 if( intra_cur || intra_top )
542 M32( bs[1][4*j] ) = 0x03030303;
544 // deblock the first horizontal edge of the even rows, then the first horizontal edge of the odd rows
545 deblock_edge( h, pixy + j*stridey, 2* stridey, bs[1][4*j], qp_top, a, b, 0, h->loopf.deblock_luma[1] );
548 deblock_edge( h, pixuv + j*strideuv, 2*strideuv, bs[1][4*j], qpc_top, a, b, 0, h->loopf.deblock_luma[1] );
549 deblock_edge( h, pixuv + uvdiff + j*strideuv, 2*strideuv, bs[1][4*j], qpc_top, a, b, 0, h->loopf.deblock_luma[1] );
552 deblock_edge( h, pixuv + j*strideuv, 2*strideuv, bs[1][4*j], qpc_top, a, b, 1, h->loopf.deblock_chroma[1] );
557 int qpt = h->mb.qp[h->mb.i_mb_top_xy];
558 int qp_top = (qp + qpt + 1) >> 1;
559 int qpc_top = (qpc + h->chroma_qp_table[qpt] + 1) >> 1;
560 int intra_top = IS_INTRA( h->mb.type[h->mb.i_mb_top_xy] );
561 int intra_deblock = intra_cur || intra_top;
563 /* This edge has been modified, reset effective qp to max. */
564 if( h->fdec->mb_info && M32( bs[1][0] ) )
566 RESET_EFFECTIVE_QP(mb_xy);
567 RESET_EFFECTIVE_QP(h->mb.i_mb_top_xy);
570 if( (!b_interlaced || (!MB_INTERLACED && !h->mb.field[h->mb.i_mb_top_xy])) && intra_deblock )
572 FILTER( _intra, 1, 0, qp_top, qpc_top );
577 M32( bs[1][0] ) = 0x03030303;
578 FILTER( , 1, 0, qp_top, qpc_top );
583 if( !first_edge_only )
585 FILTER( , 1, 1, qp, qpc );
586 FILTER( , 1, 2, qp, qpc );
587 FILTER( , 1, 3, qp, qpc );
594 /* For deblock-aware RD.
596 * deblock macroblock edges
597 * support analysis partitions smaller than 16x16
598 * deblock chroma for 4:2:0/4:2:2
599 * handle duplicate refs correctly
601 void x264_macroblock_deblock( x264_t *h )
603 int a = h->sh.i_alpha_c0_offset - QP_BD_OFFSET;
604 int b = h->sh.i_beta_offset - QP_BD_OFFSET;
605 int qp_thresh = 15 - X264_MIN( a, b ) - X264_MAX( 0, h->pps->i_chroma_qp_index_offset );
606 int intra_cur = IS_INTRA( h->mb.i_type );
608 int qpc = h->mb.i_chroma_qp;
609 if( (h->mb.i_partition == D_16x16 && !h->mb.i_cbp_luma && !intra_cur) || qp <= qp_thresh )
612 uint8_t (*bs)[8][4] = h->mb.cache.deblock_strength;
615 M32( bs[0][1] ) = 0x03030303;
616 M64( bs[0][2] ) = 0x0303030303030303ULL;
617 M32( bs[1][1] ) = 0x03030303;
618 M64( bs[1][2] ) = 0x0303030303030303ULL;
621 h->loopf.deblock_strength( h->mb.cache.non_zero_count, h->mb.cache.ref, h->mb.cache.mv,
622 bs, 4 >> MB_INTERLACED, h->sh.i_type == SLICE_TYPE_B );
624 int transform_8x8 = h->mb.b_transform_8x8;
626 #define FILTER( dir, edge )\
629 deblock_edge( h, h->mb.pic.p_fdec[0] + 4*edge*(dir?FDEC_STRIDE:1),\
630 FDEC_STRIDE, bs[dir][edge], qp, a, b, 0,\
631 h->loopf.deblock_luma[dir] );\
634 deblock_edge( h, h->mb.pic.p_fdec[1] + 4*edge*(dir?FDEC_STRIDE:1),\
635 FDEC_STRIDE, bs[dir][edge], qpc, a, b, 0,\
636 h->loopf.deblock_luma[dir] );\
637 deblock_edge( h, h->mb.pic.p_fdec[2] + 4*edge*(dir?FDEC_STRIDE:1),\
638 FDEC_STRIDE, bs[dir][edge], qpc, a, b, 0,\
639 h->loopf.deblock_luma[dir] );\
643 if( !transform_8x8 ) FILTER( 0, 1 );
645 if( !transform_8x8 ) FILTER( 0, 3 );
647 if( !transform_8x8 ) FILTER( 1, 1 );
649 if( !transform_8x8 ) FILTER( 1, 3 );
655 void x264_deblock_v_luma_sse2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
656 void x264_deblock_v_luma_avx ( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
657 void x264_deblock_h_luma_sse2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
658 void x264_deblock_h_luma_avx ( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
659 void x264_deblock_v_chroma_sse2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
660 void x264_deblock_v_chroma_avx ( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
661 void x264_deblock_h_chroma_sse2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
662 void x264_deblock_h_chroma_avx ( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
663 void x264_deblock_h_chroma_mbaff_sse2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
664 void x264_deblock_h_chroma_mbaff_avx ( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
665 void x264_deblock_h_chroma_422_mmx2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
666 void x264_deblock_h_chroma_422_sse2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
667 void x264_deblock_h_chroma_422_avx ( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
668 void x264_deblock_v_luma_intra_sse2( pixel *pix, intptr_t stride, int alpha, int beta );
669 void x264_deblock_v_luma_intra_avx ( pixel *pix, intptr_t stride, int alpha, int beta );
670 void x264_deblock_h_luma_intra_sse2( pixel *pix, intptr_t stride, int alpha, int beta );
671 void x264_deblock_h_luma_intra_avx ( pixel *pix, intptr_t stride, int alpha, int beta );
672 void x264_deblock_v_chroma_intra_sse2( pixel *pix, intptr_t stride, int alpha, int beta );
673 void x264_deblock_v_chroma_intra_avx ( pixel *pix, intptr_t stride, int alpha, int beta );
674 void x264_deblock_h_chroma_intra_sse2( pixel *pix, intptr_t stride, int alpha, int beta );
675 void x264_deblock_h_chroma_intra_avx ( pixel *pix, intptr_t stride, int alpha, int beta );
676 void x264_deblock_h_chroma_422_intra_mmx2( pixel *pix, intptr_t stride, int alpha, int beta );
677 void x264_deblock_h_chroma_422_intra_sse2( pixel *pix, intptr_t stride, int alpha, int beta );
678 void x264_deblock_h_chroma_422_intra_avx ( pixel *pix, intptr_t stride, int alpha, int beta );
679 void x264_deblock_strength_mmx2 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
680 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4],
681 int mvy_limit, int bframe );
682 void x264_deblock_strength_sse2 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
683 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4],
684 int mvy_limit, int bframe );
685 void x264_deblock_strength_ssse3( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
686 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4],
687 int mvy_limit, int bframe );
688 void x264_deblock_strength_avx ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
689 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4],
690 int mvy_limit, int bframe );
691 void x264_deblock_strength_avx2 ( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
692 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4],
693 int mvy_limit, int bframe );
695 void x264_deblock_h_chroma_intra_mbaff_mmx2( pixel *pix, intptr_t stride, int alpha, int beta );
696 void x264_deblock_h_chroma_intra_mbaff_sse2( pixel *pix, intptr_t stride, int alpha, int beta );
697 void x264_deblock_h_chroma_intra_mbaff_avx ( pixel *pix, intptr_t stride, int alpha, int beta );
699 void x264_deblock_h_luma_mmx2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
700 void x264_deblock_v8_luma_mmx2( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
701 void x264_deblock_v_chroma_mmx2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
702 void x264_deblock_h_chroma_mmx2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
703 void x264_deblock_h_chroma_mbaff_mmx2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
704 void x264_deblock_h_luma_intra_mmx2( pixel *pix, intptr_t stride, int alpha, int beta );
705 void x264_deblock_v8_luma_intra_mmx2( uint8_t *pix, intptr_t stride, int alpha, int beta );
706 void x264_deblock_v_chroma_intra_mmx2( pixel *pix, intptr_t stride, int alpha, int beta );
707 void x264_deblock_h_chroma_intra_mmx2( pixel *pix, intptr_t stride, int alpha, int beta );
708 void x264_deblock_h_chroma_intra_mbaff_mmx2( pixel *pix, intptr_t stride, int alpha, int beta );
711 void x264_deblock_v_luma_mmx2( pixel *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
712 void x264_deblock_v_luma_intra_mmx2( pixel *pix, intptr_t stride, int alpha, int beta );
714 // FIXME this wrapper has a significant cpu cost
715 static void x264_deblock_v_luma_mmx2( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 )
717 x264_deblock_v8_luma_mmx2( pix, stride, alpha, beta, tc0 );
718 x264_deblock_v8_luma_mmx2( pix+8, stride, alpha, beta, tc0+2 );
720 static void x264_deblock_v_luma_intra_mmx2( uint8_t *pix, intptr_t stride, int alpha, int beta )
722 x264_deblock_v8_luma_intra_mmx2( pix, stride, alpha, beta );
723 x264_deblock_v8_luma_intra_mmx2( pix+8, stride, alpha, beta );
725 #endif // HIGH_BIT_DEPTH
730 void x264_deblock_v_luma_altivec( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
731 void x264_deblock_h_luma_altivec( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
734 #if HAVE_ARMV6 || ARCH_AARCH64
735 void x264_deblock_v_luma_neon ( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
736 void x264_deblock_h_luma_neon ( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
737 void x264_deblock_v_chroma_neon( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
738 void x264_deblock_h_chroma_neon( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
739 void x264_deblock_strength_neon( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
740 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4],
741 int mvy_limit, int bframe );
742 void x264_deblock_h_chroma_422_neon( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
743 void x264_deblock_h_chroma_mbaff_neon( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
744 void x264_deblock_h_chroma_intra_mbaff_neon( uint8_t *pix, intptr_t stride, int alpha, int beta );
745 void x264_deblock_h_chroma_intra_neon( uint8_t *pix, intptr_t stride, int alpha, int beta );
746 void x264_deblock_h_chroma_422_intra_neon( uint8_t *pix, intptr_t stride, int alpha, int beta );
747 void x264_deblock_v_chroma_intra_neon( uint8_t *pix, intptr_t stride, int alpha, int beta );
748 void x264_deblock_h_luma_intra_neon( uint8_t *pix, intptr_t stride, int alpha, int beta );
749 void x264_deblock_v_luma_intra_neon( uint8_t *pix, intptr_t stride, int alpha, int beta );
754 void x264_deblock_v_luma_msa( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
755 void x264_deblock_h_luma_msa( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
756 void x264_deblock_v_chroma_msa( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
757 void x264_deblock_h_chroma_msa( uint8_t *pix, intptr_t stride, int alpha, int beta, int8_t *tc0 );
758 void x264_deblock_v_luma_intra_msa( uint8_t *pix, intptr_t stride, int alpha, int beta );
759 void x264_deblock_h_luma_intra_msa( uint8_t *pix, intptr_t stride, int alpha, int beta );
760 void x264_deblock_v_chroma_intra_msa( uint8_t *pix, intptr_t stride, int alpha, int beta );
761 void x264_deblock_h_chroma_intra_msa( uint8_t *pix, intptr_t stride, int alpha, int beta );
762 void x264_deblock_strength_msa( uint8_t nnz[X264_SCAN8_SIZE], int8_t ref[2][X264_SCAN8_LUMA_SIZE],
763 int16_t mv[2][X264_SCAN8_LUMA_SIZE][2], uint8_t bs[2][8][4], int mvy_limit,
768 void x264_deblock_init( int cpu, x264_deblock_function_t *pf, int b_mbaff )
770 pf->deblock_luma[1] = deblock_v_luma_c;
771 pf->deblock_luma[0] = deblock_h_luma_c;
772 pf->deblock_chroma[1] = deblock_v_chroma_c;
773 pf->deblock_h_chroma_420 = deblock_h_chroma_c;
774 pf->deblock_h_chroma_422 = deblock_h_chroma_422_c;
775 pf->deblock_luma_intra[1] = deblock_v_luma_intra_c;
776 pf->deblock_luma_intra[0] = deblock_h_luma_intra_c;
777 pf->deblock_chroma_intra[1] = deblock_v_chroma_intra_c;
778 pf->deblock_h_chroma_420_intra = deblock_h_chroma_intra_c;
779 pf->deblock_h_chroma_422_intra = deblock_h_chroma_422_intra_c;
780 pf->deblock_luma_mbaff = deblock_h_luma_mbaff_c;
781 pf->deblock_chroma_420_mbaff = deblock_h_chroma_mbaff_c;
782 pf->deblock_luma_intra_mbaff = deblock_h_luma_intra_mbaff_c;
783 pf->deblock_chroma_420_intra_mbaff = deblock_h_chroma_intra_mbaff_c;
784 pf->deblock_strength = deblock_strength_c;
787 if( cpu&X264_CPU_MMX2 )
790 pf->deblock_luma[1] = x264_deblock_v_luma_mmx2;
791 pf->deblock_luma[0] = x264_deblock_h_luma_mmx2;
792 pf->deblock_chroma[1] = x264_deblock_v_chroma_mmx2;
793 pf->deblock_h_chroma_420 = x264_deblock_h_chroma_mmx2;
794 pf->deblock_chroma_420_mbaff = x264_deblock_h_chroma_mbaff_mmx2;
795 pf->deblock_h_chroma_422 = x264_deblock_h_chroma_422_mmx2;
796 pf->deblock_h_chroma_422_intra = x264_deblock_h_chroma_422_intra_mmx2;
797 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_mmx2;
798 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_mmx2;
799 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_mmx2;
800 pf->deblock_h_chroma_420_intra = x264_deblock_h_chroma_intra_mmx2;
801 pf->deblock_chroma_420_intra_mbaff = x264_deblock_h_chroma_intra_mbaff_mmx2;
804 pf->deblock_chroma_420_intra_mbaff = x264_deblock_h_chroma_intra_mbaff_mmx2;
806 pf->deblock_strength = x264_deblock_strength_mmx2;
807 if( cpu&X264_CPU_SSE2 )
809 pf->deblock_strength = x264_deblock_strength_sse2;
810 pf->deblock_h_chroma_420 = x264_deblock_h_chroma_sse2;
811 pf->deblock_h_chroma_422 = x264_deblock_h_chroma_422_sse2;
812 pf->deblock_h_chroma_422_intra = x264_deblock_h_chroma_422_intra_sse2;
813 pf->deblock_chroma_420_mbaff = x264_deblock_h_chroma_mbaff_sse2;
814 pf->deblock_luma[1] = x264_deblock_v_luma_sse2;
815 pf->deblock_luma[0] = x264_deblock_h_luma_sse2;
816 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_sse2;
817 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_sse2;
818 if( !(cpu&X264_CPU_STACK_MOD4) )
820 pf->deblock_chroma[1] = x264_deblock_v_chroma_sse2;
821 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_sse2;
822 pf->deblock_h_chroma_420_intra = x264_deblock_h_chroma_intra_sse2;
824 pf->deblock_chroma_420_intra_mbaff= x264_deblock_h_chroma_intra_mbaff_sse2;
828 if( cpu&X264_CPU_SSSE3 )
829 pf->deblock_strength = x264_deblock_strength_ssse3;
830 if( cpu&X264_CPU_AVX )
832 pf->deblock_strength = x264_deblock_strength_avx;
833 pf->deblock_h_chroma_420 = x264_deblock_h_chroma_avx;
834 pf->deblock_h_chroma_422 = x264_deblock_h_chroma_422_avx;
835 pf->deblock_h_chroma_422_intra = x264_deblock_h_chroma_422_intra_avx;
836 pf->deblock_luma[1] = x264_deblock_v_luma_avx;
837 pf->deblock_luma[0] = x264_deblock_h_luma_avx;
838 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_avx;
839 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_avx;
840 if( !(cpu&X264_CPU_STACK_MOD4) )
842 pf->deblock_chroma[1] = x264_deblock_v_chroma_avx;
843 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_avx;
844 pf->deblock_h_chroma_420_intra = x264_deblock_h_chroma_intra_avx;
846 pf->deblock_chroma_420_mbaff = x264_deblock_h_chroma_mbaff_avx;
847 pf->deblock_chroma_420_intra_mbaff = x264_deblock_h_chroma_intra_mbaff_avx;
851 if( cpu&X264_CPU_AVX2 )
853 pf->deblock_strength = x264_deblock_strength_avx2;
860 if( cpu&X264_CPU_ALTIVEC )
862 pf->deblock_luma[1] = x264_deblock_v_luma_altivec;
863 pf->deblock_luma[0] = x264_deblock_h_luma_altivec;
865 #endif // HAVE_ALTIVEC
867 #if HAVE_ARMV6 || ARCH_AARCH64
868 if( cpu&X264_CPU_NEON )
870 pf->deblock_luma[1] = x264_deblock_v_luma_neon;
871 pf->deblock_luma[0] = x264_deblock_h_luma_neon;
872 pf->deblock_chroma[1] = x264_deblock_v_chroma_neon;
873 pf->deblock_h_chroma_420 = x264_deblock_h_chroma_neon;
874 pf->deblock_h_chroma_422 = x264_deblock_h_chroma_422_neon;
875 pf->deblock_chroma_420_mbaff = x264_deblock_h_chroma_mbaff_neon;
876 pf->deblock_chroma_420_intra_mbaff = x264_deblock_h_chroma_intra_mbaff_neon;
877 pf->deblock_h_chroma_420_intra = x264_deblock_h_chroma_intra_neon;
878 pf->deblock_h_chroma_422_intra = x264_deblock_h_chroma_422_intra_neon;
879 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_neon;
880 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_neon;
881 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_neon;
882 pf->deblock_strength = x264_deblock_strength_neon;
887 if( cpu&X264_CPU_MSA )
889 pf->deblock_luma[1] = x264_deblock_v_luma_msa;
890 pf->deblock_luma[0] = x264_deblock_h_luma_msa;
891 pf->deblock_chroma[1] = x264_deblock_v_chroma_msa;
892 pf->deblock_h_chroma_420 = x264_deblock_h_chroma_msa;
893 pf->deblock_luma_intra[1] = x264_deblock_v_luma_intra_msa;
894 pf->deblock_luma_intra[0] = x264_deblock_h_luma_intra_msa;
895 pf->deblock_chroma_intra[1] = x264_deblock_v_chroma_intra_msa;
896 pf->deblock_h_chroma_420_intra = x264_deblock_h_chroma_intra_msa;
897 pf->deblock_strength = x264_deblock_strength_msa;
900 #endif // !HIGH_BIT_DEPTH
902 /* These functions are equivalent, so don't duplicate them. */
903 pf->deblock_chroma_422_mbaff = pf->deblock_h_chroma_420;
904 pf->deblock_chroma_422_intra_mbaff = pf->deblock_h_chroma_420_intra;