Mercurial > libavcodec.hg
annotate h264dsp.c @ 11913:80916e795581 libavcodec
Bump micro version, add changelog entry for DTS-ES extension decoding support
author | mstorsjo |
---|---|
date | Tue, 22 Jun 2010 09:22:21 +0000 |
parents | 7dd2a45249a9 |
children |
rev | line source |
---|---|
0 | 1 /* |
11499 | 2 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder |
3 * Copyright (c) 2003-2010 Michael Niedermayer <michaelni@gmx.at> | |
5214 | 4 * |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3807
diff
changeset
|
5 * This file is part of FFmpeg. |
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3807
diff
changeset
|
6 * |
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3807
diff
changeset
|
7 * FFmpeg is free software; you can redistribute it and/or |
429 | 8 * modify it under the terms of the GNU Lesser General Public |
9 * License as published by the Free Software Foundation; either | |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3807
diff
changeset
|
10 * version 2.1 of the License, or (at your option) any later version. |
0 | 11 * |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3807
diff
changeset
|
12 * FFmpeg is distributed in the hope that it will be useful, |
0 | 13 * but WITHOUT ANY WARRANTY; without even the implied warranty of |
429 | 14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
15 * Lesser General Public License for more details. | |
0 | 16 * |
429 | 17 * You should have received a copy of the GNU Lesser General Public |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3807
diff
changeset
|
18 * License along with FFmpeg; if not, write to the Free Software |
3036
0b546eab515d
Update licensing information: The FSF changed postal address.
diego
parents:
3029
diff
changeset
|
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
0 | 20 */ |
2967 | 21 |
1106 | 22 /** |
11644
7dd2a45249a9
Remove explicit filename from Doxygen @file commands.
diego
parents:
11506
diff
changeset
|
23 * @file |
11499 | 24 * H.264 / AVC / MPEG4 part10 DSP functions. |
25 * @author Michael Niedermayer <michaelni@gmx.at> | |
1106 | 26 */ |
2967 | 27 |
11499 | 28 #include <stdint.h> |
0 | 29 #include "avcodec.h" |
11499 | 30 #include "h264dsp.h" |
1168 | 31 |
4594 | 32 #define op_scale1(x) block[x] = av_clip_uint8( (block[x]*weight + offset) >> log2_denom ) |
33 #define op_scale2(x) dst[x] = av_clip_uint8( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1)) | |
2415 | 34 #define H264_WEIGHT(W,H) \ |
35 static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride, int log2_denom, int weight, int offset){ \ | |
3029 | 36 int y; \ |
2415 | 37 offset <<= log2_denom; \ |
38 if(log2_denom) offset += 1<<(log2_denom-1); \ | |
39 for(y=0; y<H; y++, block += stride){ \ | |
40 op_scale1(0); \ | |
41 op_scale1(1); \ | |
42 if(W==2) continue; \ | |
43 op_scale1(2); \ | |
44 op_scale1(3); \ | |
45 if(W==4) continue; \ | |
46 op_scale1(4); \ | |
47 op_scale1(5); \ | |
48 op_scale1(6); \ | |
49 op_scale1(7); \ | |
50 if(W==8) continue; \ | |
51 op_scale1(8); \ | |
52 op_scale1(9); \ | |
53 op_scale1(10); \ | |
54 op_scale1(11); \ | |
55 op_scale1(12); \ | |
56 op_scale1(13); \ | |
57 op_scale1(14); \ | |
58 op_scale1(15); \ | |
59 } \ | |
60 } \ | |
3029 | 61 static void biweight_h264_pixels ## W ## x ## H ## _c(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset){ \ |
62 int y; \ | |
63 offset = ((offset + 1) | 1) << log2_denom; \ | |
2415 | 64 for(y=0; y<H; y++, dst += stride, src += stride){ \ |
65 op_scale2(0); \ | |
66 op_scale2(1); \ | |
67 if(W==2) continue; \ | |
68 op_scale2(2); \ | |
69 op_scale2(3); \ | |
70 if(W==4) continue; \ | |
71 op_scale2(4); \ | |
72 op_scale2(5); \ | |
73 op_scale2(6); \ | |
74 op_scale2(7); \ | |
75 if(W==8) continue; \ | |
76 op_scale2(8); \ | |
77 op_scale2(9); \ | |
78 op_scale2(10); \ | |
79 op_scale2(11); \ | |
80 op_scale2(12); \ | |
81 op_scale2(13); \ | |
82 op_scale2(14); \ | |
83 op_scale2(15); \ | |
84 } \ | |
85 } | |
86 | |
87 H264_WEIGHT(16,16) | |
88 H264_WEIGHT(16,8) | |
89 H264_WEIGHT(8,16) | |
90 H264_WEIGHT(8,8) | |
91 H264_WEIGHT(8,4) | |
92 H264_WEIGHT(4,8) | |
93 H264_WEIGHT(4,4) | |
94 H264_WEIGHT(4,2) | |
95 H264_WEIGHT(2,4) | |
96 H264_WEIGHT(2,2) | |
97 | |
98 #undef op_scale1 | |
99 #undef op_scale2 | |
100 #undef H264_WEIGHT | |
101 | |
10941
28edcc8c54c0
Mark the h264 c loop filter as av_always_inline av_flatten to make sure its
michael
parents:
10940
diff
changeset
|
102 static av_always_inline av_flatten void h264_loop_filter_luma_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0) |
2633 | 103 { |
104 int i, d; | |
105 for( i = 0; i < 4; i++ ) { | |
106 if( tc0[i] < 0 ) { | |
107 pix += 4*ystride; | |
108 continue; | |
109 } | |
110 for( d = 0; d < 4; d++ ) { | |
111 const int p0 = pix[-1*xstride]; | |
112 const int p1 = pix[-2*xstride]; | |
113 const int p2 = pix[-3*xstride]; | |
114 const int q0 = pix[0]; | |
115 const int q1 = pix[1*xstride]; | |
116 const int q2 = pix[2*xstride]; | |
2967 | 117 |
4001 | 118 if( FFABS( p0 - q0 ) < alpha && |
119 FFABS( p1 - p0 ) < beta && | |
120 FFABS( q1 - q0 ) < beta ) { | |
2967 | 121 |
2633 | 122 int tc = tc0[i]; |
123 int i_delta; | |
2967 | 124 |
4001 | 125 if( FFABS( p2 - p0 ) < beta ) { |
10940
563cb9b1a9b7
skip outer pixels if possible in h264_loop_filter_luma_c().
michael
parents:
10878
diff
changeset
|
126 if(tc0[i]) |
4594 | 127 pix[-2*xstride] = p1 + av_clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] ); |
2633 | 128 tc++; |
129 } | |
4001 | 130 if( FFABS( q2 - q0 ) < beta ) { |
10940
563cb9b1a9b7
skip outer pixels if possible in h264_loop_filter_luma_c().
michael
parents:
10878
diff
changeset
|
131 if(tc0[i]) |
4594 | 132 pix[ xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] ); |
2633 | 133 tc++; |
134 } | |
2967 | 135 |
4594 | 136 i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc ); |
137 pix[-xstride] = av_clip_uint8( p0 + i_delta ); /* p0' */ | |
138 pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */ | |
2633 | 139 } |
140 pix += ystride; | |
141 } | |
142 } | |
143 } | |
2707
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
144 static void h264_v_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) |
2633 | 145 { |
146 h264_loop_filter_luma_c(pix, stride, 1, alpha, beta, tc0); | |
147 } | |
2707
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
148 static void h264_h_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) |
2633 | 149 { |
150 h264_loop_filter_luma_c(pix, 1, stride, alpha, beta, tc0); | |
151 } | |
152 | |
10941
28edcc8c54c0
Mark the h264 c loop filter as av_always_inline av_flatten to make sure its
michael
parents:
10940
diff
changeset
|
153 static av_always_inline av_flatten void h264_loop_filter_luma_intra_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta) |
8395
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
154 { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
155 int d; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
156 for( d = 0; d < 16; d++ ) { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
157 const int p2 = pix[-3*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
158 const int p1 = pix[-2*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
159 const int p0 = pix[-1*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
160 |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
161 const int q0 = pix[ 0*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
162 const int q1 = pix[ 1*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
163 const int q2 = pix[ 2*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
164 |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
165 if( FFABS( p0 - q0 ) < alpha && |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
166 FFABS( p1 - p0 ) < beta && |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
167 FFABS( q1 - q0 ) < beta ) { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
168 |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
169 if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){ |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
170 if( FFABS( p2 - p0 ) < beta) |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
171 { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
172 const int p3 = pix[-4*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
173 /* p0', p1', p2' */ |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
174 pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
175 pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
176 pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
177 } else { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
178 /* p0' */ |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
179 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
180 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
181 if( FFABS( q2 - q0 ) < beta) |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
182 { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
183 const int q3 = pix[3*xstride]; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
184 /* q0', q1', q2' */ |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
185 pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
186 pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
187 pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
188 } else { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
189 /* q0' */ |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
190 pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
191 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
192 }else{ |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
193 /* p0', q0' */ |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
194 pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
195 pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
196 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
197 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
198 pix += ystride; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
199 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
200 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
201 static void h264_v_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta) |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
202 { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
203 h264_loop_filter_luma_intra_c(pix, stride, 1, alpha, beta); |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
204 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
205 static void h264_h_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta) |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
206 { |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
207 h264_loop_filter_luma_intra_c(pix, 1, stride, alpha, beta); |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
208 } |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
209 |
10941
28edcc8c54c0
Mark the h264 c loop filter as av_always_inline av_flatten to make sure its
michael
parents:
10940
diff
changeset
|
210 static av_always_inline av_flatten void h264_loop_filter_chroma_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta, int8_t *tc0) |
2633 | 211 { |
212 int i, d; | |
213 for( i = 0; i < 4; i++ ) { | |
214 const int tc = tc0[i]; | |
215 if( tc <= 0 ) { | |
216 pix += 2*ystride; | |
217 continue; | |
218 } | |
219 for( d = 0; d < 2; d++ ) { | |
220 const int p0 = pix[-1*xstride]; | |
221 const int p1 = pix[-2*xstride]; | |
222 const int q0 = pix[0]; | |
223 const int q1 = pix[1*xstride]; | |
224 | |
4001 | 225 if( FFABS( p0 - q0 ) < alpha && |
226 FFABS( p1 - p0 ) < beta && | |
227 FFABS( q1 - q0 ) < beta ) { | |
2633 | 228 |
4594 | 229 int delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc ); |
230 | |
231 pix[-xstride] = av_clip_uint8( p0 + delta ); /* p0' */ | |
232 pix[0] = av_clip_uint8( q0 - delta ); /* q0' */ | |
2633 | 233 } |
234 pix += ystride; | |
235 } | |
236 } | |
237 } | |
2707
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
238 static void h264_v_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) |
2633 | 239 { |
240 h264_loop_filter_chroma_c(pix, stride, 1, alpha, beta, tc0); | |
241 } | |
2707
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
242 static void h264_h_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0) |
2633 | 243 { |
244 h264_loop_filter_chroma_c(pix, 1, stride, alpha, beta, tc0); | |
245 } | |
246 | |
10941
28edcc8c54c0
Mark the h264 c loop filter as av_always_inline av_flatten to make sure its
michael
parents:
10940
diff
changeset
|
247 static av_always_inline av_flatten void h264_loop_filter_chroma_intra_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta) |
2707
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
248 { |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
249 int d; |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
250 for( d = 0; d < 8; d++ ) { |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
251 const int p0 = pix[-1*xstride]; |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
252 const int p1 = pix[-2*xstride]; |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
253 const int q0 = pix[0]; |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
254 const int q1 = pix[1*xstride]; |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
255 |
4001 | 256 if( FFABS( p0 - q0 ) < alpha && |
257 FFABS( p1 - p0 ) < beta && | |
258 FFABS( q1 - q0 ) < beta ) { | |
2707
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
259 |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
260 pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2; /* p0' */ |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
261 pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2; /* q0' */ |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
262 } |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
263 pix += ystride; |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
264 } |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
265 } |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
266 static void h264_v_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta) |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
267 { |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
268 h264_loop_filter_chroma_intra_c(pix, stride, 1, alpha, beta); |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
269 } |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
270 static void h264_h_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta) |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
271 { |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
272 h264_loop_filter_chroma_intra_c(pix, 1, stride, alpha, beta); |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
273 } |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
274 |
11499 | 275 void ff_h264dsp_init(H264DSPContext *c) |
296 | 276 { |
11499 | 277 c->h264_idct_add= ff_h264_idct_add_c; |
278 c->h264_idct8_add= ff_h264_idct8_add_c; | |
279 c->h264_idct_dc_add= ff_h264_idct_dc_add_c; | |
280 c->h264_idct8_dc_add= ff_h264_idct8_dc_add_c; | |
281 c->h264_idct_add16 = ff_h264_idct_add16_c; | |
282 c->h264_idct8_add4 = ff_h264_idct8_add4_c; | |
283 c->h264_idct_add8 = ff_h264_idct_add8_c; | |
284 c->h264_idct_add16intra= ff_h264_idct_add16intra_c; | |
857 | 285 |
2415 | 286 c->weight_h264_pixels_tab[0]= weight_h264_pixels16x16_c; |
287 c->weight_h264_pixels_tab[1]= weight_h264_pixels16x8_c; | |
288 c->weight_h264_pixels_tab[2]= weight_h264_pixels8x16_c; | |
289 c->weight_h264_pixels_tab[3]= weight_h264_pixels8x8_c; | |
290 c->weight_h264_pixels_tab[4]= weight_h264_pixels8x4_c; | |
291 c->weight_h264_pixels_tab[5]= weight_h264_pixels4x8_c; | |
292 c->weight_h264_pixels_tab[6]= weight_h264_pixels4x4_c; | |
293 c->weight_h264_pixels_tab[7]= weight_h264_pixels4x2_c; | |
294 c->weight_h264_pixels_tab[8]= weight_h264_pixels2x4_c; | |
295 c->weight_h264_pixels_tab[9]= weight_h264_pixels2x2_c; | |
296 c->biweight_h264_pixels_tab[0]= biweight_h264_pixels16x16_c; | |
297 c->biweight_h264_pixels_tab[1]= biweight_h264_pixels16x8_c; | |
298 c->biweight_h264_pixels_tab[2]= biweight_h264_pixels8x16_c; | |
299 c->biweight_h264_pixels_tab[3]= biweight_h264_pixels8x8_c; | |
300 c->biweight_h264_pixels_tab[4]= biweight_h264_pixels8x4_c; | |
301 c->biweight_h264_pixels_tab[5]= biweight_h264_pixels4x8_c; | |
302 c->biweight_h264_pixels_tab[6]= biweight_h264_pixels4x4_c; | |
303 c->biweight_h264_pixels_tab[7]= biweight_h264_pixels4x2_c; | |
304 c->biweight_h264_pixels_tab[8]= biweight_h264_pixels2x4_c; | |
305 c->biweight_h264_pixels_tab[9]= biweight_h264_pixels2x2_c; | |
306 | |
2633 | 307 c->h264_v_loop_filter_luma= h264_v_loop_filter_luma_c; |
308 c->h264_h_loop_filter_luma= h264_h_loop_filter_luma_c; | |
8395
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
309 c->h264_v_loop_filter_luma_intra= h264_v_loop_filter_luma_intra_c; |
195cba8f6257
Move filter_luma_intra into dsputil for later addition of asm.
darkshikari
parents:
8375
diff
changeset
|
310 c->h264_h_loop_filter_luma_intra= h264_h_loop_filter_luma_intra_c; |
2633 | 311 c->h264_v_loop_filter_chroma= h264_v_loop_filter_chroma_c; |
312 c->h264_h_loop_filter_chroma= h264_h_loop_filter_chroma_c; | |
2707
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
313 c->h264_v_loop_filter_chroma_intra= h264_v_loop_filter_chroma_intra_c; |
360024d31dab
H.264 deblocking optimizations (mmx for chroma_bS4 case, convert existing cases to 8-bit math)
lorenm
parents:
2696
diff
changeset
|
314 c->h264_h_loop_filter_chroma_intra= h264_h_loop_filter_chroma_intra_c; |
3645
47821be55b6c
mmx implementation of deblocking strength decision.
lorenm
parents:
3568
diff
changeset
|
315 c->h264_loop_filter_strength= NULL; |
2967 | 316 |
11499 | 317 if (ARCH_ARM) ff_h264dsp_init_arm(c); |
11506 | 318 if (HAVE_ALTIVEC) ff_h264dsp_init_ppc(c); |
11499 | 319 if (HAVE_MMX) ff_h264dsp_init_x86(c); |
0 | 320 } |