1/*
2 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
3 * Copyright (c) 2003-2010 Michael Niedermayer <michaelni@gmx.at>
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22/**
23 * @file
24 * H.264 / AVC / MPEG4 part10 DSP functions.
25 * @author Michael Niedermayer <michaelni@gmx.at>
26 */
27
28#include <stdint.h>
29#include "avcodec.h"
30#include "h264dsp.h"
31
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))
34#define H264_WEIGHT(W,H) \
35static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride, int log2_denom, int weight, int offset){ \
36    int y; \
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} \
61static 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; \
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
87H264_WEIGHT(16,16)
88H264_WEIGHT(16,8)
89H264_WEIGHT(8,16)
90H264_WEIGHT(8,8)
91H264_WEIGHT(8,4)
92H264_WEIGHT(4,8)
93H264_WEIGHT(4,4)
94H264_WEIGHT(4,2)
95H264_WEIGHT(2,4)
96H264_WEIGHT(2,2)
97
98#undef op_scale1
99#undef op_scale2
100#undef H264_WEIGHT
101
102static 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)
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];
117
118            if( FFABS( p0 - q0 ) < alpha &&
119                FFABS( p1 - p0 ) < beta &&
120                FFABS( q1 - q0 ) < beta ) {
121
122                int tc = tc0[i];
123                int i_delta;
124
125                if( FFABS( p2 - p0 ) < beta ) {
126                    if(tc0[i])
127                    pix[-2*xstride] = p1 + av_clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
128                    tc++;
129                }
130                if( FFABS( q2 - q0 ) < beta ) {
131                    if(tc0[i])
132                    pix[   xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
133                    tc++;
134                }
135
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' */
139            }
140            pix += ystride;
141        }
142    }
143}
144static void h264_v_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
145{
146    h264_loop_filter_luma_c(pix, stride, 1, alpha, beta, tc0);
147}
148static void h264_h_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
149{
150    h264_loop_filter_luma_c(pix, 1, stride, alpha, beta, tc0);
151}
152
153static av_always_inline av_flatten void h264_loop_filter_luma_intra_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta)
154{
155    int d;
156    for( d = 0; d < 16; d++ ) {
157        const int p2 = pix[-3*xstride];
158        const int p1 = pix[-2*xstride];
159        const int p0 = pix[-1*xstride];
160
161        const int q0 = pix[ 0*xstride];
162        const int q1 = pix[ 1*xstride];
163        const int q2 = pix[ 2*xstride];
164
165        if( FFABS( p0 - q0 ) < alpha &&
166            FFABS( p1 - p0 ) < beta &&
167            FFABS( q1 - q0 ) < beta ) {
168
169            if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
170                if( FFABS( p2 - p0 ) < beta)
171                {
172                    const int p3 = pix[-4*xstride];
173                    /* p0', p1', p2' */
174                    pix[-1*xstride] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
175                    pix[-2*xstride] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
176                    pix[-3*xstride] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
177                } else {
178                    /* p0' */
179                    pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
180                }
181                if( FFABS( q2 - q0 ) < beta)
182                {
183                    const int q3 = pix[3*xstride];
184                    /* q0', q1', q2' */
185                    pix[0*xstride] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
186                    pix[1*xstride] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
187                    pix[2*xstride] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
188                } else {
189                    /* q0' */
190                    pix[0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
191                }
192            }else{
193                /* p0', q0' */
194                pix[-1*xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
195                pix[ 0*xstride] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
196            }
197        }
198        pix += ystride;
199    }
200}
201static void h264_v_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
202{
203    h264_loop_filter_luma_intra_c(pix, stride, 1, alpha, beta);
204}
205static void h264_h_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
206{
207    h264_loop_filter_luma_intra_c(pix, 1, stride, alpha, beta);
208}
209
210static 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)
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
225            if( FFABS( p0 - q0 ) < alpha &&
226                FFABS( p1 - p0 ) < beta &&
227                FFABS( q1 - q0 ) < beta ) {
228
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' */
233            }
234            pix += ystride;
235        }
236    }
237}
238static void h264_v_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
239{
240    h264_loop_filter_chroma_c(pix, stride, 1, alpha, beta, tc0);
241}
242static void h264_h_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
243{
244    h264_loop_filter_chroma_c(pix, 1, stride, alpha, beta, tc0);
245}
246
247static av_always_inline av_flatten void h264_loop_filter_chroma_intra_c(uint8_t *pix, int xstride, int ystride, int alpha, int beta)
248{
249    int d;
250    for( d = 0; d < 8; d++ ) {
251        const int p0 = pix[-1*xstride];
252        const int p1 = pix[-2*xstride];
253        const int q0 = pix[0];
254        const int q1 = pix[1*xstride];
255
256        if( FFABS( p0 - q0 ) < alpha &&
257            FFABS( p1 - p0 ) < beta &&
258            FFABS( q1 - q0 ) < beta ) {
259
260            pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */
261            pix[0]        = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */
262        }
263        pix += ystride;
264    }
265}
266static void h264_v_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
267{
268    h264_loop_filter_chroma_intra_c(pix, stride, 1, alpha, beta);
269}
270static void h264_h_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
271{
272    h264_loop_filter_chroma_intra_c(pix, 1, stride, alpha, beta);
273}
274
275void ff_h264dsp_init(H264DSPContext *c)
276{
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;
285
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
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;
309    c->h264_v_loop_filter_luma_intra= h264_v_loop_filter_luma_intra_c;
310    c->h264_h_loop_filter_luma_intra= h264_h_loop_filter_luma_intra_c;
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;
313    c->h264_v_loop_filter_chroma_intra= h264_v_loop_filter_chroma_intra_c;
314    c->h264_h_loop_filter_chroma_intra= h264_h_loop_filter_chroma_intra_c;
315    c->h264_loop_filter_strength= NULL;
316
317    if (ARCH_ARM) ff_h264dsp_init_arm(c);
318    if (HAVE_ALTIVEC) ff_h264dsp_init_ppc(c);
319    if (HAVE_MMX) ff_h264dsp_init_x86(c);
320}
321