2 * Copyright (C) 2004 the ffmpeg project
4 * This file is part of Libav.
6 * Libav is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * Libav is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 * Standard C DSP-oriented functions cribbed from the original VP3
30 #define IdctAdjustBeforeShift 8
39 #define M(a,b) (((a) * (b))>>16)
41 static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
45 int A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;
46 int Ed, Gd, Add, Bdd, Fd, Hd;
50 /* Inverse DCT on the rows now */
51 for (i = 0; i < 8; i++) {
52 /* Check for non-zero values */
53 if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
54 A = M(xC1S7, ip[1]) + M(xC7S1, ip[7]);
55 B = M(xC7S1, ip[1]) - M(xC1S7, ip[7]);
56 C = M(xC3S5, ip[3]) + M(xC5S3, ip[5]);
57 D = M(xC3S5, ip[5]) - M(xC5S3, ip[3]);
59 Ad = M(xC4S4, (A - C));
60 Bd = M(xC4S4, (B - D));
65 E = M(xC4S4, (ip[0] + ip[4]));
66 F = M(xC4S4, (ip[0] - ip[4]));
68 G = M(xC2S6, ip[2]) + M(xC6S2, ip[6]);
69 H = M(xC6S2, ip[2]) - M(xC2S6, ip[6]);
80 /* Final sequence of operations over-write original inputs. */
94 ip += 8; /* next row */
99 for ( i = 0; i < 8; i++) {
100 /* Check for non-zero values (bitwise or faster than ||) */
101 if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
102 ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
104 A = M(xC1S7, ip[1*8]) + M(xC7S1, ip[7*8]);
105 B = M(xC7S1, ip[1*8]) - M(xC1S7, ip[7*8]);
106 C = M(xC3S5, ip[3*8]) + M(xC5S3, ip[5*8]);
107 D = M(xC3S5, ip[5*8]) - M(xC5S3, ip[3*8]);
109 Ad = M(xC4S4, (A - C));
110 Bd = M(xC4S4, (B - D));
115 E = M(xC4S4, (ip[0*8] + ip[4*8])) + 8;
116 F = M(xC4S4, (ip[0*8] - ip[4*8])) + 8;
123 G = M(xC2S6, ip[2*8]) + M(xC6S2, ip[6*8]);
124 H = M(xC6S2, ip[2*8]) - M(xC2S6, ip[6*8]);
135 /* Final sequence of operations over-write original inputs. */
137 ip[0*8] = (Gd + Cd ) >> 4;
138 ip[7*8] = (Gd - Cd ) >> 4;
140 ip[1*8] = (Add + Hd ) >> 4;
141 ip[2*8] = (Add - Hd ) >> 4;
143 ip[3*8] = (Ed + Dd ) >> 4;
144 ip[4*8] = (Ed - Dd ) >> 4;
146 ip[5*8] = (Fd + Bdd ) >> 4;
147 ip[6*8] = (Fd - Bdd ) >> 4;
149 dst[0*stride] = av_clip_uint8((Gd + Cd ) >> 4);
150 dst[7*stride] = av_clip_uint8((Gd - Cd ) >> 4);
152 dst[1*stride] = av_clip_uint8((Add + Hd ) >> 4);
153 dst[2*stride] = av_clip_uint8((Add - Hd ) >> 4);
155 dst[3*stride] = av_clip_uint8((Ed + Dd ) >> 4);
156 dst[4*stride] = av_clip_uint8((Ed - Dd ) >> 4);
158 dst[5*stride] = av_clip_uint8((Fd + Bdd ) >> 4);
159 dst[6*stride] = av_clip_uint8((Fd - Bdd ) >> 4);
161 dst[0*stride] = av_clip_uint8(dst[0*stride] + ((Gd + Cd ) >> 4));
162 dst[7*stride] = av_clip_uint8(dst[7*stride] + ((Gd - Cd ) >> 4));
164 dst[1*stride] = av_clip_uint8(dst[1*stride] + ((Add + Hd ) >> 4));
165 dst[2*stride] = av_clip_uint8(dst[2*stride] + ((Add - Hd ) >> 4));
167 dst[3*stride] = av_clip_uint8(dst[3*stride] + ((Ed + Dd ) >> 4));
168 dst[4*stride] = av_clip_uint8(dst[4*stride] + ((Ed - Dd ) >> 4));
170 dst[5*stride] = av_clip_uint8(dst[5*stride] + ((Fd + Bdd ) >> 4));
171 dst[6*stride] = av_clip_uint8(dst[6*stride] + ((Fd - Bdd ) >> 4));
183 ip[7*8] = ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
192 dst[7*stride]= av_clip_uint8(128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20));
195 int v= ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
196 dst[0*stride] = av_clip_uint8(dst[0*stride] + v);
197 dst[1*stride] = av_clip_uint8(dst[1*stride] + v);
198 dst[2*stride] = av_clip_uint8(dst[2*stride] + v);
199 dst[3*stride] = av_clip_uint8(dst[3*stride] + v);
200 dst[4*stride] = av_clip_uint8(dst[4*stride] + v);
201 dst[5*stride] = av_clip_uint8(dst[5*stride] + v);
202 dst[6*stride] = av_clip_uint8(dst[6*stride] + v);
203 dst[7*stride] = av_clip_uint8(dst[7*stride] + v);
208 ip++; /* next column */
213 void ff_vp3_idct_c(DCTELEM *block/* align 16*/){
214 idct(NULL, 0, block, 0);
217 void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
218 idct(dest, line_size, block, 1);
221 void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
222 idct(dest, line_size, block, 2);
225 void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/){
226 int i, dc = (block[0] + 15) >> 5;
228 for(i = 0; i < 8; i++){
229 dest[0] = av_clip_uint8(dest[0] + dc);
230 dest[1] = av_clip_uint8(dest[1] + dc);
231 dest[2] = av_clip_uint8(dest[2] + dc);
232 dest[3] = av_clip_uint8(dest[3] + dc);
233 dest[4] = av_clip_uint8(dest[4] + dc);
234 dest[5] = av_clip_uint8(dest[5] + dc);
235 dest[6] = av_clip_uint8(dest[6] + dc);
236 dest[7] = av_clip_uint8(dest[7] + dc);
241 void ff_vp3_v_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_values)
245 const int nstride= -stride;
247 for (end= first_pixel + 8; first_pixel < end; first_pixel++) {
249 (first_pixel[2 * nstride] - first_pixel[ stride])
250 +3*(first_pixel[0 ] - first_pixel[nstride]);
251 filter_value = bounding_values[(filter_value + 4) >> 3];
252 first_pixel[nstride] = av_clip_uint8(first_pixel[nstride] + filter_value);
253 first_pixel[0] = av_clip_uint8(first_pixel[0] - filter_value);
257 void ff_vp3_h_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_values)
262 for (end= first_pixel + 8*stride; first_pixel != end; first_pixel += stride) {
264 (first_pixel[-2] - first_pixel[ 1])
265 +3*(first_pixel[ 0] - first_pixel[-1]);
266 filter_value = bounding_values[(filter_value + 4) >> 3];
267 first_pixel[-1] = av_clip_uint8(first_pixel[-1] + filter_value);
268 first_pixel[ 0] = av_clip_uint8(first_pixel[ 0] - filter_value);