2 * Copyright (C) 2004 the ffmpeg project
4 * This file is part of FFmpeg.
6 * FFmpeg 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 * FFmpeg 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 FFmpeg; 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)
44 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
46 int A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;
47 int Ed, Gd, Add, Bdd, Fd, Hd;
51 /* Inverse DCT on the rows now */
52 for (i = 0; i < 8; i++) {
53 /* Check for non-zero values */
54 if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
55 A = M(xC1S7, ip[1]) + M(xC7S1, ip[7]);
56 B = M(xC7S1, ip[1]) - M(xC1S7, ip[7]);
57 C = M(xC3S5, ip[3]) + M(xC5S3, ip[5]);
58 D = M(xC3S5, ip[5]) - M(xC5S3, ip[3]);
60 Ad = M(xC4S4, (A - C));
61 Bd = M(xC4S4, (B - D));
66 E = M(xC4S4, (ip[0] + ip[4]));
67 F = M(xC4S4, (ip[0] - ip[4]));
69 G = M(xC2S6, ip[2]) + M(xC6S2, ip[6]);
70 H = M(xC6S2, ip[2]) - M(xC2S6, ip[6]);
81 /* Final sequence of operations over-write original inputs. */
95 ip += 8; /* next row */
100 for ( i = 0; i < 8; i++) {
101 /* Check for non-zero values (bitwise or faster than ||) */
102 if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
103 ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
105 A = M(xC1S7, ip[1*8]) + M(xC7S1, ip[7*8]);
106 B = M(xC7S1, ip[1*8]) - M(xC1S7, ip[7*8]);
107 C = M(xC3S5, ip[3*8]) + M(xC5S3, ip[5*8]);
108 D = M(xC3S5, ip[5*8]) - M(xC5S3, ip[3*8]);
110 Ad = M(xC4S4, (A - C));
111 Bd = M(xC4S4, (B - D));
116 E = M(xC4S4, (ip[0*8] + ip[4*8])) + 8;
117 F = M(xC4S4, (ip[0*8] - ip[4*8])) + 8;
124 G = M(xC2S6, ip[2*8]) + M(xC6S2, ip[6*8]);
125 H = M(xC6S2, ip[2*8]) - M(xC2S6, ip[6*8]);
136 /* Final sequence of operations over-write original inputs. */
138 ip[0*8] = (Gd + Cd ) >> 4;
139 ip[7*8] = (Gd - Cd ) >> 4;
141 ip[1*8] = (Add + Hd ) >> 4;
142 ip[2*8] = (Add - Hd ) >> 4;
144 ip[3*8] = (Ed + Dd ) >> 4;
145 ip[4*8] = (Ed - Dd ) >> 4;
147 ip[5*8] = (Fd + Bdd ) >> 4;
148 ip[6*8] = (Fd - Bdd ) >> 4;
150 dst[0*stride] = cm[(Gd + Cd ) >> 4];
151 dst[7*stride] = cm[(Gd - Cd ) >> 4];
153 dst[1*stride] = cm[(Add + Hd ) >> 4];
154 dst[2*stride] = cm[(Add - Hd ) >> 4];
156 dst[3*stride] = cm[(Ed + Dd ) >> 4];
157 dst[4*stride] = cm[(Ed - Dd ) >> 4];
159 dst[5*stride] = cm[(Fd + Bdd ) >> 4];
160 dst[6*stride] = cm[(Fd - Bdd ) >> 4];
162 dst[0*stride] = cm[dst[0*stride] + ((Gd + Cd ) >> 4)];
163 dst[7*stride] = cm[dst[7*stride] + ((Gd - Cd ) >> 4)];
165 dst[1*stride] = cm[dst[1*stride] + ((Add + Hd ) >> 4)];
166 dst[2*stride] = cm[dst[2*stride] + ((Add - Hd ) >> 4)];
168 dst[3*stride] = cm[dst[3*stride] + ((Ed + Dd ) >> 4)];
169 dst[4*stride] = cm[dst[4*stride] + ((Ed - Dd ) >> 4)];
171 dst[5*stride] = cm[dst[5*stride] + ((Fd + Bdd ) >> 4)];
172 dst[6*stride] = cm[dst[6*stride] + ((Fd - Bdd ) >> 4)];
184 ip[7*8] = ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
193 dst[7*stride]= cm[128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20)];
196 int v= ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
197 dst[0*stride] = cm[dst[0*stride] + v];
198 dst[1*stride] = cm[dst[1*stride] + v];
199 dst[2*stride] = cm[dst[2*stride] + v];
200 dst[3*stride] = cm[dst[3*stride] + v];
201 dst[4*stride] = cm[dst[4*stride] + v];
202 dst[5*stride] = cm[dst[5*stride] + v];
203 dst[6*stride] = cm[dst[6*stride] + v];
204 dst[7*stride] = cm[dst[7*stride] + v];
209 ip++; /* next column */
214 void ff_vp3_idct_c(DCTELEM *block/* align 16*/){
215 idct(NULL, 0, block, 0);
218 void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
219 idct(dest, line_size, block, 1);
222 void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
223 idct(dest, line_size, block, 2);
226 void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/){
227 int i, dc = (block[0] + 15) >> 5;
228 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP + dc;
230 for(i = 0; i < 8; i++){
231 dest[0] = cm[dest[0]];
232 dest[1] = cm[dest[1]];
233 dest[2] = cm[dest[2]];
234 dest[3] = cm[dest[3]];
235 dest[4] = cm[dest[4]];
236 dest[5] = cm[dest[5]];
237 dest[6] = cm[dest[6]];
238 dest[7] = cm[dest[7]];
243 void ff_vp3_v_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_values)
247 const int nstride= -stride;
249 for (end= first_pixel + 8; first_pixel < end; first_pixel++) {
251 (first_pixel[2 * nstride] - first_pixel[ stride])
252 +3*(first_pixel[0 ] - first_pixel[nstride]);
253 filter_value = bounding_values[(filter_value + 4) >> 3];
254 first_pixel[nstride] = av_clip_uint8(first_pixel[nstride] + filter_value);
255 first_pixel[0] = av_clip_uint8(first_pixel[0] - filter_value);
259 void ff_vp3_h_loop_filter_c(uint8_t *first_pixel, int stride, int *bounding_values)
264 for (end= first_pixel + 8*stride; first_pixel != end; first_pixel += stride) {
266 (first_pixel[-2] - first_pixel[ 1])
267 +3*(first_pixel[ 0] - first_pixel[-1]);
268 filter_value = bounding_values[(filter_value + 4) >> 3];
269 first_pixel[-1] = av_clip_uint8(first_pixel[-1] + filter_value);
270 first_pixel[ 0] = av_clip_uint8(first_pixel[ 0] - filter_value);