2 * Copyright (C) 2004 the ffmpeg project
4 * This library is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2 of the License, or (at your option) any later version.
9 * This library is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with this library; if not, write to the Free Software
16 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21 * Standard C DSP-oriented functions cribbed from the original VP3
29 #define IdctAdjustBeforeShift 8
38 static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
41 uint8_t *cm = cropTbl + MAX_NEG_CROP;
43 int A_, B_, C_, D_, _Ad, _Bd, _Cd, _Dd, E_, F_, G_, H_;
44 int _Ed, _Gd, _Add, _Bdd, _Fd, _Hd;
49 /* Inverse DCT on the rows now */
50 for (i = 0; i < 8; i++) {
51 /* Check for non-zero values */
52 if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
53 t1 = (int32_t)(xC1S7 * ip[1]);
54 t2 = (int32_t)(xC7S1 * ip[7]);
59 t1 = (int32_t)(xC7S1 * ip[1]);
60 t2 = (int32_t)(xC1S7 * ip[7]);
65 t1 = (int32_t)(xC3S5 * ip[3]);
66 t2 = (int32_t)(xC5S3 * ip[5]);
71 t1 = (int32_t)(xC3S5 * ip[5]);
72 t2 = (int32_t)(xC5S3 * ip[3]);
78 t1 = (int32_t)(xC4S4 * (A_ - C_));
82 t1 = (int32_t)(xC4S4 * (B_ - D_));
90 t1 = (int32_t)(xC4S4 * (ip[0] + ip[4]));
94 t1 = (int32_t)(xC4S4 * (ip[0] - ip[4]));
98 t1 = (int32_t)(xC2S6 * ip[2]);
99 t2 = (int32_t)(xC6S2 * ip[6]);
104 t1 = (int32_t)(xC6S2 * ip[2]);
105 t2 = (int32_t)(xC2S6 * ip[6]);
120 /* Final sequence of operations over-write original inputs. */
135 ip += 8; /* next row */
140 for ( i = 0; i < 8; i++) {
141 /* Check for non-zero values (bitwise or faster than ||) */
142 if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
143 ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
145 t1 = (int32_t)(xC1S7 * ip[1*8]);
146 t2 = (int32_t)(xC7S1 * ip[7*8]);
151 t1 = (int32_t)(xC7S1 * ip[1*8]);
152 t2 = (int32_t)(xC1S7 * ip[7*8]);
157 t1 = (int32_t)(xC3S5 * ip[3*8]);
158 t2 = (int32_t)(xC5S3 * ip[5*8]);
163 t1 = (int32_t)(xC3S5 * ip[5*8]);
164 t2 = (int32_t)(xC5S3 * ip[3*8]);
170 t1 = (int32_t)(xC4S4 * (A_ - C_));
174 t1 = (int32_t)(xC4S4 * (B_ - D_));
182 t1 = (int32_t)(xC4S4 * (ip[0*8] + ip[4*8]));
186 t1 = (int32_t)(xC4S4 * (ip[0*8] - ip[4*8]));
190 t1 = (int32_t)(xC2S6 * ip[2*8]);
191 t2 = (int32_t)(xC6S2 * ip[6*8]);
196 t1 = (int32_t)(xC6S2 * ip[2*8]);
197 t2 = (int32_t)(xC2S6 * ip[6*8]);
218 _Gd += IdctAdjustBeforeShift;
219 _Add += IdctAdjustBeforeShift;
220 _Ed += IdctAdjustBeforeShift;
221 _Fd += IdctAdjustBeforeShift;
223 /* Final sequence of operations over-write original inputs. */
225 ip[0*8] = (_Gd + _Cd ) >> 4;
226 ip[7*8] = (_Gd - _Cd ) >> 4;
228 ip[1*8] = (_Add + _Hd ) >> 4;
229 ip[2*8] = (_Add - _Hd ) >> 4;
231 ip[3*8] = (_Ed + _Dd ) >> 4;
232 ip[4*8] = (_Ed - _Dd ) >> 4;
234 ip[5*8] = (_Fd + _Bdd ) >> 4;
235 ip[6*8] = (_Fd - _Bdd ) >> 4;
237 dst[0*stride] = cm[(_Gd + _Cd ) >> 4];
238 dst[7*stride] = cm[(_Gd - _Cd ) >> 4];
240 dst[1*stride] = cm[(_Add + _Hd ) >> 4];
241 dst[2*stride] = cm[(_Add - _Hd ) >> 4];
243 dst[3*stride] = cm[(_Ed + _Dd ) >> 4];
244 dst[4*stride] = cm[(_Ed - _Dd ) >> 4];
246 dst[5*stride] = cm[(_Fd + _Bdd ) >> 4];
247 dst[6*stride] = cm[(_Fd - _Bdd ) >> 4];
249 dst[0*stride] = cm[dst[0*stride] + ((_Gd + _Cd ) >> 4)];
250 dst[7*stride] = cm[dst[7*stride] + ((_Gd - _Cd ) >> 4)];
252 dst[1*stride] = cm[dst[1*stride] + ((_Add + _Hd ) >> 4)];
253 dst[2*stride] = cm[dst[2*stride] + ((_Add - _Hd ) >> 4)];
255 dst[3*stride] = cm[dst[3*stride] + ((_Ed + _Dd ) >> 4)];
256 dst[4*stride] = cm[dst[4*stride] + ((_Ed - _Dd ) >> 4)];
258 dst[5*stride] = cm[dst[5*stride] + ((_Fd + _Bdd ) >> 4)];
259 dst[6*stride] = cm[dst[6*stride] + ((_Fd - _Bdd ) >> 4)];
271 ip[7*8] = ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
280 dst[7*stride]= 128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
283 int v= ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
284 dst[0*stride] = cm[dst[0*stride] + v];
285 dst[1*stride] = cm[dst[1*stride] + v];
286 dst[2*stride] = cm[dst[2*stride] + v];
287 dst[3*stride] = cm[dst[3*stride] + v];
288 dst[4*stride] = cm[dst[4*stride] + v];
289 dst[5*stride] = cm[dst[5*stride] + v];
290 dst[6*stride] = cm[dst[6*stride] + v];
291 dst[7*stride] = cm[dst[7*stride] + v];
296 ip++; /* next column */
301 void ff_vp3_idct_c(DCTELEM *block/* align 16*/){
302 idct(NULL, 0, block, 0);
305 void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
306 idct(dest, line_size, block, 1);
309 void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
310 idct(dest, line_size, block, 2);