Libav 0.7.1
libavcodec/vc1dsp.c
Go to the documentation of this file.
00001 /*
00002  * VC-1 and WMV3 decoder - DSP functions
00003  * Copyright (c) 2006 Konstantin Shishkov
00004  *
00005  * This file is part of Libav.
00006  *
00007  * Libav is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * Libav is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with Libav; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00028 #include "vc1dsp.h"
00029 
00030 
00033 static void vc1_v_overlap_c(uint8_t* src, int stride)
00034 {
00035     int i;
00036     int a, b, c, d;
00037     int d1, d2;
00038     int rnd = 1;
00039     for(i = 0; i < 8; i++) {
00040         a = src[-2*stride];
00041         b = src[-stride];
00042         c = src[0];
00043         d = src[stride];
00044         d1 = (a - d + 3 + rnd) >> 3;
00045         d2 = (a - d + b - c + 4 - rnd) >> 3;
00046 
00047         src[-2*stride] = a - d1;
00048         src[-stride] = av_clip_uint8(b - d2);
00049         src[0] = av_clip_uint8(c + d2);
00050         src[stride] = d + d1;
00051         src++;
00052         rnd = !rnd;
00053     }
00054 }
00055 
00058 static void vc1_h_overlap_c(uint8_t* src, int stride)
00059 {
00060     int i;
00061     int a, b, c, d;
00062     int d1, d2;
00063     int rnd = 1;
00064     for(i = 0; i < 8; i++) {
00065         a = src[-2];
00066         b = src[-1];
00067         c = src[0];
00068         d = src[1];
00069         d1 = (a - d + 3 + rnd) >> 3;
00070         d2 = (a - d + b - c + 4 - rnd) >> 3;
00071 
00072         src[-2] = a - d1;
00073         src[-1] = av_clip_uint8(b - d2);
00074         src[0] = av_clip_uint8(c + d2);
00075         src[1] = d + d1;
00076         src += stride;
00077         rnd = !rnd;
00078     }
00079 }
00080 
00081 static void vc1_v_s_overlap_c(DCTELEM *top,  DCTELEM *bottom)
00082 {
00083     int i;
00084     int a, b, c, d;
00085     int d1, d2;
00086     int rnd1 = 4, rnd2 = 3;
00087     for(i = 0; i < 8; i++) {
00088         a = top[48];
00089         b = top[56];
00090         c = bottom[0];
00091         d = bottom[8];
00092         d1 = a - d;
00093         d2 = a - d + b - c;
00094 
00095         top[48]   = ((a << 3) - d1 + rnd1) >> 3;
00096         top[56]   = ((b << 3) - d2 + rnd2) >> 3;
00097         bottom[0] = ((c << 3) + d2 + rnd1) >> 3;
00098         bottom[8] = ((d << 3) + d1 + rnd2) >> 3;
00099 
00100         bottom++;
00101         top++;
00102         rnd2 = 7 - rnd2;
00103         rnd1 = 7 - rnd1;
00104     }
00105 }
00106 
00107 static void vc1_h_s_overlap_c(DCTELEM *left, DCTELEM *right)
00108 {
00109     int i;
00110     int a, b, c, d;
00111     int d1, d2;
00112     int rnd1 = 4, rnd2 = 3;
00113     for(i = 0; i < 8; i++) {
00114         a = left[6];
00115         b = left[7];
00116         c = right[0];
00117         d = right[1];
00118         d1 = a - d;
00119         d2 = a - d + b - c;
00120 
00121         left[6]  = ((a << 3) - d1 + rnd1) >> 3;
00122         left[7]  = ((b << 3) - d2 + rnd2) >> 3;
00123         right[0] = ((c << 3) + d2 + rnd1) >> 3;
00124         right[1] = ((d << 3) + d1 + rnd2) >> 3;
00125 
00126         right += 8;
00127         left += 8;
00128         rnd2 = 7 - rnd2;
00129         rnd1 = 7 - rnd1;
00130     }
00131 }
00132 
00141 static av_always_inline int vc1_filter_line(uint8_t* src, int stride, int pq){
00142     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
00143 
00144     int a0 = (2*(src[-2*stride] - src[ 1*stride]) - 5*(src[-1*stride] - src[ 0*stride]) + 4) >> 3;
00145     int a0_sign = a0 >> 31;        /* Store sign */
00146     a0 = (a0 ^ a0_sign) - a0_sign; /* a0 = FFABS(a0); */
00147     if(a0 < pq){
00148         int a1 = FFABS((2*(src[-4*stride] - src[-1*stride]) - 5*(src[-3*stride] - src[-2*stride]) + 4) >> 3);
00149         int a2 = FFABS((2*(src[ 0*stride] - src[ 3*stride]) - 5*(src[ 1*stride] - src[ 2*stride]) + 4) >> 3);
00150         if(a1 < a0 || a2 < a0){
00151             int clip = src[-1*stride] - src[ 0*stride];
00152             int clip_sign = clip >> 31;
00153             clip = ((clip ^ clip_sign) - clip_sign)>>1;
00154             if(clip){
00155                 int a3 = FFMIN(a1, a2);
00156                 int d = 5 * (a3 - a0);
00157                 int d_sign = (d >> 31);
00158                 d = ((d ^ d_sign) - d_sign) >> 3;
00159                 d_sign ^= a0_sign;
00160 
00161                 if( d_sign ^ clip_sign )
00162                     d = 0;
00163                 else{
00164                     d = FFMIN(d, clip);
00165                     d = (d ^ d_sign) - d_sign;          /* Restore sign */
00166                     src[-1*stride] = cm[src[-1*stride] - d];
00167                     src[ 0*stride] = cm[src[ 0*stride] + d];
00168                 }
00169                 return 1;
00170             }
00171         }
00172     }
00173     return 0;
00174 }
00175 
00185 static inline void vc1_loop_filter(uint8_t* src, int step, int stride, int len, int pq)
00186 {
00187     int i;
00188     int filt3;
00189 
00190     for(i = 0; i < len; i += 4){
00191         filt3 = vc1_filter_line(src + 2*step, stride, pq);
00192         if(filt3){
00193             vc1_filter_line(src + 0*step, stride, pq);
00194             vc1_filter_line(src + 1*step, stride, pq);
00195             vc1_filter_line(src + 3*step, stride, pq);
00196         }
00197         src += step * 4;
00198     }
00199 }
00200 
00201 static void vc1_v_loop_filter4_c(uint8_t *src, int stride, int pq)
00202 {
00203     vc1_loop_filter(src, 1, stride, 4, pq);
00204 }
00205 
00206 static void vc1_h_loop_filter4_c(uint8_t *src, int stride, int pq)
00207 {
00208     vc1_loop_filter(src, stride, 1, 4, pq);
00209 }
00210 
00211 static void vc1_v_loop_filter8_c(uint8_t *src, int stride, int pq)
00212 {
00213     vc1_loop_filter(src, 1, stride, 8, pq);
00214 }
00215 
00216 static void vc1_h_loop_filter8_c(uint8_t *src, int stride, int pq)
00217 {
00218     vc1_loop_filter(src, stride, 1, 8, pq);
00219 }
00220 
00221 static void vc1_v_loop_filter16_c(uint8_t *src, int stride, int pq)
00222 {
00223     vc1_loop_filter(src, 1, stride, 16, pq);
00224 }
00225 
00226 static void vc1_h_loop_filter16_c(uint8_t *src, int stride, int pq)
00227 {
00228     vc1_loop_filter(src, stride, 1, 16, pq);
00229 }
00230 
00233 static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
00234 {
00235     int i;
00236     int dc = block[0];
00237     const uint8_t *cm;
00238     dc = (3 * dc +  1) >> 1;
00239     dc = (3 * dc + 16) >> 5;
00240     cm = ff_cropTbl + MAX_NEG_CROP + dc;
00241     for(i = 0; i < 8; i++){
00242         dest[0] = cm[dest[0]];
00243         dest[1] = cm[dest[1]];
00244         dest[2] = cm[dest[2]];
00245         dest[3] = cm[dest[3]];
00246         dest[4] = cm[dest[4]];
00247         dest[5] = cm[dest[5]];
00248         dest[6] = cm[dest[6]];
00249         dest[7] = cm[dest[7]];
00250         dest += linesize;
00251     }
00252 }
00253 
00254 static void vc1_inv_trans_8x8_c(DCTELEM block[64])
00255 {
00256     int i;
00257     register int t1,t2,t3,t4,t5,t6,t7,t8;
00258     DCTELEM *src, *dst, temp[64];
00259 
00260     src = block;
00261     dst = temp;
00262     for(i = 0; i < 8; i++){
00263         t1 = 12 * (src[ 0] + src[32]) + 4;
00264         t2 = 12 * (src[ 0] - src[32]) + 4;
00265         t3 = 16 * src[16] +  6 * src[48];
00266         t4 =  6 * src[16] - 16 * src[48];
00267 
00268         t5 = t1 + t3;
00269         t6 = t2 + t4;
00270         t7 = t2 - t4;
00271         t8 = t1 - t3;
00272 
00273         t1 = 16 * src[ 8] + 15 * src[24] +  9 * src[40] +  4 * src[56];
00274         t2 = 15 * src[ 8] -  4 * src[24] - 16 * src[40] -  9 * src[56];
00275         t3 =  9 * src[ 8] - 16 * src[24] +  4 * src[40] + 15 * src[56];
00276         t4 =  4 * src[ 8] -  9 * src[24] + 15 * src[40] - 16 * src[56];
00277 
00278         dst[0] = (t5 + t1) >> 3;
00279         dst[1] = (t6 + t2) >> 3;
00280         dst[2] = (t7 + t3) >> 3;
00281         dst[3] = (t8 + t4) >> 3;
00282         dst[4] = (t8 - t4) >> 3;
00283         dst[5] = (t7 - t3) >> 3;
00284         dst[6] = (t6 - t2) >> 3;
00285         dst[7] = (t5 - t1) >> 3;
00286 
00287         src += 1;
00288         dst += 8;
00289     }
00290 
00291     src = temp;
00292     dst = block;
00293     for(i = 0; i < 8; i++){
00294         t1 = 12 * (src[ 0] + src[32]) + 64;
00295         t2 = 12 * (src[ 0] - src[32]) + 64;
00296         t3 = 16 * src[16] +  6 * src[48];
00297         t4 =  6 * src[16] - 16 * src[48];
00298 
00299         t5 = t1 + t3;
00300         t6 = t2 + t4;
00301         t7 = t2 - t4;
00302         t8 = t1 - t3;
00303 
00304         t1 = 16 * src[ 8] + 15 * src[24] +  9 * src[40] +  4 * src[56];
00305         t2 = 15 * src[ 8] -  4 * src[24] - 16 * src[40] -  9 * src[56];
00306         t3 =  9 * src[ 8] - 16 * src[24] +  4 * src[40] + 15 * src[56];
00307         t4 =  4 * src[ 8] -  9 * src[24] + 15 * src[40] - 16 * src[56];
00308 
00309         dst[ 0] = (t5 + t1) >> 7;
00310         dst[ 8] = (t6 + t2) >> 7;
00311         dst[16] = (t7 + t3) >> 7;
00312         dst[24] = (t8 + t4) >> 7;
00313         dst[32] = (t8 - t4 + 1) >> 7;
00314         dst[40] = (t7 - t3 + 1) >> 7;
00315         dst[48] = (t6 - t2 + 1) >> 7;
00316         dst[56] = (t5 - t1 + 1) >> 7;
00317 
00318         src++;
00319         dst++;
00320     }
00321 }
00322 
00325 static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
00326 {
00327     int i;
00328     int dc = block[0];
00329     const uint8_t *cm;
00330     dc = ( 3 * dc +  1) >> 1;
00331     dc = (17 * dc + 64) >> 7;
00332     cm = ff_cropTbl + MAX_NEG_CROP + dc;
00333     for(i = 0; i < 4; i++){
00334         dest[0] = cm[dest[0]];
00335         dest[1] = cm[dest[1]];
00336         dest[2] = cm[dest[2]];
00337         dest[3] = cm[dest[3]];
00338         dest[4] = cm[dest[4]];
00339         dest[5] = cm[dest[5]];
00340         dest[6] = cm[dest[6]];
00341         dest[7] = cm[dest[7]];
00342         dest += linesize;
00343     }
00344 }
00345 
00346 static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, DCTELEM *block)
00347 {
00348     int i;
00349     register int t1,t2,t3,t4,t5,t6,t7,t8;
00350     DCTELEM *src, *dst;
00351     const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
00352 
00353     src = block;
00354     dst = block;
00355     for(i = 0; i < 4; i++){
00356         t1 = 12 * (src[0] + src[4]) + 4;
00357         t2 = 12 * (src[0] - src[4]) + 4;
00358         t3 = 16 * src[2] +  6 * src[6];
00359         t4 =  6 * src[2] - 16 * src[6];
00360 
00361         t5 = t1 + t3;
00362         t6 = t2 + t4;
00363         t7 = t2 - t4;
00364         t8 = t1 - t3;
00365 
00366         t1 = 16 * src[1] + 15 * src[3] +  9 * src[5] +  4 * src[7];
00367         t2 = 15 * src[1] -  4 * src[3] - 16 * src[5] -  9 * src[7];
00368         t3 =  9 * src[1] - 16 * src[3] +  4 * src[5] + 15 * src[7];
00369         t4 =  4 * src[1] -  9 * src[3] + 15 * src[5] - 16 * src[7];
00370 
00371         dst[0] = (t5 + t1) >> 3;
00372         dst[1] = (t6 + t2) >> 3;
00373         dst[2] = (t7 + t3) >> 3;
00374         dst[3] = (t8 + t4) >> 3;
00375         dst[4] = (t8 - t4) >> 3;
00376         dst[5] = (t7 - t3) >> 3;
00377         dst[6] = (t6 - t2) >> 3;
00378         dst[7] = (t5 - t1) >> 3;
00379 
00380         src += 8;
00381         dst += 8;
00382     }
00383 
00384     src = block;
00385     for(i = 0; i < 8; i++){
00386         t1 = 17 * (src[ 0] + src[16]) + 64;
00387         t2 = 17 * (src[ 0] - src[16]) + 64;
00388         t3 = 22 * src[ 8] + 10 * src[24];
00389         t4 = 22 * src[24] - 10 * src[ 8];
00390 
00391         dest[0*linesize] = cm[dest[0*linesize] + ((t1 + t3) >> 7)];
00392         dest[1*linesize] = cm[dest[1*linesize] + ((t2 - t4) >> 7)];
00393         dest[2*linesize] = cm[dest[2*linesize] + ((t2 + t4) >> 7)];
00394         dest[3*linesize] = cm[dest[3*linesize] + ((t1 - t3) >> 7)];
00395 
00396         src ++;
00397         dest++;
00398     }
00399 }
00400 
00403 static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
00404 {
00405     int i;
00406     int dc = block[0];
00407     const uint8_t *cm;
00408     dc = (17 * dc +  4) >> 3;
00409     dc = (12 * dc + 64) >> 7;
00410     cm = ff_cropTbl + MAX_NEG_CROP + dc;
00411     for(i = 0; i < 8; i++){
00412         dest[0] = cm[dest[0]];
00413         dest[1] = cm[dest[1]];
00414         dest[2] = cm[dest[2]];
00415         dest[3] = cm[dest[3]];
00416         dest += linesize;
00417     }
00418 }
00419 
00420 static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, DCTELEM *block)
00421 {
00422     int i;
00423     register int t1,t2,t3,t4,t5,t6,t7,t8;
00424     DCTELEM *src, *dst;
00425     const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
00426 
00427     src = block;
00428     dst = block;
00429     for(i = 0; i < 8; i++){
00430         t1 = 17 * (src[0] + src[2]) + 4;
00431         t2 = 17 * (src[0] - src[2]) + 4;
00432         t3 = 22 * src[1] + 10 * src[3];
00433         t4 = 22 * src[3] - 10 * src[1];
00434 
00435         dst[0] = (t1 + t3) >> 3;
00436         dst[1] = (t2 - t4) >> 3;
00437         dst[2] = (t2 + t4) >> 3;
00438         dst[3] = (t1 - t3) >> 3;
00439 
00440         src += 8;
00441         dst += 8;
00442     }
00443 
00444     src = block;
00445     for(i = 0; i < 4; i++){
00446         t1 = 12 * (src[ 0] + src[32]) + 64;
00447         t2 = 12 * (src[ 0] - src[32]) + 64;
00448         t3 = 16 * src[16] +  6 * src[48];
00449         t4 =  6 * src[16] - 16 * src[48];
00450 
00451         t5 = t1 + t3;
00452         t6 = t2 + t4;
00453         t7 = t2 - t4;
00454         t8 = t1 - t3;
00455 
00456         t1 = 16 * src[ 8] + 15 * src[24] +  9 * src[40] +  4 * src[56];
00457         t2 = 15 * src[ 8] -  4 * src[24] - 16 * src[40] -  9 * src[56];
00458         t3 =  9 * src[ 8] - 16 * src[24] +  4 * src[40] + 15 * src[56];
00459         t4 =  4 * src[ 8] -  9 * src[24] + 15 * src[40] - 16 * src[56];
00460 
00461         dest[0*linesize] = cm[dest[0*linesize] + ((t5 + t1) >> 7)];
00462         dest[1*linesize] = cm[dest[1*linesize] + ((t6 + t2) >> 7)];
00463         dest[2*linesize] = cm[dest[2*linesize] + ((t7 + t3) >> 7)];
00464         dest[3*linesize] = cm[dest[3*linesize] + ((t8 + t4) >> 7)];
00465         dest[4*linesize] = cm[dest[4*linesize] + ((t8 - t4 + 1) >> 7)];
00466         dest[5*linesize] = cm[dest[5*linesize] + ((t7 - t3 + 1) >> 7)];
00467         dest[6*linesize] = cm[dest[6*linesize] + ((t6 - t2 + 1) >> 7)];
00468         dest[7*linesize] = cm[dest[7*linesize] + ((t5 - t1 + 1) >> 7)];
00469 
00470         src ++;
00471         dest++;
00472     }
00473 }
00474 
00477 static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
00478 {
00479     int i;
00480     int dc = block[0];
00481     const uint8_t *cm;
00482     dc = (17 * dc +  4) >> 3;
00483     dc = (17 * dc + 64) >> 7;
00484     cm = ff_cropTbl + MAX_NEG_CROP + dc;
00485     for(i = 0; i < 4; i++){
00486         dest[0] = cm[dest[0]];
00487         dest[1] = cm[dest[1]];
00488         dest[2] = cm[dest[2]];
00489         dest[3] = cm[dest[3]];
00490         dest += linesize;
00491     }
00492 }
00493 
00494 static void vc1_inv_trans_4x4_c(uint8_t *dest, int linesize, DCTELEM *block)
00495 {
00496     int i;
00497     register int t1,t2,t3,t4;
00498     DCTELEM *src, *dst;
00499     const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
00500 
00501     src = block;
00502     dst = block;
00503     for(i = 0; i < 4; i++){
00504         t1 = 17 * (src[0] + src[2]) + 4;
00505         t2 = 17 * (src[0] - src[2]) + 4;
00506         t3 = 22 * src[1] + 10 * src[3];
00507         t4 = 22 * src[3] - 10 * src[1];
00508 
00509         dst[0] = (t1 + t3) >> 3;
00510         dst[1] = (t2 - t4) >> 3;
00511         dst[2] = (t2 + t4) >> 3;
00512         dst[3] = (t1 - t3) >> 3;
00513 
00514         src += 8;
00515         dst += 8;
00516     }
00517 
00518     src = block;
00519     for(i = 0; i < 4; i++){
00520         t1 = 17 * (src[ 0] + src[16]) + 64;
00521         t2 = 17 * (src[ 0] - src[16]) + 64;
00522         t3 = 22 * src[ 8] + 10 * src[24];
00523         t4 = 22 * src[24] - 10 * src[ 8];
00524 
00525         dest[0*linesize] = cm[dest[0*linesize] + ((t1 + t3) >> 7)];
00526         dest[1*linesize] = cm[dest[1*linesize] + ((t2 - t4) >> 7)];
00527         dest[2*linesize] = cm[dest[2*linesize] + ((t2 + t4) >> 7)];
00528         dest[3*linesize] = cm[dest[3*linesize] + ((t1 - t3) >> 7)];
00529 
00530         src ++;
00531         dest++;
00532     }
00533 }
00534 
00535 /* motion compensation functions */
00537 #define VC1_MSPEL_FILTER_16B(DIR, TYPE)                                 \
00538 static av_always_inline int vc1_mspel_ ## DIR ## _filter_16bits(const TYPE *src, int stride, int mode) \
00539 {                                                                       \
00540     switch(mode){                                                       \
00541     case 0: /* no shift - should not occur */                           \
00542         return 0;                                                       \
00543     case 1: /* 1/4 shift */                                             \
00544         return -4*src[-stride] + 53*src[0] + 18*src[stride] - 3*src[stride*2]; \
00545     case 2: /* 1/2 shift */                                             \
00546         return -src[-stride] + 9*src[0] + 9*src[stride] - src[stride*2]; \
00547     case 3: /* 3/4 shift */                                             \
00548         return -3*src[-stride] + 18*src[0] + 53*src[stride] - 4*src[stride*2]; \
00549     }                                                                   \
00550     return 0; /* should not occur */                                    \
00551 }
00552 
00553 VC1_MSPEL_FILTER_16B(ver, uint8_t);
00554 VC1_MSPEL_FILTER_16B(hor, int16_t);
00555 
00556 
00559 static av_always_inline int vc1_mspel_filter(const uint8_t *src, int stride, int mode, int r)
00560 {
00561     switch(mode){
00562     case 0: //no shift
00563         return src[0];
00564     case 1: // 1/4 shift
00565         return (-4*src[-stride] + 53*src[0] + 18*src[stride] - 3*src[stride*2] + 32 - r) >> 6;
00566     case 2: // 1/2 shift
00567         return (-src[-stride] + 9*src[0] + 9*src[stride] - src[stride*2] + 8 - r) >> 4;
00568     case 3: // 3/4 shift
00569         return (-3*src[-stride] + 18*src[0] + 53*src[stride] - 4*src[stride*2] + 32 - r) >> 6;
00570     }
00571     return 0; //should not occur
00572 }
00573 
00576 #define VC1_MSPEL_MC(OP, OPNAME)\
00577 static void OPNAME ## vc1_mspel_mc(uint8_t *dst, const uint8_t *src, int stride, int hmode, int vmode, int rnd)\
00578 {\
00579     int     i, j;\
00580 \
00581     if (vmode) { /* Horizontal filter to apply */\
00582         int r;\
00583 \
00584         if (hmode) { /* Vertical filter to apply, output to tmp */\
00585             static const int shift_value[] = { 0, 5, 1, 5 };\
00586             int              shift = (shift_value[hmode]+shift_value[vmode])>>1;\
00587             int16_t          tmp[11*8], *tptr = tmp;\
00588 \
00589             r = (1<<(shift-1)) + rnd-1;\
00590 \
00591             src -= 1;\
00592             for(j = 0; j < 8; j++) {\
00593                 for(i = 0; i < 11; i++)\
00594                     tptr[i] = (vc1_mspel_ver_filter_16bits(src + i, stride, vmode)+r)>>shift;\
00595                 src += stride;\
00596                 tptr += 11;\
00597             }\
00598 \
00599             r = 64-rnd;\
00600             tptr = tmp+1;\
00601             for(j = 0; j < 8; j++) {\
00602                 for(i = 0; i < 8; i++)\
00603                     OP(dst[i], (vc1_mspel_hor_filter_16bits(tptr + i, 1, hmode)+r)>>7);\
00604                 dst += stride;\
00605                 tptr += 11;\
00606             }\
00607 \
00608             return;\
00609         }\
00610         else { /* No horizontal filter, output 8 lines to dst */\
00611             r = 1-rnd;\
00612 \
00613             for(j = 0; j < 8; j++) {\
00614                 for(i = 0; i < 8; i++)\
00615                     OP(dst[i], vc1_mspel_filter(src + i, stride, vmode, r));\
00616                 src += stride;\
00617                 dst += stride;\
00618             }\
00619             return;\
00620         }\
00621     }\
00622 \
00623     /* Horizontal mode with no vertical mode */\
00624     for(j = 0; j < 8; j++) {\
00625         for(i = 0; i < 8; i++)\
00626             OP(dst[i], vc1_mspel_filter(src + i, 1, hmode, rnd));\
00627         dst += stride;\
00628         src += stride;\
00629     }\
00630 }
00631 
00632 #define op_put(a, b) a = av_clip_uint8(b)
00633 #define op_avg(a, b) a = (a + av_clip_uint8(b) + 1) >> 1
00634 
00635 VC1_MSPEL_MC(op_put, put_)
00636 VC1_MSPEL_MC(op_avg, avg_)
00637 
00638 /* pixel functions - really are entry points to vc1_mspel_mc */
00639 
00640 #define PUT_VC1_MSPEL(a, b)\
00641 static void put_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \
00642      put_vc1_mspel_mc(dst, src, stride, a, b, rnd);                         \
00643 }\
00644 static void avg_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \
00645      avg_vc1_mspel_mc(dst, src, stride, a, b, rnd);                         \
00646 }
00647 
00648 PUT_VC1_MSPEL(1, 0)
00649 PUT_VC1_MSPEL(2, 0)
00650 PUT_VC1_MSPEL(3, 0)
00651 
00652 PUT_VC1_MSPEL(0, 1)
00653 PUT_VC1_MSPEL(1, 1)
00654 PUT_VC1_MSPEL(2, 1)
00655 PUT_VC1_MSPEL(3, 1)
00656 
00657 PUT_VC1_MSPEL(0, 2)
00658 PUT_VC1_MSPEL(1, 2)
00659 PUT_VC1_MSPEL(2, 2)
00660 PUT_VC1_MSPEL(3, 2)
00661 
00662 PUT_VC1_MSPEL(0, 3)
00663 PUT_VC1_MSPEL(1, 3)
00664 PUT_VC1_MSPEL(2, 3)
00665 PUT_VC1_MSPEL(3, 3)
00666 
00667 static void put_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
00668     const int A=(8-x)*(8-y);
00669     const int B=(  x)*(8-y);
00670     const int C=(8-x)*(  y);
00671     const int D=(  x)*(  y);
00672     int i;
00673 
00674     assert(x<8 && y<8 && x>=0 && y>=0);
00675 
00676     for(i=0; i<h; i++)
00677     {
00678         dst[0] = (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6;
00679         dst[1] = (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6;
00680         dst[2] = (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6;
00681         dst[3] = (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6;
00682         dst[4] = (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6;
00683         dst[5] = (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6;
00684         dst[6] = (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6;
00685         dst[7] = (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6;
00686         dst+= stride;
00687         src+= stride;
00688     }
00689 }
00690 
00691 #define avg2(a,b) ((a+b+1)>>1)
00692 static void avg_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
00693     const int A=(8-x)*(8-y);
00694     const int B=(  x)*(8-y);
00695     const int C=(8-x)*(  y);
00696     const int D=(  x)*(  y);
00697     int i;
00698 
00699     assert(x<8 && y<8 && x>=0 && y>=0);
00700 
00701     for(i=0; i<h; i++)
00702     {
00703         dst[0] = avg2(dst[0], ((A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6));
00704         dst[1] = avg2(dst[1], ((A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6));
00705         dst[2] = avg2(dst[2], ((A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6));
00706         dst[3] = avg2(dst[3], ((A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6));
00707         dst[4] = avg2(dst[4], ((A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6));
00708         dst[5] = avg2(dst[5], ((A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6));
00709         dst[6] = avg2(dst[6], ((A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6));
00710         dst[7] = avg2(dst[7], ((A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6));
00711         dst+= stride;
00712         src+= stride;
00713     }
00714 }
00715 
00716 av_cold void ff_vc1dsp_init(VC1DSPContext* dsp) {
00717     dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_c;
00718     dsp->vc1_inv_trans_4x8 = vc1_inv_trans_4x8_c;
00719     dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_c;
00720     dsp->vc1_inv_trans_4x4 = vc1_inv_trans_4x4_c;
00721     dsp->vc1_inv_trans_8x8_dc = vc1_inv_trans_8x8_dc_c;
00722     dsp->vc1_inv_trans_4x8_dc = vc1_inv_trans_4x8_dc_c;
00723     dsp->vc1_inv_trans_8x4_dc = vc1_inv_trans_8x4_dc_c;
00724     dsp->vc1_inv_trans_4x4_dc = vc1_inv_trans_4x4_dc_c;
00725     dsp->vc1_h_overlap = vc1_h_overlap_c;
00726     dsp->vc1_v_overlap = vc1_v_overlap_c;
00727     dsp->vc1_h_s_overlap = vc1_h_s_overlap_c;
00728     dsp->vc1_v_s_overlap = vc1_v_s_overlap_c;
00729     dsp->vc1_v_loop_filter4 = vc1_v_loop_filter4_c;
00730     dsp->vc1_h_loop_filter4 = vc1_h_loop_filter4_c;
00731     dsp->vc1_v_loop_filter8 = vc1_v_loop_filter8_c;
00732     dsp->vc1_h_loop_filter8 = vc1_h_loop_filter8_c;
00733     dsp->vc1_v_loop_filter16 = vc1_v_loop_filter16_c;
00734     dsp->vc1_h_loop_filter16 = vc1_h_loop_filter16_c;
00735 
00736     dsp->put_vc1_mspel_pixels_tab[ 0] = ff_put_pixels8x8_c;
00737     dsp->put_vc1_mspel_pixels_tab[ 1] = put_vc1_mspel_mc10_c;
00738     dsp->put_vc1_mspel_pixels_tab[ 2] = put_vc1_mspel_mc20_c;
00739     dsp->put_vc1_mspel_pixels_tab[ 3] = put_vc1_mspel_mc30_c;
00740     dsp->put_vc1_mspel_pixels_tab[ 4] = put_vc1_mspel_mc01_c;
00741     dsp->put_vc1_mspel_pixels_tab[ 5] = put_vc1_mspel_mc11_c;
00742     dsp->put_vc1_mspel_pixels_tab[ 6] = put_vc1_mspel_mc21_c;
00743     dsp->put_vc1_mspel_pixels_tab[ 7] = put_vc1_mspel_mc31_c;
00744     dsp->put_vc1_mspel_pixels_tab[ 8] = put_vc1_mspel_mc02_c;
00745     dsp->put_vc1_mspel_pixels_tab[ 9] = put_vc1_mspel_mc12_c;
00746     dsp->put_vc1_mspel_pixels_tab[10] = put_vc1_mspel_mc22_c;
00747     dsp->put_vc1_mspel_pixels_tab[11] = put_vc1_mspel_mc32_c;
00748     dsp->put_vc1_mspel_pixels_tab[12] = put_vc1_mspel_mc03_c;
00749     dsp->put_vc1_mspel_pixels_tab[13] = put_vc1_mspel_mc13_c;
00750     dsp->put_vc1_mspel_pixels_tab[14] = put_vc1_mspel_mc23_c;
00751     dsp->put_vc1_mspel_pixels_tab[15] = put_vc1_mspel_mc33_c;
00752 
00753     dsp->avg_vc1_mspel_pixels_tab[ 0] = ff_avg_pixels8x8_c;
00754     dsp->avg_vc1_mspel_pixels_tab[ 1] = avg_vc1_mspel_mc10_c;
00755     dsp->avg_vc1_mspel_pixels_tab[ 2] = avg_vc1_mspel_mc20_c;
00756     dsp->avg_vc1_mspel_pixels_tab[ 3] = avg_vc1_mspel_mc30_c;
00757     dsp->avg_vc1_mspel_pixels_tab[ 4] = avg_vc1_mspel_mc01_c;
00758     dsp->avg_vc1_mspel_pixels_tab[ 5] = avg_vc1_mspel_mc11_c;
00759     dsp->avg_vc1_mspel_pixels_tab[ 6] = avg_vc1_mspel_mc21_c;
00760     dsp->avg_vc1_mspel_pixels_tab[ 7] = avg_vc1_mspel_mc31_c;
00761     dsp->avg_vc1_mspel_pixels_tab[ 8] = avg_vc1_mspel_mc02_c;
00762     dsp->avg_vc1_mspel_pixels_tab[ 9] = avg_vc1_mspel_mc12_c;
00763     dsp->avg_vc1_mspel_pixels_tab[10] = avg_vc1_mspel_mc22_c;
00764     dsp->avg_vc1_mspel_pixels_tab[11] = avg_vc1_mspel_mc32_c;
00765     dsp->avg_vc1_mspel_pixels_tab[12] = avg_vc1_mspel_mc03_c;
00766     dsp->avg_vc1_mspel_pixels_tab[13] = avg_vc1_mspel_mc13_c;
00767     dsp->avg_vc1_mspel_pixels_tab[14] = avg_vc1_mspel_mc23_c;
00768     dsp->avg_vc1_mspel_pixels_tab[15] = avg_vc1_mspel_mc33_c;
00769 
00770     dsp->put_no_rnd_vc1_chroma_pixels_tab[0]= put_no_rnd_vc1_chroma_mc8_c;
00771     dsp->avg_no_rnd_vc1_chroma_pixels_tab[0]= avg_no_rnd_vc1_chroma_mc8_c;
00772 
00773     if (HAVE_ALTIVEC)
00774         ff_vc1dsp_init_altivec(dsp);
00775     if (HAVE_MMX)
00776         ff_vc1dsp_init_mmx(dsp);
00777 }