FFmpeg: libavcodec/rv30dsp.c Source File

00001 /*
00002  * RV30 decoder motion compensation functions
00003  * Copyright (c) 2007 Konstantin Shishkov
00004  *
00005  * This file is part of FFmpeg.
00006  *
00007  * FFmpeg 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  * FFmpeg 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 FFmpeg; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00027 #include "avcodec.h"
00028 #include "dsputil.h"
00029 
00030 #define RV30_LOWPASS(OPNAME, OP) \
00031 static av_unused void OPNAME ## rv30_tpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
00032     const int h=8;\
00033     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00034     int i;\
00035     for(i=0; i<h; i++)\
00036     {\
00037         OP(dst[0], (-(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + 8)>>4);\
00038         OP(dst[1], (-(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + 8)>>4);\
00039         OP(dst[2], (-(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + 8)>>4);\
00040         OP(dst[3], (-(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + 8)>>4);\
00041         OP(dst[4], (-(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + 8)>>4);\
00042         OP(dst[5], (-(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + 8)>>4);\
00043         OP(dst[6], (-(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + 8)>>4);\
00044         OP(dst[7], (-(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + 8)>>4);\
00045         dst+=dstStride;\
00046         src+=srcStride;\
00047     }\
00048 }\
00049 \
00050 static void OPNAME ## rv30_tpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
00051     const int w=8;\
00052     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00053     int i;\
00054     for(i=0; i<w; i++)\
00055     {\
00056         const int srcA= src[-1*srcStride];\
00057         const int src0= src[0 *srcStride];\
00058         const int src1= src[1 *srcStride];\
00059         const int src2= src[2 *srcStride];\
00060         const int src3= src[3 *srcStride];\
00061         const int src4= src[4 *srcStride];\
00062         const int src5= src[5 *srcStride];\
00063         const int src6= src[6 *srcStride];\
00064         const int src7= src[7 *srcStride];\
00065         const int src8= src[8 *srcStride];\
00066         const int src9= src[9 *srcStride];\
00067         OP(dst[0*dstStride], (-(srcA+src2) + src0*C1 + src1*C2 + 8)>>4);\
00068         OP(dst[1*dstStride], (-(src0+src3) + src1*C1 + src2*C2 + 8)>>4);\
00069         OP(dst[2*dstStride], (-(src1+src4) + src2*C1 + src3*C2 + 8)>>4);\
00070         OP(dst[3*dstStride], (-(src2+src5) + src3*C1 + src4*C2 + 8)>>4);\
00071         OP(dst[4*dstStride], (-(src3+src6) + src4*C1 + src5*C2 + 8)>>4);\
00072         OP(dst[5*dstStride], (-(src4+src7) + src5*C1 + src6*C2 + 8)>>4);\
00073         OP(dst[6*dstStride], (-(src5+src8) + src6*C1 + src7*C2 + 8)>>4);\
00074         OP(dst[7*dstStride], (-(src6+src9) + src7*C1 + src8*C2 + 8)>>4);\
00075         dst++;\
00076         src++;\
00077     }\
00078 }\
00079 \
00080 static void OPNAME ## rv30_tpel8_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00081     const int w = 8;\
00082     const int h = 8;\
00083     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00084     int i, j;\
00085     for(j = 0; j < h; j++){\
00086         for(i = 0; i < w; i++){\
00087             OP(dst[i], (\
00088                   src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
00089               -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i] +72*src[srcStride* 0+i+1] -12*src[srcStride* 0+i+2] +\
00090                -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i] +36*src[srcStride* 1+i+1]  -6*src[srcStride* 1+i+2] +\
00091                   src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2] +\
00092                   128)>>8);\
00093         }\
00094         src += srcStride;\
00095         dst += dstStride;\
00096     }\
00097 }\
00098 \
00099 static void OPNAME ## rv30_tpel8_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00100     const int w = 8;\
00101     const int h = 8;\
00102     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00103     int i, j;\
00104     for(j = 0; j < h; j++){\
00105         for(i = 0; i < w; i++){\
00106             OP(dst[i], (\
00107                   src[srcStride*-1+i-1]  -12*src[srcStride*-1+i+1]  -6*src[srcStride*-1+i]    +src[srcStride*-1+i+2]+\
00108               -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i+1] +72*src[srcStride* 0+i] -12*src[srcStride* 0+i+2]+\
00109                -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i+1] +36*src[srcStride* 1+i]  -6*src[srcStride* 1+i+2]+\
00110                   src[srcStride* 2+i-1]  -12*src[srcStride* 2+i+1]  -6*src[srcStride* 2+i]    +src[srcStride* 2+i+2]+\
00111                   128)>>8);\
00112         }\
00113         src += srcStride;\
00114         dst += dstStride;\
00115     }\
00116 }\
00117 \
00118 static void OPNAME ## rv30_tpel8_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00119     const int w = 8;\
00120     const int h = 8;\
00121     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00122     int i, j;\
00123     for(j = 0; j < h; j++){\
00124         for(i = 0; i < w; i++){\
00125             OP(dst[i], (\
00126                   src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
00127                -6*src[srcStride* 0+i-1]  +72*src[srcStride* 0+i] +36*src[srcStride* 0+i+1]  -6*src[srcStride* 0+i+2]+\
00128               -12*src[srcStride* 1+i-1] +144*src[srcStride* 1+i] +72*src[srcStride* 1+i+1] -12*src[srcStride* 1+i+2]+\
00129                   src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2]+\
00130                   128)>>8);\
00131         }\
00132         src += srcStride;\
00133         dst += dstStride;\
00134     }\
00135 }\
00136 \
00137 static void OPNAME ## rv30_tpel8_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00138     const int w = 8;\
00139     const int h = 8;\
00140     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00141     int i, j;\
00142     for(j = 0; j < h; j++){\
00143         for(i = 0; i < w; i++){\
00144             OP(dst[i], (\
00145                36*src[i+srcStride*0] +54*src[i+1+srcStride*0] +6*src[i+2+srcStride*0]+\
00146                54*src[i+srcStride*1] +81*src[i+1+srcStride*1] +9*src[i+2+srcStride*1]+\
00147                 6*src[i+srcStride*2] + 9*src[i+1+srcStride*2] +  src[i+2+srcStride*2]+\
00148                128)>>8);\
00149         }\
00150         src += srcStride;\
00151         dst += dstStride;\
00152     }\
00153 }\
00154 \
00155 static void OPNAME ## rv30_tpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
00156     OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00157     OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00158     src += 8*srcStride;\
00159     dst += 8*dstStride;\
00160     OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00161     OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00162 }\
00163 \
00164 static void OPNAME ## rv30_tpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
00165     OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00166     OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00167     src += 8*srcStride;\
00168     dst += 8*dstStride;\
00169     OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00170     OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00171 }\
00172 \
00173 static void OPNAME ## rv30_tpel16_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00174     OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
00175     OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
00176     src += 8*srcStride;\
00177     dst += 8*dstStride;\
00178     OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
00179     OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
00180 }\
00181 \
00182 static void OPNAME ## rv30_tpel16_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00183     OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
00184     OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
00185     src += 8*srcStride;\
00186     dst += 8*dstStride;\
00187     OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
00188     OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
00189 }\
00190 \
00191 static void OPNAME ## rv30_tpel16_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00192     OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
00193     OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00194     src += 8*srcStride;\
00195     dst += 8*dstStride;\
00196     OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
00197     OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00198 }\
00199 \
00200 static void OPNAME ## rv30_tpel16_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00201     OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
00202     OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00203     src += 8*srcStride;\
00204     dst += 8*dstStride;\
00205     OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
00206     OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00207 }\
00208 \
00209 
00210 #define RV30_MC(OPNAME, SIZE) \
00211 static void OPNAME ## rv30_tpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
00212     OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 12, 6);\
00213 }\
00214 \
00215 static void OPNAME ## rv30_tpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\
00216     OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 6, 12);\
00217 }\
00218 \
00219 static void OPNAME ## rv30_tpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
00220     OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 12, 6);\
00221 }\
00222 \
00223 static void OPNAME ## rv30_tpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\
00224     OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 6, 12);\
00225 }\
00226 \
00227 static void OPNAME ## rv30_tpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
00228     OPNAME ## rv30_tpel ## SIZE ## _hv_lowpass(dst, src, stride, stride);\
00229 }\
00230 \
00231 static void OPNAME ## rv30_tpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
00232     OPNAME ## rv30_tpel ## SIZE ## _hvv_lowpass(dst, src, stride, stride);\
00233 }\
00234 \
00235 static void OPNAME ## rv30_tpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
00236     OPNAME ## rv30_tpel ## SIZE ## _hhv_lowpass(dst, src, stride, stride);\
00237 }\
00238 \
00239 static void OPNAME ## rv30_tpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
00240     OPNAME ## rv30_tpel ## SIZE ## _hhvv_lowpass(dst, src, stride, stride);\
00241 }\
00242 \
00243 
00244 #define op_avg(a, b)  a = (((a)+cm[b]+1)>>1)
00245 #define op_put(a, b)  a = cm[b]
00246 
00247 RV30_LOWPASS(put_       , op_put)
00248 RV30_LOWPASS(avg_       , op_avg)
00249 RV30_MC(put_, 8)
00250 RV30_MC(put_, 16)
00251 RV30_MC(avg_, 8)
00252 RV30_MC(avg_, 16)
00253 
00254 av_cold void ff_rv30dsp_init(DSPContext* c, AVCodecContext *avctx) {
00255     c->put_rv30_tpel_pixels_tab[0][ 0] = c->put_h264_qpel_pixels_tab[0][0];
00256     c->put_rv30_tpel_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c;
00257     c->put_rv30_tpel_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c;
00258     c->put_rv30_tpel_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c;
00259     c->put_rv30_tpel_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c;
00260     c->put_rv30_tpel_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c;
00261     c->put_rv30_tpel_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c;
00262     c->put_rv30_tpel_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c;
00263     c->put_rv30_tpel_pixels_tab[0][10] = put_rv30_tpel16_mc22_c;
00264     c->avg_rv30_tpel_pixels_tab[0][ 0] = c->avg_h264_qpel_pixels_tab[0][0];
00265     c->avg_rv30_tpel_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c;
00266     c->avg_rv30_tpel_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c;
00267     c->avg_rv30_tpel_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c;
00268     c->avg_rv30_tpel_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c;
00269     c->avg_rv30_tpel_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c;
00270     c->avg_rv30_tpel_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c;
00271     c->avg_rv30_tpel_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c;
00272     c->avg_rv30_tpel_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c;
00273     c->put_rv30_tpel_pixels_tab[1][ 0] = c->put_h264_qpel_pixels_tab[1][0];
00274     c->put_rv30_tpel_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c;
00275     c->put_rv30_tpel_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c;
00276     c->put_rv30_tpel_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c;
00277     c->put_rv30_tpel_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c;
00278     c->put_rv30_tpel_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c;
00279     c->put_rv30_tpel_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c;
00280     c->put_rv30_tpel_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c;
00281     c->put_rv30_tpel_pixels_tab[1][10] = put_rv30_tpel8_mc22_c;
00282     c->avg_rv30_tpel_pixels_tab[1][ 0] = c->avg_h264_qpel_pixels_tab[1][0];
00283     c->avg_rv30_tpel_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c;
00284     c->avg_rv30_tpel_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c;
00285     c->avg_rv30_tpel_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c;
00286     c->avg_rv30_tpel_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c;
00287     c->avg_rv30_tpel_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c;
00288     c->avg_rv30_tpel_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c;
00289     c->avg_rv30_tpel_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c;
00290     c->avg_rv30_tpel_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c;
00291 }