X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fh264idct.c;h=78fe431eb6e3194cd3aca4aee580c31499d62f4c;hb=2e63619ff9643c6f5ec220da8b896ffeaf87f49c;hp=3506418ad6804a411f437621a56ddc21dfb71f09;hpb=55fde95e3bfc0e337a482e2508c3ce60009d8c7b;p=ffmpeg diff --git a/libavcodec/h264idct.c b/libavcodec/h264idct.c index 3506418ad68..78fe431eb6e 100644 --- a/libavcodec/h264idct.c +++ b/libavcodec/h264idct.c @@ -17,18 +17,17 @@ * You should have received a copy of the GNU Lesser General Public * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - * */ /** - * @file h264-idct.c + * @file libavcodec/h264idct.c * H.264 IDCT. * @author Michael Niedermayer */ #include "dsputil.h" -static always_inline void idct_internal(uint8_t *dst, DCTELEM *block, int stride, int block_stride, int shift, int add){ +static av_always_inline void idct_internal(uint8_t *dst, DCTELEM *block, int stride, int block_stride, int shift, int add){ int i; uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; @@ -73,58 +72,57 @@ void ff_h264_lowres_idct_put_c(uint8_t *dst, int stride, DCTELEM *block){ void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride){ int i; - DCTELEM (*src)[8] = (DCTELEM(*)[8])block; uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; block[0] += 32; for( i = 0; i < 8; i++ ) { - const int a0 = src[i][0] + src[i][4]; - const int a2 = src[i][0] - src[i][4]; - const int a4 = (src[i][2]>>1) - src[i][6]; - const int a6 = (src[i][6]>>1) + src[i][2]; + const int a0 = block[0+i*8] + block[4+i*8]; + const int a2 = block[0+i*8] - block[4+i*8]; + const int a4 = (block[2+i*8]>>1) - block[6+i*8]; + const int a6 = (block[6+i*8]>>1) + block[2+i*8]; const int b0 = a0 + a6; const int b2 = a2 + a4; const int b4 = a2 - a4; const int b6 = a0 - a6; - const int a1 = -src[i][3] + src[i][5] - src[i][7] - (src[i][7]>>1); - const int a3 = src[i][1] + src[i][7] - src[i][3] - (src[i][3]>>1); - const int a5 = -src[i][1] + src[i][7] + src[i][5] + (src[i][5]>>1); - const int a7 = src[i][3] + src[i][5] + src[i][1] + (src[i][1]>>1); + const int a1 = -block[3+i*8] + block[5+i*8] - block[7+i*8] - (block[7+i*8]>>1); + const int a3 = block[1+i*8] + block[7+i*8] - block[3+i*8] - (block[3+i*8]>>1); + const int a5 = -block[1+i*8] + block[7+i*8] + block[5+i*8] + (block[5+i*8]>>1); + const int a7 = block[3+i*8] + block[5+i*8] + block[1+i*8] + (block[1+i*8]>>1); const int b1 = (a7>>2) + a1; const int b3 = a3 + (a5>>2); const int b5 = (a3>>2) - a5; const int b7 = a7 - (a1>>2); - src[i][0] = b0 + b7; - src[i][7] = b0 - b7; - src[i][1] = b2 + b5; - src[i][6] = b2 - b5; - src[i][2] = b4 + b3; - src[i][5] = b4 - b3; - src[i][3] = b6 + b1; - src[i][4] = b6 - b1; + block[0+i*8] = b0 + b7; + block[7+i*8] = b0 - b7; + block[1+i*8] = b2 + b5; + block[6+i*8] = b2 - b5; + block[2+i*8] = b4 + b3; + block[5+i*8] = b4 - b3; + block[3+i*8] = b6 + b1; + block[4+i*8] = b6 - b1; } for( i = 0; i < 8; i++ ) { - const int a0 = src[0][i] + src[4][i]; - const int a2 = src[0][i] - src[4][i]; - const int a4 = (src[2][i]>>1) - src[6][i]; - const int a6 = (src[6][i]>>1) + src[2][i]; + const int a0 = block[i+0*8] + block[i+4*8]; + const int a2 = block[i+0*8] - block[i+4*8]; + const int a4 = (block[i+2*8]>>1) - block[i+6*8]; + const int a6 = (block[i+6*8]>>1) + block[i+2*8]; const int b0 = a0 + a6; const int b2 = a2 + a4; const int b4 = a2 - a4; const int b6 = a0 - a6; - const int a1 = -src[3][i] + src[5][i] - src[7][i] - (src[7][i]>>1); - const int a3 = src[1][i] + src[7][i] - src[3][i] - (src[3][i]>>1); - const int a5 = -src[1][i] + src[7][i] + src[5][i] + (src[5][i]>>1); - const int a7 = src[3][i] + src[5][i] + src[1][i] + (src[1][i]>>1); + const int a1 = -block[i+3*8] + block[i+5*8] - block[i+7*8] - (block[i+7*8]>>1); + const int a3 = block[i+1*8] + block[i+7*8] - block[i+3*8] - (block[i+3*8]>>1); + const int a5 = -block[i+1*8] + block[i+7*8] + block[i+5*8] + (block[i+5*8]>>1); + const int a7 = block[i+3*8] + block[i+5*8] + block[i+1*8] + (block[i+1*8]>>1); const int b1 = (a7>>2) + a1; const int b3 = a3 + (a5>>2); @@ -166,3 +164,55 @@ void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){ dst += stride; } } + +//FIXME this table is a duplicate from h264data.h, and will be removed once the tables from, h264 have been split +static const uint8_t scan8[16 + 2*4]={ + 4+1*8, 5+1*8, 4+2*8, 5+2*8, + 6+1*8, 7+1*8, 6+2*8, 7+2*8, + 4+3*8, 5+3*8, 4+4*8, 5+4*8, + 6+3*8, 7+3*8, 6+4*8, 7+4*8, + 1+1*8, 2+1*8, + 1+2*8, 2+2*8, + 1+4*8, 2+4*8, + 1+5*8, 2+5*8, +}; + +void ff_h264_idct_add16_c(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){ + int i; + for(i=0; i<16; i++){ + int nnz = nnzc[ scan8[i] ]; + if(nnz){ + if(nnz==1 && block[i*16]) ff_h264_idct_dc_add_c(dst + block_offset[i], block + i*16, stride); + else idct_internal (dst + block_offset[i], block + i*16, stride, 4, 6, 1); + } + } +} + +void ff_h264_idct_add16intra_c(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){ + int i; + for(i=0; i<16; i++){ + if(nnzc[ scan8[i] ]) idct_internal (dst + block_offset[i], block + i*16, stride, 4, 6, 1); + else if(block[i*16]) ff_h264_idct_dc_add_c(dst + block_offset[i], block + i*16, stride); + } +} + +void ff_h264_idct8_add4_c(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){ + int i; + for(i=0; i<16; i+=4){ + int nnz = nnzc[ scan8[i] ]; + if(nnz){ + if(nnz==1 && block[i*16]) ff_h264_idct8_dc_add_c(dst + block_offset[i], block + i*16, stride); + else ff_h264_idct8_add_c (dst + block_offset[i], block + i*16, stride); + } + } +} + +void ff_h264_idct_add8_c(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){ + int i; + for(i=16; i<16+8; i++){ + if(nnzc[ scan8[i] ]) + ff_h264_idct_add_c (dest[(i&4)>>2] + block_offset[i], block + i*16, stride); + else if(block[i*16]) + ff_h264_idct_dc_add_c(dest[(i&4)>>2] + block_offset[i], block + i*16, stride); + } +}