Mercurial > libavcodec.hg

comparison fdctref.c @ 635:3e0f62e5eed6 libavcodec

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
dct cleanup more accurate mmx dct (dont discard bits for fun) fixing mmx quantizer bug for qscale%2==1 (bias was slightly wrong)
author michaelni
date Sun, 01 Sep 2002 16:52:33 +0000
parents c77207dc78f4
children 1e39f273ecd6
comparison
equal deleted inserted replaced
634:be1cb0e1f276 635:3e0f62e5eed6
101 s += c[i][3] * tmp[8 * 3 + j]; 101 s += c[i][3] * tmp[8 * 3 + j];
102 s += c[i][4] * tmp[8 * 4 + j]; 102 s += c[i][4] * tmp[8 * 4 + j];
103 s += c[i][5] * tmp[8 * 5 + j]; 103 s += c[i][5] * tmp[8 * 5 + j];
104 s += c[i][6] * tmp[8 * 6 + j]; 104 s += c[i][6] * tmp[8 * 6 + j];
105 s += c[i][7] * tmp[8 * 7 + j]; 105 s += c[i][7] * tmp[8 * 7 + j];
106 s*=8.0;
106 107
107 block[8 * i + j] = (short)floor(s + 0.499999); 108 block[8 * i + j] = (short)floor(s + 0.499999);
108 /* 109 /*
109 * reason for adding 0.499999 instead of 0.5: 110 * reason for adding 0.499999 instead of 0.5:
110 * s is quite often x.5 (at least for i and/or j = 0 or 4) 111 * s is quite often x.5 (at least for i and/or j = 0 or 4)