avfilter/dctdnoiz: use 32-bit (float) operations instead of 64 (double) for DCTs
authorClément Bœsch <u@pkh.me>
Fri, 8 Aug 2014 17:59:15 +0000 (19:59 +0200)
committerClément Bœsch <u@pkh.me>
Fri, 8 Aug 2014 18:00:14 +0000 (20:00 +0200)
commit1ba7c6ead24f1485ad537cfdedb2a6cf5ca8e869
treee99c35fb8505859332c29a100698957172c40fb8
parenteb16a6d2294cb81d60fb592c4d4439522f0b21eb
avfilter/dctdnoiz: use 32-bit (float) operations instead of 64 (double) for DCTs

This makes the code about 1.5x faster without any noticeable difference
in the output.
libavfilter/vf_dctdnoiz.c