[FFmpeg-devel] [PATCH] add ME_T/ESA to avcodec.h
Wed Jun 4 09:41:19 CEST 2008
>> * Motion estimation algorithm used for video coding.
>> * 1 (zero), 2 (full), 3 (log), 4 (phods), 5 (epzs), 6 (x1), 7 (hex),
>> - * 8 (umh), 9 (iter) [7, 8 are x264 specific, 9 is snow specific]
>> + * 8 (umh), 9 (iter) [7, 8, 10, 11 are x264 specific, 9 is snow specific]
>> * - encoding: MUST be set by user.
>> * - decoding: unused
Is there any reason to distinguish ESA from FULL? Shouldn't the two
input strings just map to the same internal option? Yes, x264's ESA
is actually Sequential Elimination (SEA), but given that is
mathematically equivalent to a full search, there is no possible
reason anyone would want to FULL if x264-style ESA was available, so
there is no good reason to imply that there is any real distinction
between FULL and ESA, I think.
More information about the ffmpeg-devel