1
0
mirror of https://github.com/mpv-player/mpv synced 2025-04-01 00:07:33 +00:00

document x264's option subq=0, plus a bit of factoring and added details

git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@27906 b3059339-0415-0410-9bf9-f77b7e298cf2
This commit is contained in:
gpoirier 2008-11-13 17:18:40 +00:00
parent 5dbcc861f5
commit acd3b32cc8

View File

@ -10050,34 +10050,38 @@ exhaustive search (very slow, and no better than umh)
radius of exhaustive or multi-hexagon motion search (default: 16)
.
.TP
.B subq=<1\-9>
.B subq=<0\-9>
Adjust subpel refinement quality.
This parameter controls quality versus speed tradeoffs involved in the motion
estimation decision process.
subq=5 can compress up to 10% better than subq=1.
.PD 0
.RSs
.IPs 1
.IPs 0
Runs fullpixel precision motion estimation on all candidate
macroblock types.
Then selects the best type.
Then refines the motion of that type to fast quarterpixel precision (fastest).
Then selects the best type with SAD metric (faster than subq=1, not recommended
unless you're looking for ultra-fast encoding).
.IPs 1
Does as 0, then refines the motion of that type to fast quarterpixel precision
(fast).
.IPs 2
Runs halfpixel precision motion estimation on all candidate macroblock types.
Then selects the best type.
Then selects the best type with SATD metric.
Then refines the motion of that type to fast quarterpixel precision.
.IPs 3
As 2, but uses a slower quarterpixel refinement.
.IPs 4
Runs fast quarterpixel precision motion estimation on all candidate
macroblock types.
Then selects the best type.
Then selects the best type with SATD metric.
Then finishes the quarterpixel refinement for that type.
.IPs 5
Runs best quality quarterpixel precision motion estimation on all
candidate macroblock types, before selecting the best type.
Also refines the two motion vectors used in bidirectional macroblocks,
rather than reusing vectors from the forward and backward searches.
Also refines the two motion vectors used in bidirectional macroblocks with
SATD metric, rather than reusing vectors from the forward and backward
searches.
.IPs 6
Enables rate-distortion optimization of macroblock types in
I- and P-frames (default).