[FFmpeg-devel] [PATCH] libx264 psyrd update V4

Jason Garrett-Glaser darkshikari
Thu Jan 22 17:39:39 CET 2009

On Thu, Jan 22, 2009 at 4:07 AM, Marcus Blomenkamp <mblomenk at gmx.de> wrote:
>> This could be done more accurately (for 0-5 at least) simply by
>> copying x264's --longhelp:
>> - 0: fullpel only (not recommended)
>> - 1: SAD mode decision, one qpel iteration
>> - 2: SATD mode decision
>> - 3-5: Progressively more qpel
>> - 6: RD mode decision for I/P-frames
>> - 7: RD mode decision for all frames
>> - 8: RD refinement for I/P-frames
>> - 9: RD refinement for all frames
>>>Note that psyrd requires rate distribution,
>>Rate-distortion optimization.
> Thanks very much for this kind of input. Actually, while packing up this small patchset, i spent most of the time searching for documentation and wondering whether the found terms were correct and uptodate.
> So in fact, considering you as the reference person, i have not found a definitive comprehensive explaination of the psyopt variables written by you. You may have done it, but one can not find it. The reasoning behind the chosen defaults, backed up with pictures or streams showing the effects of these variables, are nowhere to be found. Often due to trivial reasons such as hosting expirations.
> To make a long story short: Could you please either take a look at my documentation attempts and point me to incorrect or unprecise terms OR could you write up a summary as a sticky topic/posting on doom9? It better not be too technical, but mostly focused on preventing misconceptions.


> Terms that i found questionable/explainable/replaceable:
> psyrd vs psyrdo?
> rate-distortion optimization == RD mode decision != RD mode refinement (which the average user would assume because 'optimization' sounds similar to 'refinement')?

Rate-distortion optimization is a general term for all types of RD
optimization.  RD mode decision and RD refinement both fall under this

Dark Shikari

More information about the ffmpeg-devel mailing list