[MEncoder-users] Performance on multi core machines - a specific enquiry re avisynth mt input
mroz at ukgateway.net
Sat Sep 29 18:05:56 CEST 2007
James Hastings-Trew wrote:
> Michael Rozdoba wrote:
>> Aside: How would I replicate the avisynth script I posted using just
>> Thanks for the feedback
> Honestly, I don't know what half of that stuff in the AVISynth script is
> about, but there is a comment line in there about cleaning up a
> "blocky/noisy" mpeg2 source. That sounds like a job for the hqdn3d
> filter (high quality denoise 3d). This filters image noise in both
> spacial and temporal domains. There are also horizontal and vertical
> deblocking filters (pp=hb,pp=vb). pp=de will give you a default
> deblock/dering filter set.
I'll add looking into that to my todo list.
> I've no idea what kind of output file you are trying to create, so this
> example just uses default settings for lavc encoders. You can get much
> better results by reading the manual and following the examples. (for
> example, by specifying the codecs you want to use, codec options to
> optimize size/quality how many CPU threads to use, etc.):
> mencoder INPUT -vf pp=de,hqdn3d -oac lavc -ovc lavc -ffourcc DIVX -of
> avi -o OUTPUT
Thanks, but it's only a temporary file using a lossless codec that I
need at that point. The purpose of the encode is purely to allow the
avisynth script to do its stuff & only have to do so once. Without this
pre-rendering the subsequent x264 encode, as it uses two passes, would
require the avisynth processing be performed twice on the original
source, one for each pass.
Can anyone comment on why avisynth is being restricted to one core
/only/ when it's being used as input to mencoder?
I could really use a solution as I've just discovered my alternative
solution - vdub & ffdshow's huffy, is introducing a yellow colour cast
(probably as it requires an additional converttorgb step at the end of
the avisynth script, as vdub will only accept rgb).
If no one here knows, any suggestions where I could ask? The issue does
seem to be specific to mencoder.
More information about the MEncoder-users