[MPlayer-users] Re: Low bitrate using 2-pass with libavcodec

[Michael Niedermayer]

Hi

On Sunday 21 April 2002 01:38, Arpi wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> Hi,
>
> > Don't use vhq on the first pass. The second pass will still be encoded
> > with a lower bitrate than the first, but the difference will be
> > minimal (less than 10%). Adjust the bitrate accordingly. For example,
> > on a 10 minutes sample I specified vbitrate=858 and got 797 on the
> > second pass. The ratio may depend on what you encode.
>
> i'll fix it next week. the current 2-pass code was made for divx4, and
> seems to always overestimate bitrate of libavcodec.
perhaps, but there is another vhq stats bug in lavc (is fixed in my local 
tree and ill commit ASAP)

> i've already wrote code for libavcodec 2pass which works better, at least
> it get +-1mb diff at 700mb filesize, but i'm not staisfied with the bitrate
> control, sometimes produce blocky scenes.
ehh, iam working on 2pass lavc bitrate stuff currently, should be finished 
today, but another alternative 2pass code cant hurt :)

[...]

Michael