[MEncoder-users] Accuracy of the -endpos option

[Grozdan]

On Wed, Jun 30, 2010 at 11:24 PM, Josef Wolf <jw at raven.inka.de> wrote:
> On Wed, Jun 30, 2010 at 03:59:59AM -0500, Andrew Berg wrote:
>>  On 6/30/2010 3:34 AM, Josef Wolf wrote:
>> > > frameref=15
>> > > Looks rather/very high to me
>> >
>> > This is the highest possible value AFAIK. I am trying to get the highest
>> > possible quality with file size as short as possible because I encode for
>> > archival. I don't really care much about encoding time.
>> Unless you are encoding very flat cartoons, a high frameref is not
>> beneficial. It increases the power needed to decode (it affects decoding
>> power needed much more than encoding power needed) and only increases
>> efficiency for cartoons/CGI.
>
> OK, thanks for the hint!
>
>> You can expect the level to go up from
>> High at 3.0 to High at 3.2 (or possibly High at 4.0 if you have other things that
>> raise the level).
>
> What do those parameters mean? are this preset values?

No, those are profiles. A profile in H.264 dictates certain things
that are allowed. F.ex, Baseline does not allow b-frames or
interlacing (among other things). Main does not allow custom matrices
and 8x8 transform while High allows everything. The numbers above are
levels, which carry restrictions on fps, resolution, macroblocks per
second and frame, etc

>
>> I use a high frameref only for cartoons on DVD and
>> limit it when doing live action or HD. If you want to increase
>> efficiency and don't care about encoding speed, there are other, more
>> sane things you could change (e.g. increase subq to 9).
>
> OK, I'll try subq=9. But what frameref do you recommend for live recordings?

Between 4 and 7. I personally use 4

>
>> If you want to
>> burn through CPU cycles for no reason, you could use me=tesa (look into
>> the placebo preset for more).
>
> Well, I want to burn cycles to get better quality with smaller files. Does
> this qualify as "good reason"? ;-)

That's for you to decide ;)

> _______________________________________________
> MEncoder-users mailing list
> MEncoder-users at mplayerhq.hu
> https://lists.mplayerhq.hu/mailman/listinfo/mencoder-users
>