[MPlayer-DOCS] complete documentation

[Diego Biurrun]

Guillaume POIRIER writes:
> Le ven 10/09/2004 à 12:48, Diego Biurrun a écrit :
> 
> > Can I get you interested in documenting some of the other encoding
> > options?
> 
> That's possible, but not right now as I have postponed quite a lot of
> personal work in order to complete French manpage synch and XviD
> documentation.

Don't worry, this is not a short-term thing.

> I'm also the maintainer of a DVD-rip script based on
> MPlayer/Mencoder and transcode called tuxrip (http://tuxrip.free.fr if I
> recall right).
> It was originally written by someone else (Jotenakis) in bash, but since
> it had no design, it just can't be extended or it would explode ! ;-)

And you didn't tell yet and were missing from our projects page ?!?
Shame on you ;-)

Tuxrip added.

> > Now that you have finished XviD we can safely tell you that
> > lavc is much faster and even better ;-).
> 
> Umh... talking about speed, it all depends on which encoding options you
> use! With lavc and qns=3, plus other fancy HQ options (qpel, tellis), I
> had a film (about 1h40) that took around 2 days to encode on a
> P4-1.8Ghz! I agree that this option improves visual quality quite a bit,
> and isn't supposed to be particularly fast, but still!
> Don't get me wrong. Libavcodec is a feature-rich amazing codec, but it
> isn't foolproof!
> I agree with you, it does need to be better documented so that "Joe
> users" don't say that it is slow. It is not. It needs to be properly
> used, though.

Yes, lavc is a tool for experts.  I doubt it is harder to use than
XviD, though, if there is good documentation.

> Maybe that isn't the right place to talk about features, but I'd like to
> point out a really cool XviD feature that could maybe get ported in
> lavc.
> It's called... turbo mode!
> Basically, what it does is during first pass of 2-pass mode, it uses
> simpler MV search fonctions and I think it also disables some CPU-hungry
> options. On my machine, the first pass will run at 40fps and 9fps, with
> not that much of a PSNR reduction actually.
> I wish it could be implemented in lavc.

Well, Michael, what do you think?  Looks like Guillaume will improve
the documentation in return ;-)

> > The documentation is very
> > terse, though, and it would benefit a lot from better documentation.
> > I think we should try to promote lavc a bit, it deserves to be known
> > better.  Good user documentation should help boost it quite a bit.
> 
> Right. I remember when I read doom9 codec comparison, about 9 month from
> now that they were quite confused with how lavc works... and that's
> maybe due to the lack of documentation, but also the fact that XviD has
> a strong Windows base, so documentation is available everywhere.
> ...it's kinda the same situation for DivX: here's a crappy encoder
> that's still heavily used, because many people now it, and use it.

Yes. :-(

> > Guillaume, you've done such a good job with XviD, could you repeat
> > that for lavc?
> 
> Thanks Diego, what you're saying me is going right to my heart ;-)

:-)

> As I said, I don't plan to do that for at least a week... but after
> that, I should have more time to harvest some docs around the net to
> better document lavc. I'd also need the original devs of new options to
> help me, just like XviD devs kindly helped me, but I see no reason why
> wouldn't! ;-)

LOL, a week is not considered a long time around here :-)

Glad to hear you are willing to tackle this.  To learn more about
lavc's multitude of options you should ask on the ffmpeg-devel mailing
list.  Just subscribe to it at http://ffmpeg.org.  The people on that
mailing list will be able to help you out.

Diego