[MPlayer-dev-eng] [RFC] aspect scaling, panscan etc. in windowed mode

[Reimar Döffinger]

On Fri, Sep 14, 2012 at 11:09:36AM +0200, Nicolas George wrote:
> Le nonidi 29 fructidor, an CCXX, Reimar Döffinger a écrit :
> > The reason is simple: that the letterbox/panscan code produces anything
> > even remotely sensible in window mode is a relatively recent development.
> > And one that involved a lot of preparatory work, like making sure all vos
> > actually provide the current window size in a somewhat consistent form.
> > And when all was done I didn't really feel like changing behaviour when I
> > couldn't decide what is better.
> 
> Ok, thanks for the explanations. I have no idea how backends other than X11
> work, so I had not realized that e.g. getting the window size could be a
> problem.

X11 was mostly the problem. And the problem is for example getting the
window size to the code doing the calculations, it's not enough that
that you can get it "somehow" after all.
There was no consistency between e.g. what different vos expected in
vo_dheight: some would expect window height, some the height of the area
they should draw the video into.

> That would give something like that:
> 
> -nokeepaspect: do not ask the WM to preserve the ratio in windowed mode.
> 
> - default behaviour: letterbox or panscan to preserve the aspect ratio, both
>   in full screen and windowed mode, the latter only relevant if
>   -nokeepaspect or bogus WM;
> 
> - with -noaspect: always fill the window, both in full screen and windowed
>   mode;
> 
> - to disregard bogus aspect ratio in the video and/or disable prescaling,
>   "-vf dsize=-1:-1".
> 
> That seems reasonable, IMHO.

I'll try to think about it.
Right now I don't much like the idea of -noaspect no longer only
affecting a stream property but vo behaviour.
It also would become completely meaningless for mencoder (well, I guess
it can keep its old behaviour and do the vo specific behaviour only
in addition, but even then it might still be a bit confusing).