[FFmpeg-devel] [RFC] Filters and encoding commandline options interaction

[Christian Ebert]

* Luca Barbato on Tuesday, February 15, 2011 at 13:14:00 +0100
> On 2/15/11 12:48 PM, Christian Ebert wrote:
>> While the version @videolan offers definite improvements - thank
>> you! -, from a poor _user_'s perspective the interaction between
>> -vf and -aspect and -s is utterly obscure and confusing,
>> especially when the source has non-square pixels (eg. vob,
>> pal-dv). I tried to note some of my findings at
>> 
>> http://roundup.ffmpeg.org/issue2377
>> 
>> latest:
>> 
>> http://roundup.ffmpeg.org/msg13631
>> 
>> It was marked solved and closed, but, again from user's
>> perspective, this feels almost like a joke. Add to it that the 2
>> heads of the ffmpeg hydra now also _behave_ differently, "solved
>> and closed" is, errmh ...
> 
> Could you please open another bug

Why not reopen the bug? I really tried as best as I could to
describe what is happening, what works with certain versions and
what doesn't.

> and help us figure out a good way to come up with documentation

I wish I could, but how can I document something where I do not
understand what's going on?

I can describe, in layman's words what mencoder does in
comparison to ffmpeg:

[ assuming a pal source ]

1) prescales to spal (it _seems_ this step can be omitted - is it
   a good thing to omit it?? - with ffmpeg from videolan)
2) the filters are applied in the order given with -vf (ffmpeg
   from videolan ignores order in -vf; atm I can't find a way to
   make -vf from ffmpeg.org create correct results - it used to
   work via "prescaling" ? la mencoder with -s and -vf scale
   later, but not anymore)
3) I'm losing it ;-)

I sort of understand what mencoder does, and I can bring it to
give me the results I need; I can bring ffmpeg at videolan to give
the results I need, albeit with less understanding of what's
going on and whether it's right thing (is it good to enforce
yadif before crop? might be). ffmpeg at ffmpeg.org's -vf, as I said,
doesn't work for me[tm] at all, at least atm.

I sort of prefer the mencoder way, because it seems to do what I
tell it to. So if I make a mistake in the filter chain order, it
is _my_ mistake, and I may be able to learn from people in the
know how to improve it.

> and a better solution?

I'm stuttering at describing the problem, I don't see it as
solved, as I said, a better solution would need some solution
first. As a user I become insecure because on the one hand I am
told that mencoder is basically unmaintained and I should switch
to ffmpeg because the filters are ported to it, or at least the
ones I use most often (I still need mcdeint sometimes), but then
I spent hours to make heads or tails of the ffmpeg -vf handling,
and when I try to give, admittedly clumsy, feedback it is brushed
away, or I am asked to just review again ... sorry for this rant,
but this is how it comes across, even when I take into account
the complexity of the project and the current tense situation.

c
-- 
\black\trash movie           _COWBOY  CANOE  COMA_
            Ein deutscher Western/A German Western

--->> http://www.blacktrash.org/underdogma/ccc.php