[Mplayer-advusers] mencoder crash/fail with some DVDs/VOBs

[Arpi]

Hi,

> >>> This looks like interlaced video.  I think 3 pass encoding does not
> >>> work with these.  If that's the problem you are facing, just rm
> >>> frameno.avi and do -oac mp3lame during all passes and it should be 
> >>> ok.
> >>  the video is not interlaced! (transcode can detect that)
> > the better word is 'field-coded', not nececssary interlaced, although 
> > thsi
> > coding method has no much sence for progressive (non-interlaced) video.
> > anyway i doubt that any app (be it transcode or other) can reliable 
> > detect
> > interlacing, sometimes even teh human eye+mind combination fails...
> hm, if an app can render MPEG2 video, it must also know it is 
> interlaced or not!? am i wrong here?

yes you're wrong
the format/codecs itself doesnt know about the original video being
interlaced. there are flags noting if its' field-coded (a sin your case) or
telcined (3:2 pulldown) which both suggests interlacing but doenst force it.

you can encode progressive video as interlaced and vice versa.
(by losing quality)

> >> any idea how i can draw the developers attention to that problem? the
> >
> > if you mean mplayer developers, they do read this list.
> > this is why we really hate beginner questions and incomplete 
> > bugreports here.
> well i made a complete bugreport in my initial post (well with a macosx 
> backtrace instead of a gdb one, but i can of course supply an gdb one 
> too)
> 
> about beginner question....i searched both the whole documentation, FAQ 
> and mailing list archives and could not find anything useful about this 
> issue

dunno if it's well documented, i'm not docs author.
but it was discussed more than enough at all teh mailing lists.

> >> sample .VOB and problem description are still in the incoming dir on
> >> the ftp server...?
> >
> > probably.
> > but it's a known limitation, and not possible to fix in mencoder
> exactly what is the limitation here? i have 14 DVDs here that work fine 

without decoding the video stream (ie. 1st pass of 3pass encode) mencoder
don't know if it's field-coded, so it have to read 2 packet to get 1 output
frame, while in normal (frame-coded) case it's one packet for one frame.

we could either hack a partial header decoder into demuxer/mencoder to
detect it without decoding, or redesign teh whole demuxer-decoder-encoder
pipeline API to pass such info back somehow. it just doesnt worth it.

> with 3-pass-mencoder and one that does not (crashes mencoder just like 

it means that 14 dvd are frame-coded and this one is field-coded
(field-coded mpeg is rare enough)

> reported)...the issue doesn't seem to be the interlaced video (because 
> transcode says it isn't interlaced)... is there any way i can know 
> (prior to encoding) if a DVD will work or mencoder will just crash?

no
then mencoder could know too...

> > , without
> > major hacks or architecture changes. i mean this isn't a bug.
> if this isn't a bug, mencoder should print out a message that it can't 
> encode this DVD (or .VOB) instead of silently mysteriously failing, or 
> crashing!

it shouldnt crash (it doesnt crash for me, just drop everty 2nd field
resulting bad quality jumpy output file)


A'rpi / Astral & ESP-team

--
Developer of MPlayer G2, the Movie Framework for all - http://www.MPlayerHQ.hu