[FFmpeg-devel] [PATCH] probetest.c

[Michael Niedermayer]

On Tue, Sep 22, 2009 at 08:35:12PM +0200, Ivo wrote:
> On Tuesday 22 September 2009 17:52, Michael Niedermayer wrote:
> > On Tue, Sep 22, 2009 at 03:08:41AM +0200, Ivo wrote:
> > > On Tuesday 15 September 2009 19:39, Vitor Sessak wrote:
> > > > pva
> > >
> > > I checked it, and I actually use return AVPROBE_SCORE_MAX/2 exactly for
> > > the reason to avoid misdetections. I could lower it to /3 though if
> > > that's wanted.
> >
> > cant you check more bytes?
> >
> > > > Others weak probes:
> > >
> > > [..]
> > >
> > > > lmlm4
> > >
> > > This one returns either /3 or /5 depending on "how much" it finds.
> >
> > same ...
> 
> Sadly not really. Both are are near-raw formats, encapsulating an MPEG 
> stream. pva.c checks five bytes of which three must be identical, one 
> non-zero and one less than three. LMLM4 has no header at all and probe  
> checks whether the the first few bytes look like a valid LMLM4 packet and 
> after that checks whether the packet contains something MPEG like 
> (v&0xfffe==0xfffc return MAX/3 or a PES signal, return MAX/5). Both file 
> formats are easily mistaken for mpeg streams, that's why I chose the values 
> to be low as there's not much else to check and be sure.
> 

> Hmm, I could check if there's a second packet at the end of the first. How 
> large is the probebuffer? 

RTFS :)
its increased if no high score detection happens


> Perhaps I could even keep checking till the end 
> of the probebuffer.

yes

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Into a blind darkness they enter who follow after the Ignorance,
they as if into a greater darkness enter who devote themselves
to the Knowledge alone. -- Isha Upanishad
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090922/c5afc232/attachment.pgp>