[Ffmpeg-devel] Re: swscaler integration

Luca Abeni lucabe72
Sat Mar 18 12:50:31 CET 2006

Hi Baptiste,

On Sat, 2006-03-18 at 01:14 +0100, Baptiste COUDURIER wrote:
> Hi
> Luca Abeni wrote:
> > [...]
> > 
> > Uhmm... It seems that we wrote two similar patches (see the email that I
> > wrote this afternoon).
> Im sorry about that. Im slow.
Don't worry... It's my fault too. Anyway, it's not a big problem (it's
better to have two patches than not having any patch at all ;-)

> > I still like my version better, because I think it simplifies the
> > interaction between the padding and rescaling code. Anyway, tomorrow
> > I'll try to see how I can rebase my patches on your one.
> You are lovely :>
> Feel free Luca.
Well, I'm just trying to find a compromise for having the code in CVS in
a short time. I have my design fixed in my mind, and maybe I am liking
it too much... Trying to use your patch as a base will help me to "open
my mind" considering different designs ;-)

> >>It fill dst picture if src picture is NULL and copy src picture if src
> >>is not NULL.
> > 
> > Maybe img_pad() can be simplified if you call img_copy for copying src?
> Well I thought that it would be quicker to fill the new picture with old
> one in one pass, maybe not.
Ok. I was not thinking about speed, but about simplifying the code. But
it was just an idea...

> > Also, I see that your patch modifies the cropping code too... I assume
> > such changes are unrelated, right?
> It does not change that much, only assigns final_picture before the process.
Yes, I see... My point was: why is this needed? Maybe this change can be
put in a different patch?

If you agree, I'll split the patch into two or more parts, and integrate
it in my patchset. I'll post it soon (hopefully).


 Email.it, the professional e-mail, gratis per te: http://www.email.it/f
 Gioca con la DOG-MACHINE di Pedigree: vinci ogni giorno simpatici premi per il tuo cane. Vai subito!
 Clicca qui: http://adv.email.it/cgi-bin/foclick.cgi?mid=4876&d=18-3

More information about the ffmpeg-devel mailing list