[XviD-devel] Display Aspect Ratio -> Pixel Aspect ratio solution

Christoph Lampert chl at math.uni-bonn.de
Fri Dec 12 14:21:06 CET 2003

On Fri, 12 Dec 2003, Dirk Knop wrote:
> I was looking into aspect ratio addition to VfW GUI and started with a 
> small hack - setting frame.par to the #defines in xvid.h. Unfortunately 
> this isn't really what (windows-)users expect when they choose an aspect 
> ratio.
> So I want to make it Display Aspect Ratio in the GUI and abuse the 
> par_height and par_width for that. (I.e. encoder gets fed with a 640x240 
> image and a 4:3 DAR - set par_height to 2 and par_width to 1.). This 
> would work with mplayer (verified). Hardware players should work with 
> that, too. Only some windows players/decoders would have problems with 
> such anamorphic encodes due to the rescaling necessary after building a 
> directshow filter graph.
> On the other hand, people who want anamorphic encoding are capable of 
> using a correct player/decoder. Other users won't be affected since the 
> default will be 1:1 PAR/DAR.
> Is that too much of a "hack" of the standard or can we agree on such an 
> abuse of the PAR?

Why exactly do you consider this an "abuse"?  I thought that was what PAR
is for. The question is if it should really be so simple, to let users
chose AspectRatio and resolution independently. They might thinks it's
just another way of scaling the 40 inch image of their TV to fit on this
small 12cm disc. 

I guess it's very often that users who decode and user who encode aren't
the same users (What??? How could this happen?), and people playing with
broken decoders will of course complain.

And of course most users don't know what AspectRatio is, even less the 
difference between PAR and DAR. 

The only reasons I see for PAR are: 
a) encode anamorphic DVD material without rescaling, then the PAR from
   DVD should of course be asked for and used, 
b) poor man's deinterlace with par_height==2 and par_width==1. 

Both could be GUI option. Arbitrary values, hm, better hide that very very


More information about the XviD-devel mailing list