[MPlayer-dev-eng] [PATCH] add 8 bit Bayer format and filter

[Michael Niedermayer]

On Fri, Mar 07, 2008 at 01:54:12AM -0500, Rich Felker wrote:
> On Fri, Mar 07, 2008 at 05:42:43AM +0100, Michael Niedermayer wrote:
> > What this thread is about is entirely different, there is no lowpass
> > filter at all. 
> 
> If the data is insufficiently sampled, it's aliased 

yes


> and already
> damaged beyond all hope of repair. 

If the data is random yes, images are not random. As example for repair
think of a human artist looking at the available samples and filling in
missing ones, iam sure you will agree that he is able to guess
them much better than a linear filter could. And if so why do you think
a linear filter is the best "non human" choice.


> Fortunately the optics provide a
> decent amount of band-limiting so it's probably not that bad..
> 
> > > I agree. There's nothing nonlinear about this interpolation. All it
> > > requires is the right filter coefficients.
> > 
> > This reminds me of something mans said about you being an expert in
> > everything you never heard of before.
> > Anyway ill leave this to you to deal with, i dont have the energy for this
> > fight, write some linear interpolation and let the world have some
> > entertainment with the resulting images.
> 
> Linear filter and linear interpolation have nothing to do with one
> another. I guess it was poorly worded since it sounded like I was
> talking about linear interpolation, but what I meant was of course
> that, whatever the proper interpolation is, it's given by a linear
> transformation, i.e. there's no need for terms that involve products
> of distinct sample values (or higher powers of single samples).

Didnt you study math? Why dont you try to proof what you claim?


> 
> > Ive seen the difference, ill post a link to some of the papers if i find
> > them again.
> > The difference from linear to non linear methods is like that of a 320x240
> > to a 640x480 image. (ignoring the aliassing artifacts the linear variant
> > inevitably has)
> 
> I'll happily read the papers, but if you really mean to say that the
> transformations are not linear, it sounds like magic to me, and highly
> heuristic. 

It is heuristic, that is the point, some "shapes" occur much more often than
others, straight lines vs staircases aligned to samples with moire
artifacts being an example. An algorithm which fills in samples so as to
favor straight lines without moire artifacts will be closer to the real
data than one just running a linear filter over the image.

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

Its not that you shouldnt use gotos but rather that you should write
readable code and code with gotos often but not always is less readable
-------------- 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/mplayer-dev-eng/attachments/20080307/86516151/attachment.pgp>