Re: [Bino-list] green-magenta

[Alexey Osipov]

В Пнд, 14/02/2011 в 22:00 +0100, Martin Lambers пишет:
> >>> Maybe it will be even better, if user could adjust anaglyph filter
> >>> colours by himself. Perhaps with some calibration method like already
> >>> implemented crosstalk ghost busting.
> >>
> >> Do you think this is worth the trouble? A user would have to tweak two
> >> 3x3 matrices (18 floating point values). This might work to produce an
> >> anaglyph from a single stereo image, but for a whole video? Different
> >> values would be optimal for each different scene.
> > 
> > 18 values is too many indeed. What about entering 6 values (RGB for left
> > eye filter and RGB for right eye filter) and calculating other values
> > from there? Perhaps with some predefined other constants.
> 
> Do you have a detailed suggestion for such a method?

No, I don't. We can get Dubois calculations as a starting point.


> > Why do you think it will not work for movie? Red-cyan anaglyph matrices
> > are constant during all movie, aren't they? Why we can't calculate such
> > a matrices but for different user-supplied colours?
> 
> It can work, I just think that nobody wants to tweak this for a movie,
> since when you finished tweaking it for one scene, it may look bad in
> the next scene. 

> The current matrices are intended to be ok for most
> video content, 

It's for predefined filter colours. What I suggest is that Bino
automatically calculate such "ok for most video content" matrices, but
for slightly (significantly? does it matter?) different filter colours.

As for my "broken" glasses - their ghosting effect is always identical
for every scene. So I don't think there could be much difference between
scenes. If I could tune them for one scene, I suppose it will be much
better (not ideal however) for other scenes too.

> but of course constant matrices cannot be perfect for
> every scene.

Agree.


Alexey.