color correction & vignetting [Re: [ptx] Hugin wishlist, RFC]

[Glenn Barry]

It would be great if you could incorporate the flat field correction at some
stage as I think this will shortcut the blending process for people with
calibrated systems. May even make it possible to have single layer images as
the output if the calibration is precise enough

I know that my corrections on the originals before stitching have simplified
my workflow after stitching to choosing my seams and not having to do colour
correction on the stitched images and seams for the most part.

Glenn
----- Original Message -----
From: "Pablo d'Angelo" <pablo at mathematik.uni-ulm.de>
To: "PTX mailing list" <ptx at email-lists.org>
Sent: Sunday, February 08, 2004 9:48 AM
Subject: Re: color correction & vignetting [Re: [ptx] Hugin wishlist, RFC]


> On Wed, 04 Feb 2004, Glenn Barry wrote:
>
> > On the subject of colour correction I have a Fuji S1 which has uneven
colour
> > from one side of the CCD to the other, I've made photoshop actions to
> > correct for this and vignetting when using the Sigma 8mm fisheye, but
the
> > way I did it was to photograph a white evenly lit piece of paper and use
the
> > RGB channels as masks for the colour correction.
> >
> > I don't know how prevalent uneven colour across CCD's is but what would
be a
> > great feature would to be take a calibration shot of a piece of evenly
lit
> > white paper and use the light falloff/vignetting to calculate values for
a
> > particular lense aperture combination and save those settings for radial
> > luminance correction.
>
> Hmm, with a flatfield correction (for each channel separately), this
should
> be possible, much like you do it by hand right now.
>
> > Now sorry for the rant, but if you have a calibration shot for a certain
> > lens/aperture you can adjust for any differences in light falloff and
> > vigentting from a known reference point. This may also help with
calculating
> > colour correction for shots all taken with the same settings for any
given
> > panorama. I know this adds one more step to maing panoramas, but given
that
> > FOV, A, B and C calibration is a good idea anyway, the extra step to
record
> > colour variation may make totally seamless output a reality.
> >
> > What does anyone think?
>
> While I haven't heard of the problem your camera has before, it might be
> nice if these errors could be corrected as well.
>
> And, yes, this correction is needed for color correction and stitching,
but
> it'll take some time until the current SIFT feature stuff etc. has been
> integrated.
>
> >
> > BTW I am going to try and Port Hugin to SGI Irix as I've got an SGI Box
here
> > that I want to make more use of(meaning fumble around and try to get it
to
> > compile, about which I know next to nothing, but you have to start
somewhere
> > ;-)
>
> Theoretically it should be quite straightforward, if wxwindows and vigra
are
> supported.
>
> ciao
>   Pablo
> --
> http://wurm.wohnheim.uni-ulm.de/~redman/
> Please use PGP
>