Re: SANE V2 - again...

abel deuring (a.deuring@satzbau-gmbh.de)
Mon, 30 Aug 1999 20:31:22 +0200

Hi Oliver!

> If we want to add color correction as a "midend" it could be interesting
> to add a well known option for the color correction matrix (3x3).
> But this could also be done in a frontend, that does not make much
> work so I would'nt write a midend for it.

I would like to extend the topic of colour correction: Having a quite
simple option for a linear transformation is of course fine in many
situations, but it is not enough for high quality scans. [OK, the
following mixes the topics of colour correction and calibration...] Look
for example into some documents of the ICC
(http://www.color.org/profiles.html or
http://www.color.org/iccprofiles.html): The ICC profiles allow for
non-linear transformations in order to achieve better quality. A quite
complicated but perhaps useful approach to use ICC profiles within Sane
qould be a transformation

raw
scan -----------------> CIELab ------------------------> output
data mapped using data mapped using an data
an ICC profile ICC profile for
for the scanner CRT tubes
(or whatever)

[This develops into another example for a "midend", therefore I won't
elaborate on that any longer in order to avoid contradictions to my last
mail...]

Abel

--
Source code, list archive, and docs: http://www.mostang.com/sane/
To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com