spyro at f2s.com
Wed Nov 24 16:37:56 PST 2004
> I'm setting up to be able to print digital photographs - using Gimp with
> gimp-print as the basis. However, I get unwanted color and density
> problems - differences between the on-screen image and the printed
> version, it varies with paper type too.
Oh dear... this is a non-trivial area (on any OS) ;-)
You have three things you need to gety in calibration, the camera,
monitor, and printer. each of the three work in very different ways,
colour wise (monitors emit light, paper prints reflect it, cameras
quantise it...) and the whole process is slow. ideally you want three
calibrations - camera and monitor output, monitor output and printer,
and camera and printer (the last is a good test of wether you got the
previous two right).
The nightmare only starts there though - you need to realise that the
sensors in the camera have peak sensitivities at wavelengths that will
differ from those your CRT emits, and also from those the inks (and
paper) you use reflect.
the ambient lighting conditions will not help you either, and any
calibration you do really needs to be done using a while light source
(eg. the sun on a bright day, which, if you live in the UK at present,
could be a challenge...
other problems you will encounter include the differing dynamic ranges
of all three devices - the printer will be worst and the monitor and
camera not very much better.
Oh, and the camera loves to 'help' you by selecting different white
balances, of which your software may well have no indication of, thus
meaning that you wont be able to use a single colour correction for all
And none of this even begins to take into account the problems of
performing tests unbiased by your personal preferences (its so tempting
to make the image look nice, until you realise that whilst the one image
improved, you trashed all the others...
Getting lifelike prints is not easy.
> I'm looking for methods of calibrating the gimp-print system to
> eliminate the problem. At the moment I'm working by trial and error
> printing of test grayscales and standard test pics.
> Anyone have any suggestions as to better ways of doing it? Even software?
I think its been done. the problem is in getting the data to use with
such a system - many cameras dont provide even white balance settings,
and profiles for common printers / monitors arent common.
Are you screaming yet?
More information about the lfs-chat