[sane-devel] gamma correction procedure

[Karl Heinz Kremer]

On Feb 9, 2004, at 12:32 PM, Christopher Marshall wrote:

> --- abel deuring <adeuring at gmx.net> wrote:
>> Christopher,
>>
>> this way, you measure (and correct to some extent) the colour 
>> deviations
>> of the *entire chain* image colour values -> printed colour -> scanned
>> colour. But both '->', i.e. each "transformation" of the colours into 
>> a
>> new "representation", introduces new errors. Especially, you must 
>> expect
>> serious deviations by the print process. This does not help if you 
>> want
>> to have "true colours" for a scan of a photo, since the photo has not
>> been produced by the printer.
>>
>> I'd recommend to follow Gerard's recommendation to use IT8 targets for
>> colour correction. With an IT8 target, you have a sort of a 
>> "calibrated
>> image"; which you can scan, and using programs from lcms or gcms, you
>> can "calibrate" your scanner, i.e. create an ICC profile for it. ICC
>> profiles provide a far better way to get something similar to "true
>> colours" than the comparatively simple gamma correction.
>>
>> Abel
>
> I see what you (and Gerard) are saying now.
>
> Assuming I used an IT8 target to calibrate my scanner, could I then 
> use the procedure I outlined
> to calibrate my printer?  Or I am still not getting everything you are 
> saying?
>
> I take it there are at least three gamma distortions I need to worry 
> about: my monitor, my
> scanner, and my printer.
>
> Using an IT8 target would allow me to separate the scanner's gamma 
> distortion from the other two.
> The procedure I proposed could then let me calculate the gamma 
> distortion for my printer.
>
> Correcting my monitor would be the hardest of all.  The simplest thing 
> to do would be to display
> an image of the IT8 target on my screen, hold a physical IT8 target in 
> my hand next to the
> monitor, and make adjustements to the monitor's gamma settings until 
> my eye felt the two agreed
> largely.
>
> Chris Marshall
>
Actually, calibrating your monitor is probably the easiest of the three 
(at least for a pretty good approximation), and the printer is the 
hardest. To adjust the three gamma values for the three primary colors 
of your monitor, you can use two color fields per color, one with a 
solid color, the other with lines of the full color and white. If you 
three gammas are symmetrical, you can do the same with just a pair of 
gray patches.

http://www.jasc.com/monitor1.asp
http://epaperpress.com/monitorcal/
http://www.normankoren.com/makingfineprints1A.html
http://www.easyrgb.com/calibrate.php

LittleCMS (http://www.littlecms.com) does provide a monitor profiler 
that works this way.

Profiling a printer is a major task. You would not be able to do this 
without a pretty heavy dose of math. Look at the Argyll project 
(http://web.access.net.au/argyll/argyllcms.html) for how this can be 
done.

Karl Heinz