Monitor correction chart?

[Peter]

On Mon, 8 Aug 2005, Walter Dnes wrote:

> On Mon, Aug 08, 2005 at 04:27:56AM +0300, Peter wrote
>
>> If you calibrate the screen to the printer output all your edited files
>> will only print right with your printer, with the paper you are using,
>
>  What I was trying to say was... go out and buy (yes, pay cash) for
>
>  1) a reference-chart/test-pattern printout that is guaranteed correct

You may find that that will cost you about as much as a cheap computer. 
The 'cheap' charts linked to by another poster were just about USD100 
per and that *is* cheap for charts. A professional logarythmic grayscale 
chart is not printed, but assembled (with glue) from strips and masks of 
standard (chosen and manufactured by the maker) gray cardboard or 
similar. Each shade of gray is from a different material. It may set you 
back $500, new, depending on size and quality. Color charts are also 
assembled from dozens or hundreds of color fading-proof squares and 
other pieces of material. In a way it's just expensive origami but too 
many people depend on it for their work so the price is appropriate. 
They are also expected to last 5 years in studio use (i.e. under 
several 2000+Watt lights and in the sun) without fading or changing.

Also many will argue that the professional chart is useless without a 
densitometer (colorimeter). And they are likely right. For 24 bit color 
each color must be resolved to about 0.5% to justify the system. You 
can't do that with your eyes. And you don't really need it for normal 
photo work, only in professional printing work.

>  2) the digital file that was used to generate the reference-chart
>
> ...and then load the digital file (gif, jpeg, whatever) with any viewer
> program and twiddle knobs as necessary until what I see on the screen
> matches what I see on the paper.

The point is that if you really want quality, there will not be enough 
knobs to twiddle. You have to define what you need, because there is no 
end to it. The ICC correction map uses a large dataset to achieve 
correction. It maps each R,G,B value to another R,G,B value either 
directly or by interpolation. You are talking about editing a dataset 
that can be 1MB in size and larger. I have not seen a tool that does 
this manually but it could be done (likely using bezier curve 
manipulations in a gui that represents an input/output cartesian space 
for each color - like gimp uses). I think that proper color correction 
requires more than that (like NAM) but it may be enough for visual 
purposes. By more I mean color interdependency especially at high 
saturation levels. The pigments in the printer ink overlap in absorption 
spectra and it is necessary to back off certain colors to get accurate 
shades depending on color mix and on which color gets printed over 
which. CRT screens have very narrow emission lines which are unable to 
render certain colors (the human eye integrates them so you see what you 
want to see but that's not what's there). LCD colors have broader 
spectra and their white backlights have a wide white spectrum so they 
have the same problem as the printers (overlap requires backoff). So 
it's a huge can of worms and it all depends on how much you want to 
spend and how good it has to be.

The way I have shown, to find a chart file as pdf, go to a digital 
printer and have it printed on the best matte photo paper he has, A4 
size (letter), and use it as I have shown will get you as near as 
possible to what you need for well under $50 and half a day's worth of 
work. It will also make you perfectly compatible with that digital 
printer should you ever need it. That's just my opinion.

Peter
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml