Monitor Curves and Colour Profiles

[tomczak]

I've been having a long string of printing mishaps lately - some coused by the printer, some by me, and some by general evil. I thought I could wing it knowing a few things about colour profiles, but I couldn't. So I turned to the Monitor Curves, that seem to work much better than I expected (and seem to work reasonably under more general circumstances than just comparing the callibrated monitors and test print in exactly the same place as the final print would be).

But I have a question, an observation and a suggestion.

I prepere 48bit images having AdobeRGB as a working space (but sRGB would do just as well), then when all is done, I convert the ICC profile to the Durst Lambda/matte paper profile suplied by the lab (they could do that too, but I figured one less thing to go wrong if I do it right...), convert them to 24 bit and give them to the lab to print. The prints do loose contrast especially in shadows under a wide range of practical illuminance (including subdued daylight) but also acquire colour cast (most annoying purplish in bluish areas, especially in almost white highliths, dull greens, but others as well).

Jonathan explained many times how things conspired to make it happen: squizing the monitor gamut to the printer gamut which doesn't have such saturated blues, associated Abbney's effect since monitor blues get desaturated during profile conversion, and a steep gradient between cyan and magenta in most printer gamuts that makes it easy to slip into magenta.

So the question is this: when using monitor curves, when would be the best point to apply them: before or after conversion to printer colour profile?

The suggestion: when I tried to prelighten and match colour cast of the print using monitor curves, it woud have been helpful if I could probe the image and see near which curve control point am I at. It can be estimated indirectly by probe tool for instance, but it makes it harder to match the tones and the colours.

The observation: when I constructed the monitor curves, I used Inverse Monitor Curves in Colour Curves to pre-correct the print. In monitor curves, I darkened and added color tint to the white control point - which looked about right in the monitor curves preview, but I didn't realized that applying inverse curves will blow highlights while pre-lighting, accordingly. So it's probably better not to mess up with the black and white point in Monitor curves, even if it looks like extreme highlights and shaddows need help, but rather add more control points and manipulate those close to the ends.

[jsachs]

Monitor curves don't affect the image data at all so there is no connection between using them and converting the image to another color space. The only effect occurs when you print and check the inverse monitor curves box. Thus the files you mail to a lab are not changed by monitor curves and there is no advantage to using them. Most color photographic printers don't produce really dense blacks so some loss of contrast is to be expected. It sounds like the supplied printer profile is not very good.

[tomczak]

Thanks Jonathan,

Though the monitor curves doesn't affect the image data before printing, there is another way of using them when sending a file to the printer: using Inverse Monitor Curves in Color Curves Transformation/Options (for which I'm grateful). And this is what I do: use Monitor Curves to match the test print, then I prelighten the image/reverse colour casts using their inverse in Colour Curves, before sending it to the printer.

But I'm not sure what would work better though (given for instance the purple cast I'm fighting): making the Inverse Colour Curves transformation before translating AdobeRGB to Printer's profile or after the image data refers already to Printer's gamut?

The printer's profile is probably OK (or at least not worst than others). Cheers.

[jsachs]

I would guess that applying the inverse monitor curves on the Adobe RGB file but using the printer profile as a proofing profile might work the best, but it might be worth experimenting.