Re: ScannerRGB to WorkingSpaceRGB (rendering intents)
Re: ScannerRGB to WorkingSpaceRGB (rendering intents)
- Subject: Re: ScannerRGB to WorkingSpaceRGB (rendering intents)
- From: Henrik Holmegaard <email@hidden>
- Date: Sun, 8 Jul 2001 21:31:04 +0200
Terry Wyse <email@hidden> wrote:
I'm thinking that moving the
ScannerRGB data to a working space that absolutely preserves the color gamut
of the scanner may be a better choice, particularly for archiving the RGB
scan, but with the need to move the scan data to "well-behaved" working
space instead of archiving the scan with the ScannerRGB profile embedded.
In order to preserve the full gamut of the film + lamp + filter space
(: scanner space), and avoid editing in the scanner space, then IMO
you need to do the RGB to Lab step, and leave the colors in Lab. The
problem with this is that Heidelberg is taking an awfully long time
moving 16 bit Lab into its capture software, and Adobe is taking an
awfully long time 16 bit ICC Lab support.
In converting from ScanRGB to WorkingSpaceRGB, I've generally gone the
route of converting to a more "print-friendly" RGB working space (usually a
smaller space than the scanner) such as AdobeRGB or BruceRGB. I assume the
preferred intent here would be perceptual not relative colorimetric.
An RGB working space does not support gamut mapping of out of gamut
colors. There is no block of numbers in the profile that does a
perceptual remapping. If you get a different behaviour with
Perceptual and Relative Colorimetric, then it's just the CMM or
software or both that's twitching a leg -:).
The
subsequent WorkingSpaceRGB to PressCMYK is then *usually* done with a
relative colorimetric intent (just seems to work better for most images). In
any case, you're needing to deal with out of gamut colors in same way.
Well, you're not dealing with out of gamut color this way -:).
If you want the people who built your profiling application to help
you take a shot at it, use the Perceptual intent.
If you want to take a shot at it yourself, use the Relative
Colorimetric intent *WITH* black point compensation.
If your software does not support RelCol with BPC, but RelCol ICC
classic, then your choice is Perceptual.
If you use RelCol ICC classic from RGB to CMYK, your not exactly
doing a good job on the images. Adobe RGB has a dynamic range of L 0
to L 100. Anything from L 0 to the black point of the CMYK space is
clipped to the black point of the CMYK space, if you use RelCol
classic ICC.
If you implemented "RelCol for gamut mapping" out there anywhere, go
back on a moonless night, by the back door, and change this anywhere
and everywhere. And don't forget to fix the e-mail memos, too -:).
Hope this helps.