"Preferred" perceptual greyscale gamut mapping
"Preferred" perceptual greyscale gamut mapping
- Subject: "Preferred" perceptual greyscale gamut mapping
- From: Steve Kale <email@hidden>
- Date: Tue, 06 Dec 2005 15:33:49 +0000
- Thread-topic: "Preferred" perceptual greyscale gamut mapping
Title: "Preferred" perceptual greyscale gamut mapping
Hi
Probably a bad title for the question I am going to ask but here goes. I’d like if possible to get some input from you colour gurus to the narrower question of greyscale output profiles, in particular managing the compression of image file greyscale values into a printer’s narrower dynamic range. Roy Harrington has written a lovely little piece of software to create an ICC profile for a greyscale workflow, an area long forgotten about by the giants of colour management. (It is part of the Quadtone RIP shareware product, which is generously provided at the non-profit shareware cost of $50.) In essence, the droplet program is fed data from the reading with any spectrophotometer of a n-patch greyscale step wedge printed using a B&W workflow of choice, eg a particular set of Epson Adv B&W settings. Only one intent is accommodated at the moment and this is labelled the “perceptual” intent. The B2A0 tag is single channel gray. As a result the profile does not attempt to manage hue, that is left to the tint picker in Adv B&W or the selected RIP tone “curves” or whatever. Only the lightness L* axis is managed for printing. (The A2BO tag is three channel colour and so the profile provides full colour soft proofing ability. Quite useful to preview the tone of the print settings.) My interest is in exploring the “ideal” management of the lightness axis. Obtaining insight into this is rather difficult despite the fact that it is (part of) the bread and butter of colour management engineers.
I’m aware that the perceptual intent is entirely up to the profile designer and this is where companies such as Gretag-Macbeth deploy their “secret sauce”. Currently Roy’s little greyscale ICC profiling application is based on (a) scaling all the recorded observations input to the profile for media white point (ie media relativity) and (b) a scaling for black point based on an understanding of Adobe’s black point compensation as outlined in their somewhat opaque white paper, available at www.color.org. Given the extreme importance of managing black point differences to B&W printing it’s worth exploring whether this represents a suitable/sound/sensible/preferred algorithm given one can deploy any algorithm one feels like for the Perceptual intent. It is interesting, for example, to take a look at soft-proofs using a Gretag-Macbeth colour ICC profile on a B&W image. Using an EyeOne-generated profile for my Epson 4800 for HPR paper, for example, Relative Colormetric with black point compensation checked tends to block up the shadows too much. The GM profile’s Perceptual rendering is much better. From what I understand of the Relcol intent, when applied to B&W images we simply get a scaling for media relativity. Add Adobe BPC and, assuming Roy has deployed his black point scaling in line with Adobe, the current QTR-Create-ICC application deploys an algorithm consistent with Relcol+BPC. The GM ICC profile example suggests that perhaps there is a more optimal or preferred method of managing the tonal compression. After all, when given the freedom by the Perceptual intent to deploy whatever algorithm they like they clearly deployed something which in greyscale ends up being quite different from Relcol+BPC. So it begs the question that I’d like to ask of the colour gurus out there: given the freedom to apply your vast knowledge and experience in gamut mapping algorithms to the much narrow case of B&W (L* axis only) what might represent a “preferred” method of scaling image data on a single axis for media black point and white point?
Cheers
Steve
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden