Very sorry... but actually lcms has no way to make printer profiles.
Just to make an idea on what you need to do a RGB-RGB
- Make a direct profile (this is what is doing lprof actually)
- inferring printer gamut
- invert by Newton or something else the LUT inside gamut
- extrapolate outside gamut
- gamut remapping for perceptual & saturation intents
- Gray balance.
- Undercolor removal
- Black generation
- Ink limiting
- Back to RGB
RGB input mean you have no control on K generation, so actually is an
additional difficult instead of a benefit. A grayscale patches will only
result on K ink and you got no information or C, M, Y colorants.
Sorry again. I intend to do some printer profilers, but in a far future.
If you really need it, take a look on Graeme Gill's argyll.
Anyway, be warned you will need a hardware measurement device
to do this.
----- Original Message -----
From: "Mark Rubin" <markrubn@...>
Sent: Friday, February 28, 2003 7:29 PM
Subject: [Lcms-user] Making prniter profile
> I need to generate a printer profile. Help! :-)
> I know that in the past you've said that printer profiles are
> difficult to make. But unless you tell me that "difficult" means
> "impossible", I have to try.
> I need the profile because currently I have to make several test
> prints before getting a decent final print. The printer driver I'm
> using (GIMP, gimp-print, espson/print-escp2 ) has controls for cyan,
> magenta, yellow, and gamma. Every print requires different,
> unpredictable, settings for these, and even then it's often impossible
> to match all the colors in the print to the CRT. I'm not surprised:
> Color management systems *are* necessary, and matrix color correction
> (not just R/G/B lutting) is required.
> I finally found a profile from Epson, but it's worthless. It's
> obviously for their Windows driver and for an unknown type of paper,
> but I had hoped it would have some useful colorspace shaping. In
> fact, it's simply a monochrome gamma adjustment. From experience
> (above) I know that significant color shifts are required.
> Back to LCMS. I know you've talked about the difficulties of CMYK
> undercolor removal, and also about error concatenation when using an
> input device (scanner in my case) plus profile rather than a real
> colorimeter to measure an output device. But I need an RGB-to-RGB
> (not RGB-to-CMYK) profile, and even an approximate profile would be
> better than none.
> I've been reading the lprof cmsscn.c source. Is this a useful
> starting point? From what I see, I'd need to change from
> icSigInputClass to icSigOutputClass, and all the AToB0 tags to BToA0.
> (I still find these confusing. Why couldn't the ICC have called them
> DevToPCS and PCSToDev?) (Can a CMS "invert" a profile -- turn an AToB
> transform into a BToA -- if the profile doesn't have both?)
> Again, I'm desperate to make a profile, and am willing to
> code/hack/experiment to do so. But if you can say in advance that it
> just won't work given the current state of LCMS, I don't want to waste
> the time.
> As always, thanks for any advice. Or do you have a pre-pre-alpha
> version of the printer profiler that you'd like to have tested? :-)
> Do you Yahoo!?
> Yahoo! Tax Center - forms, calculators, tips, more
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> Lcms-user mailing list