Hi Mike,
I've taken my time to check what you are reporting. After comparing results
with Photoshop, both gives same RGB values, so the problem, if any, is not
on lcms but in the profile. At that point, I wondered why you got such big
differences and here are some hints.
* First and most important: you should not trust on RGB differences. They
are meaningless. Use Lab and DeltaE instead.
* ICC states that some "pre cooking" should be applied to measured data,
specially on emissive devices:
- You need to remove veiling glare (flare)
- You need to scale the luminance (across 'lumi' tag).
I've done those on the excel file you provided and the results, again in
RGB contone, are:
http://www.littlecms.com/test%20chart%20simplified2.xls
MIN
-5
-1
0
MAX
5
13
16
AVG
0.161616
3.69697
3.959596
If you compare those with your initial results, there is some difference:
MIN
-4
-1
0
MAX
21
22
28
AVG
8.080808
11.83838
15.77778
But again, that is meaningless. Two contone values may differ 10 or more
units and be just on 0.1 dE colorimetrically.
Finally, this is a Matrix-shaper profile, which may model a CRT quite
closely, but would probably fail on LCD. I think a LUT-based profile would
be more adequate, anyway that is up to the profile manufacturer.
Hope this helps
All the best
Marti Maria
The littleCMS project
www.littlecms.com
---Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Michael A.
Litscher
Sent: Friday, December 14, 2007 10:22 AM
To: Little CMS Mailing List
Subject: Re: [Lcms-user] Lab to RGB not yielding expected values
Marti,
Thank you for replying.
I did attempt to apply a scaling factor to the L* value as I was feeding
the L*a*b* values into lcms for conversion to RGB after reading the
replies of Kai-Uwe and yourself. I found that no matter what scaling
factor I used, all it changed was which swatches had the most error, but
it didn't really reduce the sum of the errors.
So, I graphed L* against the R error, G error, B error, the average of
the RGB error, and the luminance of the RGB error to see if there was
any correlation. Here are the graphs:
http://www.flickr.com/photos/michael-litscher/sets/72157603452068860/detail/
If there was a correlation I would see a line or at least a curve in the
data, but there is no discernible trend.
Really, I do want this to work. I am not looking forward to converting
everything over to use Microsoft's color management system.
The spreadsheet is still available for download here:
http://files.colormetrix.com/lcms_test.zip
The Visual Basic code I use to call lcms from within the spreadsheet can
be viewed in Excel by clicking on the Tools menu, selecting Macro, and
then the Visual Basic Editor.
Thank you for your time.
Michael A. Litscher
CTO, ColorMetrix Technologies, LLC
[EMAIL PROTECTED] wrote:
> Hi,
> Kai-Uwe is right, ICC absolute intent does not account for intensity,
> that would
> be the CIE-absolute. There is a tag for obtaining such values, but lcms
> does not
> support that (nor any other CMM I'm aware of)
>
> Regards
> Marti
>
>
-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Lcms-user mailing list
Lcms-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/lcms-user
-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Lcms-user mailing list
Lcms-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/lcms-user
