On Saturday 15 January 2005 14:42, Gerhard Gau▀ling wrote:
> For example, if you convert from adobeRGB into sRGB and viceversa 
> times, you wouldn't receive the original color impression never more, 
> lost, and there for in poor quality (comparable with the jpeg lossy
> compression, keep that in mind).
> So, we have to convert only the displayed version of the data, not 
> original data (Jan-Peter, Hal and others, please correct me if I'm 

This is exactly correct.  The conversion between color spaces will 
always introduce quantization errors.  If only one conversion is done 
these have a minimal effect on the image assuming the the gamut of the 
original fits into the new color space with minimal loses.  This is 
not likely to be the case when converting from a wider space such as 
AdobeRGB to a narrow space like sRGB.  But if this happens many times 
the image will suffer significant degradation.  In other words the 
color space conversion is NEVER lossless.

> Okay, so I assume, that my (and Hal's and Jan-Peter's) suggestions 
have to
> wait for the release after GEGL? For a new rendering-engine further 
> GEGL? GIMP 3.0?

I am not sure that I want to wait that long but if it will take that 
long to get it right then so be it.  I would rather wait and get a 
correct implementation then get an incorrect one sooner.  But I think 
that it should be possible to do at least some of this sooner.  For 
example perhaps a color aware printer interface could be added sooner 
by leveraging Alastair M.Robinson work on PhotoPrint. 

> > We currently have the ability to do color proofs with external
> > ICC profiles. THe interface to the loading of the profiles isn't
> > perfect yet, but it's there.
> Desired is a 'On the fly' Softproof.

I have little use for soft proofs since my custom profiled printer 
gives results that are almost identical to what I see on the screen in 
Photoshop.  But for those that are working with low gamut printers 
this is likely more useful. 
> I admit, that this is a very complex subject, and it is much work to
> implement all this color stuff into the GIMP, but I'm shure it's 
worth it.

I would also like to add that although I am not a color management 
professional I did struggle with color issues in my digital darkroom 
and as a result I have spent many hours setting up a proper (but 
perhaps somewhat basic) color management work flow.  In the process I 
studied many sources and learned a lot.  Gerard is correct this is not 
going to be trivial and it will take a lot of effort to get this in 

For those that are new to color management and would like to understand 
this better from a CM users perspective I would like to recommend that 
a good starting point is 
%20again!%20%20I%20am%20learning%20how%20to%20do%20the  This web site 
is from a color management users perspective and it starts out with a 
basic overview of how color management works and what all of the 
pieces are and how they all work together.  He give examples of how to 
setup things in both Photoshop and Picture Window Pro.  So this has 
lots of info about how two different apps have set up the user 
interface for this.  

Also one of the interesting things on this site is that he has the 
GretagMacbeth ColorChecker test pattern in both SMPTE-240M (same as  
AdobeRGB 1998) and sRGB color spaces.  One of the patches (out of 24) 
is out of gamut in the sRGB version of the image but is in gamut in 
the SMPTE-240M image.

Hal V. Engel

Attachment: pgp0hVLaJ6x0W.pgp
Description: PGP signature

Reply via email to