Re: CMM's and strange banding in Gradients
Re: CMM's and strange banding in Gradients
- Subject: Re: CMM's and strange banding in Gradients
- From: Graeme Gill <email@hidden>
- Date: Wed, 22 Jan 2003 19:41:17 +1100
Job Kuipers wrote:
>
If I make a Gradient from Black to White (in Photoshop) and I take a
>
close look, I see some sort of banding. That is not new to me, but I
>
discovered that when I change my CMM in the ColorSettings-Pane, the
>
banding pattern also changes. How is that possible, I always thought
>
that the CMM was for converting from my RGB-workspace to my
>
CMYK-workspace.
The CMM is for converting one colorspace (ie. Monochrome) into another
colorspace (ie. CMYK). Unless your input colorspace (eg. working
space) is the same as your output space (eg. display/printing
device) then a color conversion is required.
Different CMM's can go about converting from one colorspace
to another in different ways. They may link input and
output profiles using different algorithms and using
different precision (numerical and lookup table
resolution). They may treat different parts of the
profile differently (ie. convert per-channel luts
separately or lump them into an overall clut transform).
There may be different rounding errors. There may be
different interpretations of the ICC profile format.
They may sub-sample the information within a profile
for a smaller memory footprint or better performance.
Interpolation of the multidimensional transform
may use different approaches to do the sub-grid cube
clut interpolation (ie. n-linear or simplex decomposition
of a grid cube, simplex decomposition may be done
in a variety of ways, for instance in 3 dimension
a cube can be divided into 5 or 6 simplexes.).
Some CMM's may choose different sub-grid cube clut
algorithms for different colorspaces (ie. for RGB to
Lab conversion, a diagonal simplex decomposition
is good, because the L value is then along the
edges of the simplicies). Some CMM's may use
higher order interpolation rather than linear,
to minimize gradient artefacts.
All these choices, plus many in the way the profiles
are constructed, will have an influence on whether
there are visible color conversion artefacts in the output.
>
And a second question, which CMM can I best use? I've been told the
>
Heidelberg CMM was the best there is, but when looking at my Gradient,
>
I find that Adobe ACE gives me a better result.
Systems like Colorsync have quite elaborate policy in regard
to which CMM they use by default. They generally try and use the
CMM that created a profile with a particular profile for instance.
You can of course overide such policy.
If by your criteria the Adobe ACE gives the best results,
then perhaps it is sensible for you to use Adobe ACE ?
Graeme Gill.
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.