Re: Photoshop Gamut warning vs ColorThink
Re: Photoshop Gamut warning vs ColorThink
- Subject: Re: Photoshop Gamut warning vs ColorThink
- From: Lars Borg <email@hidden>
- Date: Fri, 22 Feb 2008 22:51:46 -0800
Hello Graeme,
There is a more accurate method, but it isn't "better" for checking
a raster image, since it's too slow and needs extra information.
The method is to directly invert the colorimetric A2B table, since
it is the closest definition of the device response.
("invert" meaning "locate the CMYK color in the A2B table that
produces the target PCS value. If no combination of CMYK can
create it, it's out of gamut.")
That approach has problems. A2B does not represent the printable
gamut or the TAC limit.
For example, there is an entry for 400% CMYK in the A2B table, but
most prints are limited to far less.
In my view, the A2B tag is useless as a gamut indicator.
It seems there's a lack of agreement on what "gamut warning" means.
Here's my interpretation. Please let me know how you disagree.
A "gamut warning" indicates whether a specified source color can be
"reproduced accurately" when converted to a target color space.
"Gamut warning" is used only in the context of converting from a
source color space to a target color space.
"Gamut warning" is not used for color values already in the target
space, such as 400% CMYK exceeding a TAC of 240%. Is there a commonly
used term for this case?
"Gamut warning" is not used to indicate or measure the size of a
device's gamut.
"reproduced accurately" is of course a loose term and has issues. A
color accuracy metric could be "within xx dE of the original source
color".
This is a challenge with profiles where the B2A to A2B transforms
don't round-trip for core colors. The lack of round-trip implies that
core colors shift (intentionally or not) when reproduced. How do we
determine a relevant deltaE in such a case? Does the A2B tag really
reflect the output color? Note that it's not meaningful to invert the
B2A tag, as in perceptual intent this may actually round-trip for all
colors.
A deltaE accuracy metric is not meaningful in cases where the
reproduced colors are intentionally made lighter by BPC, or
compressed by perceptual, or shifted by paper color.
Thus, we may want a more refined error metric "within xx dE of the
desired reproduction of the original source color", but how do we
know what this expected reproduction is?
An additional goal could be "reproduced uniquely", that is, if the
source color appears distinct from other source colors, then the
reproduced color should also appear distinct from the reproductions
of the other source colors. A heavy gamut compression or say a
quantization to 5 bits per channel may have a modest delta E but
still make colors non-distinct.
If the device or profile is deficient (say, a LUT point is off), then
colors that are wholly within the gamut can be reproduced
inaccurately. Should this result in a gamut warning or not?
For the target gamut, I consider only the gamut reachable through the
B2A tags, not usable through direct coding of device values.
The B2A tags represent the reachable gamut, the set of CMYK colors
that can be obtained through a conversion through B2A to CMYK. 400%
CMYK is usually not within this gamut.
The B2A tag imposes TAC limits and would generate 400% ink only if
that's within TAC.
The B2A tag also reveals the gamut limits imposed by various GCR
methods. If B2A tags have no or very low GCR, then many dark colors
are out of reach, and thus out of gamut. This means that different
profiles for the same press and with the same A2B tags, but with
different GCRs, have different reachable gamuts. A high-GCR profile
usually can deliver a higher number of unique colors in the dark
area than a low-GCR profile. It is then quite appropriate that these
profiles also provide different gamut Warnings.
Given that the gamut warning indicates the device's, the profile's
and the **conversion's** ability to reproduce a source color, the
gamut warning should change when the conversion changes.
Thus, even when the output device stays the same, you should get
different gamut warnings, if you change profiles (different GCR
levels, different TACs) or rendering intents (relative, absolute,
perceptual, BPC)
Some implications of the above:
- Even if a profile had a perfect gamut tag, it would not be useful,
as it would not be applicable to all rendering intents.
- A2B tags don't reflect the reachable gamut.
- Rendering intent must affect gamut warning.
- GCR and TAC must affect gamut warning.
- Reproduction accuracy goals are vague and subjective.
- Reproduction errors are, for some profiles, difficult to separate
from intentional color shifts.
- Determining meaningful gamut warning metrics for perceptual and
other compressing transforms is a challenge.
Thanks for reading this far.
Lars
_______________________________________________
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