Re: Black Point Comp yes or no ?
Re: Black Point Comp yes or no ?
- Subject: Re: Black Point Comp yes or no ?
- From: Chris Murphy <email@hidden>
- Date: Tue, 14 Sep 2004 10:53:43 -0600
On Sep 13, 2004, at 7:18 PM, Steven Kornreich wrote:
When working in a complete icc color managed workflow, wouldn't using
black point comp in PhotoShop kind of screw things up? It's not part
of the icc spec is it?
I am fine tuning my workflow right now and have noticed some big
differences in output when using or not using BPC in PhotoShop. Any
thoughts?
Some profiles don't get along well with black point compensation.
Depending on your point of view these profiles are either improperly
built, or the ICC specification is vague on how to properly build a
profile in regards to roundtrip black point behavior.
If the destination device has an L*=13, you almost always want the
detail that lives below L*=13 to be preserved instead of clipped unless
you just don't like shadow detail. Clipping is what Relative
Colorimetric does. Black point compensation dynamically compensates for
differences in source-destination dynamic range. Perceptual can have it
built-in to some degree, but it's not dynamic. It's one size fits all.
When it goes bad:
If I ask for L*=0 from such an output profile, it will respond back
with the RGB or CMYK values necessary to get the darkest value. At
least it should. Using a CMYK profile as an example, I get 91,86,68,95
(CMYK respectively) without black point compensation. Those numeric
values piped back through the same profile get me L*=13.
For black point compensation to work properly, sending any L* value
below the blackpoint of the device should get me approximately the same
CMYK values (or RGB for an RGB device) with the colorimetric intent.
For example when I submit L*=10 to the profile, I should get back
approximately the same values I do for L*=0. The device can only
produce an L* of 13 so I should get nothing LIGHTER than that when I
ask for any L* value LOWER than the device black point. However a
surprising number of profiles don't behave this way. This example
profile reports back 93,90,74,75. Passing those numbers through the
profile I get an L* of 15. Higher than the black point.
This fakes out black point compensation. In particular it adversely
affects black ink coverage in separations. For example without Black
Point Compensation you'll get a 95% black (the ink limit for black in
the profile), but with Black Point Compensation you'll get a 90% or
maybe as bad as a 75% black. And it is noticeable in print.
If you have such profiles:
1. Tell the profile vendor. It is definitely not the right behavior.
That the ICC spec doesn't tell us how to build them correctly in this
respect isn't an excuse, the ICC spec is vague on a great many other
things also. We need profiles that have accurate round tripping, and
the same numbers called for L* values below the device black point.
2. You will have to decide if you want to preserve shadow detail, with
a lighter black point than you could otherwise get. Or clobber the
detail and get a nice black black point. Or use perceptual rendering -
one size fits all "Black Point Compensation" and some slight and
potentially unnecessary desaturation.
Chris Murphy
Color Remedies (TM)
www.colorremedies.com/realworldcolor
-------------------------------------------------------------
Co-author "Real World Color Management, 2nd Edition"
Published by PeachPit Press (ISBN 0-321-26722-2)
_______________________________________________
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