Re: How do we tolorance a "match"
Re: How do we tolorance a "match"
- Subject: Re: How do we tolorance a "match"
- From: Chris Murphy <email@hidden>
- Date: Tue, 7 Nov 2000 18:02:53 -0700
>
Presses lay down color differently than proofs (we were referencing "fixed
>
density" halftone proofs at the time).
>
So, at press it is unlikely that press sheet and proof will be a 1 to 1
>
measured match.
I think if you eliminate characteristics of proofs that make them
difficult to match to a press sheet, in terms of colorimetry and color
this can be (and largely is) a solvable problem. If I'm printing on a
nice coated paper, and have a very similar proofing stock, I can get a
solid color on press to match a solid color on an inkjet. It doesn't
matter that the inks are different, or they are laid down differently. It
terms of color and colorimetry, I can make them look the same. (Well
actually a color management system does this, and does so within the
limits of device gamuts of course.)
The problem comes into play with the fact the two use different screening
algorithms so gradients and transitions are NOT inherently predictable by
a color management system.
Get me 50% cyan and 50% yellow on a press sheet and I assure you it CAN
be matched on an inkjet. But as soon as you print that same color, and
draw a red line through it, the color management system will NOT be able
to accurately predict the colorimetry of the transition area generated
going from the green background to the red line. That small little
transition area is primarily what makes up images - an image is entirely
transition areas. A very small fraction of neighboring pixels have the
same value. This is NEVER tested or measured when making *any* kind of
profile (be is separation tables, ICC profiles, or proprietary color
management systems).
The proprietary systems have a bit of an edge because their screening
process is fixed. They can incorporate visual analysis into their code,
thereby compensating for the effects of their screening algorithms. But
wide open color management systems like ICC based workflows depend on
people doing this manually - for the time being at least.
>
So the problem is: is there an objective way to define degree of mismatch so
>
that we can put a number to the amount of msmatch. Then we might be able to
>
define a tolerance range of acceptable mismatch.
What if the proofer does a great job in every case except for certain
violets and blues? Only those colors are off? How do you rate that system
compared to one that has only slight error in everything?
What about a system that does a great job matching uniform areas (like a
background, or a border); but not images? How do you rate that system?
What color model is going to be used to base these measurements? L*a*b*
predicts artificially low delta E's for blues. That's why it's possible
for Lab to predict a close hue match between a saturated blue and a
desaturated blue that actually looks like purple to the human eye. Lab
says it's good. My eyeball says something isn't right.
>
If we had tolerancing a customer could say something like: "This job is a
>
level 2 job" and everyone in the production chain would understand in
>
objective measurable terms what degree of mismatch the customer is willing
>
to accept assuming it is not one-to-one between proof and press sheet.
>
I had mentioned System Brunner in my earlier post because that problem is
>
exactly what that system is designed to do, and is being used for, in
>
Europe.
It's an apples and oranges comparison. There are a lot more printers
using ICC profiles and digital proofing in Europe than in the U.S. They
also have more pressroom training in process control. Since a majority of
design and printing isn't on a level playing field in the U.S., I think
for the near future and attempt to create a marketwide system of
qualifying color matching is futile. It will be helpful to a small
minority of printers and their customers, and will essentially be a
closed loop process. Nothing wrong with that, and certainly a worthy
goal. But it's not going to become common lingo when we don't even follow
the printing specifications and standards we have today.
Chris Murphy