Re: Attn: Graeme Gill regarding lineraization
Re: Attn: Graeme Gill regarding lineraization
- Subject: Re: Attn: Graeme Gill regarding lineraization
- From: Graeme Gill <email@hidden>
- Date: Fri, 08 Jun 2007 13:08:57 +1000
Mark Rice wrote:
I presently have a Mutoh Toucan printer with Onyx RIP, and I am very
unhappy. I used to work with photographic devices such as the Durst Lambda
and the LVT film recorder. The rips used to drive those used "iterative
linearization". The has a set of target density aim points, and it would
compare the delta e of the read in grayscales to the aim points, and make
and adjustment toward the aim point from the presently read grayscale until
all the outputs were nearly exactly on the aim point targets.
I believe the Colorbus worked in a similar fashion - generating iterative
linearization, and the ICC profiles were applied after linearization - am I
correct?
No, it was a single pass (all that's needed if the original
measurement is accurate and fine enough), but the target curves
(defined in the end point target file) were by default absolute
rather than being relative to the maximum density of each channel.
You lost a little gamut, but got printer to printer matching
so that the same (possibly hand tuned) profile could be applied,
and day to day calibration was fast.
generated from what one specifies as the ink restrictions as a base point
for the curve. The curve cannot be altered or even predicted. When one
generates a linearization, and it is close but not dead on, re-linearizing
may make it better, or it may make it worse, since the previous
linearization does not exist in the computer's algorithm.
Sounds fairly useless unless you are just after "pretty pictures".
Anyway, my linearizations are extremely inexact compared to the photographic
images I used to output. Can you think of any solutions for this problem? I
talked to the head colorscientist at Onyx, and he acted as if I were being
ridiculously picky.
A calibration/linearisation system can be applied external to a device,
piggybacking on the top of ICC profiles, by adjusting just the
profiles B2A output per channel curves. I once sketched such a system
out, and it should be workable, the only disadvantage is that the
output resolution available to the profile may be inferior to
the built in system. It was certainly that way on the Cyclone,
since the calibration system output values were 12 bit (4096 steps),
whereas the ICC profile/image path was all 8 bit (256 steps).
Graeme Gill.
_______________________________________________
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