Re: Instruments and models Vs. Accuracy.
Re: Instruments and models Vs. Accuracy.
- Subject: Re: Instruments and models Vs. Accuracy.
- From: Graeme Gill <email@hidden>
- Date: Wed, 14 Sep 2005 18:14:12 +1000
edmund ronald wrote:
I'm just your average journo, but this discussion about instrument
precision has caught my eye - Is instrument precision really relevant
these days ?
The construction of a behavioural model of a device (a profile in CMS
jargon) implies considerable assumptions about the physics of the
device.
Not necessarily. It depends on the details of the profiling software.
General purpose profilers make very few assumptions about the
physics of a device. They are the ones that needs lots of
sample patches. If you start making models for specific types
of devices, then you can get away with fewer patches, and
your profiles are only good for devices that the model represents.
In either case, the accuracy of your model is going to depend
on the instrument.
So, now please convince me that existing mathematical models of LCD
screens and inks on paper, the size of the measure samples, and the
training and interpolation algorithms in the profiling software are
good enough that a percent here or there in instrument precision or
repeatability actually matters *today*.
It depends a lot on what you are trying to do. If you want a profile
to convert between fundamentally different color spaces (such as
monitor to print), for a pleasing result, then instrument
differences aren't going to have much effect. There's lots
happening (gamut mapping), and nothing to reference to
(subjective "pleasing").
If you're trying to do proofing, then instrument differences
can be highly significant. (ie. if you were to use different
instruments for the source and destination profiles.)
Near the white point, errors of < 1 delta E are rather visible.
My feeling is the instrument is probably the strongest link in the
chain these days.
Not really. It's hard to make a good, accurate, stable instrument
for a low price.
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