Measurement Geometry was: Big problems using Bill Atkinsons new Eye-1 profiles
Measurement Geometry was: Big problems using Bill Atkinsons new Eye-1 profiles
- Subject: Measurement Geometry was: Big problems using Bill Atkinsons new Eye-1 profiles
- From: Ken Fleisher <email@hidden>
- Date: Wed, 15 Feb 2006 22:16:25 -0500
Randy,
You've raised an very interesting issue which is often overlooked. A
profile can only be as good as it's measurements, so taking proper
measurements of your samples is essential. Most, if not all,
spectrophotometers used in the graphics industry have a 45/0
measurement geometry. I won't go into all of the details, but this is
not necessarily the best geometry for measuring canvas or other course
materials. An integrating sphere would probably be better.
But that's not to say you must go out and get an integrating sphere. To
understand the problem, think about how a 45/0 instrument measures.
When you have a very course sample such as canvas, light will scatter
in many different directions and a 45/0 measurement geometry is very
sensitive to this. That is why you found that taking more samples
greatly improved the results. I don't have any hard data on this, but
I'd be willing to suggest that an average of even more measurements,
perhaps 6-8, could improve the results even more when dealing with that
type of material. Of course, multiple measurements in this case implies
"with replacement" meaning you must remove the media after measuring
the chart once, then place in back on the table again so that the
alignment is not exactly the same. There is little point on measuring
the exact same location multiple times (at least in this case). The
multiple measurements will allow your average measurement to be closer
to "actual" spectral reflectance.
Glossy materials can benefit from this as well. It all depends on how
accurate your instrument is. How many measurements (with replacement)
you will need to take an average of in order to improve your profile's
accuracy will depend on how precise your instrument is--that is, how
closely do the measurements match on successive readings of the same
sample "without replacement". In general, the less expensive devices
will be slightly less precise than the more expensive ones. As for the
typical ones used by those on this list, three measurements with
replacement is not a bad standard practice. Of course, you would have
to weigh the benefits of a "potentially" slightly more accurate profile
versus the extra effort to generate it. After all, even the affordable
spectrophotometers today are pretty darn good!
I hope this helped a little...
Ken Fleisher
_______________________________________________
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