[argyllcms] Re: Profile Evaluation
- From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sun, 12 Jun 2016 12:18:45 -0700
On Jun 12, 2016, at 11:32 AM, Roger Breton <graxx@xxxxxxxxxxxx> wrote:
First, are the reported units in "DeltaEs" of some kind?
The goal of "Delta E" is an unit which represents the smallest perceptible
difference between two different colors. There have been at least a few
attempts at refining it. The best-known is simply the distance between two
points plotted in 3-D L*a*b* space. For many applications, that's "good
enough," and it's simple to conceptualize and program. Most other formulations
you're likely to encounter (such as DE2000) will be closer to the original
(Have to study the
concept of RMS which, according to some videos, comes from the world of
"Root Mean Square," a statistical function of deviation. Think of bell curve
Argyll may already have the analytical tool to answer this question?
Assuming you're not doing anything exotic, profcheck is the tool you're looking
for. Sort the output and you'll get the worst deviations at the top or bottom.
When trying to figure out what went worng, keep in mind the possibility of
measurement error -- especially if you've only got a couple outliers, as is
often the case with a low average error.
Other related posts: