Re: Color Measurement
Re: Color Measurement
- Subject: Re: Color Measurement
- From: Rolf Gierling <email@hidden>
- Date: Mon, 20 Jul 2009 13:10:32 +0200
Hello Robin and Graeme,
it is important to note that it's not the Just Noticable Difference
JND that is in question here,
but the non-uniformity of the color spaces used that lead to other
results.
If CIECAM02 is a nearly visual uniform space, one could say that one
unit represents one JND.
CIELab definitively isn't, so a JND might be represented by other
unit values (bigger or
lesser than one, depending on the color).
some quotes of Jan Morovic, Color Gamut Mapping, Wiley 2008
page 27:
2.6.2 Color difference and uniform color spaces
While being able to tell whether two stimuli will match is a useful
first step, it is even more important
to predict how different stimuli are when they do not match. This is
precisely what color difference
equations, typically denoted by dE, set out to do with two key aims:
1. that a color difference equal to one (i.e. d E = 1) should
represent a just noticeable difference (JND),
i.e. the smallest stimulus difference that results in a different
response by a viewer, for all color pairs;
2. that color difference predictions should relate uniformly to
judgments made by observers
(i.e. a color pair with d E 10 should be seen as 10 times as
different as another pair with dE = 1).
page 58:
For simple originals, the perceptibility threshold (for 50% of the
population of observers) is one d E,
since d E metrics are set up so that their unit is a JND (Section
2.6) when applied to spatially uniform stimuli.
Note, however, that this is the intention for all dE metrics and
that it is, therefore, important to use as accurate
a d E metric as is available, which currently is CIEDE2000 (CIE, 2001).
Rolf Gierling
Am 20.07.2009 um 03:33 schrieb Robin Myers:
I do not have a reference for the rule of thumb that dE2000 is
about half of dE1976. It was a verbal conversation at the Color
Imaging Conference a few years ago. Like all rules of thumb, large
tolerances are often involved.
However, the dE1976 of 1 unit being a JND has never really worked.
In the near grays, it has often been my experience that 1/2 to 1/4
unit is a JND depending on the difference direction vector and
position in the color space.
What all of this boils down to for me is that there is a large body
of work left to be done on color difference metrics.
Robin Myers
On Jul 18, 2009, at 12:21 AM, Graeme Gill wrote:
Robin Myers wrote:
Just as the scale changes in going from dE 1976 to dE 2000, the
tolerance range also changes. Where 1 dE 1976 was "roughly"
assumed to be a Just Noticeable Difference (JND), now that number
should be about 0.5 dE for a JND in dE 2000. As the scale
changes, so does the interpretation.
Do you have a reference for this ?
Looking thought the "The Development of the CIE 2000 Colour
Difference Formula: CIEDE2000"
and "The CIEDE2000 Color-Difference Formula: Implementation Notes,
Supplementary Test
Data, and Mathematical Observations" it isn't obvious that there
is any intention
that the JND threshold be different.
This plot also seems to contradict this idea:
<http://www.digitalcolour.org/Difference.htm>
ie. differences of about 1 CIE 1976 DE are also close to 1 CIE
DE2000 in
near the neutral axis.
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:
40rmimaging.com
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40multitools-online.de
This email sent to email@hidden
-----------------------------------------------------
Multitools
Dipl.-Ing. Rolf Gierling
Colormanagement
Development - Consulting - Distribution
Mariabrunnstr. 16
52064 Aachen
Germany
Tel. +49 (0) 2 41 / 4 01 25 79
Fax. +49 (0) 2 41 / 4 01 25 83
email@hidden
www.multitools-online.de
Author of "Farbmanagement"
3rd Edition July 2006
mitp Verlag, Bonn
ISBN 3-8266-1626-X
_______________________________________________
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