Re: ColorChecker
Re: ColorChecker
- Subject: Re: ColorChecker
- From: Robin Myers <email@hidden>
- Date: Tue, 16 Sep 2008 14:26:37 -0700
I have the original 1976 paper and spectra derived from the published
graphs. I also have direct spectral measurements from two
ColorCheckers from 1977 and at various dates in its evolution. In the
years since its release, the ColorChecker has been reformulated at
least twice, according to personal measurements. In talks with people
involved in its production, the reformulations were made in order to
use pigments that were less toxic than some of the original pigments
or to replace pigments that were unavailable or too costly. These
reformulations may have reduced the number of pigments involved, which
would make sense from a business point of view, i.e. reduced
manufacturing costs, not having to place toxic warnings on the
product, etcetera. In spite of the spectral differences between the
original and the current ColorChecker, the folks at X-Rite have done
an excellent job of keeping the color appearance, i.e. colorimetry,
very close to the original.
So Eric Walowit may be correct that the current ColorChecker shows
more correlation amongst the spectra than expected. I also agree with
him that trying to make a chart that reflects the wide range of
colorants in the real world would be an incredibly difficult, if not
impossible, task. However, within some specific application areas,
custom charts made from similar pigments found in the subjects have
been successful. For instance, Roy Berns of RIT published a paper
about some work with the National Gallery of Art in Washington, D.C.
where they used a custom chart made from artist's pigments for
calibrating and profiling for fine art reproduction.
As for the ColorChecker SG, comparison of the spectra in my collection
show that although the 24 patches in common with the ColorChecker have
similar spectral shapes, they are not at the same reflectance factors
as the originals. This is to be expected from the surface alone since
the ColorChecker SG has a semi-gloss surface and the ColorChecker has
a matte surface. Thus the ColorChecker SG will show less diffuse
reflectance with a correspondingly more saturated color appearance.
I would caution everyone to double-check the statements on websites,
even from companies that should know better. The case in point is X-
Rite where their description of the ColorChecker SG states that it has
a 17 step gray scale on the chart where it only has a 15 step gray
scale (i.e. E5, F5, G5, H5, I5, J5, E6, F6, G6, H6, I6, J6, K6, K7,
K8; the outer white, gray and black patches are duplicates of interior
gray scale patches).
Robin Myers
On Sep 16, 2008, at 12:24 , edmund ronald wrote:
Can't someone dig out the original colorchecker paper ?
Edmund
On Tue, Sep 16, 2008 at 8:38 PM, Lars Borg <email@hidden> wrote:
Hi Eric,
So maybe there are only 6 pigments in CC24??
That would be disappointing
Lars
At 10:56 PM -0400 9/15/08, <email@hidden> wrote:
Eric,
Actually the first 6 principal components account for nearly all the
variation, so the 24 patches are highly correlated. And despite
claims to
the
contrary, there is no way that a reflective target, no matter how
well
designed can
reproduce the complex interreflections, translucence, and other
non-reflective
effects that make up most of the scenes we wish to reproduce.
Colorants
behave
very differently spectrally depending on the mode of illumination.
The other Eric
Eric Walowit
Tahoe
_______________________________________________
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