Re: ColorChecker DC Question
Re: ColorChecker DC Question
- Subject: Re: ColorChecker DC Question
- From: "Bruce J. Lindbloom" <email@hidden>
- Date: Mon, 30 Jul 2001 17:04:23 -0500
Since the DC chart seems to be undergoing some scrutiny, I'll throw in my
two-cents worth. Here is a critique that I forwarded to GretagMacbeth about
a year ago (some items are nit-picky, others are more important):
1) The data format of the custom reference files is different from the data
format of the generic reference file.
2) The data formats are similar to, but not the same as, the data format for
IT8 targets. Why not use an existing standard?
3) It is unfortunate that GretagMacbeth chose to label columns with letters
and rows with numbers, when IT8 targets, which have been around for some ten
years, use the opposite convention.
4) The case in which to store the chart is lined with a black flocking
material that sheds profusely. Not only will this mess up a profile, but I
have observed that this gives people an irresistible urge to blow on (read
"spit on") the chart or to rub the patches with their greasy fingers.
5) The DC chart has 237 patches, but only one (S8) has an L* value less than
20 (and this one gets thrown out if you choose to ignore the glossy
patches). So dark colors have a very poor representation. Where this
becomes a big problem IMO is with cameras using a tone curve to compress
16-bit samples down to 8-bits. A user defined curve is often used to do
this task, and without any patches for dark colors, there is no way in the
world for profiling software to figure out the shape of the curve after the
fact. This can cause poor rendition or color crossovers in shadow detail.
For this reason, I recommend shooting high-bit linear raw images (which will
look very dark when initially viewed on a monitor) and then assign the
profile to make it look right.
6) The DC chart does not include the colors of the original ColorChecker
chart. I think it would have been nice to have that embedded reference.
7) It would also have been nice to have a light trap as one of the patches.
This would be a square hole in the chart with a small box lined with felt
(maybe here's where GretagMacbeth could use up their black flocking
inventory!) stuck on the back of the chart. The box could be folding for
storage purposes. A light trap is very helpful in measuring the bias of the
camera and optics (dark current and flare). This is _very_ useful
information -- it's easy to build your own light trap out of two empty sets
of nesting 4 x 5 Ektachrome film boxes.
On top of these are the issues of glossy patches and polarized light that
have been topics of recent discussion on this list.
--
Bruce J. Lindbloom, Pictographics Intl. Corp.