Color calibration targets (color correction of scanned paper images)

Hello,

I'd like to get recommendations on color calibration targets for the
archival scanning of printed paper materials (books, magazines,
photographic prints, etc.)

The purpose is to color-correct the digital scan images.

In addition, I may in the future employ a camera-based scanner
(exemplified by the Internet Archive's "Scribe" machine), and curious
if there is also a standardized target for measuring and even
correcting any lens distortion (such as barrel distortion)? Are
there calibration targets that could be used for both color and
geometric distortion correction?

Thanks.

Jon Noring

"Jon" <jon@noring.name> wrote in message
news:eanb82tinraelul0b30vq0mhfbapsvlplg@4ax.com...
> Hello,
>
> I'd like to get recommendations on color calibration targets for the
> archival scanning of printed paper materials (books, magazines,
> photographic prints, etc.)
>
> The purpose is to color-correct the digital scan images.
>
> In addition, I may in the future employ a camera-based scanner
> (exemplified by the Internet Archive's "Scribe" machine), and curious
> if there is also a standardized target for measuring and even
> correcting any lens distortion (such as barrel distortion)? Are
> there calibration targets that could be used for both color and
> geometric distortion correction?
>
> Thanks.
>
> Jon Noring

Paper targets are available.
You can get color charts from Photographic supply houses.

There is the MacBeth color chart and the Kodak Color Bars and Grey scale.
I have a tif image of scans of those targets.
http://www.carlmcmillan.com/Test_images.htm


B&H link to the (current) MacBeth colorchecker.
http://www.bhphotovideo.com/bnh/con...oughType=search

B&H link to TecNec MCS-1 Test Chart Set.
http://www.bhphotovideo.com/bnh/con...oughType=search

There are some Resolution Test Patterns that could give perspective on
geometric distortion checking.
http://www.bealecorner.com/trv900/respat/

There are some really good PDF test charts in the above link that can be
printed at very high resolution on a real good inkjet printer on
photographic glossy paper.
This one is black and white.
http://www.bealecorner.com/trv900/respat/respat.pdf

--
CSM1
http://www.carlmcmillan.com
--

"CSM1" wrote:
> Jon Noring asked:

>> I'd like to get recommendations on color calibration targets for
>> the archival scanning of printed paper materials (books, magazines,
>> photographic prints, etc.)
>>
>> The purpose is to color-correct the digital scan images.
>>
>> In addition, I may in the future employ a camera-based scanner
>> (exemplified by the Internet Archive's "Scribe" machine), and
>> curious if there is also a standardized target for measuring and
>> even correcting any lens distortion (such as barrel distortion)?
>> Are there calibration targets that could be used for both color and
>> geometric distortion correction?

> Paper targets are available. You can get color charts from
> Photographic supply houses.

One issue, obviously, is getting color charts with known RGB values,
plus some sort of color management (or mapping) system, which takes
a scan of the color chart, and from that creates a profile which may
be used to color correct page scans from the same scanner.


> B&H link to the (current) MacBeth colorchecker.
> http://www.bhphotovideo.com/bnh/con...oughType=search
>
> B&H link to TecNec MCS-1 Test Chart Set.
> http://www.bhphotovideo.com/bnh/con...oughType=search

Gulp. The prices of these charts are quite high. I was looking for
something in the $30 range or less. My goal is not necessarily to get
color correction to 0.1% accuracy, but something reasonably close for
archival (preservation) purposes. Obviously, what is "reasonable",
without providing a number, is open to a lot of interpretation -- one
person's "reasonable" is another person's "unacceptable."

Is it difficult to produce a color calibration target that is
reasonably priced, or does each one (or those from a batch) require
individual calibration?


> There are some Resolution Test Patterns that could give perspective on
> geometric distortion checking.
> http://www.bealecorner.com/trv900/respat/
>
> [snip]

Thanks!

Anyone else want to weigh in on this topic?

Is there anyone out there who color scans printed paper items (like
books), who uses color calibration targets (and color correction) for
every job?

Jon Noring

"Jon" <jon@noring.name> wrote in message
news:jqrb82pq26rdjmm5vt29aff1jdqnr5m24n@4ax.com...
SNIP
> One issue, obviously, is getting color charts with known RGB
> values, plus some sort of color management (or mapping)
> system, which takes a scan of the color chart, and from that
> creates a profile which may be used to color correct page
> scans from the same scanner.

Correct, that's the idea.

http://www.targets.coloraid.de/ is a source for good quality, yet
reasonably priced 'IT-8' targets, including reflective versions. They
come with a reference file that gives the actual colors, but at a
predefined reference spectral illumination. You'll need an application
that can build a profile based on a standard illumination level and
as little postprocessing as possible.

At the risk of attracting the resident VueScan basher :-(, VueScan
(professional) <http://www.hamrick.com/vsm.html> allows to build
profiles with such an IT-8 target, and probably can drive your scanner
as well. The profiles are of a simplified type, but will get you close
to the result you want.

Other options include experimenting with Little CMS
(http://www.littlecms.com/).

Bart

If I understand your post (and I apologize if I do not):
You are asking irrelevant questions.
If you work in a color managed environment these issues can be easily
addressed by your workflow.
You do not necessarily need to calibrate your scanner even for archival work
as most decent flatbed scanners will ship with a canned but reliable
profile. As the light source ages the canned profile will not likely be
correct but in truth it is irrelevant.
Unless your material is archived with a color profile and downloaded by
users into a color managed program that can use that information all is for
naught anyway.
Whoever downloads your material can use it as is or apply whatever
corrections they deem necessary. There is in fact no 100% accurate color
scan or way to reproduce any image with 100% accuracy or distribute it that
way to end users. End users will do whatever they want with the image
anyway.
There are several programs and plugins that purport to have a database of
distortion correction information for commonly used lenses but in practice
even these are not necessarily reliable (e.g. DxO) in that you may not agree
that the applied correction yields a better image. There is nothing of that
sort for flatbed scanners. Unless you transmit all info about how the camera
image was obtained these programs would be useless anyway.

[Some comments on the scanning of public domain books at end.]


Bart van der Wolf wrote:
> Jon Noring wrote:

>> One issue, obviously, is getting color charts with known RGB
>> values, plus some sort of color management (or mapping)
>> system, which takes a scan of the color chart, and from that
>> creates a profile which may be used to color correct page
>> scans from the same scanner.

> Correct, that's the idea.
>
> http://www.targets.coloraid.de/ is a source for good quality, yet
> reasonably priced 'IT-8' targets, including reflective versions.
> They come with a reference file that gives the actual colors, but at
> a predefined reference spectral illumination. You'll need an
> application that can build a profile based on a standard
> illumination level and as little postprocessing as possible.

Thanks! I should have realized the role illumination level plays on
color reproduction. *slapping hand on head*


> At the risk of attracting the resident VueScan basher :-(, VueScan
> (professional) <http://www.hamrick.com/vsm.html> allows to build
> profiles with such an IT-8 target, and probably can drive your
> scanner as well. The profiles are of a simplified type, but will get
> you close to the result you want.
>
> Other options include experimenting with Little CMS
> (http://www.littlecms.com/).

Thanks for the informative feedback.

****

Now, I'd like to address the reply from "bmoag" by providing a little
more background to my initial request.

My interest with color calibration is to setup a local "distributed
scanning" group to scan public domain books, and do so in association
with the Internet Archive and OCA (now talking with Brewster Kahle.)
It is likely we'll start off using the fairly inexpensive Plustek
OpticBook 3600 which can do bound books. Page scans will be at 600
dpi, 24-bit color, and losslessly compressed, burned to DVD-ROM and
sent in to IA as well as archived on our end (some illustrations we
might go as high as 1200 dpi.) Later on, after we better understand
the real-world work flow issues and become comfortable with our QA
procedures, we may get a professional-level sheet feed scanner to do
books which can be chopped -- this should greatly increase our
throughput. Of course, the Internet Archive has developed the Scribe
scanner, which uses digital cameras rather than a flatbed arrangement
for scanning, and if our group thrives, we may add that scanner, or
something similar like the Atiz, to our fleet of scanners (thus my
interest in geometric lens distortion issues.)

Obviously, in a volunteer group setting like this, using multiple
scanners, we need to color normalize our scans, or provide the means
for someone else to do so by including scans of a standardized color
target along with each book scan set. For our scanning workflow, the
approach I'm thinking of is to scan the color target first, then scan
the book pages, then rescan the color target at the end of the job.
The color target scans will always be included with the book's raw
scan set.

One issue we have, of course, is QA. And one QA item (among several)
is if a scanner has technical problems. This should be seen by our
QA volunteers when they look at the scan of the color target. If too
dark, or the colors are obviously way off, or there's "banding" across
the solid color areas, means the scanner is having some kind of
problem, or the person scanning may have used the wrong scanner
settings.

We don't initially plan to cleanup ("restore") the raw page scans,
which includes color normalization, deskewing, cropping, etc. That's
for another volunteer group to do (anyone here want to volunteer to
set that up and lead it? <smile/>) Certainly having a scan of the
color target for each book set will aid with both QA and with color
normalization.

Now, one has to remember that in a volunteer setting with "amateurs",
many of whom are not trained Rocket Scientists (tm), we have to
minimize as much as possible all the technical details they have to
contend with during scanning -- they will have enough on their plate
as it is. Keep this in mind if you decide to reply with suggestions of
your own (of course, suggestions are welcome!)

Thanks.

Jon Noring


(p.s., yes I'm aware of the large-scale scanning activities taking
place. After all, I'm in contact with Brewster Kahle at IA, etc. So
don't bring this up -- there is interest in trying the "distributed
scanning" idea for reasons off-topic to this group.)