RE: HowTo? Objectively measure quality of print profiles?
RE: HowTo? Objectively measure quality of print profiles?
- Subject: RE: HowTo? Objectively measure quality of print profiles?
- From: Roger Breton <email@hidden>
- Date: Fri, 03 May 2013 06:46:42 -0400
Dear LdaSignup,
The test of Output Profiles can be arduous and not necessarily reducible to
simple colorimetric accuracy.
The iSis or some other automated instrument -- I'm still fond of my
Spectroscan tables -- can and should be used to assess the colorimetric
accuracy for proofing but no more. The rest, the appearance of neutrals, the
rendering of skintones and other memory colors should be done visually. For
what it's worth, I am a die-hard ColorChecker user, I like checking the
accuracy of the colorimetric table of my Output Profiles with a custom
measured chart. It's fast, simple, not too many samples and the samples are
large enough to have a good stab at visual comparison between the original
and its reproduction. Heck! Everyone should have a ColorChecker chart on its
shelves -- thank you so much Cal McCamy! BTW, have you ever seen the
ColorChecker made up of yarn that they have at the Munsell Institute of
Color Science, in Rochester? Wow...
Best / Roger
-----Original Message-----
From: colorsync-users-bounces+graxx=email@hidden
[mailto:colorsync-users-bounces+graxx=email@hidden] On
Behalf Of LdaSignup
Sent: Thursday, May 2, 2013 8:29 PM
To: 'colorsync-users?lists.apple.com' List
Subject: HowTo? Objectively measure quality of print profiles?
Starting a new thread to avoid hijacking another, unrelated thread about
picking a camera ...
>> I'm very curious how Andrew Rooney (aka "The Digital Dog")
>> tests/evaluates the quality of profiles.
> Everything was visual. Lots and lots of real world and synthetic images.
> Some of Bill Atkinson's test images are wonderful (28 Balls).
> The 28 Balls image is a ball breaker and shows a lot about the color
> engines.
> The Roman 16's are always used.
>> Is this test/evaluation something that a well calibrated X-rite
>> i1iSis could do "by the numbers" with "apples and apples" De2k's
>> using the X-Rite MeasureTool's Compare freeware (2.0.# on a PC, 1.5.# on
a Mac)?
Hypothetically, suppose someone claimed that profiles from ArgyllCms are
typically as good or better than those from ProfileMaker-5, and almost as
good as i1Profiler 1.4.2. Is that something that could be objectively
measured by an automated spectro, rather than an informed, but still
subjective, opinion?
* Could there be some kind of weighted "figure of merit" of, for example,
De2k + Adjusted-Gamut-Volume?
* Are De2k accuracy + gamut-volume the primary considerations when
evaluating profile quality? What else? Banding? Is there a way to measure
that, and reduce to a number grade by a machine?
* How important is it to have "flexibility" so that the same profile to
handle more situations? Or would a pro printer be inclined to use multiple
separate profiles for the same printer + paper + ink for different photos
... like portraits vs landscapes? (ignoring b/w which I have no experience
with)
* Depending on photo and customer, should flesh tones be given extra weight
in a "De2k + gamut-volume based figure of merit"?
* Should "Intent" be a consideration: Perceptual vs RelativeColor vs
Saturation vs Absolute?
* What are characteristics of a series of good test prints? Lots of gamut vs
low-gamut vs reasonable gamut?
* I would think that one profile might be better for flesh tones, but not so
good with synthetics. One profile might "win" 4 out of 8 print tests, but
the other profile would "win" flesh tones, and they would "tie" on the other
three. Which profile "wins"? Depends?
...
* For those with access to an i1iSis, how useful would it be to have aligned
i1iSis patches integrated into a test print? The top part of the letter
sized test-print would be about 150 to 1000+ carefully chosen patches, and
the separately printed bottom part could be "28 balls" or "Roman 16". Or the
bottom half was a series of human "Shirleys" with carefully exposed,
non-synthetic flesh tones? Or i1iSis patches on the equivalent of the
OutbackPhoto test print?
...
* Obviously, the correct answer to many or most of the above questions would
be: "it depends".
* Here are some examples that this admittedly CM newbie has been trying out:
http://berean.zenfolio.com/isis_test_prints
http://berean.zenfolio.com/isis_test_prints/h5ddc538a#h5ddc53b6
...
* What if the evaluation broke out different "categories" of patches, like
Thomas Fors / Rags Gardner's AcrCalibrator freeware .jsx scripts do for
camera profiles:
** blacks and dark grays, mid grays, light grays and white
** dark reds, mid-reds, saturated reds
** dark blues, mid-blues, saturated blues, multiple sky blues,
** dark greens, mid-greens, saturated greens, multiple grass greens
** etc ...
** orange, cyan, pink, violet/purple, periwinkle, "sky purple", etc.
** light flesh tones, mid flesh-tones, dark flesh-tones, by continent?
** etc, etc.
_______________________________________________
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
_______________________________________________
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