Re: maclife.de
Re: maclife.de
- Subject: Re: maclife.de
- From: Uli Zappe <email@hidden>
- Date: Mon, 1 Sep 2008 22:37:38 +0200
Part 5, which deals with camera profiling, has now been published
(again, German only):
http://www.maclife.de/index.php?module=pdfarchive&func=download&pid=2078
This part of the review differs from the rest; since it's so
controversial whether camera profiling is feasible at all, much of the
test tried to answer this question more than evaluate specific
profiling software.
I used a Canon EOS 40D as my test camera (as it's situated right
between "amateur" and "pro" models, so I hoped I'd get results
representative for both). I built 125 camera profiles out of the
various combinations of profiling software (i1 Match, ProfileMaker,
MonacoProfiler, basICColor input, SilverFast IT8 option), camera
targets (Digital Color Checker SG, Mini ColorChecker, Wolf Faust IT8
camera target and basICColor dcam target), target lighting conditions
(2 studio flashs, 1 studio flash, camera flash) and image formats (RAW
and JPEG). I then applied these profiles to 8 test images including a
Digital ColorChecker SG target and shot in different lighting
conditions (daylight (shadow), daylight (sunlight), daylight
(interior), dusk, halogen, neon, camera flash and a Granger Rainbow
screenshot) so I got 1000 test images that my statistical evaluations
were based upon.
Using the Digital ColorChecker SG for validation is not ideal since
it's also one of the targets used for building the profiles, but since
color deviations are much bigger for camera profiles, anyway, and the
question is more if camera profiling is possible at all, I figured
that would be OK.
My focus was on building a "generic" camera profile that would work
for all lighting conditions since I feel that the applicability of
lighting specific profiles is very limited. I'm quite aware of the
arguments that this cannot possibly work, but I wanted to see for
myself.
Here are some interesting results:
- Depending on the selected "picture style" of the Canon EOS 40D, mean
delta E varied between roughly 6 and 16 for the default results of the
camera (there are detailed measurement diagrams in the review)
- Almost all RAW converters I tried (Aperture, Capture One Pro,
iPhoto, Preview (Mac OS X generic), Lightroom and SilverFast DCPro)
produced results between the best and the worst Canon default results,
but closer to the best. Only SilverFast DCPro was worse than the worst
Canon default result. Aperture was clearly the best RAW converter (but
still not better than the best Canon default)
- Of all tested profiling software, *only* i1 Match and ProfileMaker
(which use the same Logo code for their engines) were capable of
producing profiles that could be used in all lighting conditions
without producing artifacts or color distortions (like "solarisation"
etc.)
- The profiles produced with i1Match/ProfileMaker produced better
(smaller) delta E's than the best Canon default in *every single* test
lighting condition (roughly between 3 and 7). This means that with
these two applications, it is indeed possible to build universally
applicable profiles that improve the color reproduction of a camera -
contrary to what some theoretical arguments say. But note that 2
different profiles for e.g. daylight and halogen may produce even
better results - I didn't test that
- There is some dispute as to what is the best lighting of the camera
profiling targets. The manuals usually say you should use two flashs
right and left of the target in a 45 degree angle, but a minority
position argues one flash (again in a 45 degree angle) is better
because no two flashs have exactly the same color temperature, while
uneven lighting can be compensated for by the profiling software by
analyzing the white color patches all around a target like Digital
ColorChecker SG. I found the minority position to be correct, though
not by much (roughly 10% better delta E values). The only exception is
the Wolf Faust IT8 target which produces slightly better results with
two flashes (since it has no white color patches all around it)
- Very surprisingly to me, I found only *one* RAW converter software
capable of a correct color management workflow, and that is Capture
One Pro. It's the only one that uses ICC (as opposed to proprietary)
profiles by default for the RAW color conversion and thus allows you
to drop in your self-built profile instead of the default at exactly
the right place in the workflow. Neither Lightroom nor Aperture allow
the usage of individual camera profiles at all (if you use them to
import your images). :-((
- One common argument against the usage of ICC camera profiles is that
proprietary formats such as that of Adobe implement a "dual matrix"
approach by actually incorporating two profiles for different color
temperatures and interpolating between them for actual images. This is
again something that sounds good in theory but that my tests revealed
to be wrong at least for my test camera. Not only are the Adobe
profiles not among the best default profiles (clearly worse than the
Aperture default profiles), I also used Adobe's DNG Profile Editor to
build individual profiles in the Adobe format. These were a tiny bit
better than Adobe's defaults, but still worse than e.g. Aperture's
defaults and much worse than ICC profiles built by i1Match/
ProfileMaker. But even more interestingly, a single matrix profile
built with Adobe's DNG Profile Editor produced slightly better results
than a dual matrix profile ...
So this was the last part of the Mac Life review, and I wonder if
there's a life after color management ... ;-)
Bye
Uli
________________________________________________________
Uli Zappe, Solmsstraße 5, D-65189 Wiesbaden, Germany
http://www.ritual.org
Fon: +49-700-ULIZAPPE
Fax: +49-700-ZAPPEFAX
________________________________________________________
_______________________________________________
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