Re: Media Testing for maclife.de
Re: Media Testing for maclife.de
- Subject: Re: Media Testing for maclife.de
- From: Uli Zappe <email@hidden>
- Date: Tue, 16 Sep 2008 07:27:49 +0200
[ooops, should have scrolled up to not overlook yet another thread.
These threads have become sooooo long that it's easy to miss something
- sorry.]
Am 15.09.2008 um 22:53 schrieb Eric Chan:
I'm not denying that it is important how close the profiling
process can approach the "raw device", but that's not different
from profiling a printer, a scanner or a display.
It's different with raw converters, compared to printers & displays,
because different raw converters treat color differently.
The same is true for different printer drivers (e.g. you cannot use a
profile created for an EPSON driver with the Gutenprint driver for the
very same printer), but to be close, let's stick to scanners as a
comparison. Scanner drivers certainly differ as much as RAW converters.
For example, Camera Raw's color profiles are all based on scene-
referred colorimetry and deliberately ignore tone curves (i.e.,
since everything is linear in the raw stage, any required exposure
adjustment is just a simple scaling of the raw data by a constant).
The idea being that the profile sets up the initial scene-referred
color and it's then up to the user to guide the rendering process,
mostly through the use of a non-linear tone mapping curve (looks
like a big gamma adjustment with an S-curve, by default). So if
you're evaluating the final output, which it seems like you're doing,
Yes.
it's not at all surprising that you're seeing large mismatches
For me, it is. For all this talk about scene-referred vs. output-
referred, as long as I strictly adhere to colorimetric renderings, I
fail to see the difference to at least a scanner. And nobody seems to
take issue with evaluating a scanner the way I evaluated cameras (at
least I'm not aware of it).
, esp. if you're including L* differences in your error metric.
Of course I adjusted for luminosity.
In your terminology, is this still scene-referred (still original
camera color space) or already output-referred (white balance and
tone curve set)?
In my opinion if the data has been tone mapped it's no longer scene-
referred.
Which within the "scene vs. output" framework means it's already
output-referred which I'd consider a complete misnomer at this state.
Ok, but the reference XYZ/Lab values provided in the PM text files
are all relative to D50. As I recall you are testing under multiple
different lighting conditions.
Yep, that was a very important part of my testing goals.
So, I assume you are starting from the reference __spectral__ values
of the chart and then, for each lighting condition that you're
shooting under, you derive the final XYZ/Lab values based on the
actual spectrum of your light (measured with your Eye-One).
True?
Yep; X-Rite's reference data are spectral; I exported the Lab values
for the different lighting conditions with MeasureTool.
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