Re: Media Testing for maclife.de
Re: Media Testing for maclife.de
- Subject: Re: Media Testing for maclife.de
- From: "Fleisher, Ken" <email@hidden>
- Date: Wed, 10 Sep 2008 14:42:20 -0400
- Thread-topic: Media Testing for maclife.de
Uli,
Thank you for your reply. My comments were merely out of curiosity about
your method since I have not seen your report. Regarding the flat-fielding
that I spoke of, I should have emphasized that it has the potential to be
problematic, but is not necessarily so, depending on your setup. For
example, if you used a long lens and were far from the target, then there
probably won't be a significant lens fall-off at the edges. You would see a
much stronger effect if you were close with a short focal length lens, and
therefore this setup would affect your measurements to a greater degree.
I would not assume that profiling software will automatically compensate for
uneven lighting. I know the SG chart has the white/gray/black patches around
the edges, as well as an entire neutral step wedge in the center (by the
way, it's located in the center precisely for the reasons I mentioned--this
is where you will have the least amount of light falloff due to the lens.
Since the gray wedge is responsible for the tone curve and is a critical
part of profiling, they decided to keep all gray patches in the center
instead of in a strip or along an edge of the chart). But just because the
chart has these patches doesn't mean the software uses them for this
purpose. I have always used them myself as a guide to see whether I need to
adjust my lighting to be more even.
In any case, I don't know which software, if any, use these patches for
flat-fielding. Can anyone who knows the answer for any particular package
comment on that? I am curious to find out the answer. Thanks!
> In any case, this does not seem relevant for our discussion at hand,
> because the disputed fact is that I wrote I was able to achieve better
> results with ICC profiles than Adobe etc.
Actually, I think it's completely relevant. My point is that if the lighting
is uneven, and perhaps inconsistently between the various setups, then this
could skew not only the profiles that are built, but your results as well.
You assume that all profiling software will react in a similar way to the
uneven lighting, but this is not a fair assumption. What if one package
compensates for the lighting but another does not. I think you have
introduced an extra uncontrolled variable into your experiment that has the
potential of giving you false results. For example if you did perform
flat-fielding first (or took care to make your lighting as even as
possible), can you say for certain that the same packages would give the
same relative performances? I don't think you can.
On 9/10/08 1:00 PM, "edmund ronald" <email@hidden> wrote:
> The Colorchecker SG has white squares in the center and along the
> edges, which one can use to verify the correctness of the exposure. I
> find that about 1 or 2 points point in 250 luminosity difference
> across the target are attainable in practice. Of course, as you
> indicate, no real-world lens and lighting system will allow a target
> to be *perfectly* uniformly exposed. This is one reason why
> target-based camera profiling will only get one so far. But then who
> ever said it should work ?
I'd say if you are achieving 1-2 points difference then you are doing very
well. I can't imagine anyone expecting a perfect result.
--
Ken Fleisher
Photographer
Imaging & Visual Services
National Gallery of Art
Washington, D.C.
Phone: (202) 712-7471
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