On Sun, Jun 10, 2007 at 06:34:07PM -0700, Moose wrote:
> Nils Frohberg wrote:
> > On Sat, Jun 09, 2007 at 08:42:23PM -0700, Jay Drew wrote:
> >
> >> I read Moose's post to mean that you take a shot of an IT8 target with the
> >> negative film, not that there is an IT8 target source as a negative film.
> >>
> >
> > Missing out on the rest of the thread, I just wanted to note that this
> > method is a bit flawed IMO.
>
> Do you literally mean in your opinion, i.e. as a matter of personal
> logic/speculation? Or as a result of personal experimentation? Or from
> some other source to which you can refer?
I literally mean in my opinion, mainly based on personal logic. No
real-life experimentation done.
We both agree that we can achieve color accuracy _inside_ the target's
range. What's going on outside of it is where opinions differ.
Please also note that all I'm saying is that _color_accuracy_ is lost
outside of the range; (normally) you still will get good colors, but you
simply can't be sure that they're accurate. (Color accuracy meaning that
the color doesn't differ by more than a certain set threshold from the
original color, since we need interpolation even inside the range.)
If, e.g., the target's range falls within your film's linear
(approximation for simplicity) portion of its response curve, how do
you extract the precise shoulder/toe values? With CN film you suddenly
have more than one curve you have to approximate, and this is where your
colors can get skewed.
> This does leave open the treatment of values on scanned film that are
> above or below the limits of the profile. I suspect that the
> international standards creators have a standard for this for a couple
> of reasons. First, they are smart enough to see the problem should have
> a resolution. Second, they are smart enough to know that the more nooks
> and crannies they can find that need definition and standards, the
> longer they remain on the payroll and staying in the sort of nice
> locations chosen by international standards organizations (ISOs, sound
> familiar?) for their work.
>
> I can easily imagine a few ways to do this, but have no idea what they
> may have done. Nor do I know whether any or all software apps that use
> icc profiles use the same method.
Maybe I'm missing something completely obvious here, how would you
suggest extrapolating the response curves?
> > Or you take various exposures of the target, with different
> > shutter/aperture settings, resulting in several shots ranging over the
> > entire dynamic range of your film. You can now calculate a color profile
> > for the full dynamic range.
> >
> Something like this would indeed be necessary if your above assumption
> is correct. For it to be correct does posit a level of technical
> incompetence and general lack of intelligence that I find unlikely even
> given the politics and general foolishness that are a part of the
> establishment and staffing of any ISO.
>
> As a practical matter as one who has regularly used icc film profiles
> created in this way for some time, I haven't seen a problem. Using
> profiles has made such a HUGE improvement in scanning results that I've
> been more inclined to just get on with the work than pick theoretical
> nits. And it just doesn't seem like I am losing any highlight or shadow
> detail compared to before the use of profiles. Hmmm, I wonder if I saved
> the RAW scan files from the roll that I scanned with and without
> profiles....
I agree with you that you'll gain a lot using your method, and that the
results will improve drastically. Therefore I also suggest that one
should do at least that step. But one has to know that it still is an
approximation. (OK, so are the values within the target's range, but you
know what I'm saying...)
Nils
--
The Moon is Waning Crescent (19% of Full)
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|