Filmscanners mailing list archive (email@example.com)
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: filmscanners: Infrared scan
>> I just tried scanning a slide and outputting a colour TIFF and an
>> IR one. It was very educational. Any sort of mark, scratch or
>> dust spot is utterly black in the IR scan.
> If you look closer, you'll see that the dust spots aren't pure black,
> but instead are just quite dark.
Um, OK, I didn't bother checking out the actual RGB values. In
the Fujichrome slide, the lowest value in the infra-red channel was
only 33. After conversion to 8 bits per channel of greyscale,
anyhow. While my comment above was not scientifically correct,
I was simply pointing out that the effect of the IR channel with
normal ektachrome or C41 film was to make the image almost
invisible and the dust *very* visible. This makes it easy to use
as a mask.
> > Some of the image is also visible as is some of the
> > grain, which probably explains why the image is softened by ICE.
> No, this has nothing to do with the softening in the ICE
> algorithm. This is just a flaw in the algorithm (it does
> unnecessary smoothing).
Which leads me to wonder how ICE determines the difference between a
scratch and a darker pattern in the latent image or grain. But I don't need
an explanation. :) I'm not planning to reverse engineer it! What this
exercise *does* make me wonder is whether it would be worthwhile
separating the dust filter from the scratch filter (not to mention the grain
filter, but we've been there before :). Judging by the opacity of objects
in the IR channel it looks to me like dust (or completely removed
emulsion) would be the easiest thing to remove without affecting the rest
of the image.
Thanks for the feedback Ed! It is as I mentioned earlier, educational. :)
I hope I answered the question which was asked on the list recently about
ICE and B&W film - that it's unlikely to *ever* work unless the film is
chromogenic. Maybe someone will come up with another method of
detecting surface defects like bouncing light (or another EMR) at an
angle off the surface of the emulsion. Who knows? Then again, by the
time such technology was developed, digital cameras may have reached
a stage of development where we'll give up on film. :)
(who still doesn't have a digital camera)