----- Original Message -----
From: "Tony Sleep" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Tuesday, September 19, 2000 11:45 PM
Subject: Re: Group Scan site has posted SS4000 scans
> (a) The gamma and black point of the Polaroid (Halloween pic) are set
> lower than with the others (and the highlights clipped) by overexposure,
> and the Elite is set higher than the Canon and Acer. As with *any* scanner
> this exposes CCD noise in the Polaroid which would not normally be seen,
> and conceals noise in the Elite which would be. I suggest you try
> normalising them by adjusting Levels in PS. Doing so loses the noise
> without losing shadow detail. The noise ranking then changes to Pol4000
> (least noise), Elite, Canon, Acer (most).
Finally, some thoughtful comments rather than knee jerk reaction...
I agree with your main points. In fact, I described all this in my reponse to
Jim Levitt post on the subject some days ago. I reached slightly differenct
conclusions but don't have a strong view at this point given there is so much
going on in that picture.
My initial impressions remain that Polaroid probably has too much noise for my
taste. But then so does the Canon and every other "home" scanner I've used.
So I'll probably wait for the next generation and hope they do a little better.
A nice upgrade to the Polaroid would be multiscanning to further reduce noise.
> (b) The Polaroid Halloween pic changes appearance - for the better -
> dramatically when viewed in PS5 on a calibrated screen, rather than in a
> non-ICM aware browser on an uncalibrated screen. The other scans of the
> same slide change very little. There is probably some arcane ICM tagging
> issue here involving the monitor profile of the system which created the
> Polaroid scan. I don't know what.
>
Now that is an interesting comment. Are we talking about the Halloween slide
or negative? I only had time to check the slide and didn't see anything
unusual (small difference).
Ian understand this color profile stuff better than either of us...perhaps he
can tell us if something weird is happening here (and if so what).
I note Photoshop opens it without any profile mismatch complaints so it appears
to be saved in the proper color space. And of course, we all know Vuescan
doesn't tag the data so it isn't obvious where other weird tagging would come
from. Anyway, I don't see anything unexpected so I guess it is over to the
incomparable Mr Lyons.
Perhaps you are just seeing the difference between calibration and
non-calibration on your own system? That would have this effect I would think.
Anyway, all my comments are based on viewing in Photoshop.
> (c)The USAF test slide is designed for microdensitometer evaluation and
> not easily visually comparable on screen due to the varying pixel
> densities of the different scanners. Variation in contrast and gamma
> exists between the samples and affect apparent sharpness.
The easiest is to resample the lower resolution image up to the 4000 dpi size.
This actually penalizes the lower resolution scanner since bicubic resampling
is not perfect and has the side-effect of reducing resolution. Still the
effect isn't large and for our purposes within the margins of error.
Contrast and gamma have an impact, but the target has a farily low sensitivity
to those errors so as long as it is similar the "score" is not affected in any
significant way. It is easy to check this simply by changing those parameters
in the downloaded image and see how much it takes to move the "score" up or
down by one block. And of course it is easy to adjust these as close as one
would like. This is where experience comes in, it helps put the various
theoretical issues into perspective.
I agree the target works best visually for the upper reaches of the MTF. But
this is a significant parameter so worthwile. Lower parts of the MTF curve
are also important and the target allows for that too but it is harder to test
and thus not addressed on the group scan site. Best to get a full and proper
MTF curve defined. Perhaps that is something you can do on your new site that
will help distingquish it from what the average person can do.
Anyway, good stuff although I didn't have time to read the rest of your post so
caveat emptor... :))
Cheers,
Byron
====================================================================
The filmscanners mailing list is hosted by http://www.halftone.co.uk
To resign, <mailto:[EMAIL PROTECTED]> with UNSUBSCRIBE FILMSCANNERS in the
title, or UNSUBSCRIBE FILMSCANNERS_DIGEST if you are reading the Digest.
