Jerry Oostrom wrote:
> I have a 2720 that does not function properly. The problem is probably
> background noise in the CCD. During calibration only the responsiveness to
> the white light is calibrated for each CCD pixel. I think you can compare
> it to determining the whitepoint for each pixel. I would like to have
> added the blackpoint too. (I.e. read the response from the CCD with no
> light reaching the CCD). Of course this would imply changes to the
> firmware if at all possible.
>
> I mailed Ed Hamrick on this issue and he agreed after several mails that
> the error I was seeing is indeed a CCD response issue, but he could not
> help me with added functionality to calibration (blackpoint calibration)
> as the hardware of the scanwit does not offer a means yet to do so.
I think blackpoint calibration would be useful on most scanners. The CCD in my
Minolta Elite seems pretty good but in the very depths of the shadow range there are a
few CCD elements (particularly in the green channel) which seem to return slightly too
high a base blackpoint value compared to the rest. The result is faint tracks if the
shadow detail is really pulled up. (Oh, and it's not dust in the lightpath, as the
effect is
still there if I scan a piece of opaque card in the slide holder, and is not a problem
once out of the deep shadows where the CCD response seems much more even.)
I know Ed said that (at least with the Scanwit) blackpoint calibration might not be
possible because of hardware/firmware limitations but why rely on the scanner's
capabilities to do this? Why not just do one calibration pass for the white point and
then an (optional) second pass with a piece of opaque card in the slide holder to
calibrate the black point? Low tech maybe, but it should work.
Certainly, with the Elite, Vuescan's normal white point calibration isn't foolproof.
It
does prompt you to remove the media holder before continuing but doesn't (cannot?)
check that this has been removed. Leaving it in does the calibration no good at all!
Seriously, though, if Vuescan requires manual intervention to remove the holder before
doing the whitepoint calibration, why not trust the user to be able to put an opaque
target in for blackpoint calibration? I know Ed will want to minimise potential user
error
but the blackpoint calibration could be an advanced option which has to be actively
switched on every calibration. (And could even include a "health warning" for
improper
or sloppy usage.)
Maybe it isn't as simple as that (if the CCD blackpoint/whitepoint response and
adjustments for anomalies are not linear) but I would still expect it to be better
than
having no CCD blackpoint calibration at all.
Just a thought :)
Al Bond
