Re: filmscanners: Bad CCD elements - was Scan Dual II Bad Elements

["Al Bond" <al@greenspace.freeserve.co.uk>]

> > Norman Unsworth wrote:
> > 
> > How do the bad elements in the CCD evidence themselves?
> > 

Art replied: 
 
<snip>
> However, individual pixels or CCD elements can also be defective or
> miscalibrated.  The best way I have found to check for these is to use a
> slide with areas of darker colors, perhaps even a near black slide will
> work.  You want something that doesn't have a lot of lines or detail in
> it.

This is all very familiar.  When I got my Scan Elite 18 months ago these sort 
of CCD 
defects were very obvious in the green channel with only very little gamma and 
white 
point adjustments.  I got it repaired under warranty and it seemed much 
improved 
(although not perfect) and generally usable.

However, several months later, I started to do some scans of night scenes on 
Kodachrome 64 which needed the shadow detail boosted and multiscanning to 
reduce 
noise.  That highlighted some dodgy green channel CCD elements but, more 
worryingly, a shift in the whole CCD response in longer duration scans.  
Basically, if 
the black edge of the frame was returning, say, an average value of 25 in the 
green 
channel at the start of the scan, by the end of the scan it might be well over 
60!  The 
greater the level of multiscanning, the worse it got so x8 and x16 
multiscanning 
introduced a haze over all the deep shadows across all but the first few pixels 
across 
whole frame.  Certainly, there was no hint of noise but no detail either......

After many mails to Minolta UK (and the inevitable "it's been referred to Japan 
but they 
haven't replied" black hole), they recently replaced the unit.  It, too shows a 
couple of 
"lazy" green channel CCD elements but nothing too bad.  It does seem to have 
slightly 
more noise generally than my old scanner but, as it was defective, it's rather 
hard to 
compare.

Anyway, if there aren't too many lazy CCD elements, they can be fixed at the 
post 
scan stage relatively easily.  (The ones on my Elite seem to be due to poor 
calibration, 
rather than being broken, with the black end response in the green channel 
starting too 
high hence the green tracks in the deep shadows.)  After finishing adjusting 
levels, 
curves etc in 16 bit, convert the image to 8 bit.  In Photoshop, use the single 
row or 
column marquee tool, depending on orientation, to select the offending CCD 
element.  
Use Select and Color Range to select just the shadows.  Then, in the green 
channel 
(or whatever the affceted channel is) adjust the black point in levels to so it 
matches 
the neighbouring elements but keep the mid-point slider in the same position.  
If done 
well, there is *no* evidence of the lazy CCD.

If there aren't too many lazy elements and the scanning exposure is relatively 
constant, it should be possible to record a PS action to do this 
automaitically.  (Not 
that I've done this yet!)  Of course, assuming that there is a true black 
reference point 
in the scan like the frame edge, what would really be good would be a bit of 
software 
which took each row (in a 16 bit raw gamma 1 file) and set black point.  I know 
the 
scanner calibration process should do this but, from the problems some Scanwit 
and 
Minolta owners have had, the scanner calibration and/or software just ain't 
cutting the 
mustard!



Al Bond