[filmscanners] RE: rebuild your scanner and get better results

["Austin Franklin" <austin@darkroom.com>]

Andras,

> > That is a bad design in my book.  You can get banding by doing
> that, as well
> > as inaccurate color.  They SHOULD have calibration for all three colors,
> > whether it's the light source that changes for each color, or they use a
> > tri-linear sensor...or in the case of the 8000, they do both I believe.
>
> Calibration for all colours is not necessary, and lack of it doesn't
> cause banding.

Well, that's VERY debatable.

> The problem in the LS-8000 is that only one of three
> CCD lines is calibrated.

Correct, and that's the same thing as I said.  If you are using a tri-linear
sensor, whether it's used as three monochromatic sensors, or separate RGB
sensors, and only calibrate one line, the calibration will not be accurate.

> > I'm not a fan of LED light source, and I believe the 8000 uses
> a monochrome
> > tri-linear sensor, and changes the color of the light source
> and takes three
> > exposures per line...and as I said above, not calibrating, and providing
> > calibration correction for, each line of the tri-linear sensor
> by it self
> > can lead to banding...and apparently that's what was discovered
> as the cause
> > of banding on the 8000.
>
> True. The LS-8000 exposes all three lines with the same colour at a
> time, thus banding has nothing to do with the differences between the
> different LEDs.

I disagree that banding (in general) can't be caused by differences not
accounted for in calibration between each of the LEDs, both of the same
color, and between the different colors.

Typical scanner calibration does calibrate all three colors, and calibrates
each individual sensor for all three colors.  Every scanner I've designed
does that, and every one I've done work on does.  I KNOW the Leafscan does
it that way.  I can't vouch for any other desktop scanners though, but it
makes absolutely no sense that a competent programmer wouldn't do it that
way, because not doing it that way gives less accurate results.

> > I don't quite know what you are saying above...as LEDs have a
> rather wide
> > angle of dispersion, and that really has nothing to do with the
> > frequency/wavelength.  I'm talking about physical coverage, not
> > frequency/wavelength.  Again, I don't believe there is a
> physical light gap
> > between adjacent LEDs.  If they did, as I said, then there
> would be gaps in
> > the illumination...even with a diffuser, the middle area would
> have slightly
> > degraded illumination...and as I said, that could cause very noticeable
> > banding.
>
> Ah, OK, illumination only affects colour balance along the x direction
> of the scan, can't cause banding (in the y direction, that's where
> banding occurs).

Something and/or someone is confusing/ed here.  For example, you have an out
LED, which does leave a gap...that is in the X direction, but the banding is
in the Y direction, just like a clogged print nozzle.  Film staging problems
are what typically cause banding in the X direction on a film scanner.

> > We should possibly ask Ed Hamrick about that, but needless to say, I'm
> > skeptical as far as the 8000 goes.  Though you may have "a" scanner
>
> He doesn't know more than I do, he has the same documentation. I not
> only have documentation but have also had contact with Nikon's
> developers.
>
> My points are these, I'm absolutely sure about these and not going to
> argue any further:
>
> - banding is caused by the failure of the LS-8000 to calibrate three
>   CCD lines separately, and has nothing to do with the illumination.
>
> - the LS-8000 has fixed-intensity LEDs, and there's no reason to run
>   LED sources at anything less than full intensity anyway.
>
> You may disagree, of course.

These are really superfluous to the main issue I commented on.  My initial
comment was in response to your claim that varying the intensity of the LEDS
was "technically difficult", and it is not, as you further agreed.

The first statement above I agree with, and the second I have no first hand
knowledge of, so I really don't know, but am still very skeptical.  I find
it "odd", to say the least, that they don't have the ability to vary the
LEDs in the hardware.  There is a reason for doing it, as it can decrease
smear and bloom, which can lead to a better scan.  Now, they may have tuned
the intensity of each color set of LEDS so that it is provides an intensity
that is a compromise of scan time vs bloom/smear, but still, having control
over the LED intensity would be a very useful thing in my book.  But, as I
said, I'm not a fan of LED light sources, as they have their own inherent
problems.

Regards,

Austin

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body