[filmscanners] RE: scanner dmax discussion

["Chuck Cole" <cncole@earthlink.net>]

> -----Original Message-----
> From: filmscanners_owner@halftone.co.uk
> [mailto:filmscanners_owner@halftone.co.uk]On Behalf Of
> austin@darkroom.com
> Sent: Thursday, July 17, 2003 5:41 AM
> To: cncole@earthlink.net
> Subject: [filmscanners] RE: scanner dmax discussion
>
>
>
> Chuck,
>
> > > There are LUTs (Look-Up Tables) in the scanner that compensate for the
> > > non-linearities that are determined during calibration.  These are
> simply
> > > offset values that are added/subtracted from the actual data values.
>
> > That is likely,
>
> Likely?  It's what scanners go through a calibration sequence for!

I presumed so (thus "likely"), but do not know the exact and complete design
data for any actual model.  The info may be in the HP lit I keep referring
to, but I haven't had time to dig for it - yet  :-)


> > but the table values would be offsets and not mathematical
> > non-linearities per se.
>
> They absolutely are to correct for non-linearities, and addition (which is
> what an offset is) IS how you correct for non-linearity in a system like
> this.

Again, my use of "non-linearity" is mathematicsally pure, while I think
yours is less formal.  The CCD signal is fully described by a y = mx + b
form of equation which is linear by the most comprehensive of formal
definitions.  The offsets we speak of are "b" values in this, so they are
not non-linearities per se.


>
> > > Basically, they simply count photons, and the output is directly
> > > proportional to the number of photons counted.
>
> > These do not count photons except in the gross aggregate where
> the actual
> > count is unknowable.
>
> They DO in fact "count" photons.  The voltage output is DIRECTLY LINEARLY
> proportional to the number of photons that the sensor detects.  Go read a
> CCD spec and you'll see this to be true.

I have about 50 pounds of detector specs and tech papers here.  These CCDs
do not "count" in any way as in quantum detection.  There are other
detectors and detection systems that actually do single photon counting,
however.  These scanner CCDs do respond to an aggregate photon flux where
the stream of photons is uncountably large (beyond "tera-") and signals vary
as huge (uncountable) number changes occur.  There is no case where these
CCDs can resolve a signal difference attributable to a count difference of
as few as a million or so photons.  The spec sheets are  referring to a
generality of photon flux and not a specied ability to do photon counting.
Again, my terminology is formal in order to separate the cases where signals
and noise permit these different detection schemes and issues.


>
> > > >  Nearly all non-linearities and most noise effects are introduced
> > > > after the detectors and are minimal effects.
> >
> > > Hum.  I disagree with that.  You aren't taking into account
> > > non-linearities across sensor elements.
>
> > That is not a mathematical non-linearity at all: it's only a linear
> offset.
>
> The statement I commented on is incorrect, for the reason I said.  The
> sensor non-linearities ARE a mathematical non-linearity, addition is
> mathemeatical, and yes it is simply an offset, which is addition, and
> addition is mathematical.  I'm not sure what the distinction/point you are
> trying to make here is.

Ditto: why isn't a formal definition that this is linear good enough to
negate "non-linear"?  True nonlinearities require some very different signal
handling and math because those signal types involve transcendantal math
functions like logarithms, sines, etc.

Regards,

Chuck



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