[filmscanners] Re: scanning at less than optical res

["Arthur Entlich" <artistic@telus.net>]

Laurie Solomon wrote:

> Art,
>
> Bob clarified what he was referring to in a later post, which you may have
> seen.  The gist of it was that the post on this list was a repost of a
> response he made on another list where the original poster said that they
> were essentially cropping a smaller portion of an image from a larger one
> during the scan.  His reference to the large format scan was to a scan of
> the total image versus reducing the scan to the cropped portion only.
>

OK, I must have skimmed over that particular reference.  Thanks for the
heads up.

>
>>My UMAX Astra 1200S (Flatbed) definitely, at least in the direction of
>>the scan "head", doesn't scan full optical if you choose a lower than
>>optic resolution.  I can both see and hear the stepper motor change
>>number or size of the discreet steps it uses.
>
>
> My understanding is that the horizontal line of sensors define the optical
> resolution and this definition is uneffected by the step motor operation
> which controls the verticle movement of the horizintal line of sensors,
> which are fixed.  Thus the size of the steps taken may change and probably
> do, but the effective optical resolution remains the same and the reduction
> in the size of the sample is a software instituted operation performed on
> the full optical scan.
>

I am not sure if this is accurate or not.  Yes, the number of scan
elements cannot change physically on the scan bar/ CCD chip, but one has
to know where and how the data is reduced.  Does the software simply
record every other, or third or fourth data sampling point from the line
of CCD sensors in each color, ignoring those it skips?  Or does it
capture all of them and do some of the many possible algorithmic
interpolations?  And if so, which does it do?  Does it work only within
a horizontal line, or does it use nearest neighbor, or bilinear, etc, or
some other concept?  In Photoshop and some scanner software, I can
choose, but many do it without user input.

I also believe optical resolution is only half determined by the maximum
   CCD chip element count.  The other vector also has an optical
resolution, regardless if it is expressed in some weird unbalanced
manner like "600 x 1200 dpi".  Each CCD element has a defined dimension
- width (x) and length (y).  If the sample is taken every "y" (or y -2%,
to prevent banding, for example) that is an optical resolution in that
direction. In the case of the above scanner it could conceivably be
scanning at 600 x 1200 (dpi) if the element were twice as wide as long,
but more likely the elements are close to square, so the scanner's
"optical" resolution would be 600 x 600 dpi in that case.   That is not
to say that 1200 x 1200 dpi with that same scanner won't supply more
accuracy (it does (although not as much as a true 1200 x 1200 dpi
scanner would, but the prior does so with the benefit of no additional
expense other than a slightly higher resolution stepper motor and a bit
of extra software coding).  However, that does introduce some need to
interpolate due to the need to double the data points the CCD can offer
as the true maximum optical resolution of the CCD chip in that direction.

If one were to not consider the stepper motor placement resolution as
significant in the discussion of optical resolution, the scanner could
scan at, say, 600 x 1 dpi.  That would make for a pretty awful looking
scan, interpolated to 600 x 600 dpi, or not.

Art

> filmscanners_owner@halftone.co.uk wrote:
>
>>Maybe it should have been called "larger file scanning" or Larger
>>  format output scanning?
>>
>>It would be nice to know which scanners did what when lower resolution
>>modes are selected.
>>
>>My UMAX Astra 1200S (Flatbed) definitely, at least in the direction of
>>the scan "head", doesn't scan full optical if you choose a lower than
>>optic resolution.  I can both see and hear the stepper motor change
>>number or size of the discreet steps it uses.  I don't know what it
>>does
>>in the other direction, but based upon the speed difference, I'm
>>guessing it is not interpolation from a full optical scan, but instead
>>just ignoring the intermediate ones.
>>
>>Art
>>
>>Laurie Solomon wrote:
>>
>>
>>>Yes, I did understand that; but I did not think that too many
>>>scanners out there used this approach except to produce low
>>>resolution preview scans. But I could be wrong.  What I did not get
>>>and may have been a terminological confusion was his reference to
>>>"large format scanning" with respect to the discussion.  I suppose
>>>he really meant to say - now that I re-examine it - "high resolution
>>>scanning" as contrasted to "large format scanning."
>>>
>>>
>>>----- Original Message -----
>>>From: "Arthur Entlich" <artistic@telus.net>
>>>To: <laurie@advancenet.net>
>>>Sent: Friday, December 10, 2004 5:58 AM
>>>Subject: [filmscanners] Re: scanning at less than optical res
>>>
>>>
>>>I believe what Bob is stating is that some scanners literally skip
>>>lines
>>>or sensors and just record the spaced information, rather than taking
>>>the full resolution and then averaging the pixels out via a series of
>>>algorithms.  This, of course, would introduce a great many sampling
>>>errors, since it is a much coarser sample of the actual data.
>>>
>>>Some scanners do seem to do just this, as the resulting scan takes as
>>>little as one third the time to be scanned and produced as a raster
>>>image when the resolution is quartered.
>>>
>>>Art
>>>
>>>LAURIE SOLOMON wrote:
>>>
>>>
>>>
>>>>>If you scan at 1200dpi, the scanner usually either samples all the
>>>>>4800 possible data points per inch and throws three out of every
>>>>>four away, or only samples every fourth possible point. So you are
>>>>>only getting one quarter of the possible data from the film. So
>>>>>why scan at large format if you are throwing three quarters of the
>>>>>film data away?
>>>>
>>>>
>>>>Bob, I beliee you are correct; but I do not understand your
>>>>question. What do you mean by "scan at large format" in this case?
>>>>I must have missed something in the discussion.  The first
>>>>method,which you note, involves the actual sampling of original
>>>>data using sampling algorithms and does result in a loss of
>>>>ortiginal data; but the second method, which you speak of with
>>>>respect to Vuescan and is available in almost all other scnning
>>>>applications, involves resampling of the original sample data using
>>>>formulas for combining and recombining data on the basis of all
>>>>existing data and the formulas.  Both methods, however, would
>>>>involve the scanner reading during the scan all 4800 points; so
>>>>both would involve a "scan at large format" - using your terms - or
>>>>whatever optical format is used by the scanner. After the scan,
>>>>everything else by way of sampling or resampling is either digital
>>>>conversion via hardware or software generated.
>>>>
>>>>-----Original Message-----
>>>>From: filmscanners_owner@halftone.co.uk
>>>>[mailto:filmscanners_owner@halftone.co.uk]On Behalf Of Bob Frost
>>>>Sent: Thursday, December 09, 2004 4:10 AM
>>>>To: laurie@advancenet.net
>>>>Subject: [filmscanners] scanning at less than optical res
>>>>
>>>>
>>>>Is this correct?
>>>>
>>>>
>>>>"As I understand things, a scanner with an
>>>>optical resolution of 4800dpi can take a sample reading every
>>>>1/4800 of an inch. If you scan at the optical resolution, that is
>>>>what is does and you get 4800 readings per inch along that axis
>>>>(usually a different resolution on the other axis).
>>>>
>>>>If you scan at 1200dpi, the scanner usually either samples all the
>>>>4800 possible data points per inch and throws three out of every
>>>>four away, or only samples every fourth possible point. So you are
>>>>only getting one quarter of the possible data from the film. So why
>>>>scan at large format if you are throwing three quarters of the film
>>>>data away?
>>>>
>>>>With Vuescan software, you can set it to scan all 4800 data points
>>>>per inch, but then to take the average of every four data points
>>>>and reduce them to one, so that the file you get out is the
>>>>equivalent of a 1200dpi scan, but all the data points have
>>>>contributed to the final result."
>>>>
>>>>Bob Frost.
>>>>
>>>>------------------------------------------------------------------------
>
> ----
>
>>>>------------
>>>>Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
>>>>filmscanners' or 'unsubscribe filmscanners_digest' (as appropriate)
>>>>in the message title or body
>>>>
>>>>
>>>>--
>>>>No virus found in this incoming message.
>>>>Checked by AVG Anti-Virus.
>>>>Version: 7.0.289 / Virus Database: 265.4.8 - Release Date: 12/8/2004
>>>>
>>>>--
>>>>No virus found in this outgoing message.
>>>>Checked by AVG Anti-Virus.
>>>>Version: 7.0.289 / Virus Database: 265.4.8 - Release Date: 12/8/2004
>>>>
>>>>
>>>
>>>
>>>-------------------------------------------------------------------------
>
> ---------------
>
>>>Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
>>>filmscanners' or 'unsubscribe filmscanners_digest' (as appropriate)
>>>in the message title or body
>>>
>>>
>>>
>>>
>>>--
>>>No virus found in this incoming message.
>>>Checked by AVG Anti-Virus.
>>>Version: 7.0.289 / Virus Database: 265.5.0 - Release Date: 12/9/2004
>>>
>>>
>>>
>>>
>>>
>>>--
>>>No virus found in this outgoing message.
>>>Checked by AVG Anti-Virus.
>>>Version: 7.0.289 / Virus Database: 265.5.0 - Release Date: 12/9/2004
>>>
>>>
>>
>>--------------------------------------------------------------------------
>
> --------------
>
>>Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
>>filmscanners' or 'unsubscribe filmscanners_digest' (as appropriate)
>>in the message title or body
>
> --
> No virus found in this outgoing message.
> Checked by AVG Anti-Virus.
> Version: 7.0.289 / Virus Database: 265.4.8 - Release Date: 12/8/2004
>
>

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