Re: filmscanners: Nikon Scan DOES Vary LED Brightness

[Julian Robinson <jrobinso@pcug.org.au>]

I have watched this thread with my own slight sense of disbelief - ISTM 
that it makes no difference at all whether the LED brightness or 
integration time is changed.  There is no difference in changing 
integration time or brightness to change what Nikon call "analog gain" 
except for second order effects on noise performance.

If I was designing the thing, I would use a combination of both.  Normal 
gain would use the LED's at their maximum long-term brightness level - to 
give shortest possible scan time.

- Any *reduction* in "analog gain" would be by decreasing LED brightness 
from that point or decreasing amplifier gain, so the scan time would 
not  change.
- Any *increase* in "analog gain" would be by increasing integration time, 
thus increasing scan time but not increasing LED brightness or analog gain 
beyond the normal maximum.  This might fit with Jawed's observations 
although I take Ed's point that I don't know how you can tell whether an 
apparent brightness change is due to actual brightness change or change in 
the duty cycle of the LED (to accommodate changed integration times).

But really I don't see why we are even thinking about this  - the fact is 
that the gain should be defined from optical input (i.e. film density) to 
volts out, and will be changed the same, irrespective of whether light 
brightness or integration time (or for that matter amplifier gain as Austin 
was wondering, but this would make a mess noise-wise with higher gain) are 
changed.

As an aside, if I were trying to write a manual to be used by 
photographically knowledgeable but not electronic knowledgeable people I 
would say that "analog gain varies the LED brightness" because this EXACTLY 
describes the effective situation and is all you need to know in order to 
use the control properly.  It seems that I agree with Nikon on this.

Julian

At 09:50 27/11/01, you wrote:
>Well, in disbelief, I have just done a test.  With Nikon Scan 3.1.1 running
>stand-alone.
>
>Nikon Scan has four sliders for Analog Gain: Master, Red, Green and Blue.
>They work additively, i.e. Red brightness equals Master + Red.
>
>In my test I used a slide that I have lying around (i.e. I didn't use the
>motorised film strip adaptor that I normally use, as I normally scan only
>negs).  No autofocus.  No ICE, no GEM, no ROC.  No tonal curves.  No USM.
>No crop.
>
>12-bit scan depth.
>
>CMS On.
>
>Performed on an Athlon 1.2GHz, 512MB, Win98 PC.  Timings are from the time
>the scan button was pressed until I could hear the scan mechanism start
>retracting (which coincides with the "Performing Scan..." progress bar
>coming to a halt, the message being blanked and the heading "No scanning
>items" appearing).
>
>Test 1: Analog gain set to 0.  Time 77s.
>Test 2: Analog gain set to -4 (-2 on Master, plus -2 on each of the colour
>channels).  Time 77s
>Test 3: Analog gain set to +4.  Time 87s.
>
>Looking into the scanner ***I could clearly see that the brightness of the
>alternating Red, Green, Blue light differed with each scan***.  The light is
>seen reflecting off the slide mount.
>
>In test 3 the sound the mechanism makes is different from the other two
>tests.  It has a more "distinctly-stepped" quality to it.
>
>The sounds are like a continuous burr from a fine stepper motor and a
>less-frequent thrum from a coarse stepper motor.  Each thrum sound is
>accompanied by the light going off.  The light is off for longer in test 3
>than in tests 1 and 2.  I estimate around 300 seperate thrum sounds are made
>per scan (I counted!).
>
>Maybe Nikon Scan 3.0 varied integration time.  Its scans of slides have been
>documented as clearly inferior to 3.1 (Mike Duncan published some nice
>graphs) in terms of OD and the shape of the curve produced.  3.1.1 certainly
>isn't solely varying integration time to vary exposure.
>
>Also NS, version 3.1.1, seems to write the file to disk much more rapidly
>now than it did in 3.1.
>
>Those legs were tasty - char-grilled with a mustard sauce.
>
>Jawed
>
>
> > -----Original Message-----
> > From: owner-filmscanners@halftone.co.uk
> > [mailto:owner-filmscanners@halftone.co.uk]On Behalf Of Arthur Entlich
> > Sent: 26 November 2001 13:13
> > To: filmscanners@halftone.co.uk
> > Subject: Re: filmscanners: RE: filmscanners: VueScan clipping & flat
> > images
>
> >
> > Boy, this thing has grown legs... I was the person who indicated that
> > Nikon's manual states that the brightness of the LEDs is altered.  Ed
> > came back with information that the info is not correct, and that the
> > CCD exposure time is altered instead, in spite of what Nikon's manual
> > states and he was able to prove this via his monitoring of the command
> > flow between the scanner and software adjustments.
> >
> > I've quoted myself and Ed below.
> >
> > Art
> >
> >
> > > In a message dated 11/21/2001 8:07:15 AM EST, artistic@ampsc.com writes:
> > >
> > >
> > >> Nikon might be approaching the limits of linearity in the LEDs.  They
> > >> also need to be able to have a range of brightness available
> > to them for
> > >> the "analog exposure" they offer.
> > >
> > >
> > > No, Nikon scanners don't vary the brightness of the LEDs.  The
> > > "analog gain" option in NikonScan only changes the CCD exposure
> > > time.  I've traced the commands that NikonScan sends to the
> > > scanner, and the field it changes is definitely the CCD exposure
> > > time field.
> > >
> > > In addition, the scan speed is proportional to the "analog gain"
> > > setting, which it wouldn't be if NikonScan were only changing
> > > the brightness of the LEDs.
> > >
> > > Regards,
> > > Ed Hamrick
> > >
> >
> >
> >
> >
> >