Re: emulsion side down and histograms

In article <d9pc06$nkj$1$8302bc10@xxxxxxxxxxxxxxxx>, John <warthog@xxxxxxxxxxxxxxxxxx> writes 
"Kennedy McEwen" <rkm@xxxxxxxxxxxxxxxxxx> wrote in message
KtzcCpEsFgvCFwo$@kennedym.demon.co.uk">news:KtzcCpEsFgvCFwo$@kennedym.demon.co.uk...

In article <42BD3785.31E59460@xxxxxxx>, JJackson382@xxxxxxx writes
>
>
>Kennedy McEwen wrote:
>>
>> In article <230620051942375501%lrcullen@xxxxxxxxxxxxxxxxxxxx>, Linda
>> <lrcullen@xxxxxxxxxxxxxxxxxxxx> writes
>

In *both* cases the adjustment is made by changing the exposure time.
There is *NO* capability in any of the Nikon scanner series to change
the LED intensity, only the exposure time.

Are you sure?

Yes, absolutely sure. Linearly changing the intensity of an LED is not trivial, requiring electronic control of a precision far beyond what is capable of building into a consumer scanner.

Try the following test: In NikonScan, set analogue gain as
follows:
Red  =  +2; Green = Blue = -2.

Load a slide with a white mount, so that you can see the reflected light
from the scanner LEDs when the scanner head approaches the front of the
scanner. Now do a scan (NOT a preview - the preview scan doesn't seem to be
affected) and watch the colour of the reflected light from the slide mount.
If your scanner is like mine (Coolscan 4000) then you will see red light,
proving surely that the analogue gain controls at least alter the LED
intensities.

No, it doesn't prove that at all. This same "proof" that the LED intensity changes has been raised in this group several times so I assume it must either be coming from a common origin or be a rather trivial misconception that many people make without thinking through the detail of what they are looking at.

So, for the umpteenth time on this forum, let me explain what you are seeing with your own eyes to you!

The scanner scans a typical 35nn frame in around 40seconds with about 6000 scan lines. During each scan line, the scanner cycles the LEDs through red, green and blue (as well as infrared, if ICE is enabled). That means that the scanner conducts about 18,000 typical exposures in 40 seconds so, neglecting the readout time from the CCD and the time to mechanically step from one row to the next, that makes the exposure to each colour around 2mS long. Even if the exposure is increased by +2 under analogue gain control, it is, at most 8mS long. If the exposure is reduced by -2 in the analogue gain then the LED pulse is only 0.5mS long.

Now, the persistence of vision in your eye means that the minimum that you perceive any pulse of light to last is around 50mS-200mS depending on the ambient illumination level. Darker ambients produce longer persistence. This is the entire principle that movie films and television rely on - that you cannot distinguish short bright pulses of light from long dark pulses once the pulse time is short - they both look the same. A short pulse lasting twice as long as normal with the same intensity will appear to your eye as a single pulse which is twice as intense as normal.

So, look at the figures and *think* about what you are seeing. You cannot discern the intensity or the width of the pulses which are shorter than typically 100mS long, yet the scanner is pulsing its LEDs at 0.5-8mS according to the analogue gain control. So the longer pulses simply *appear* to your eye to be brighter - but they are simply the same LED intensity for a longer time.

Hence, when you increase the analogue gain of the scanner in one colour, the apparent colour of the light leaking from the front aperture at the end of the scan appears to change colour - just as the colour of a pixel on a plasma TV changes colour depending on the amount of time each cell is illuminated: there is no amplitude control on the illumination of the scanner LED or the plasma pixel.

Now, there is *one* way to view the illumination time of the LEDs in the scanner and confirm that time control, and not intensity control, is used. View the scanner aperture (and thus the light from the LEDs) through a spinning mirror or polygon. This will cause the LED pulses to spread out over the angle that the polygon turns during the exposure, the size of that angle is proportional to the pulse time and the image intensity directly viewable. You will see red, greed and blue images of the scan aperture at different positions and smeared over different angular subtenses. Of course, the polygon needs to be synchronised with the LED cycling rate, but with a variable speed control it is possible to do that over a typical scan period.

A simpler test is just to measure the scan time with different analogue gain settings and confirm that it takes longer to make a scan at +2 than at -2. The readout time of the CCD and the stepper motor time, both ignored above, means that this is not a simple proportional relationship with analogue gain, but it is possible to determine the "dead" time between exposures and the constant of proportionality by a simple linear curve fit.

I have been confused about this for some time, 

Clearly.  I hope the explanation has cleared it up for you.

in fact since I read the
Vuescan Help file - Ed Hamrick makes a similar statement to you along the
lines of "Analogue Gain is Nikon's term for CCD exposure time". It is a
little odd, therefore, that Vuescan's analogue gain control has exactly the
same effect!

Why would you think it should be different? The software you scan with has not affected the way your eyes have evolved!
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
.

Relevant Pages

  • Re: Long range IR signal
    ... Assuming I wanted a 1A pulse ... likely get any benefit from shrinking the carrier frequency duty cycle. ... If you continuously run the LEDs at 38kHz the receiver IC ... > capacitor across the supply close to the circuit. ...
    (sci.electronics.design)
  • Re: Cant get enough current for 1W Luexon LED even without resistor
    ... Then I tried without the resistor but ... the LED I have to raise the voltage to about 11V and of course I ... LEDs for a long time, and never suffered a problem with a switch-on ... max forward current quoted in specs, is for a short duration pulse. ...
    (sci.electronics.repair)
  • Re: Coolscan 8000 v. 9000?
    ... 5400II has much more in common with a Nikon scanner than with a 5400. ... both use 4 colored LEDs and no filters on the CCD. ... >step the scan head motion had not settled when the samples were taken. ... >is and if that accounts for the amount of misalignment that you see. ...
    (comp.periphs.scanners)
  • Re: Cant get enough current for 1W Luexon LED even without resistor
    ... resistor but still only about 100ma current. ... want to be thinking in terms of voltage. ... life from any high power LED, is to pulse drive it. ... The level of these that some high power LEDs can ...
    (sci.electronics.repair)
  • Re: Flashing LEDs proportion to Servo position
    ... At 0 degrees full brakes are applied, ... > transformer, some resistors, and a pot, to blink a LED at different ... The LEDs are ... Convert the pulse width to a voltage using an RC filter. ...
    (sci.electronics.basics)