Question for Ed Hamrick: Turning a film scanner into a Super8 scanner

  • Thread starter Thread starter jojax14
  • Start date Start date
J

jojax14

Hi Ed,

Since you are the foremost expert on film scanners here, I have an
idea (and a couple of questions) for you....

I am looking for a way to modify a film scanner so that it can scan a
movie on Super8 film. I am aware that this would be a very slow
process, but if it can be automated, I don't really care about that.

In order for this to work, I have a couple of requirements:
* The scanner must have a stationary CCD, and must move the film
through the scanner (for all I know, all film scanners may do this).
I realise that the stepper motors are probably not built for
'continuous' operation, but if any potential heat build-up could be
dealt with, I can't see how that would be a problem.
* The scanner must allow film to enter from the front and exit at the
rear (to be respooled onto a reel). For me, it does not matter if I
have to remove the casing or do something else that invalidates the
warranty.
* The scanner must allow the computer software to control total
movement of the stepper motor.
* I would have to work out some way of loading the Super8 film (a
'gate'), but I figure this should be the easy bit if the other
requirements could be met.

If you are aware of a scanner that meets those requirements? If so,
would it be a big stretch to add a 'continuous scan' mode to Vuescan?
The next challenge following this is splitting up the resulting scans
into separate frames. There is a piece of software called 8mmtoavi
that does this by analysing the position of the sprockets. There is
plenty of different pieces of software that will then recombine
individually scanned images into an AVI.

Ed, if any of this is at all possible I strongly suggest you consider
implementing it. There is a very large number of people who are
looking for ways to get their treasured family memories onto DVD with
a reasonable level of quality, but there is no way for the 'consumer'
to do this, short of spending a couple of hundred dollars to get a
professional to do the job (which I would much rather spend getting
the scanning hardware myself!).

I have in the past tried scanning a couple of Super8 frames on my
Epson Perfectin 1650 scanner. Despite the low resolution and the
'fuzziness' of a flatbed, after some USM (OK, a lot!) and levels
adjustment, I was able to produce images that far exceeded the quality
of the old 'video the screen' trick. I keep thinking - if that is
what I could do with a flatbed, what could I do with a 'real' scanner!

JJ
 
jojax14 said:
I am looking for a way to modify a film scanner so that it can scan a
movie on Super8 film. I am aware that this would be a very slow
process, but if it can be automated, I don't really care about that.

I have tried this with a Nikon LS-4000 and scanned rolls of Standard 8
with mixed success. ;-)
In order for this to work, I have a couple of requirements:
* The scanner must have a stationary CCD, and must move the film
through the scanner (for all I know, all film scanners may do this).

The method I used was to construct an external film transport mechanism
using an old film editor to hold the feed and take-up spools. The
internal feed of the Nikon scanners is only suitable for 35mm wide film
and requires the perforations of the 35mm film for accurate frame
positioning. So an external approach seemed the simplest at the time.

I modified a standard GePe 35mm slide mount to mask an 8mm frame and
permit the film to feed through without touching the mount. Then I
glued this to an aluminium support which sat in front of the LS-4000
with the slide mount inside the scanner in the MA-20 adapter. I
temporarily modified this by clipping it open and removing the slide
ejector mechanism - which would have obstructed the film transport and
scratched the film. It is easy to remove with a couple of screws and
just as easy to replace when you are finished.

The aluminium support also held a stepper motor and gearbox, which had
an 8mm sprocket feed fitted to the gear output. The stepper motor was
driven from the PC serial port and experiments determined the exact
number of steps necessary to move the film by exactly one frame.
I realise that the stepper motors are probably not built for
'continuous' operation, but if any potential heat build-up could be
dealt with, I can't see how that would be a problem.

Shouldn't be a problem even if you used the internal scanner motors
instead of an external one that I used.
* The scanner must allow film to enter from the front and exit at the
rear (to be respooled onto a reel). For me, it does not matter if I
have to remove the casing or do something else that invalidates the
warranty.

It doesn't actually matter, but in my case the film was fed from the
rear of the scanner to the front, with the sprocket pulling the film
through.
* The scanner must allow the computer software to control total
movement of the stepper motor.

I wrote a little windows application to produce the exact number of
steps for a single frame. This applet also manipulated the mouse cursor
to initiate a numbered frame scan in NikonScan after the required number
of steps had been cycled on the serial port.
* I would have to work out some way of loading the Super8 film (a
'gate'), but I figure this should be the easy bit if the other
requirements could be met.
Now, this is where things start to get difficult - and not for the
reasons that you might expect.

Getting the film into the Nikon and feeding through is relatively
straight forward, if fiddly.

The biggest problem you encounter though is calibration - for which the
Nikon requires a clear gate. ie. no film or slide present. That means
that you have to perform a manual calibration before feeding the film
into the scanner - otherwise the scanner objects. It also means that
you cannot recalibrate the scanner until the film spool has been
removed.

Unfortunately, the Nikon scanners seem to have an automatic
recalibration which cuts in after a fixed period of time, so after a few
feet of film have been scanned the scanner requests the slide be removed
from the scanner for recalibration - and that means rethreading entire
spool of film. :-(

The other problem I encountered was frame to frame registration.
Although the stepper moved the film on average by a frame pitch, there
was a cyclic error of around 3Hz, most likely due to a slight
eccentricity in mounting the sprocket wheel on the gearbox or the hub of
the gearbox itself. On top of this there were irregular glitches where
I suspect the torque on the feed or takeup spools reached sufficient to
overcome stiction and moved to take up or feed more film through.

The gate I prepared in the 35mm mount was deliberately larger than an
8mm frame so that I could implement software realignment of the result
without cropping, but I never really found adequate software to do this
accurately without manual intervention, which was impractical. Finally,
the gradual degradation of the calibration was actually quite visible
over a small period of time. Consequently the results were less than
satisfactory.

Nevertheless, the quality of the 8-mm movies scanned at 4000ppi with ICE
removing all of the scratches and defects that had accumulated over
decades of projection was quite impressive. With Standard-8 I was
getting a frame of about 768x576 pixels, which matched PAL video
standard exactly. Super-8 should be around 900 pixels wide if scanned
at 4000ppi. Image quality was much better than anything I have seen
from available cine-2-video transfers. If I could get the problems of
calibration, stability and frame registration solved then I would
transfer all of my old cine stock to DVD.

It is something I intend to go back to try and improve once I get the
time.
 
I wrote a little windows application to produce the exact number of
steps for a single frame.

Does that mean you have the developer kit?
Before I get it and risk wasting more time, two quick questions:

1. Is it possible to modify exposure and focus at each scan line?
2. Is there any documentation on the diagnostic mode?
The biggest problem you encounter though is calibration - for which the
Nikon requires a clear gate. ie. no film or slide present. That means
that you have to perform a manual calibration before feeding the film
into the scanner - otherwise the scanner objects. It also means that
you cannot recalibrate the scanner until the film spool has been
removed.

Unfortunately, the Nikon scanners seem to have an automatic
recalibration which cuts in after a fixed period of time, so after a few
feet of film have been scanned the scanner requests the slide be removed
from the scanner for recalibration - and that means rethreading entire
spool of film. :-(

I'm not quite sure I understand this. I have had a slide in for a very
long time and there was no explicit request by the scanner for
recalibration. I could just go on scanning.

Now, when I eventually remove the slide then - indeed - if the
internal recalibration counter has counted down, the scanner will
recalibrate.

However, if you have a roll of 8mm film you will only be advancing the
film and never really completely remove it. So I don't understand when
and how does the recalibration kick in.

Do you actually get a modal dialog on the 4000 preventing you from
further scanning until you recalibrate?

I find that after the scanner has been on for a while and the
temperature has stabilized subsequent calibration seems to make little
difference. There may still be slight deviation but nothing major. On
the other hand, for a contiguous roll of 8mm film that may be
noticeable...

Don.
 
Hi,

Thanks for your very informative response.

The issue with calibration does sound like a problem. Maybe Ed
Hamrick can advise whether calibration can be bypassed somehow in
Vuescan. I suppose it all depends on how 'dumb' the internals of the
Nikon are.

Is there any chance that you could make some examples of the video (or
even just still frames) available for viewing?

JJ
 
Back
Top