WinAit/Wolverine/Reflecta/etc. 8mm digitisers - getting 1:1 pixel ratio?

[J. P. Gilliver]

I've finally acquired one of these machines, to convert my old standard
and super 8 films.

They produce 1080p (1440×1080 20fps) files. (The very early models
produced 720p files.)

The sensor is stated to be 3.53 Megapixels (2304 × 1536) 1/3" CMOS. (I
believe the Kodak Reelz machine has a bigger sensor, but still produces
1080p files, so one wonders why they bothered with a bigger sensor. The
other machines - Wolverine, WinAit, Reflecta, Digisomething, and own
brands - are all the same machine under assorted badges. I think WinAit
is the actual manufacturer - Chinese, of course.)

They provide X, Y, and Z controls - X and Y for framing, and Z (may be
called something else) for zooming in and out, so you can eliminate or
include the frame border, sprocket holes, etcetera. (Some cameras used
to actually shoot into the space between the holes.)

It seems pretty certain that these adjustments are in software only -
I'm pretty sure there is no physical zoom movement, let alone very fine
stepper motor control for framing: in other words, they just do the X
and Y by selecting different parts of the image, and Z by "digital
zoom".

Since framing/cropping/zoom can be done afterwards in post, some have
suggested one should capture fully zoomed out, on the basis of not
throwing away any information at the scanning stage. (There's also some
implication that by including a fair amount of the sprocket hole [which
appears as a white patch], one can to some extent override certain
elements of the automation - exposure and colour control - which these
machines' firmware insists on inserting. (Such corrections being better
implemented in post.)

It occurs to me, however, that the least resampling distortion would be
a "zoom" that causes the machine to use exactly 1440×1080 of the
sensor's pixels - neither interpolating ("zoomed" in) nor combining
("zoomed" out) - and that, indeed, using a "zoom" that was just one side
or the other of these would in fact be the worst.

Anyone know how to determine what "zoom" level hits this? (Maybe the
"default setting"?)

I presume, when broadcasters use home movie footage (e. g. in
"documentaries" about some celebrity [such as Julie Andrews]), they use
a conventional continuous-motion telecine machine - right? Presumably
with _optical_ enlargement for the smaller format, so that the
line-of-cells sensor is still used for at least a fair amount of the
image.

[John Williamson]

Post by J. P. Gilliver
The sensor is stated to be 3.53 Megapixels (2304 × 1536) 1/3" CMOS. (I
believe the Kodak Reelz machine has a bigger sensor, but still produces
1080p files, so one wonders why they bothered with a bigger sensor. The
other machines - Wolverine, WinAit, Reflecta, Digisomething, and own
brands - are all the same machine under assorted badges. I think WinAit
is the actual manufacturer - Chinese, of course.)

All else being equal, bigger pixels produce less noise (More photons hit
the sensor per exposure, smoothing out the random errors.), and a bigger
sensor allows more latitude for errors in the optical side of things.

[J. P. Gilliver]

Post by John Williamson

Post by J. P. Gilliver
The sensor is stated to be 3.53 Megapixels (2304 × 1536) 1/3" CMOS. (I
believe the Kodak Reelz machine has a bigger sensor, but still produces
1080p files, so one wonders why they bothered with a bigger sensor. The
other machines - Wolverine, WinAit, Reflecta, Digisomething, and own
brands - are all the same machine under assorted badges. I think WinAit
is the actual manufacturer - Chinese, of course.)

All else being equal, bigger pixels produce less noise (More photons
hit the sensor per exposure, smoothing out the random errors.), and a
bigger sensor allows more latitude for errors in the optical side of
things.

Sorry, I was imprecise: I meant the Kodak machine claims a sensor with
more pixels - I don't know if it's physically bigger. (If the same size,
then more pixels will of course mean they are actually smaller.)

[Graham.]

Two questions if I may.
What's the maximum reel size, and how long does it take to scan
each frame?

When I looked into these scanners some time ago, I decided they
were too slow plus I would need to re-spool some 400 foot (7 inch
dia) reels onto smaller ones.

Instead I just used a projector and my camcorder with all the
automatics disabled, pointing at a small screen.


I'm quite pleased with the results.

[J. P. Gilliver]

Post by Graham.
Two questions if I may.
What's the maximum reel size, and how long does it take to scan
each frame?

The original and still commonest model - 5 inches. (Which is the biggest
I have, so fine by me.) Basically because the spindles are at the corner
of the device. Many people in the early days used various means - the
most common being, I think, being a hand-winding machine - of allowing
bigger reels. (With either ingenious ways of driving the takeup spool,
or just feeding into a clean box.) Some models now come with a swing-out
arm that allows a bigger feed reel.

All these machines - both the WinAit variations (Wolverine, Reflecta,
etc.), and the Kodak Reelz - scan at 2 f. p. s., i. e. one eighth (for
standard 8) or one ninth (for super 8 silent) nominal speed. (One
twelfth for super 8 sound, but I don't know if anyone's doing that on
these, as they don't do the sound anyway.

Post by Graham.
When I looked into these scanners some time ago, I decided they
were too slow plus I would need to re-spool some 400 foot (7 inch
dia) reels onto smaller ones.

They're certainly slow, see above. Also - allegedly, I haven't used mine
yet - tend to jam on splices or any similar irregularities, so I doubt
you'd get a 400 through in one go.

Post by Graham.
Instead I just used a projector and my camcorder with all the
automatics disabled, pointing at a small screen.

That (or use of a mirror device) was the only way available before these
machines came along.

Post by Graham.
I'm quite pleased with the results.

The main concern has been the inevitable loss of sync., resulting in a
beat in brightness (and perhaps jerkiness/blur in frames at the minima
of sync). Not really having tried it, I can't say. (Plus I don't have
any HD camcorder [and my projectors are on their last legs], and I feel
the format is quite capable, at its best, of far better than SD; I'm
sure most of my films are far from perfect, but I think in resolution
terms they're quite good. (Mostly from the 70s, with mostly 10 ASA film
for the Std8 ["25" - really 40 with a filter - for the Sup8].))

[Graham.]

Post by J. P. Gilliver
The main concern has been the inevitable loss of sync., resulting in a
beat in brightness (and perhaps jerkiness/blur in frames at the minima
of sync). Not really having tried it, I can't say. (Plus I don't have
any HD camcorder [and my projectors are on their last legs], and I feel
the format is quite capable, at its best, of far better than SD; I'm
sure most of my films are far from perfect, but I think in resolution
terms they're quite good. (Mostly from the 70s, with mostly 10 ASA film
for the Std8 ["25" - really 40 with a filter - for the Sup8].))

Thanks for that.
I bought a Dixons Prinz Magnon Lv cheaply off Ebay and used my Sony
Handycam, seting the up close together to reduce trapizoidal and
keystone effects and of course with manual focus and exposure.

I used my osciloscope with a photocell (actually a PV solar cell from
a garden light) to pick op stray light, and set the X timbase to
"Line" (ie 50Hz mains) and adjusted the motor speed to get a roughly
stationary Lissajous figure on the scope
The projector shutter has 3 blades
16.6666x3=50
it meant the film was running slightly slow, 16.666 rather than 18 FPS
but strobing was at an acceptale level

In post I applied a small ammount of cropping and colour correction