Sprocketless Film Camera

[Nicholas Kovats]

A more mechanical approach relative to a film scanner, i.e.
https://www.google.com/patents/US576155 ... CDQQ6AEwAA

[carllooper]

I am very excited to report that I found a very interesting practical link for your project. A working prototype from a very prolific US inventor, Stephen Hines. He appears to have actually developed a working capstan S8 prototype for Kodak in 1977 with a fiber optic viewfinder. Very thin design, i.e. http://www.hineslab.com/opto-mechanical ... ie-camera/

That's brilliant. A 70s Super8 spy camera.

Design wise I'm interested in something that could be made cheaply on a 3D printer, with some help from some cheap stepper motors and cheap postage stamp computer such as Arduino, or Rasberry Pi. It's not so much that it needs to be cheap but it's a nice design constraint to work within. Allows more experimentation and different directions one can go without having to be too anxious about the cost. But at the some time one doesn't compromise quality of the resulting image.

The stepper motor makes possible the elimination of almost all of the complicated mechanics associated with analog (continuous) motors. When you think about it, film is basically analog in the spatial domain but digital in the time domain. All the complicated mechanics in a traditional camera are about moving the film in discrete steps. A stepper motor basically provides for that. The registration diode takes care of any residual vagaries in image positioning. Of course this all presupposes that the film will have a digital only release (or will have a digital intermediate).

However it's also conceivable one could bypass a digital intermediate (or digital release) and do either an analog print (eg. to 16mm), or indeed projection on a custom projector.

How might a custom projector work? This could be an interesting project in itself and one I've got in the back of my head somewhere. The positional accuracy of a stepper motor isn't bad, especially if you are using a small radius capstan to drive it. But in any case, how might the film be mechanically re-registered during projection? The projector can use a small radius capstan and high accuracy stepper motor in conjunction with a sensor to monitor the registration marks as they pass by. An appropriate feedback mechanism, conceptually similar to a phase locked loop, can ensure the film is positioned where it needs to be when the shutter opens. The registration marks become it's sprocket holes or "clock signal" in computer terms. Can be, mechanically, extremely simple, yet deadly accurate. Error cancellation. Instead of trying to elliminate mechanical errors through costly precision engineering, one allows them, but in such a way that they can cancel each other out.

[Nicholas Kovats]

Carl,

If your interested I have started a thread on motion picture capstan technology on a Turkish/English cinema technologist Facebook page called "Sinema Teknolojisi|Motion-Picture Technology|Film-Technik|Кинотехника "

Cheers!

NK

[carllooper]

Carl,

If your interested I have started a thread on motion picture capstan technology on a Turkish/English cinema technologist Facebook page called "Sinema Teknolojisi|Motion-Picture Technology|Film-Technik|Кинотехника "

Cheers!

NK

thanks Nicholas. Will check it out.

[carllooper]

For a custom projector at nominal frame rate of 24, the film needs to advance in 1/6th of 1/24th a second, ie. 1/144th sec

Assuming a capstan with a nominal 64mm circumference (roughly 10mm radius) and a stepper motor with a step angle of 0.9 degrees, that means it would have a positional resolution of 0.9/360 x 64mm = 0.16 mm, or about 1/47th the screen height of 16mm. That's not a very good resolution for re-positioning of the film. Indeed nowhere near it. However it can get the film into the ballpark at which point one needs something else to fine tune it.

But about the speed of the ball park positioning? The 0.9 degree stepper motor I'm looking at has an RPM of 2344, or 14064 degrees/sec. Given the step size is 0.9 degrees that works out to be 15,626 steps a sec, or 108 steps within the time it would otherwise need to position the film (ie. within the 1/144th second that the shutter is closed). So it can move the film 108 times, in 0.16mm increments, within that time constraint,

Now for fine tuning I figure a mirror on the lens can do that (but see below for a better idea). The film is positioned to within 1/47th of the frame height, of where it needs to be, and a mirror, also on a stepper motor, does the precise fine tuning. For 1/3000th frame height definition the mirror only needs to accommodate a minimum of 65 steps, within the time constraint, where each step moves the projection by 1/3000th of the frame height. But if the stepper can do 108 steps (like the film drive) then the system's total positional resolution (for standard 16mm) becomes 1/(108 x 108) = 1 / 11,664th of the frame height. By decreasing the capstan radius by half one can double this to 1/23,328th positional resolution, for standard 16mm film.

Indeed if one allows a few seconds at projection startup, to obtain rough phase lock on frame pitch, the required adjustment per frame becomes much much smaller, and the number of steps the stepper needs, to correct the positioning, is correspondingly a lot smaller - so the overall positioning resolution can be made orders of magnitude higher.

In fact, thinking about this more, the mirror idea can be thrown out. Instead, one can use two or three capstans in a row (each on solenoids and differently geared), where each acts to fine tune the previous. The first gets the film into the ballpark while the other two are disengaged. The second then engages (by solenoid) to fine tune the positioning, and then the third locks in, to do the insanely nano sized positioning. All within 1/144th sec. Indeed while the third is holding the film in place the first can go about lining up the next frame into rough position. And the second it's contribution. As soon as the third releases it's hold, the film snaps forward by the positioning preset on the first two. A small wave propagates between the capstans. Three fingers in a digital dance ...

The entire thing is regulated by the registration mark. Move it this way a bit. Move it that way a bit ... a bit like the Dambusters inching their aircraft to the correct altitude, by watching lights reflected off the water, or Tom Hanks at the window of the Apollo 13, using the Earth as a reference, to control retro rocket firings.

The Dambusters has to be one of my all time favourite movies. This scene sums up it's central charm:

https://www.youtube.com/watch?v=qr48QxzyG98

[MovieStuff]

The projector is the easiest part. Presuming the side to side position of the film is mechanically constant, just use the clear registration marks next to each frame to trigger a high powered strobe as the film passes by continuously. Not likely you'll be trying to fill a large theater screen so a strobe should be plenty bright to see in a living room.

Roger

[carllooper]

The projector is the easiest part. Presuming the side to side position of the film is mechanically constant, just use the clear registration marks next to each frame to trigger a high powered strobe as the film passes by continuously. Not likely you'll be trying to fill a large theater screen so a strobe should be plenty bright to see in a living room.

Roger

Like a film to digital transfer system. One 'problem', though, is that it would flicker - the image would appear for a fraction of a second and then nothing for the remainder of the 1/24th sec.

BUT - it's definitely worth doing, because (apart from just the fun of it) it would work really really well in a live performance context. A bit of a disco vibe. And, as you say, it's the easiest to put together.

cheers
Carl

[MovieStuff]

The projector is the easiest part. Presuming the side to side position of the film is mechanically constant, just use the clear registration marks next to each frame to trigger a high powered strobe as the film passes by continuously. Not likely you'll be trying to fill a large theater screen so a strobe should be plenty bright to see in a living room.

Roger

Like a film to digital transfer system. One 'problem', though, is that it would flicker - the image would appear for a fraction of a second and then nothing for the remainder of the 1/24th sec.

BUT - it's definitely worth doing, because (apart from just the fun of it) it would work really really well in a live performance context. A bit of a disco vibe. And, as you say, it's the easiest to put together.

cheers
Carl

Yeah, I realized it would flicker but, not to put too fine a point on it, I never thought one would be cutting together anything for formal projection. I just figured it would be a way to preview the footage before telecine.

Roger

[carllooper]

The projector is the easiest part. Presuming the side to side position of the film is mechanically constant, just use the clear registration marks next to each frame to trigger a high powered strobe as the film passes by continuously. Not likely you'll be trying to fill a large theater screen so a strobe should be plenty bright to see in a living room.

Roger

Like a film to digital transfer system. One 'problem', though, is that it would flicker - the image would appear for a fraction of a second and then nothing for the remainder of the 1/24th sec.

BUT - it's definitely worth doing, because (apart from just the fun of it) it would work really really well in a live performance context. A bit of a disco vibe. And, as you say, it's the easiest to put together.

cheers
Carl

Yeah, I realized it would flicker but, not to put too fine a point on it, I never thought one would be cutting together anything for formal projection. I just figured it would be a way to preview the footage before telecine.

Roger

oh no - the idea of a projector is for live performance work. Art installation environment. That sort of thing. The idea will work perfectly fine in that context. But would also like a more traditional projection effect of the relax in your seat and drift away type ...

But the idea also provides another way of looking at the problem which I hadn't quite worked through. I was imagining I'd need some kind of multi-sensor array (digital camera) to pick up where the registration mark was, but that's not necessarily required at all. A single sensor could do the job. As the film is being transported the registration mark will sooner of later cross the sensor and it's at that precise moment (a timestamp) that the system then knows the film's actual position relative to the current state of the stepper's advancement of the film. Although would need a fast acting sensor to read the mark, since a single step operates very fast ... hmmm ... one approach would be to encode a continuously varying signal as the registration "mark". That way, as the signal crosses the sensor it can deliver an amplitude proportional to the distance between that point and an 'origin' for the frame ... a triangle shape ... something like that ... something that works for both live projection and digital transfer. Encoded at an odd rate relative to the sprocket pitch (and roughly twice frequency of the sprocket pitch) ... because those sprocket holes are going to get in the way. Have two sensors - the idea being that when one is out of action due to sprocket hole, the other won't be.

Indeed I've just thought of another problem that is going to throw a spanner in the works: any edge markings already on the film ... will need to ensure encoding is away from such.

Can always encode on the opposite side to the sprockets and it's edge markings. But just looking at some film now - some Tri-X, and looks like there's enough room, on the sprocket side, between the edge markings and the frame edge, to print a decent signal.

[S8 Booster]

Just gimme some Fuji F500T and i would not give a fu*k about the capturing system.

.

Shoot..whatever film

[carllooper]

Just gimme some Fuji F500T and i would not give a fu*k about the capturing system.

Yes sir, who needs a fo*king camera ... Henry Fux Talbot did without ...



ha ha ha



Well I did.

[carllooper]

Sprocketless projector to match the sprocketless camera.

The projector needs to have a capstan with the same diameter as the camera. Prior to projection a hi-rez stepper would move the film into correct launch position, ie. into correct phase with the projector aperture. It doesn't need to do this at high speed. It would occur prior to projection. One then dials in the same pitch that was used to shoot the film (at whatever aspect/pitch was used: Ultrapan8, 16mm, sideways 16:9, 'vistavison', etc). Indeed the projector could conceivably read the registration signal on the film to determine in what mode it was shot and automatically initialise itself. Anyway, once the film is started it will be in phase, but any small differences between camera capstan and projector capstan will mean the image can start to drift out of phase: ie. the image can start to slowly migrate across the projector frame. It is this migration, as much as any frame to frame jitter, that needs to be mitigated (or rather: that I want to mitigate)

Frame to frame jitter will be caused by very small inaccuracies in the stepper motor, however these should be statistically zero over one revolution of the stepper, so shouldn't, on their own, cause any global drift. It will be micro-differences between the capstans (and any micro-vagaries in film traction) that can emerge as a global drift.

The simplest solution I have at the moment is to have a fast low res stepper (the same used in the camera) mounted on the shaft of a slower hi res stepper. The fast stepper handles advancing the film at the same pitch as the film was originally shot, and with identical diameter capstans the image drift should be quite slow (close to zero). In any case the purpose of the slower hi-res stepper is to cancel out any residual image drift and frame to frame jitter that otherwise occurs, regulated by the frame registration signal. When any image drift manages to reach the distance traveled by one step of the faster stepper (but cancelled out by the slower stepper), it can then skip a step, allowing the slower stepper to skip back by the same angle in the opposite direction, the purpose of which is to avoid cables eventually starting to wrap around the assembly. The image on the screen remains unchanged. These skips will probably be few and far between but in any case handled without interrupting one's viewing pleasure. The main purpose of the hi-res stepper is to handle the frame to frame jitter, which I calculate is capable of keeping jitter down to about 1/5076 of a 16mm frame height. In digital terms this would be equivalent to bumping a 6768 x 5076 scan of 16mm film, by only one pixel.

In other words I reckon it should provide close to a rock solid image on the screen.

Now the projector could also be made to handle traditional sprocket registered film. A sensor just reads the sprockets to regulate transport. No point ruling out the projection of traditionally shot film.

That all said, it can't be as good as a properly implemented pin registration system (across camera and projector). Pin registration across camera and projector would keep jitter near to absolute zero (limited only by any wear and tear on sprockets). Nevertheless a cheap stepper based solution looks like it should do at least as well as a standard 16mm projector and probably a lot better.

-

[carllooper]

Calling a camera a "sprocketless camera" is like calling a car a "horseless carriage" - anyway ...

Here's a drawing of what I'm thinking for the projector.



What I need to resolve is the mounting between the bottom stepper and the top stepper. From the drawing I can see the main problem is going to be wobble. Both steppers have to be axially aligned and there will be some sort of wobble introduced here. The wobble doesn't need to be elliminated - just minimised. The film gate can constrain the effect of wobble to the direction in which the film is travelling. This wobble can then be corrected by the lower stepper motor which does the fine tuning on transport.

Here is where I'm depicting the actual size of the stepper motors. They are quite small. The film depicted here is 16mm film. I've put a face plate in there which can act as a stabiliser. The optical encoder I was imagining is redundant as the stepper angle is defined by the motor controller. And the film provides it's own registration signal (be it optically written at the time of exposure or otherwise given by the sprocket holes). The capstan can be defined in relation to the fast stepper to give a default pitch equal to 16mm pitch.



The fast stepper (and capstan) are the same as used in the camera, so will deliver the same transport rate (mm/sec) as was used to expose the film, be it at UP8 rate, 16mm rate or "vistavision" rate (or anything in between). The slow stepper (the bigger one behind the fast stepper) is only on the projector, and is used to do the frame to frame registration (minimising jitter) and in conjunction with the fast stepper initialises the starting phase offset.

The projector would need to be built in such a way that facilitates both normal or sideways projection, ie. having two floors.

C

[carllooper]

Here's an animation showing the fast stepper advancing the film at 1 frame per quarter turn.

Behind that is the slow stepper, fine tuning or correcting the fast stepper, by smaller offsets. I've exaggerated the correction the slow stepper would normally be doing. In reality the corrections would be very small, ie. in the order of 1/5000th of a frame height.

http://members.iinet.net.au/~carllooper ... erAnim.mov

[carllooper]

Here's how I figure the motors could be mounted in such a way that ensures any tendency to wobble can be constrained.