8mm DIY Telecine

The problem

Silent 8mm cine films come in two flavours - single or super 8, and regular 8. Regular 8 was the first format and consisted of 16mm film with double the number of perforations. Half the width was exposed at a time (to expose the other half one turned the film over 1/2 way through), then the film was split down the middle and spliced into one length by the processing company. Usually exposed at 16 fps (I think, though it isn't completely clear whether it was usually 16 or 18 fps as both were used), the frame size is approximately 4.8 x3.6mm.

Super 8 came along later and gave an increased frame of approximately 5.8 x 4.2mm, exposed at 18 fps. (Commercial variants exposed at higher speeds.) The film is usually colour reversal and Kodachrome or Ektachrome, tungsten variant, with an 85B or equivalent filter for daylight use.

The true viewable frame sizes in practice depend on the size of the gates in the camera and projector. The films came in 50 foot lengths (after processing).

Regular and super 8 (from Kodak) were actually different films. The super 8 film base seems thinner (it is certainly easier to tear the sprocket holes), and the the images are often colder (a very slight relative blueish cast).

Single 8 is a Fuji variant of the same dimensions as super 8, the one significant difference being that tape splices must be used as opposed to film cement owing to the different film base.

In all cases the aspect ratio provided by the gate is usually 1.33:1.

There are two instant problems to overcome:

Contrast

Cine film is usually positive transparency film, and as anyone who has tried to duplicate 35mm slides will know, transparency film is very contrasty and difficult to copy. It is very easy to lose either the blacks or the whites. Kodak used to make a special low contrast transparency duplication film. It is also tricky stuff to expose right in the first place, with reasonable latitude being little more than 1/2 a stop. It is obviously possible to do, and modern film scanners do a pretty good job.

No-one ever used an incident light meter to get the exposure right for any of our films, it was all through the lens average weighted metering. Some cameras appear to have been better than others: over the years we went through several, being unknown (probably Prinz), Nizo, Bolex for regular 8, then Bauer and Beaulieu for super 8. Still got the latter - no reasonable offer refused! The standard 8 Bolex K2 nearly always seemed to get it right, whereas the Beaulieu super 8 had a strong tendency to pay too much attention to the sky and underexpose by anything up to a stop - or even two. (Half our films were taken with the latter).

The result is that we have a lot of films with variable exposure, tending towards underexposure. It's a bad start - underexpose transparencies and lots of detail disappears into the blacks.

Frame rate and synchronisation

Home cine films are typically taken at a frame rate of 16 or 18 fps. The projector usually provides three light 'fields' per frame using a three bladed shutter, presumably to reduce flicker. PAL TV runs at 25 fps, 50 fields per second. It is not going to be possible to synchronise the two.

Some telecine systems appear to capture at 16 2/3 fps, capturing three PAL fields per frame. It is doubtful one can tell much difference between 16, 16 2/3 and 18 fps in reality, especially with no sound.

However, clearly one either needs to synchronise the camera to the projector, or the projector to the camera. Or does one? What a lot of people appear to do is use a free running projector, a free running camcorder, use a long shutter speed and rely on the long exposure to smooth over any flickering. This was after all effectively what I did in the 1980's with the vidicon camera. Maybe it now works well, maybe it doesn't - it didn't for me then.

The concept

Starting from scratch and making all the transport mechanics just isn't on - have you ever looked inside a cine projector? No - well it is full of all sorts of cranks and whizzits that would do Heath Robinson proud. Therefore, this needed to be based on some physical hardware to get the transport mechanism, and I have a Sankyo Dualux 1000 (switchable standard and super 8) sitting around. Therefore, for reasons of expediency, that is to be the base to build on. It may be good, bad or indifferent, but it is what I have. Having said that, internally it is beautifully made, as so many things from that era were. These days it would be all plastic (but no doubt a lot cheaper).

Rightly or wrongly, I didn't consider the free running approach, after the 1980's experience it seemed better to have a frame locked system.

The Dualux also does have variable speed, though on closer inspection it is not very variable or controllable. It is more variable torque than variable speed - remove the drive belt and fan so there is no friction and the motor speed is virtually unaffected by the speed control. It relies on the friction in the mechanism (the fan, rubber belt, cam for driving the frame pulldown mechanism etc). The effect of this is that when cold it runs down to 6-7 fps, but once warmed up, won't run much below 11-12 fps.

In addition to that, it is a mains motor so awkward to produce controlling electronics for. One option was to replace the motor with a DC servo (or stepper), but it's expensive and good accurate control over a system with variable friction isn't easy without a lot of surplus power. It is also generally easier to lock an electronic system to a mechanical system than the orher way round. For all those reasons it was decided to let the projector motor do its own thing, speed controlled as far as possible with the Sankyo speed controller, and synchronise a camera to it (though as can be seen later, some better form of control is likely to be needed at some point).

That ruled out any reasonably cheap modern camcorder, or a CCTV camera (all locked to PAL or NTSC frame rates). They are consumer products and not usually designed, unless one pays a lot of money, to be externally controlled.

Both VideoFred and Frank Vine use machine vision cameras, and that is what I decided to go for. Providing one can trigger it externally one can get the film to the right place, capture a frame, move it on, capture another and so on. Having once captured the stream (on to a PC) into some form of video file format, one can then post process it into whatever one wants. That is the theory, anyway.

So, in summary:

• a consumer 8mm projector, modified as necessary
• machine vision camera
• capture on demand to PC as video stream
• post process on PC