CRT film-outs

[Paul Bruening]

Yes, Keith. I'd settle for simple stereo analog at this point. All I need is a stereo light gate. I could mount it against the large, continuous sprocket in the Mitchell. Actually, a drum and rollers out of a Mitchell SS would do better and be easily fit into my Mitchell. The motor is digitally controlled for perfect 24 fps. How does the light splash around from one of those gates? Could I run each roll of print stock through twice, once for picture at 3 fps then a second time at 24 fps for sound? If so, that would bypass the sound roll/contact printing operations. It can't be that hard scrounging the amps to drive the gates. The sound would come off the computer in perfect time as the source, cued from a blip of projectionist's cuing tape at the head of the roll. Actually, the software Paul Moore wrote for my scan rig counts motor rotations. Something like it could count a predetermined and standardized number of rotations and start the sound at perfect sync point as well as monitor timeline numbers to check sync continuously.

[Keith Walters]

All I need is a stereo light gate. I could mount it against the large, continuous sprocket in the Mitchell. Actually, a drum and rollers out of a Mitchell SS would do better and be easily fit into my Mitchell. The motor is digitally controlled for perfect 24 fps. How does the light splash around from one of those gates? Could I run each roll of print stock through twice, once for picture at 3 fps then a second time at 24 fps for sound? If so, that would bypass the sound roll/contact printing operations. It can't be that hard scrounging the amps to drive the gates. The sound would come off the computer in perfect time as the source, cued from a blip of projectionist's cuing tape at the head of the roll. Actually, the software Paul Moore wrote for my scan rig counts motor rotations. Something like it could count a predetermined and standardized number of rotations and start the sound at perfect sync point as well as monitor timeline numbers to check sync continuously.

" All I need is a stereo light gate."

Ye-e-ess... And where are you going to get one of those?

There's still a demand for them, since virtually all release print still have an analog sound track, and so virtually all the ones ever made are still in regular use. They're built to be reconditioned, rather like 35mm movie cameras, and just like movie cameras, there aren't all that many new ones made per year because there is little requirement to expand the existing fleets. There are probably old mono ones floating around but that would be it.

 

The laser scanner idea is a suggestion for a cheap alternative that should work well if you have someone to do the electronics for you.

 

"How does the light splash around from one of those gates? "

It doesn't really matter in a duplicating house because the release print only "sees" the sound track area, the rest is masked off. I suspect extra masking would be needed.

 

"Could I run each roll of print stock through twice, once for picture at 3 fps then a second time at 24 fps for sound?"

 

Why not slow the sound down to 1/8 speed (even Audacity will do a good job of that for you) and then do the whole thing in one go.

[Phil Rhodes]

Even if you could build a sound camera, there's still complexities - optical sound has a very strange EQ curve in it ('pre-emphasis') which you'd need to implement.

 

Not a trivial little project - I'd be more nervous about getting that to work than I would the picture side of it - but possibly more likely to work than trying to shoot the audio in an intermittent device!

 

P

[Paul Bruening]

Well, fellas,

 

You've just about got me talked out of doing the sound side of this. I'm still holding out for an eventual registration test before I totally give up on the Mitchell delivering on the DG5 track idea. But, I'm definitely on the highly skeptical side of the project instead of the optimistic-idea side of it.

 

I do, very much, appreciate ya'll helping me avoid the hazards involved in this. Thank you.

[Phil Rhodes]

Y'know how I'd do it is, I'd get an oscilloscope, y'see, and I'd build a basic little horizontal deflection circuit, to scan an amplitude-dependent horizontal line, set up an optical system to focus the resulting trace on the soundtrack area of the neg, and...

 

P

[Paul Bruening]

But... I rarely stop thinking.

 

If mono gates are findable, what about running the print stock through twice with a two position gate?

 

Or, in the spirit of broke-a**-broke, can modern projectors read a mono track?

[Paul Bruening]

If this keeps up I could end up settling for silent prints sent out with a guy and a microphone to do all the dialogue during shows.

 

Ironically, my current project, with some slight rewriting could actually be done that way.

[John Sprung]

I could mount it against the large, continuous sprocket in the Mitchell.

 

That sprocket, gear driven with film moving on it, wouldn't turn anywhere near smoothly enough. There'd be vibration. For analog sound, you needed a smooth drum with a huge flywheel, like about 10 pounds. I have a brass one somewhere.

 

 

 

 

-- J.S.

[Keith Walters]

But... I rarely stop thinking.

 

If mono gates are findable, what about running the print stock through twice with a two position gate?

 

Or, in the spirit of broke-a**-broke, can modern projectors read a mono track?

Trouble is, the mono track is twice the width of a stereo track.

Stereo optical pickups were specifically designed to be compatible with mono, because when the format was first introduced, there were still lots of mono films around.

[Keith Walters]

Even if you could build a sound camera, there's still complexities - optical sound has a very strange EQ curve in it ('pre-emphasis') which you'd need to implement.

 

P

That's mostly for the benefit of the electromagnetic drivers in the light gate. The actual slit modulation is linear (or as close to linear as can be achieved). Apart from the Dolby-A 4-band de-emphasis, no special response shaping is needed for the output of the optical pickup. I've run the "Solar Cell" outputs directly into a set of amplified computer speakers and it sounded quite OK. Like Dolby-B for audio cassettes Dolby-A was also designed to produce acceptable sound in old non-Dolby-equipped projectors.

 

A keen and capable electronics enthusiast could produce a workable Dolby A encoder clone. All you need to do is split the signal into 4 bands and apply individual AGC to each of them. Might not exactly meet the original Dolby Specs, but I doubt too many people would notice.

 

The trouble with your oscilloscope idea is that the phosphor decay is far too too long in your average ZnS phosphor tube.

Also, the thin bright line would soon get burned into the phosphor. No doubt special tubes could be made to overcome these problems, but I suspect that industrially, Light Gates were more practical.

[Keith Walters]

If this keeps up I could end up settling for silent prints sent out with a guy and a microphone to do all the dialogue during shows.

 

Ironically, my current project, with some slight rewriting could actually be done that way.

Hey! I've just had an idea.

What if you hired someone to play - I don't know - an organ or a piano?!

You could get different types of music to match the on-screen mood and have the pianist play faster or slower in time with the action.

That could really catch on...

[Paul Bruening]

The guy at Magnatech was kind enough to pass on the name of the folks who still make light gates. A new one is a mere $12,000. I guess this is going to be beyond my level of sticks and string engineering. I'll do some overlap tests trying this recommended Audacity software. Eventually, I'll do the registration tests on the Mitchell and compare the variances in the two systems. That's about all the hope I'm holding out right now.

 

Thanks again, guys.

[Phil Rhodes]

While we're pursuing crazy ideas...

 

The trouble with your oscilloscope idea is that the phosphor decay is far too too long in your average ZnS phosphor tube.

 

Well, yeah, but with all that hi-con development, who's gonna notice?

 

Or you could just modulate a laser with a galvanometer, which would have zero delay.

 

P

[Keith Walters]

Well, yeah, but with all that hi-con development, who's gonna notice?

 

P

 

Erm, that was the case in the old days when they used to add a strip of new monochrome emulsion after the colour picture development, and then run the film through a monochrome chain to process the sound track. More finesse is required when you're processing the picture at the same time.

 

Actually, the oscilloscope idea would be great for making a master sound negative but it's the "wrong way round" for print film. Plus stereo might be a bit of an ask.

 

Also modern prints use a cyan sound track which requires red light to produce, not the green from a oscilloscope tube. Cyan works better than using all three layers as slight mis-registration introduces noise. This required the retrofitting of Red LEDs in place of the old heavy filament exciter lamps in projectors.

[Keith Walters]

The guy at Magnatech was kind enough to pass on the name of the folks who still make light gates. A new one is a mere $12,000.

Thanks again, guys.

That sounds awfully cheap for such a specialized product. Are they the same ones used by major processing labs?

Actually I've often wondered whether you could make one out of a pair of ordinary loudspeakers with razor blades glued to them to form the slit. (Stereo would be difficult to engineer of course).

If you could get hold of a standard projector pickup you could monitor the slit output to produce a negative feedback signal. In fact with a DC-coupled amplifier you could even have a slit-width servo to keep the razor blade spacing constant.

Actually you could use carbon-fibre rods to couple the speaker signals to the razor blades.

 

I guess you could do stereo by simply delaying one of the channels so it could be "burned on" by a second unit a couple of inches further down the film

[Keith Walters]

Wa-a-a-aidaminnit!

How did I manage to get diverted so far off the track?

 

A dual-slit light gate won't be any good to you, because that will put two squiggly cyan lines on a colourless background, whereas you want two colourless wiggly lines on a cyan background!

 

That is the whole point of my modulated laser assembly: it will produce the aforesaid cyan stripe with clear wiggly lines on it. Light gates are designed to produce a negative on monochrome film, which is then used to lay the sound track on the print film.

 

I can see I'll have to start digging in dumpsters for some pensioned-off laser printers.

I've got a collection of precision laser modules, the electronics isn't too complicated; maybe I can demonstrate the principle.

Edited December 15, 2009 by Keith Walters

[Paul Bruening]

I haven't found anything on the internet (I heard it's a series of tubes or something) that can convert audio files into images. So far, screen grabs of maximised Goldwave or Audacity workspace is all I've come up with. I guess the resolution of my CRTs is high enough to create a sharp enough image of a single frame's worth of sound image. I could run the whole process off macros in and out of a suitable audio program and Photoshop. Seems a bit clumsy. But, doable. I don't know what kind of image resolution I'll actually have to obtain. I could make it even higher using the DG5 as the resolution determining device, I suppose. That would be the upper limit on that little trick.

[Keith Walters]

I doubt that will be very useful

Even with a 1920 pixel wide screen your entire peak-to-peak audio range is going to be "quantitized" into something less than 2,000 steps by the screen pixels, giving you the equivalent of less than 11-bit audio, and that's for mono.

 

Stereo would give something less than 10 bit audio equivalent.

 

It might be just workable, but you'd need to use pretty heavy compression.

[Phil Rhodes]

Well I didn't go down the "let's image a TFT into the soundtrack area" route for exactly that reason... it wouldn't have nearly high enough resolution. I don't think 10-bit audio would be good enough. There were tricks to obtain 10-bit audio out of old Amiga computers and it never sounded that good - the quantised "fizziness" was still there.

 

This does make me blanch slightly at how good this needs to be. I don't know what the distortion numbers are supposed to be like for optical soundtracks, but that's a precision item.

 

P

[Paul Bruening]

Grabs off the DG5 would give me something more like 3600+ steps per frame. Is that closer? Actually, if it's all going through precise, repeatable macros, I could chop out smaller sub-steps of butt-join segments of each frame, grab them, recombine them with the overlaps on each end segment for a recombine back on the DG5 for shooting. The resolution could be pushed to whatever level you think is needed. Even fixing resolution at that step, the limit is the 2400 height pixels on the DG5, shot by the Mitchell. Are 2400 steps per frame enough for passable sound?

[Paul Bruening]

I couldn't get the precise measurements of 4 perfs length of film. I took Wiki's VV frame numbers and did the math with that. So, these numbers are pretty rough. The 4-perf frame came out around 0.75" long. If I add 10% as my overlap amount and divide that by 2400 steps, I get a step thickness of .00034375 (When I say, "line or thickness" I mean a unique, detectable peak of horizontal exposure). Just how thick is the average line of analog audio? Can a light gate cut a line any thinner? Am I thinking wrong that film can't resolve that high anyway. Doesn't it already exceed the technical demands of film? 3.5/10,000" (1/2,857") per line?

[Phil Rhodes]

A 16-bit audio stream has 65,535 levels - and broadcast audio is often 24 (16.7 million).

 

It's an interesting question, though. Even the 16-bit number suggests that, with two stereo channels, the thing is capable of resolving over 1,300,000 DPI (1/10" over two 65k code-value images).

 

Which I don't think is the case.

 

P

[Paul Bruening]

A 16-bit audio stream has 65,535 levels - and broadcast audio is often 24 (16.7 million).

 

It's an interesting question, though. Even the 16-bit number suggests that, with two stereo channels, the thing is capable of resolving over 1,300,000 DPI (1/10" over two 65k code-value images).

 

Which I don't think is the case.

 

P

 

Do you think all that can manifest in an analog film track? How do you think the actual manifestation of analog audio on a conventional release print would stack up with my numbers on a 1st gen print?

[Phil Rhodes]

Not the faintest idea. You need to talk to audio guys.

 

Presumably the SMPTE has paperwork on the required spec.

 

P

[Paul Bruening]

What about the cyan parts? I figure I could arrive at a dependable color to represent the eventual accurate color on film. The concern I have is in the overlapped parts of the exposure. The overlapped section will be twice as thick with cyan (1 stop darker). Wouldn't that merely filter the red light better and not pose any variation in the sound?

 

As far as the resolution comparison of my system and a conventional print, I'll scan a few frames of a trailer and count them out in a measurable environment like Photoshop. My Epson can scan native up to 9,600 dpi which will be enough for an analysis.