SMPTE Timecode

[carllooper]

So the challenge I've set myself is to design a setup where we can shoot film (16mm) while recording sound separately (on digital) and then subsequently project the film on a film projector (not a video projector) with sync sound playing from a computer.

A number of options have been thought through.

The first was a setup where some sort of sensor would be mounted to the projector which would send a frame count signal to the computer. The setup would be such that the film could be threaded and then inched to some sort of start position from which the frame count would be zeroed - and then the projection started. I'd write some custom software to read the projector signal and regulate the sound playback accordingly, ie. to keep the sound in sync with the projector. However the downside of this is that if the projector had to be stopped mid screening, for any reason, we'd lose the count and have no way of re-establishing sync without rewinding the film back to the beginning.

After considering an number of things (such as a "graycode" sensor on the projector) I'm now thinking what might be a more robust idea (but more involved) is to print an optical soundtrack on the film - not for carrying the audio as such, but carrying a timecode signal. This signal would just go out through the speaker lead to the computer, where the software would use that signal regulate the soundtrack. The beauty of this approach is that if the film is ever stopped or skips a beat the timecode would immediately re-establish sync regardless of where the film was.

Which brings to me my question.

Is it possible to buy 16mm film anywhere with a generic timecode already printed on the film (ie. SMPTE timecode). The idea being that we would do a double pass contact print where we would print the camera original in one pass and then just print the timecode soundtrack in a second pass.

The alternative I'm looking at, at the moment, is getting hold of a Super16 camera where I'd generate a timecode pattern on the computer (frame by frame) and print that to the soundtrack area. However I don't have a Super16 camera. There's some on ebay at the moment I'm considering but I figure if one can just buy an already printed timecode film that would be a cheaper solution.

C

[S8 Booster]

from the soft disk - long time ago but i think this may meet your requirement in some way or another:

from a member/former member of this board:
http://www.super8sync.com/super8sync/un ... 008-M.html



shoot......

[carllooper]

from a member/former member of this board:
http://www.super8sync.com/super8sync/un ... 008-M.html

Thanks S8 Booster. That's the sort of thing I'm working on for sure.

I've got the digital side worked out (since I'm writing the software) - just need to work out some way for the projector to produce the syncronisation signal - but I need a sync signal that is recoverable from any mechanical error during projection - eg. were the film to stutter, skip a beat or otherwise (god fordbid) jam. In other words the best sync signal would be a time code signal, basically where each frame would have it's own unique number, and I'm steering towards burning that into the optical sound track.

Found this site on SMPTE timecode: http://www.philrees.co.uk/articles/timecode.htm

I don't need to use this standard. But the encoding idea is a good one. For a binary digit 0 you would have one pulse, and for a binary digit 1 you would have two pulses (but at twice the frequency).

So, for example, to encode the number 001101 ( = 13) the optical sound track across the frame would look something like this, if you know what I mean:

1 ======
0 --------
1 ======
1 ======
0 --------
0 --------

I think I might just find a cheap Super16 camera and generate the required patterns on a computer screen, exposing the sound track area to the patterns on the screen. Can then use the result of that for printing the time code onto the release film.

The other solution is to open up a projector and mod it with some sensors but it wouldn't be as portable. Using the optical sound track as the sync signal generator means I can then use any 16mm projector - with the only thing needing to ship from one place to another being the film, software and computer interface (ie. for the projector's speaker cable).

C

[carllooper]

I've gone ahead and acquired a Super16 modded K3 off ebay. Hopefully it's working okay.

So the plan is to design some time code format (similar in concept to the SMPTE timecode format) and then generate this as a pattern on a computer screen, frame by frame, and use the camera (frame by frame) to 'print' this pattern into the sound track area of the film. The rest of the frame can be just left blank or perhaps generate/print some big numbers in the image area, as an eyeball guide to what is otherwise encoded in the soundtrack area: basically a computer generated optical sound track. Not sure how the K3 might be modded for computer control of frame advance but will work that out when the camera arrives - doing it by hand (with a cable release) would drive me completely and utterly insane ... But anyway, can then use this film as a reusable printing master for printing time code onto any release prints (ie. in a second pass during contact printing).

So the workflow would be

1. Shoot image and sound double system (ie. with clapperboard)
2. Do a frame accurate transfer of film to video as a visual reference (doesn't need to be high quality transfer - just frame accurate)
3. Sync up sound to image in a video edit program and otherwise rearrange shots (if required). Image will look crud but it's just as a reference for composing the sound (and as an EDL for the next step).
4. Go back to the original film and edit the original to match any edit decisions made in the video edit (or not as the case may be). Add some leader at the head.
5. Print the edited (or unedited) original, burning the time code into the soundtrack area of the print (in a second pass).
6. Project the print in a film projector, with sound track (from Step 3) playing from the computer, in sync with the film.

In Step 5, the timecode pass doesn't need to be exactly aligned to the image pass so long as the timecode starts talking at some point during the leader interval (and the start point is visually identifiable in the resulting print) because one can always dial that start point into the software that handles the sound playback.

I reckon this setup should provide the ground for doing something that looks and sounds pretty awesome: analog image and digital sound. Could even do some sort of surround sound thing with enough speakers.

C

[Will2]

Why not make it simple and do everything you say in the first few steps, but then do a film out with audio track as SMPTE code whereby you can then sync the film to a computer or audio tape machine with SMPTE sync ability? (you could also just do a standard optical soundtrack although they sound pretty bad).

Guessing the obvious answer would be cost, but film outs are getting less expensive all the time. You could also frame your film as Cinemascope and do a film out in Cinemascope where you'd need a fairly common scope lens to show widescreen.

[carllooper]

Why not make it simple and do everything you say in the first few steps, but then do a film out with audio track as SMPTE code whereby you can then sync the film to a computer or audio tape machine with SMPTE sync ability? (you could also just do a standard optical soundtrack although they sound pretty bad).

Guessing the obvious answer would be cost, but film outs are getting less expensive all the time. You could also frame your film as Cinemascope and do a film out in Cinemascope where you'd need a fairly common scope lens to show widescreen.

Hi Will

I'm very keen to keep the image ala naturale - ie. using traditional analog printing methods rather than it going through a DI. And the different result I find quite visible - there's nothing quite like analog film, analog projected. It's almost a mystical thing. Or a spooky thing. It has what Walter Benjamin might have otherwise meant by "aura". It's not just a psychological thing but is out there on the screen, or in the gallery space, with the chatter of the projector in the background, and the flickering light on the wall - an alternative reality ...

Anyway it's not a cost issue as such - as indeed I've actually built a setup (if somewhat temporarily dismantled at the moment) to do a decent DI: 5k film scan and (at least for now) a 2K print back - and with the S16 K3 now, can print the soundtrack timecode as well. So will certainly be doing the DI approach and with a cinemascope lens on the projector - in fact the DI setup was built to support an UltraPan8 project (ie. cinemascope work) but for the moment (for this particular project) just wanting to keep the image firmly in the analog domain, using 4:3 16mm. Basically we've got a colour contact printer and it needs some love. Or rather, it has plenty of love to give us.

I was looking at some Bach Auricons on ebay to do optical sound recording but it's hard to tell on some if they have the optical sound head, or if they did, whether they worked. And the one that stated it had optical sound and was all apparently working didn't quite fill with me enough confidence. In the end I figured the Super16 K3 on ebay offer would give me more control over the process - I could guarantee the optical sound track was printed correctly rather than having to worry about how the light galvo might muck up the signal (and how to compensate for that).

As a technical aside: one could actually print just completely random patterns on the optical sound track and so long as the computer had a copy of the same random patterns, indexed against frame numbers, it could read those patterns and convert them back into the corresponding frame numbers, or into SMPTE timecode, or into MIDI etc. There's no requirement that the patterns have any deterministic order at all - other than each frame of information (or noise!) be different from any other frame. Printing a determinstic set of patterns such as SMPTE is just one way of ensuring the information changes from one frame to the next. Using a random generated GUID is another. Of course some consideration needs to be given to the the way the signal is electrically propogated - some signal shapes are better than others. For example SMPTE timecode (regardless of what information is actually encoded at any point) is first and foremost designed for signal stability.

Carl

[Will2]

I'm very keen to keep the image ala naturale - ie. using traditional analog printing methods rather than it going through a DI. And the different result I find quite visible - there's nothing quite like analog film, analog projected. It's almost a mystical thing.

I agree totally. I love pulling out the Kodachrome 16mm and projecting it. Nothing comes close. Try a film out once though, just for comparison.

[carllooper]

Try a film out once though, just for comparison.

For sure. I'll definitely do that in due course. The film-in>DI>film-out is actually a bit more work (or takes more time) than the film>film process - but only because I'm doing it all myself.

I originally misread your comment as "trying film once" (as distinct from: trying "film out" once) but while trying to make sense of that misreading, it did inspire this formula:

1. Try out film once.
2. Repeat Step 1


Addendum.
I often find the DIY process of transferring film to digital is not unlike the process Wile E Coyote undergoes transferring a rock to the road runner:

http://www.youtube.com/watch?v=5aCgSwmm5Ho

C

[Will2]

film-out. Hyphens can change the world.

[MovieStuff]

You don't really need a constantly updating address track like SMPTE, which identifies every frame. All you need is a reference point that does not repeat within an isolated time frame relative to the film being watched.

Let's say that you have a film that is 10 minutes in length and you have two outputs on the projector. One is a once-per-frame contact closure or sensor that marks the passage of each frame discreetly. The other is a timing disk with one, two, three, four, five and six slots in ascending order that is geared in reduction to the film advance across, say, one minute. In your program, you know that every 10 seconds you will get, in order of appearance, one pulse, then two, then three, then four, then five, etc until the pattern starts over. But when the pattern starts over, you are into minute number two. Repeat for minute number 3 on up to the end of your 10 minute film. So your audio track has these two sync streams. One stream has a pulse for every expected frame of film. The other stream has a unique event every 10 seconds of each minute and then repeats. No matter where you are in the audio track, these two streams accompany it.

Now, for projection, you align the start of the film with the beginning of both timing discs. During projection, everything happening is relative to a PC's internal clock (which never stops) and you know that the pulses coming from the ascending timing disc will never, never, ever hit the same spot in time during the run of the film. If you stop the projector, the audio program stops because it receives no per frame pulses. When the projector starts again, the audio starts and will realign itself within 10 seconds because the ascending timing disc creates a unique event every 10 seconds within the minute that the projector stopped.

So what happens if the projector stopped but the audio kept running?

Well, because the PC's clock never ceases counting time, the software knows how long the projector has been off before it starts up again and can quickly calculate where the pattern should be falling if the projector had continued. So if the projector were off for, say, a minute and then started again, the PC's internal clock knows that a minute has passed and needs to stop and hold the audio until the frame count/event log coincides and then get in step which, again, will happen within 10 seconds of the audio being released.

Basically, you are creating a repeating pattern generated by the projector that never repeats itself within a given 60 second period and the PC keeps track of the minutes as divided by the predicted frame count. Because the PC's internal clock never stops, it is the reference against which everything is fitted. Even though the projector isn't running crystal sync, the unique events created by the ascending timing disc give it a reference point within each 60 second period. Ultimately, the ascending timing disc will provide a "re-sync" if necessary while the per-frame pulses keeps in in sync from frame to frame.

Make sense? ;)

Roger

[carllooper]

You don't really need a constantly updating address track like SMPTE, which identifies every frame. [...] Make sense? ;) Roger

Thanks Roger. I understand. Its a great idea. The issue for me (I was using a graycode disc in an earlier design to accomodate any rewind necessity - but otherwise a similar problem) is if some mechanical error were to occur where you had to stop the projector and rejig the film in the gate - ie. destroying the relationship between film and projector. God forbid that happen of course. But the other (which is more the clincher) is portability. I want to be able to send (or take) the film to alternative venues without having to necessarily send (or take) a custom projector along with the work. Or at least that's the idea. On the other hand a cool projector would deserve some travel.

C

[MovieStuff]

You don't really need a constantly updating address track like SMPTE, which identifies every frame. [...] Make sense? ;) Roger

Thanks Roger. I understand. Its a great idea. The issue for me (I was using a graycode disc in an earlier design to accomodate any rewind necessity - but otherwise a similar problem) is if some mechanical error were to occur where you had to stop the projector and rejig the film in the gate - ie. destroying the relationship between film and projector. God forbid that happen of course. But the other (which is more the clincher) is portability. I want to be able to send (or take) the film to alternative venues without having to necessarily send (or take) a custom projector along with the work. Or at least that's the idea. On the other hand a cool projector would deserve some travel.

Well, how about a small box with a sprocket that sticks to the output of any given projector with velcro. The film just threads through the sprocket and the gearing is internal and it has two RCA jacks for the data pulses. A simple "reset" button on the box aligns the discs to the start position once the sync frame is in the gate. I mean, ideally, this same box could simply have an SMPTE time code generator in it but I'm thinking that they are going to be pricey and hard to come by and difficult to service. A simple gear box would be pretty cheap and easy to reproduce. But I understand what you mean about the projector breaking down but I think that's an anomaly that should not be a deciding factor in design.

Roger

[carllooper]

You don't really need a constantly updating address track like SMPTE, which identifies every frame. [...] Make sense? ;) Roger

Thanks Roger. I understand. Its a great idea. The issue for me (I was using a graycode disc in an earlier design to accomodate any rewind necessity - but otherwise a similar problem) is if some mechanical error were to occur where you had to stop the projector and rejig the film in the gate - ie. destroying the relationship between film and projector. God forbid that happen of course. But the other (which is more the clincher) is portability. I want to be able to send (or take) the film to alternative venues without having to necessarily send (or take) a custom projector along with the work. Or at least that's the idea. On the other hand a cool projector would deserve some travel.

Well, how about a small box with a sprocket that sticks to the output of any given projector with velcro. The film just threads through the sprocket and the gearing is internal and it has two RCA jacks for the data pulses. A simple "reset" button on the box aligns the discs to the start position once the sync frame is in the gate. I mean, ideally, this same box could simply have an SMPTE time code generator in it but I'm thinking that they are going to be pricey and hard to come by and difficult to service. A simple gear box would be pretty cheap and easy to reproduce. But I understand what you mean about the projector breaking down but I think that's an anomaly that should not be a deciding factor in design.

Roger

Yep - I reckon that's the goer. I'll implement that.

It just has to convert sprockets passing by (and the direction of travel) into a two bit code (sprocket or not/ direction) and send that to the computer - the rest is just done in software on the computer. The computer reads the incoming codes and for any sprocket bit, it inc or dec the frame count as a function of the direction bit:

00 Between sprockets/Forward (no count)
01 Between sprockets/Reverse (no count)
10 Sprocket/Forward (increment count)
11 Sprocket/Reverse (decrement count)

The software I'm writing regulates the sound playback in realtime as a function of frame count - or SMPTE or MIDI - doesn't actually matter as I'm writing the software myself anyway. The sound engineer (musican) I'm working with has some ideas that need MIDI, but the framecount to MIDI conversion can be done in software.

cheers
C

ps. I'll still do the optical soundtrack idea but as a second iteration of the project.

[S8 Booster]

now,here is what you need - no fuzz



btw wouldnt it be better to let the sound system control the projector?

shoot....

[carllooper]

btw wouldnt it be better to let the sound system control the projector?

For dialogue it doesn't matter so much but for music it would. However controlling the projector doesn't solve the portability issue. For music I'll try some real time pitch correction. However music can also be done in such a way as to go off a cue system - where drift isn't so critical it can be dynamically mixed (ie. at runtime) as a function of cue points. Music keeps playing while projectionist is frantically fixing a jammed projector ...

C