there are modules for detecting position consisting of a LED and Photo-transistor. in a 16mm projector their was just a lamp, a slit and a lens to project a narrow line of light across the film, and a phototube (or photodiode/Transistor) to pick up the variance of the total light passing.
RECENT 35mm prints have a cyan track which works best with red light. and proably use a red laser, like in a laser pointer.
I got a laser level at the hardware store one time which has a diffraction grating to spread the laser light out in a line. that sort of hardware coeld proably be used to fake the light source, which would have to be powered from DC. (some of the old projectors used a ultrasonic AC source, but just because that let them filter out the AC hum by using a small transformer.)
A photo-transistor could easily detect the beam, and give audio out. BUT the trick as mentioned is teh film has to run smoothly at the correct speed.
For that matter, one could just use the position detector to key off a sprocket on a working projector and pick the audio out of the amplifier. send both the audio and the sproket track to a digital recorder, and use the sproket track to keep the sound in sync with the pictures.
The reason AEO-Light isn’t used is that the sound quality isn’t great, and you really need to be scanning at 4K or better with great optics for the sound processing to work well, so it is out of the realm for some people.
As mentioned before and again by cmacd, if you either record the sprocket holes as ‘ticks’ on a third channel, or use the ‘ticks’ to control a recording device, you can compensate somewhat for variance in speed.
One more idea- and this is being used somewhere, though I don’t know what software actually makes it work. There have been several systems created that use a separate video camera, probably line scan to capture a file of some sort, that can be used to recreate the audio. I have seen it on a sound follower system at Chace Audio, now part of Deluxe, as well as on the experimental Walde scanner. I am sure it would take some work to get the right camera, the right optics, the right backlight, and of course it would depend on the speed being accurate and constant, which is not trivial. Actual, I just remembered that Darren Walde told me he was using a special version of AEO-light with this system, one that could process a continuous track, as opposed to stitching frames together. It seems a camera system could be chosen with enough resolution to make this viable.
Certainly could be done, but if the speed needs to be constant anyway, then a photocell and light source would be much cheaper and less complex to achieve the same end.
I am currently writing my master thesis about restoration of optical soundtracks.
The development of an extraction and restoration software is part of this.
At the moment I have implemented plenty functions without gui.
So far I am able to extract audio signals from images with higher quality than AEO-Light does.
I want to improve my software, but my material sources are limited.
This means that I only have three uncompressed videos from the AEO-Light home page to work with.
So I am searching for all sorts of digitized film material with different types of optical audio, with damages and dirt on the audio.
Can someone at this forum help me?
Short clips of 35mm, 16mm and 8mm film would do it (uncompressed if possible).
Here is a first comparison of AEO-Light and my software at the current state:
Thank you for your samples! @dan74: Whew, your sample is really a worst case scenario.
There are several technical aspects, which make it hard to get a “great” result here.
First of all, there is virtually no significant overlap between neighbour frames, so there are gaps in the audio. AEO-Light and also my software (at this point) are not able to fill gaps in the audio signal.
This causes the stammer artefacts in the audio signal of the sample.
But that is a new point that should be considered in the future, an algorithm to interpolate gaps in the audio.
The next problem is small resolution of the images, they are only in 480p.
So (without overlap) there are still 480 audio samples per frame, this results in a sampling rate of 11520Hz (at 24fps) and according to the sampling theorem the audio signal could then only have a bandwidth within 0Hz-5760Hz.
Next problem is a high jpeg compression. The audio signal is stored in the transition between the brighter and darker areas of soundtrack. Jpeg compression is DCT based, so the image is transformed to the frequency domain and is quantised to reduce data.
This also affects the data on transitions and areas in form of ringing and blocking. So while maybe areal noise is reduced, new artefacts are brought in the audio signal.
Then, there seem to be line artefacts and some kind of ghosting artefacts, maybe caused by a suboptimal CCD sensor.
Uneven background illumination and so on…
Some of this artefacts could be reduced, but it would be much better to avoid them.
But generally, worst case scenarios are absolutely welcome to push the limits of the software.
That brings me to some theoretical aspects, which should be considered during the digitisation of optical soundtracks, to make a better result.
An optimal usage of image area would be the green frame. This is the best compromise between audio overlap and movie picture resolution.
Since my software uses the image information to generate an ideal overlap, it is good to have some characteristic details like sprocket holes on it.
This is superior to stitching after audio analysis, because it avoids misinterpretations of periodic audio signals. Also the resolution between both overlap positions can be interpreted as information about shrinkage of the film.
In fact, a correct scan with 2K resolution would be absolutely enough to cover the full bandwidth of the audio signal. However 4K would be a better compromise to the movie picture resolution.
Due to the area needed for the audio overlap, the movie picture resolution would fall under HD resolution if a 2K sensor is used.
It would be best to use a monochromatic image sensor, because of the lack of a Bayer-pattern and as a result of this, a better filling factor and anti-aliasing.
For variable density soundtracks a higher bit depth is needed, because the audio information is stored in the luminance. So a smaller bit depth is a lower precision quantisation of the audio signal.
An uneven background illumination causes hum in the audio signal, so it should be as even as possible.
The sensor which is used to digitise, should not have a fixed pattern noise and the optics should be clean and free of dust (even dust causes an audible periodic noise).
Some of the resulting artefacts can be reduced or completely removed, but it is still better to avoid them.
@ dan74
I will try to enhance your clip.
I plan to make the software open source, once it is in better state of development.
But it will only happen after I have finished my thesis in August.
@dan74: Thank you very much!
Ok, here you go, this is a test of audio extraction from your clip:
Some artefacts that the AEO-Light version had are gone (uneven illumination, stitching, stammer), but others are much more evident now, because my software does not apply noise reduction to the audio.
The most noise in your video is caused by dirt on the left part of the soundtrack (those little dots), other noise sources are the jpeg compression, the ghosting artefacts also a rotation and distortion of the image and so on.
The next planned step for my software is to automatically remove the dirt, that should improve the audio further, but expect no wonders the source is far away from the optimum.
And another thing I have noticed is, that the AEO-Light version is some kind of pitched deeper and I can’t explain why that is.
Hello everyone…as per my experience the soundtrack varies within a fixed width, the light source is a very
thin line, so it results in the photocell receiving an amount of light proportional to the width of that section of the soundtrack. It is very simple really.The difficulty is in the calibration, the speed has to be constant,
the slit perfectly focused and aligned to the soundtrack, i.e. the line has to fall horizontally across the soundtrack, and even a slight angle will cause the sound to be inferior.
True on all accounts. I’m looking at continuous servo motors that are used on tape recording machines to ensure constant speed. As for focus and alignment - definitely things that can be solved. It’s the electronics I’m clueless about. I have no experience with that kind of circuitry, as I deal mostly with MCU circuits. Got any recommendations?
It is still work in progress.
I am focused on the documentation and
on developing and testing cleanup algorithms at the moment.
Updates will follow, when I have finished the thesis.
