How I made Steamboat Willie's visual soundtrack (Patreon)

After this, I realized it probably makes as much sense as a text post as a video. Nevertheless, I made a video (above).

As I explain there, I couldn't find a period-appropriate scan of SW (scans available are from later on, using different sound methods). I needed variable density and, happily, found a way to make it.

My steps were:

1) Select a Wav for turning into optical sound.

2) Visit the project: https://github.com/sixteenmillimeter/SoundtrackOptical 

These mad geniuses have developed a way to make it all work.

3) Download Processing: https://processing.org/ 

Free software that you can import the library to and use to make your soundtrack. I have no idea of competing ways to do this - again, this is sort of an Optical Sound for Dummies by a Dummy.

4) Hack it together! I ran into some issues because I had to regulate access to files and was using the wrong type of Waveform. You'll want a mono wav that matches the example files provided by the Sountrack Optical people. I also ran into issues because I was doing the wrong number of frames - so mess with that. It was commented in my copy of the code.

I just asked and insulted ChatGPT into helping me along. I'm pretty anti-Chat GPT for anything but the most basic research tasks (it is very dumb), but to help a non-coder become a terrible coder, I think it's a great tool.

5) Watch the output! I believe Chat GPT actually gave me code to spit out the PNGs which I then made into a video sequence in Premiere or After Effects. Your mileage with this code may vary, but it worked.

void draw () {

  if (currentFrame < totalFrames) {

    soundtrack.draw(0, 0);

    saveFrame("/Users/philipedwards/Desktop/Phil Edwards INC/Mickey/Processing Outputs/new/output-###.png");

    currentFrame++;

  } else {

    noLoop(); // Stops the draw loop

  }

6) Check your work - or watch other sound. It was important to me that before publishing I verified that I was making it work right. I think if I were doing this for a, like, client, I would probably do a bit more. But I was pretty satisfied with this. I checked my work via this project: https://usc-imi.github.io/aeo-light/#about 

It is free and awesome. You set the bounds you want it to read - I tested it on a film strip as well and used it that way. And then it gets to work and plays you a sample! I was delighted to find that from my image alone it successfully read the soundtrack. It'd be awesome to find a filmstrip that hadn't been scanned for sound and try this.

That's the procedure! If you have questions or errors, let me know here or via email.