Banana pixel synth question


First post here - been using TD for a few weeks.

I have made a pixel to audio synth. It seems to work well except only along a thin portion of the window’s x axis.

When I move the geometry up or down the y axis it stops triggering the synth. I tried tinkering around with the resolution but must be missing something.

I’ve attached 2 networks that I tried this approach.



bananasynth.toe (14.1 KB)
Modulated Planar_instancing.toe (73.2 KB)

1 Like

According to the TOP to CHOP, yes, you are sampling a single row of pixels at v = .5 (middle of image vertically). It’s tricky how you sample images for sound. Often you are bound by the fact you are re-sampling your image 60 times per second, so there is often a 60Hz base to it all. The art is in how you create, reduce and sample images in converting to audio. What may look like it sounds cool usually doesn’t and vice versa.

Thanks Greg, that’s helpful.

My goal is to be able to move the contents of a geo around the output window and have it triggering the synth wherever it is. I’m then going to map the position of the contents of the geo to an external ambisonic sound panner which will spatialise the sound in relation to the position of the geo contents.

So if I want to sample all the pixels do I need a cascading row of TOP to CHOPS with v= .1, v=.2, v=.3 and so on? Or a more elegant way of doing in?

Also similar for the horizontal axis. Do you know how I would sample every pixel?

Thanks again.

I would crop/rotate the image and then scale the resolution down to whatever samples you want out, using a Resolution TOP, choose the Input Smoothness parameter to whatever looks good. Say you want 5 rows with 200 samples in each row, you scale down to 5x200 resolution. The set your TOP to CHOP sample the entire image, not one row.

1 Like

Great thanks Greg that made a big difference