I saw the cool seismic readout from the earthquake that occurred in Norway on the 26th March 2026 and thought… hmm? Whatever happened to Metasynth?
Its not gone, but its tricky to set up, so i made this… ok so it’s not the same, I doubt I will make whole albums with it, but, its a start.
Work in Progress.
If you want to make sound from an image, then try this:
https://markfrancombe.com/image_sound_lab/index.html
When making videos for Cranes recent re-releases, Im always working solely with old material. For Dada 331 I used footage from the Paradiso, and for “Wires” and “Here comes the snow”, both the videos used footage from a show in teh earlier 2000´s which I didn’t play at. I didn’t originally start with the footage of snow and landscape near my home in Norway.
As the song comes from the album “Particles and Waves” I first tried to build a video out of the DATA from that concert footage, and the audio waveforms of the song.
The technique I wanted to use was using a program called Touch Designer which I have been wanting to try for a while. Normally used in VJ effects or artist gallery installations, its particularly good at taking data and transforming it in realtime into visuals.
So I tried, using the video and audio for “Light Song.
So its two main elements. Visuals created from a particle system powered from teh audio that has been split into 3 frequency bands, a bit like an old Sound to Light system in school discos from the 70s, and 3d video generated from the gig footage.
Here what I wanted to do was, by using the same particle system, but this time moving teh particles based on depth in the video. If a particle was on Alisons arm, it would be infront of a partical that was on her face. So first I had to get depth information from teh video. iPhones generate it, (If they have the Lidar lens) and 3d modelling can generate a depth map, but not old video footage So I used an AI too from artificial labs to “guess” a depth map. This worked, but only on a frame by frame basis, so I had to create a way to
1. chop up the video into single frames
2. upload one image to artificial labs to do the depth map
3. download it when finished and upload the next
As the site only did one frame at a time I programmed using Claude a system to do this automatically.
When I finished, it hadn´t quite worked. It was a cool experiment, but we decided not to upload it to the Cranes channel.
But it WAS a fun experiment, so rather than bury it on a hard-drive heres first, I decided to put it here:
This is VERY NERDY. But the full and geeky explanation video complete with instructions and code follows on the second page of this post.