ohai.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A cozy, fast and secure Mastodon server where everyone is welcome. Run by the folks at ohai.is.

Administered by:

Server stats:

1.8K
active users

#cellularautomata

1 post1 participant0 posts today
Pasture permutation

Double exposure of farmland in Weesp, Netherlands, processed with a reversible cellular automaton based tracing algorithm. Original picture is a double exposure using a Zeis Jena Tessar 50mm f2.8 in a Pentacon Vebur leaf shutter for exposing on a Canon EOS M. March 2021.

#farmland #experimentalphotography #computationalart #cellularautomata #weesp #abstractlandscape #abstractphotography #netherlands #imageprocessing

One from the archives (2018): A live recording made from only two samples, using generative sequencing (using my own tools) and probably the (by far!) most amount of effects, layers & signal chains I've ever used on a track... Granular synthesis, time stretching/slicing, resonators, filters, erosion... it gets pretty wild & textured from ~2:30 once the cellular automatas controlling various granular synth params had a bit of tantrum for a minute or so... :)

Always wanted to hear that on a large PA. Some of the sounds almost remind me of the high voltage zapps produced by tesla coils...

The samples:
1) A needle skipping on a dusty vinyl record
2) An (unintelligible) voice-like AI sample...

soundcloud.com/toxi/20180922-m

Replied in thread

@zeda
There are 2+ ways to derive the Thue-Morse sequence using #CellularAutomata, and as far as I can tell, nobody has formally verified this. They just point to it and say "this works out to an arbitrary number of terms". I want to prove that it always works. One is a regular automata. The other is one based on the Firing Squad Synchronization Problem, which uses very weird CAs that I don't fully understand

For what I'm reading now the system I've independently developed is closest in spirit to Reservoir Computing, Artificial Life, and Agent-Based Modeling, and to a certain extent, to the broader field of Decentralised AI. It shares their focus on emergent behaviour from local interactions, but it's differentiated by its lack of explicit learning mechanisms, optimisation goals, and symbolic reasoning, making it a distinct kind of complex system simulation rather than a direct mapping to any AI technique.

The core idea is using dynamical systems, with their own complex local rules and interactions, to observe the system-level behaviour. It's a simulation model designed to explore a particular set of dynamics rather than a direct implementation of a learning model. This makes it a valuable tool in complex systems research, even if it is independently developed and not aligned with the mainstream of modern AI techniques.

The animation shows it's stable after 30000 cycles, with no new energy injected into the system.