A repeating pattern can feel mechanical or alive. The difference often lies in how much variation it permits. A metronome repeats one event. Rain repeats a family of events, with irregular spacing, different impacts and changes in density. Both contain rhythm, but they give us very different things to pay attention to.
That distinction makes Order from Chaos, by Max Cooper with visuals by Maxime Causeret, interesting beyond the usual relationship between music and animation. Cooper describes building the track around recordings of rain, while Causeret's accompanying imagery explores systems that resemble living structures. The sound and image share an idea without having to illustrate each other literally.
I like this approach to generative work. A bass note doesn't need to become a bigger sphere every time it arrives. That mapping is readable, but it can exhaust itself quickly. A more useful connection might be between the density of percussion and the frequency of new growth, or between a slow musical transition and a change in the conditions under which a visual system operates.
Consider a simple imaginary animation. Points move through a field and leave trails. Each point follows three rules: move forward, turn a little toward nearby points, and avoid getting too close. Individually these instructions say nothing about ribbons or branching structures. Together they may create both. The result depends on the balance between the rules, the initial positions and how long the system is allowed to run.
Now connect that system to sound. Making every beat change the direction of every particle would reveal the audio connection immediately, but it would also destroy the accumulated motion. A smaller intervention can be more expressive. Let the music change the attraction radius slowly. Preserve the particles' momentum. A passage becomes a change in behavior rather than a series of interruptions.
There is a useful separation here between simulation time and editing time. The simulation produces possibilities. The edit decides which possibilities deserve attention, how long to hold them and when a cut should change our reading of what we see. Procedural generation doesn't remove composition. It moves part of composition into the choice of rules, then leaves another part in the selection of results.
This also explains why a technically elaborate system can produce a dull film. Complexity is not a substitute for contrast. If everything changes at once, there is no stable reference from which to notice a change. A quiet region, a repeated shape or a temporarily fixed camera can make a small disturbance much more visible.
What stays with me about this kind of audiovisual work is the relationship between restraint and surprise. The rules need enough consistency for us to sense a system, and enough freedom for the next moment to remain uncertain. That's a useful design problem whether the output is a music video, an installation or a few hundred lines of graphics code.