Sorry, it's a part of a commercial code that I wrote :/ But if you want to hear some tech details, feel free to ask me 🙂
but I code this entirely inside my “engine” 😊
github.com/altunenes/cu...

Note that no RGB involved here, only wavelengths :-P I assigned every particle a physical wavelength in nanometers and accumulate them in the CIE XYZ spectral space.
see for tech:
en.wikipedia.org/wiki/Spectra...

example (entirely on CPU):

😁👍 thank you for your contributions!

Curious what was the blocker that caused it to be so late from first place

☑️ VOTED

I think this comment could provide a similar insight into the illusion in our example. Beyond that, I don't have any further comments at this time. If anything comes to your mind, don't hesitate to share the knowledge 😁

comment from the paper: "The reason is that the black/white contrast signals between adjacent dots along the length of the line are stronger than black/grey or white/grey contrast signals across the line, and the motion is computed as a vector sum of local contrast-weighted motion signals"

well I just discovered this illusion, and it doesn't really resemble the other illusions I've experiment 🙃. The one it most resembles is the following:

bsky.app/profile/altu...

You are right, but when there is no "movement", the effect—at least in my perception—seems quite weak :-)
Also, a bit to test what you said: When I increased the bg brightness, this time I definitely needed the motion now...😁

codes:

WGSL compute shader (with my Rust backend):
github.com/altunenes/cu...

and our beloved Shadertoy version:

www.shadertoy.com/view/3Xj3zz

You can render anything with this tech, for instance, my sdf neuron compute:

also ported on my rust + wgsl -wgpu based- engine too 🦀:

rust back:
github.com/altunenes/cu...
wgsl shader:
github.com/altunenes/cu...

#rustlang #wgpu