Steven DollinsScatter 2 million particles and move them as
P += sin(P.yzx) - 0.19 * P
and the shape they form is strangely attractive.
Here's a (nearly) looping animation of that attractor once the points have settled down.
The - 0.19 * P term pulls the particles back towards the center. By weakening that constant, we can see more of the sine wave motion. Too weak just looks like noise, but 0.08 looks particularly nice.
P += sin(P.yzx) - 0.08 * P
Strangely attractive lingerie
P += sin(P.yzx) - 0.13 * P + 0.09 * P.zxy
ijk64✅@scdollins I wonder what happens if the constants are slowly changed over time - does it morph or does it simply become chaotic jumble
Steven Stadnicki (he/him)@scdollins These are incredibly cool! It doesn't look like the points move that much each tick — is there an implicit delta-t in the expression somewhere? What's the approximate viewing region for these?
@sstadnicki Yes, there is a small dt multiplier on the integration. I actually do a few hundred steps between draws, though I think that's overkill. The attractor is a thin curve. The surface, I think, is a combination of loose integrations steps and not enough time to fully converge.
The camera is looking at -1 to 1, so I scale the points up or down to that volume. I think these are scaled by 1.75 * b. Smaller b lets them spread out more, so I scale down by that amount.
nnenov@scdollins this is so cool, thanks for sharing, I like when it starts from a small local area, and ends up in the same distribution
KloudWhen I was playing with 2d attractors, animating the variables in the formulas looked very cool. You can also try slowly changing P while it's running.