100 million balls in superellipse billiards
Further elaboration in maxing out the number of particles in dynamical billiards... PyTorch was used this time as the next natural step, turning the graphics rendering the real bottleneck. The billiards shown are |x/1.5|^n + |y|^2 = 1 for n = {2, 4, 8}. Further, rasterisation by 2D histograms were necessary for bitrate reduction and rendering time. Log scaled density for emphasizing low-density caustics. Still a bit crappy from a visual point of view imo though. Next time, 1B points and better graphics. Music by @gpcbass . The song is called No Way Out. Shoutout to my brilliant colleagues Björn Lindenberg and Jonas Nordqvist for discussions, inspiration and motivation. Frames from Python and video from FFmpeg

▶︎
Ellipse reminds you of its focal points

▶︎
What's Beyond the Screen Borders of the NES?

▶︎
100 Million Particle Ball Dropped in 4k60fps

▶︎
Billion ball billiard

▶︎
Don't Hang Up On AI Scammers. Do THIS Instead.

▶︎
The most ridiculous worker fails caught on camera

▶︎
Ten million balls in a superellipse billiard (chaotic)

▶︎
Driven Under

▶︎
Casting SALT like Metal - What Happens?

▶︎
This Orbit is the WORST

▶︎
The Game You Can't Fail

▶︎
100 Incredible Lightning Strikes Caught on Camera

▶︎
134217728 balls falling on a right-angled wedge

▶︎
Balls symmetrically dropped in a parabola

▶︎
Best Hydraulic Press Moments of 2024

▶︎
The Conway Multiverse

▶︎
Balls of varying bounciness dropped on flat surface

▶︎
Total Idiots at Work Caught on Camera | Best of 2024

▶︎
The most beautiful formula not enough people understand

▶︎
