Loading video...

Video Failed to Load

Go Home

I'm slowly porting my approximate grid SPH simulation to VRChat. It did require few improvements to be stable enough, now its a hybrid neighbor index/particle average grid. Still need to implement d4rkpl4y3r's mip counting sorting method for improved performance.

27,812 views • 1 year ago •via X (Twitter)

6 Comments

Mykhailo Moroz's profile picture
Mykhailo Moroz1 year ago

The grid now stores the minmax id's of the particles in this cell, its done by rasterizing points with vec2(id, N-1-id) with min blending (to get min and max). If a particle didn't get into this grid, its afterwards additively rasterized into another average particle grid.

Mykhailo Moroz's profile picture
Mykhailo Moroz1 year ago

But other than that its very similar to my old VRChat fluid sim, except now I can have proper separate particles with their own data that isn't lost (like color in the video) (also I no longer use Custom Render Targets (they are cursed))

Mykhailo Moroz's profile picture
Mykhailo Moroz1 year ago

cc @shr_id P.S. the (arguably cursed) mip map prefix sum method

Ivan Bozhenko's profile picture
Ivan Bozhenko1 year ago

@d4rkpl4y3r_vr Дуже круто , кожного разу дивуюсь від того що ви робите

Nuance's profile picture
Nuance1 year ago

@d4rkpl4y3r_vr That acronym is taken Don't google it

CodeRabbit's profile picture
CodeRabbit1 year ago

AI-first pull request reviewer with context-aware feedback, line-by-line code suggestions, and real-time chat.

Related Videos

Elon Musk just told you exactly how the AI race ends. Not with a better chip. Not with a smarter model. With the Sun. Musk: “So to do that, we need to harness the power of the Sun.” That is not a metaphor. It is a thermodynamic blueprint for the next thousand years. The market thinks this race ends at a data center. It does not. Every watt generated on this planet is a rounding error compared to what the Sun outputs in a single second. You cannot run superintelligence on a grid that struggles to keep the lights on. The bottleneck of this decade is not silicon. It is not software. It is raw energy. And Earth does not have enough. Musk: “We want to be a civilization that expands to the galaxy with spaceships that anyone can go anywhere they want at any time.” That is not aspiration. That is an engineering specification. A single-planet species is a dead species on a long enough timeline. Every civilization that stopped expanding did not plateau. It collapsed. The organizations fighting for control of Earth’s power grid are fighting over a grid that is already obsolete. The real play is orbital. Space-based solar. Zero atmospheric loss. Direct capture from the source. When you decouple your compute engine from terrestrial physics, the ceiling disappears. The regulatory class wants to slow the grid down. The builders want to abandon the grid entirely. That is the fracture point of this century. The United States either captures the orbital board or watches someone else do it. There is no middle position. Musk is not building rockets. He is engineering the escape velocity for an entire species. While Washington debates permits, he is calculating how to swallow the output of a star. The Sun puts out more energy in one second than humanity has consumed in its entire history. Whoever captures even a fraction of that first does not win the AI race. They win everything.

Dustin

18,203 views • 5 months ago