Which physics engine is prefered in this case?

Questions
41

Had a hard time getting a good video capture, but a wall/stack of 300 cubes in Jolt… dips down to ~50fps then back up to 60 pretty quicky once it settles.

42

That’s great!
You makes me can’t hesitate Jolt today.

And also your game looks cool!
Is this multiplayer game?

43
  • use Instanced Mesh for fewer draw calls
  • make floor thicker, so area mass deflects collisions
  • iteration updateCubes should decrement index dynamicBodies reference or the mesh may be inaccurate.
  • use MeshPhongMaterial
  • don’t update bodies lower than -100(?) if you prevent sleep
44

Thanks for your detailed suggestions! :heart:
Btw, I don’t understand some of them.

and…

Could you explain in more detail?

45
  • Decrement… so as the array grows you avoid the mistake of indeterminate Mesh updates. The step should be less error-prone and faster. It’s not a requirement but fundamentally may improve results. Best practice in a race condition.
  • You disabled canSleep… a way to regain some performance is to skip update of blocks that are out of bounds. Maybe you could hide them or reuse the geometry?
46

i dont know why it doesnt work for me,

here my code :slight_smile:

export class Template{
  constructor(){
  ...
  this.#physique();
  }

  async #physique() {
        await RAPIER.init();
        const gravity = new RAPIER.Vector3(0.0, -9.81, 0.0);
        this.worldRapier = new RAPIER.World(gravity);
        console.log(this.worldRapier)
  }
}

btw im using nuxt3 + three js native.

47

Thanks for your detailed explanation.
I tried to follow your tips, but can’t see any improvement…

48

Reduce shadows and scene resolution. Maybe you are comparing performance of fullscreen UHD with that of a letterboxed example pen? And force a limit in RAF frameloop to 60fps or 45fps!