Physical engine drive problem

[spencer17x]

Playground：https://playground.babylonjs.com/#ZF86YL#6

Real effect：
https://github.com/noname0310/babylon-mmd/assets/32794413/8a78ed2e-935d-4ec4-9151-f2a8ef2e4bd3

The example in Playground looks like the clothes are flowing strangely

[noname0310]

Physics has not yet been implemented properly and is just operational.

[noname0310]

The solution currently available is physics bake and is available using NexGiMa.

https://sites.google.com/view/nexgima/Home

[noname0310]

Related bug:
https://forum.babylonjs.com/t/havok-physics-rigidbody-dont-move-when-the-damping-values-of-several-bodies-are-different/43072/2

In the next release(0.23.0), the physics will probably work on most models

[spencer17x]

[noname0310]

update @babylonjs/havok to 1.1.2 and @babylonjs/core to 6.15.0

[spencer17x]

same problem.

2023-08-08.16.57.39.mov

[noname0310]

I'll let you know when there's another improvement.

[noname0310]

babylon-mmd 0.24.0 is released now. see CHANGELOG.md

[spencer17x]

2023-08-17.11.37.24.mov

Perhaps the problem remains.

[noname0310]

There will be more improvements in the future.

[noname0310]

There was an improvement in the 0.27.0 release. Give that a try.

[spencer17x]

Or has it already been solved?

2023-09-19.10.44.56.mov

[noname0310]

If you think it produces usable results, use it.

I am not sure how to exactly reproduce the results of bullet physics(which used by mmd) with havok.

[noname0310]

The ammo.js physics backend will be available from babylon-mmd 0.42.0.
Although ammo.js is not very good performance, you will get very similar results to mmd.

The usage is the same as before, see this test code
https://github.com/noname0310/babylon-mmd/blob/main/src/Test/Scene/physicsTestScene.ts#L63
https://github.com/noname0310/babylon-mmd/blob/main/src/Test/Scene/physicsTestScene.ts#L111-L113