What about temperature

[FSA]

I'm currently just brainstorming about features nice to have. What about temperature for bodies? Features like heat distribution per vertex, or temperature increase because of friction. It's not something that is urgent, but maybe nice to have for simulations.

[JoeJ]

You could use this: https://github.com/GPUOpen-Effects/FEMFX, and integrate it with Newton.
I have use heat diffusion on surface meshes to calculate things like approx. geodesic distance or cross fields. It's pretty expensive and a bottleneck even for my offline needs. There is a need to setup cotan weights to have a laplacian (for deformable stuff we need to update this every frame), and for the solve itself i use backwards euler, which i guess has similar costs than rigid body constraint solver, depending on resolution.

In practice you also need a volumetric representation of the models, which is hard to generate, and you need a way to upscale low poly simulation to high poly visuals.

So a lot of effort, though AMDs results look quite good and interesting:

Maybe, modeling heat or destruction is subject of the planned soft bodies in Newton?
To me, it's all a bit questionable - we want soft bodies, cloth, fluid, destruction... but we have not yet utilized the potential of rigid bodies alone in games. Most games either completely lack physics, or use it only for some decoration.

[FSA]

That looks interesting, thanks. I think I could use that.

To me, it's all a bit questionable - we want soft bodies, cloth, fluid, destruction... but we have not yet utilized the potential of rigid bodies alone in games.

Well rigid body physics was cool back in 2010, but 11 years later, particle and cloth physics should be easily available. My focus is more on developing a simulation engine, than a game engine, and these things have become very important over the years. My current approach with using bullet for cloth and soft bodies is just dirty, if newton would support all that it would be much cleaner.

[Julio Jerez]

Actually ther is a very good cloth solver in the engine.
It is kind of incomplete because I was not happy with the was the graphic was to be presented, plus the collision was practically none.

But after the sph I see if I can reinstate it.
It is very very fast and stable. So good that it is easy to make soft bodies with it by connecting rods.