Sandbox Like Program of DMM

[Spartan_344]

Ok, I actually I take it back. It won't be ready at the end of the week. We want to release something nice, not just a green and white viewer so it's needs some more work!
To answer macattack: it will have a grab feature so you can grab the table other there and smash it against a wall

Aww u dont want to realese a beta vertion ??? lol

[Dawlight]

Pure solides are one object - one velocity (well actually two - positional and rotational ;P) that needs to be calculated - and collisions just affect this whole object.
DMM splits components up into a grid like mesh where every single vertex (point in the mesh) got it's own velocity. Those vertices are connected by springs which allows the whole object to bend, fracture et cetera.
This obviously needs way more calcuations than simple rigid body dynamics.
Oh and I won't think DMM will make use of PhysX (Even though being able to combinate both would be quite cool - Simulating rigid bodies with dmm is quite heavy)
As our friends from Pixelux spend quite much time writing their own solvers and converting the whole thing to another solver would be quite senseless now.
As for the multi-core support - isn't it already multithreaded? It however pushes my dual core machine to a steady 100% cpu usage.

Yeah, but I figured that maybe they used the model-swap method to swap from a pure solid to a DMM object. That is, once the object is put under such stress that it would be affected at all. To me that sounds like a more CPU and GPU friendly method as the GPU doesn't have to render so many polygons.

But then again, I'm no expert at all.

[David]

The inner faces aren't displayed as long as the object isn't broken. But yes, your model still has to have more polygons than an "usual" solid-model if you want it to be breakable. But I suppose there are some ways to optimize the way the objects are rendered when you implement DMM in your rendering engine (as DMM itself has nothing to do with the rendering, it's only simulating the way the physical object are simulated. The rest depends of your implementation of the library in the rendering engine you're using).

[Spartan_344]

Will we be able to import the user generated maya dmm simulations into the veiwer ??? or is it going to be whatever you give us ???

[zalo]

We currently have a viewer that is a simple 3D engine that displays DMM objects in real time. We can export a complete Maya scene (only the DMM objects at the moment) and load using this real time viewer and then move around and throw stuff.

[dmmNOW]

Here are some answers:

• DMM is multi-threaded. so it has multi-core support built-in
  DMM can interact with rigid bodies (and vice versa), and the game programmer can have the engine to swap between rigid bodies and soft bodies at run time
  Inactive objects don't take any computation time

[Dawlight]

I didn't realize DMM was a complete physics engine. But yeah, I tall makes sense now.
Thanks for the answers.

How is the viewer going?

[Spartan_344]

I noticed that if I run the dmm plugin at real time i only get 8-12 FPS. im assuming that will change with the veiwer??

[Dark Matter]

This sounds great, again, thanks for the updates.

[David]

I noticed that if I run the dmm plugin at real time i only get 8-12 FPS. im assuming that will change with the veiwer??

Yes don't worry, the framerate is way higher in the viewer than in Maya