You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Now we generate both the volume and the surface on CPU. First the volume is generated, then geometry for the surface is generated on CPU. We did that because we thought we probably would need surface geometry for physics simulation. But it would probably result more performant to simulate the physics based on voxels.
So we keep volume generation host-side, and move surface generation device-side:
Remove any code involved in surface generation on CPU
Upload chunk octrees on GPU
Create a huge octree for the WorldView: individual octrees are dynamically allocated on a buffer while low resolution nodes can be statically allocated.
Ray trace the WorldView octree on GPU
Render an initial blocky surface
This issue will be followed by a Continuous Surface generation issue that will employ trilinear interpolation to make the surface continuous.
Now we generate both the volume and the surface on CPU. First the volume is generated, then geometry for the surface is generated on CPU. We did that because we thought we probably would need surface geometry for physics simulation. But it would probably result more performant to simulate the physics based on voxels.
So we keep volume generation host-side, and move surface generation device-side:
This issue will be followed by a Continuous Surface generation issue that will employ trilinear interpolation to make the surface continuous.
References:
The text was updated successfully, but these errors were encountered: