Skip to main content
Computer Graphics

Return to Question

added 146 characters in body
Source Link
Tau
Tau
  • 121
  • 5

@notch has recently been posting on an engine he is developing. He claims that it is voxel-based and supports deformable terrain, with a render distance of almost 1000 000 voxels and real-time reflections, running at 99fps in WebGL. Even with heavy LOD reduction, such a thing sounds too good to be true. He has not published any source code as of now, so does anyone have an idea how this is done? Is it just many small optimizations or could this be a new approach to real time 3D?

EDIT: He also mentions a "signed density field". What does that mean exactly?

@notch has recently been posting on an engine he is developing. He claims that it is voxel-based and supports deformable terrain, with a render distance of almost 1000 000 voxels and real-time reflections, running at 99fps in WebGL. Even with heavy LOD reduction, such a thing sounds too good to be true. He has not published any source code as of now, so does anyone have an idea how this is done? Is it just many small optimizations or could this be a new approach to real time 3D?

@notch has recently been posting on an engine he is developing. He claims that it is voxel-based and supports deformable terrain, with a render distance of almost 1000 000 voxels and real-time reflections, running at 99fps in WebGL. Even with heavy LOD reduction, such a thing sounds too good to be true. He has not published any source code as of now, so does anyone have an idea how this is done? Is it just many small optimizations or could this be a new approach to real time 3D?

EDIT: He also mentions a "signed density field". What does that mean exactly?

Source Link
Tau
Tau
  • 121
  • 5

How does notch's new graphics engine work?

@notch has recently been posting on an engine he is developing. He claims that it is voxel-based and supports deformable terrain, with a render distance of almost 1000 000 voxels and real-time reflections, running at 99fps in WebGL. Even with heavy LOD reduction, such a thing sounds too good to be true. He has not published any source code as of now, so does anyone have an idea how this is done? Is it just many small optimizations or could this be a new approach to real time 3D?