Experimenting with Concurrent Binary Trees for Large Scale Terrain Rendering

Speaker Bios:
Thomas Deliot is a senior research engineer working at Unity Technologies. His work focuses on computer graphics topics and improving real-time rendering of 3D content, by bringing new papers into rendering engines and bridging the gap from research to production. This includes GPU/parallel programming, post-processing, level-of-detail, materials, lighting and machine learning.

Jonathan Dupuy is a senior research scientist working at Unity Technologies. His research interests are primarily oriented towards high quality real-time rendering. This encompasses a wide range of topics including antialiasing, level-of-detail, analytic models for both materials and lighting, and GPU/parallel programming.

Kees Rijnen is an engineering manager working at Unity Technologies. His work focuses on helping his team research and develop new environment technologies and workflows.

Xiaoling Yao is a senior graphics programmer working at Unity Technologies. His work mainly focuses on improving the terrain system, integrating with new artist workflows and adding graphics features like instancing and virtual texturing.

[slipster]

It would be interesting to compare these results to hardware tessellation, as that shares many of the same properties in terms of results. At that triangle density, I'd expect the shading to become the primary bottleneck due to microtriangle issues, and that going to some type of index buffer approach rasterizing the index buffer in a compute shader (ala nanite) might be a huge win.

[BLBTER]

11:33 What if the terrain isn't derived by a heightmap and instead has 3D features like caves or overhangs? 2.5D works, but I don't see how you'd easily sample vertices in 3D space.

What videogame is at lower right at 15:20? (looks like a cat holding a lamp in isometric scenery)

[pacozalky]

It's not clear to me how to parallel propagate subdivision into neighboring triangles. Is there a way how to avoid recursive subdivision on half-edges of main mesh triangles?

[tall_guy]

So guys basically reinvented voxels but for polygons. Sounds crazy.