By using this site, you agree to our Privacy Policy and our Terms of Use. Close
HoloDust on 15 March 2026
Edit Thread
TallSilhouette said:
HoloDust said:

Hopefully to some sort of volumetric world representation, instead of polygonal shells that are still dominating. This will solve both lighting and physics a the same time. 

Like voxels but on a finer scale or something? Maybe, but I haven't heard much new in that space. The current rage in CG (besides AI) are gaussian splats and point clouds (sort of an upgrade to photogrammetry), though that won't do much for physics. 

Yeah, SVDAG, and Dynamic SDFs for terrain, or something like that. I'm a bit out of the loop recently, but it seems, apart from Teardown that allows for what I'm thinking, albeit on smaller scale and still with pretty rough minimal units, not much is happening in that space.

Bofferbrauer2 said:
HoloDust said:

 @bold Hopefully to some sort of volumetric world representation, instead of polygonal shells that are still dominating. This will solve both lighting and physics a the same time. But for that to happen, someone has to be bold enough to introduce hardware blocks into GPUs that are optimized for such tasks.

Considering that Donkey Kong Bananza uses this technique to render the destructible environment in the game on a Switch 2, either this means that there are already some ways to do so, or if done correctly needs so little CPU power that we won't need GPUs anymore.

DKB does it nice, for what it needs to do, but it's very chunky and physics in it is very limited (compared to something like Teardown). That said, if I think of anyone who would be bold enough to go the route of custom silicon that supports such features in hardware, it would be exactly Nintendo, since they have so much 1st party IPs that could be built around it as to stand distinctly apart from all competition.

Last edited by HoloDust - on 16 March 2026