https://jms55.github.io/posts/2026-04-12-solari-bevy-0-19/

[giscus[bot]]

https://jms55.github.io/posts/2026-04-12-solari-bevy-0-19/

Bevy, Rust, Graphics, etc

https://jms55.github.io/posts/2026-04-12-solari-bevy-0-19/

[Lex4art]

Hi. Recently I wrote an article ("The Great PCVR challenges - cost, friction and dated paradigm"), it's focused on VR but there are quite some performance benefits that may be possible to gain, when rendering a normal 3D game (non-VR) on a virtual flat screen inside a virtual space. That ability to track where eyes are focused now for rendering only a small portion of the screen - foveated rendering - has much more potential left to tap, it feels like the whole "foveated tech stack" may appear because how much rendering workload is redundant when rendering the whole screen in a uniform way. And it feels like pure path tracing as renderer (without any rasterization) may be the one that benefits most from this and easiest to implement. But I'm not a graphics programmer, so take it with the grain of salt.

Here is the link to this article on the Reddit, if all this sounds sensible:
https://www.reddit.com/r/virtualreality/comments/1trrqfe/the_great_pcvr_challenges_cost_friction_and_dated/

[astrilux1]

What are your thoughts on ReSTIR PT Enhanced, the recent paper released by Lin et al. at NVIDIA? The performance gains look super generalized and robust.

[JMS55]

Lots of cool stuff in it! Stay tuned for my next blog post around the Bevy 0.20 release :)