Hey guys, really excited to dig into the toolkit tonight once I wrap up work, wondering in the mean time if you saw that great GDC presentation by the people behind Coral VR and what your thoughts might be on if conemarching is a good fit as an optional optimization to squeeze some more complexity into (especially VR) scenes?
https://s3-us-west-1.amazonaws.com/coral-vr.com/downloads/VRDC-CoralConemarching...
http://www.fulcrum-demo.org/wp-content/uploads/2012/04/Cone_Marching_Mandelbox_b...
Big downsides:
- Kind of a pain to use with arbitrary resolutions (but not as big of a deal with VR where the render target resolution is almost always 1080x1200 for the time being?)
- Lot more knobs for the end user to turn in order to get a good performance/visual quality balance, can introduce frustrating artifacts
I haven't tried implementing conemarching on top of a sphere tracer yet myself but I might give it a shot with this in a couple days. The VR specific reprojection and foveated rendering optimizations seem a little over my head right now though :B