Modern textureless deferred rendering techniques

Do all or part (or none) of the X360 optimizations carry over to the new consoles? Or are they to different in architecture?
 
Random thought:
"The smallest amount of data you need to store is a triangle id per pixel (32 bits). Triangle id allows you to reconstruct the depth"
... makes me wonder if you could also store triangle id for shadow maps = more depth precision, no more shadow achne
 
Triangle/object ID shadow maps are a known method. Though you don't use them to reconstruct depth (it's much cheaper to store depth directly if that's all you're interested in). Instead you compare the current triangle ID with those in the shadow map directly (since all you really want to know is whether a given surface point is identical to the closest surface from the light source in that direction).

ID shadow maps have their own set of issues, though.
 
Triangle/object ID shadow maps are a known method. Though you don't use them to reconstruct depth (it's much cheaper to store depth directly if that's all you're interested in). Instead you compare the current triangle ID with those in the shadow map directly (since all you really want to know is whether a given surface point is identical to the closest surface from the light source in that direction).

ID shadow maps have their own set of issues, though.

Yup, see Nvidia's Hybrid Ray Traced Shadows for a recent example.
 
Back
Top