Since the Nvidia conf i have been checking a bit the idea and it seems to be a great way to render stuff for high resolution, VR or doing some async raytraced shaders. is it fair to think UE4 will move to TSS?
a good intro paper :
conference from oxide :
https://youtu.be/QJOIvACRY6g
http://32ipi028l5q82yhj72224m8j.wpengine.netdna-cdn.com/wp-content/uploads/2017/04/Capsaicin-Cream-GDC2017-DanBaker.pdf
ppt from oxide:
any reason to think they would not go this way?
There is a limitation: the UVs need to be completely unique to work. You would need a new UV channel, which increases draw count again.
It sounds a good fit for caching landscape layer blending results.
do you see like this would be a big issue? my understanding is since DX12 the number of draw call significantly went down so maybe we should still be at reasonable drawcall count?
How uv channels and draw count are connected?
Lightmapping need unique uv’s anyway and ue4 can generate those for static meshes.