3 ms·
> No doubt nVidia could make a card that enabled games to support models with double the polycount of today's games [...] The solution is to use tessellation o
by virtualritz 6y ago
> No doubt nVidia could make a card that enabled games to support models with double the polycount of today's games [...]
The solution is to use tessellation on the GPU, on demand, based on projected size of features.
In blockbuster VFX heavy models are a recent phenomenon because mircopolygon renderer cores were replace with ray-tracing ones.
Models used to be much lighter, subdivided at render time and displacement, both textured as well as resolution unlimited procedural, added the final details.
The same offline rendering techniques that enabled the VFX of the 90's and early 2000's could be utilized in GPUs today to push geometric detail to the next level in realtime graphics.
- gmueckl 6y agoMicropolygons work beautifully for local only shading, but that is a very limited model and a thing of the past. The milage that you can get out of this model that was designed for severely memory constrained systems (high resolution cinematic animations on computers with 4MB if RAM) is enormous. But once you bounce rays randomly through a geometrically complex scene, you either re-tesselate all possibly intersected geometry dynamically again and again with a comsiderable runtime overhead or you pre-tessellate the entire scene based on camera position and film resolution. The later approach is the basis for VFX at the moment, but some renderers like Renderman still can do dynamic on demand tessellation where required AFAIK.