I believe it's much better performance-wise to have traditional model with LODs than to rely on OpenSubdiv tesselation. You can make a very detailed retopology and it will still have much less triangles than smoothed OpenSubdiv. I can think of a few cases where it can be useful but generally It's a waste of GPU power.
We'll get there eventually. I fully expect opensubdiv and displacement maps to replace what we use now some day. I think the next big technology push will be in dynamic light rendering though.
leleuxart
i just worked with the guy that made the in game subdivision pilot helmet from that cod demo a while back, if you're referring to that.
he said it was a pain in the ass and took forever so set up so the uv's didn't break when it divided. dealing with uvs on sub-d geo isn't the funnest to begin with.
wish i had more info, he's not here anymore, but if were chatting ill ask about it.
and yea i hope count vertsalot is right, better lighting (and alpha for fuck sake) is really needed. pbr is getting all the hype right now, hopefully that shifts into the tech needed to support pbr better.
Replies
i just worked with the guy that made the in game subdivision pilot helmet from that cod demo a while back, if you're referring to that.
he said it was a pain in the ass and took forever so set up so the uv's didn't break when it divided. dealing with uvs on sub-d geo isn't the funnest to begin with.
wish i had more info, he's not here anymore, but if were chatting ill ask about it.
and yea i hope count vertsalot is right, better lighting (and alpha for fuck sake) is really needed. pbr is getting all the hype right now, hopefully that shifts into the tech needed to support pbr better.