[w]tech - Physics

That looks like three-dimensional moss. Rockin' textures! How much would it cost to get a computer that could run that. Could a medium-end laptop run it real-time, assuming it's a demo?

It isn't that hungry. I'm always making sure that it still runs on the old Geforce 8400GS Go my laptop has in, next to my GTX570.
Of course you have to turn some effects off, but it still works.

The video above was captured on a HD4890 by the way, in Full HD. Unfortunately Nvidias HBAO is very costly. I'm working on an option for slower cards.

Also this is not a demo. It's the game engine we will use for our upcoming titles.

Thanks for replying. I'm very interested in mathematics and computer graphics.

Update: I watched your demos. I can't believe we've come to the level of per-polygon collision! You are one smart programmer.

Looks good! Question: what's special about per polygon collision? I understand what I would call "traditional" collision detection (i.e. SAT, etc.), but I haven't heard of per polygon collision before. Does it have something to do with how collision detection against the terrain mesh is done?

Oh, Perpoly-collision has been around for quite some time now. AFAIK every modern physicsengine supports it, and so does Bullet, which we are using.
Also the perpoly-collisions you are seeing are all static and can't move. Actually the Bulletphysics-Manual suggests to make such static objects collide per poly( If they aren't a heightmap terrain, Bullet got it's own hulls for that).

Aside of that, dynamic perpoly-collision does exist as well, just not that fast as for static objects.