Slashdot Mirror

← Back to Stories (view on slashdot.org)

NVIDIA Shakes Its Flowing Mane With Life-Like HairWorks 1.1 Demo

Posted by timothy on from the only-her-hairdresser-knows-for-sure dept.

MojoKid writes: Previously, you might not have thought much about a wig on a manikin, but checking out NVIDIA's latest tech demo, as a gamer or 3D graphics artists, hair can be pretty interesting. The video is of NVIDIA HairWorks 1.1, a simulation and rendering tool for creating lifelike hair and fur in video games. In the clip, NVIDIA shows off a Fabio-style hairdo with about 500,000 hairs that bounce and sway as the camera circles and forces move the hair. If this was a real wig, it might unseat one of the most boring videos ever. However, as an example of what modern 3D graphics can do with hair physics, it's pretty darn cool. Previous demos of HairWorks showed up to 22,000 strands of hair, making the jump to half a million much much more significant. The video was recorded with ShadowPlay on a GeForce GTX 980, which has some serious muscle, though it's not the most powerful card in NVIDIA's lineup. What's cooler than making life-like human hair? Putting flowing manes on vicious monsters, of course. Apparently, NVIDIA HairWorks simulation technology also plays a role in bringing more than a dozen creatures to life in The Witcher 3: Wild Hunt.

3 of 61 comments (clear)

  1. My idea for HairWorks 2.0 by HideyoshiJP · · Score: 3, Interesting

    I think for HairWorks 2.0, they should simulate that beautiful mane getting tangled, because we all know you don't have much time for hair care when you're fighting orcs or dungeon creatures.

  2. Dirty hairs by manu0601 · · Score: 3, Interesting

    Now can they do dirty hairs dynamics?

  3. Still using guide hairs by Anonymous Coward · · Score: 2, Interesting

    The simulation still uses guide hairs. You can still recognize the hair strands following the individual guides around. This is one of the most prevalent artefacts of hair simulations to date and I'm a bit disappointed that it's that obvious in the demo. However, they do seem to render each hair seperately, which is a lot of effert and potentially a ton of overdraw. And they need to do some kind of dynamic depth sorting - probably per pixel. That's where it actually gets impressive.