Fastest and best looking realtime fur..IMO.

[broli]

http://cgi3.tky.3web.ne.jp/~tkano/

That's an old demo of a realtime fur implementation. I came across it some years ago when I was doing research in fur shaders. It is still to this date the best RT fur implementation I've seen and it runs even on a gf3 ti (obviously the amounts of shells matters) and supports angular momentum. In my case the demo doesn't run anymore on my ati mobility 9600 with latest drivers since it used to run on an older driver.

I'm pretty sure fur will be a big challenge for overgrowth, but I hope this will help even a little bit.

[Jeff]

Doh, doesn't work for me either.

[Ozymandias]

Ya, that's too bad. Are you planning on putting fur on rabbot or something?

[Skofo]

Ya, that's too bad. Are you planning on putting fur on rabbot or something?

Well, most rabbits have fur, don't they?

Also, idea for the full game: make rabbot an easter egg character.

EDIT: Direct link http://gfs.austindev.com/amd/GPU/zip/to ... .13.02.zip Doesn't work for me under WINE.

[broli]

Oke I just installed some old Ati driver (7.11) and it works now. I recorded it with fraps which brought down the fps more than half. The main problem with shell fur is obviously the count since it multiplies the amount of polygons of the model by the shell amounts. Obviously some distance based shell LOD can be easily achieved.

Here's the recorded video and some still shots:
http://vimeo.com/3003255

[Ozymandias]

Wow, that looks great. it's shiny and stuff. I wonder if you can make it less shiny to show raggedness...

[Viking Zippy]

Also, idea for the full game: make rabbot an easter egg character.

Not only that, but I've been playing with a11, and the grab function is ridiculously powerful (though oh-so-much fun!), so when it's scaled down for the final, I'm thinking there should be an option for the original power somewhere; it could be very handy for getting around the map in debug mode, as is the flying in that's turned on by default in Lugaru when debug is enabled.

Oke I just installed some old Ati driver (7.11) and it works now. I recorded it with fraps which brought down the fps more than half. The main problem with shell fur is obviously the count since it multiplies the amount of polygons of the model by the shell amounts. Obviously some distance based shell LOD can be easily achieved.

Here's the recorded video and some still shots:
http://vimeo.com/3003255
http://ziosproject.com/NJ/FSth1.PNG

That does indeed look awesome; if something like that could be integrated into Overgrowth, it would be crazy-awesome

[Johannes]

Oke I just installed some old Ati driver (7.11) and it works now. I recorded it with fraps which brought down the fps more than half. The main problem with shell fur is obviously the count since it multiplies the amount of polygons of the model by the shell amounts. Obviously some distance based shell LOD can be easily achieved.

Here's the recorded video and some still shots:
http://vimeo.com/3003255
http://ziosproject.com/NJ/FSth1.PNG

looks niiiice. would love to see that on some rabbots.

[BunnyWithStick]

It's so… Huggable! I want to hug a shader. What has the world come to? Or more importantly, what have I come to?

[Alorwin]

The source is included, so anyone with coding-skill could in theory figure out how to pull the furshader from it and make an OpenGL shader, I suppose.

But keep in mind, this is DirectX, Overgrowth is OpenGL. Different languages.

[Skofo]

But keep in mind, this is DirectX, Overgrowth is OpenGL. Different languages.

I'm sure that it's a portable algorithm.

Also, **APIs (sorry).

[marmorek]

Looks wonderfully fuzzy. You could probably use a variation of it for grass. That would look great.

[Johannes]

Looks wonderfully fuzzy. You could probably use a variation of it for grass. That would look great.

if you read back to the early blog posts by david, he used a grass shader for the first version of the fur.

[marmorek]

Looks wonderfully fuzzy. You could probably use a variation of it for grass. That would look great.

if you read back to the early blog posts by david, he used a grass shader for the first version of the fur.

Yeah, that's what I was thinking of.

[Brad]

Not sure why you didn't just link to the video I recorded a few years ago of the Tomohide technique, Najib. Here's a direct link to the demo on the ATI site (no registration needed) plus another video I captured of the same demo/approach added to the pot (bit closer views). Video has a wireframe comparison as well.

[Demo] [Video]

From what I comprehend, the idea overall is to use a dynamically updated offset map. Various attributes are stored in each channel based on velocity/gravity/dampening calculations of the object per vertex before finally relating everything over to tangent space. The map is then knocked over to the second step which does a standard shell approach. However, during this step it uses previously generated texture data more or less as a normal map as well as to control how the layers respond.

There's actually more involved to it than that, but you can find all the information in the briefly documented .doc file included with the zip. I've implemented my own fur shader based on the idea, but I more or less just opted with a standard shell shader while passing in angular velocity per frame from movement. I went with a simple bezier, gravity vector, and a damp float to control fur curvature, but the result was nowhere near as impressive without the self-shadowing and specular effects.

I may try and toss it what I did at least over to the Overgrowth engine to see how it fairs. Haven't looked yet, but does Overgrowth use GLSL or CG shaders?