Snow surface

If you don''t want any volumetric techniques, just a textured snow surface, then a BRDF pixelshader would be the perfect way to simulate the lighting behaviour of such a surface. The only problem is to get a good BRDF for snow. Your best bet is to find some procedural approximation of the multispectral reflectance found on ice crystals. If you have access to a goniometer, then you could ofcourse also measure a real snow surface...

Thanks Yann,

though I haven''t found any approximation on the net. Does every good and complete household have to had a goniometer?

Any additional ideas? Any ideas how to simulate the wind ripples?

Thanks alot.
- Stephan

---

OpenMountains | Open source snowboard simulation - some day.

quote:

---

though I haven't found any approximation on the net. Does every good and complete household have to had a goniometer?

---

We had one when I was at uni. It was used to measure prismatic refractions. It was very expensive, and I don't think they'd appreciate it if some students were going to take it outside in the snow

Seriously though, I'm pretty suer there must be academic papers about the optical properties of snow surfaces.

quote:

---

Any additional ideas? Any ideas how to simulate the wind ripples?

---

Well, still under the assumprion that you don't like a volumetric approach, I think the BRDF approach would look best, in terms of realism. Wind ripples can be simulated by procedurally animated bumpmaps: use some wind model (or some over fluid dynamic model) on a heightmap texture, load that texture as bumpmap onto the snow surface. On highend hardware, you could even use this heightmap in a displacement map, giving you true 3D wind ripples (perhaps a LOD scheme ? Near the camera = displacement map, further away = bumpmap, very far away = just the BRDF alone).

Adding some glare to parts of the snow (where it is exposed to the sun, as a function of the per-vertex reflection vector) might increase the realism (use the alpha buffer to store per-vertex interpolated glare values, blur it, add it onto the final rendered image).

Oh, here are two papers about snow synthesis I found on the web:
The Synthesis of Snowcovered Terrains
Geospecific rendering of alpine terrain (PDF)

Interesting stuff. Very tempting to experiment with

/ Yann

[edited by - Yann L on January 10, 2003 7:29:19 AM]

Can''t forget about those surfaces where the snow doesn''t quite cover the whole ground.. the grass blades stick out of the snow, or the snow is blown so that between the ripples you can see dirt. I suppose you could use an alpha map of where the snow may lay that interpolates between the snow and the original ground, and render the whole scene with the techniques discussed above. But doing an alpha map for a whole terrain is not feasible and you couldn''t account for various snow heights. I guess if you wanted to show snow along with ground you''d have to go volumetric so that you could see piled up snow against a rock while there''s a thin layer stretched across the dirt, and in some places no snow at all.

Thanks again Yann.

I know these 2 paper, i especially found the geospecific rendering of alpine terrain interesting, but they provide far too less information how they do the classification. And it's not suitable for closeup i think.

I go with BRDF rendering, though i have to read into it.

okonomiyaki:
I assume on glaciers and the like there's not that much grass , but just out of curiousity, what does it take if i want to go volumetric? All my traditional texture blending won't work anymore, will it?

cheers,
Stephan

---

OpenMountains | Open source snowboard simulation - some day.

[edited by - stephanh on January 10, 2003 10:30:58 AM]

I''ve found an old screenshot of Transworld Snowboarding. The snow looks pretty fluffy but it''s definitely done by an artist and not generated procedurally.

http://www.eisscholle.de/20011119_182629_transworld2.jpg

---

OpenMountains | Open source snowboard simulation - some day.

Nishita has papers on snow (online) and sand dunes with wind ripples (in a Japanese journal).

http://nis-lab.is.s.u-tokyo.ac.jp/~nis/


[edited by - greeneggs on January 15, 2003 2:11:54 PM]