Cirrus tests.

[René]

Some experiments that should lead to a certain type of cirrus clouds.
Usually I pay less attention to clouds because of the longer rendering times, but in this case I can't avoid it because they will be the focal point.
This is what I have so far and I still have some problems to solve. And yes, the rendering times are soaring.

[Matt]

Alpine Fractal? I tried that once. Reducing the number of octaves helps a great deal with the render times. This is more important with the Alpine Fractal than with other fractals. Whatever your method is I would love to hear more about it.

[René]

Yes it is the Alpine shader, two even: one "soft" with little detail and one sharp with lots of detail. The sharp Alpine is through the Merge shader combined with a shifted version of itself to obtain extra sharp plumes. The sharp and soft Alpine are then combined again by means of a mask.

I hope this makes sense.

[Kadri]

Looks interesting René.

[Dune]

Indeed. Maybe a faster way is possible...

[cyphyr]

OOh, that looks like a cool technique.

[René]

A screen capture of a 3D version of these Alpine clouds. As can be seen on the cropped image with a longer RTP rendering time, higher quality settings are required to avoid layering.

Wouldn't it be possible to write a cloud density fractal that produces this result? How hard can it be?

[Matt]

Wouldn't it be possible to write a cloud density fractal that produces this result? How hard can it be?

If it uses the same algorithm as the Alpine Fractal then it will be the same speed.

Have you tried reducing the number of octaves as low as you can before it loses detail? The Alpine Fractal render times are non-linear with respect to the number of octaves, unlike the Power Fractal / Cloud Fractal, so it's critical to keep the number as low as you can.

It's probably best to bake these textures into a texture map if you're using them for clouds. The Alpine Fractal is 2D anyway.

[WAS]

I could try using Voronoi Ridges subtracted by Voronoi Billows.

You could use a surface layer and vector disp to give windswept look, and you can also warp by the same voronoi noise on X/Z to create a more stretched look.

Here is a quick example. RTP is crippling my system for some reason though. Could hardly move the mouse to pause it just to return to the browser. :O

I also feel with how faint the cirrus are, that V3 may be overkill, to get the edges looking right you need a pretty low density/sharpness which kills v3 lighting imo.

[Hannes]

Interesting thread, guys! Cool clouds.

[René]

I could try using Voronoi Ridges subtracted by Voronoi Billows.

You could use a surface layer and vector disp to give windswept look, and you can also warp by the same voronoi noise on X/Z to create a more stretched look.

Here is a quick example. RTP is crippling my system for some reason though. Could hardly move the mouse to pause it just to return to the browser. :O

I also feel with how faint the cirrus are, that V3 may be overkill, to get the edges looking right you need a pretty low density/sharpness which kills v3 lighting imo.

Your Cirrus looks good. I've experimented with Voronoi billows and ridges, but my version, unlike yours, looked too obvious as a fractal. I think the solution might be a combination of Alpine and Perlin noise.

[René]

Alpine and Perlin noise.

[Dune]

These look good! I've also experimented with taking a stretched noise through a surface shader, breaking (the greys) up by a 90º rotated, warped and reduced version of itself, then warping the whole lot. But never ended these experiments...

[René]

What I would really like to make is clouds like these. But that might be too ambitious.

[René]

If it uses the same algorithm as the Alpine Fractal then it will be the same speed.

That makes sense.

The Alpine Fractal is 2D anyway.

If the Alpine shader is 2d, does that mean that it is basically useless for 3D clouds? That would be a pity, because the Alpine shader does exactly what you see on the edges of some clouds.