You could try interpolating between two pregenerated noise textures in HLSL. Just generate two textures with random noise and lerp between those over time in a pixel shader. I'm using this for my cloud rendering and it seems to work out just fine for that.

I tried googling for "midnight club menu", but there seem to be entirely to much midnight dinners out there, so I don't know if this technique is of use to you ;)

Quote:Original post by remigius
I tried googling for "midnight club menu", but there seem to be entirely to much midnight dinners out there, so I don't know if this technique is of use to you ;)

Your googling powers are weak young jedi [lol]

Midnight Club II Screenshots, including this one that might be the menu:



Although, I can't see any noise in that image though [oh]

Anyway, regarding the original question...

I've seen some tricks similar to what remigius described - sample from multiple different noise textures using different resolutions and animated offsets. I'm not sure, but you could probably compute some sort of simple noise in a lengthy VShader and interpolate it using the pixel shader (possibly mixing in some "static" noise from a texture)...

hth
Jack

Quote:Your googling powers are weak young jedi

Thanks, Obi Jack Jolly [wink]

Quote:Nice suggestion, I've tried it before. It looks too... "repeatish"

Well, for my clouds I actually use 3 textures. Two textures are used by the shader at any given point for the interpolation, giving me about 2 seconds to fill the 3rd one with some new noise. Once the interpolation is finished, I swap the textures and interpolate again... Like this:

1) [1] -- Int(0.5) --> [2], [3] being filled2) [1] -- Int(1.0) --> [2], [1] is unused so swap and put in texture [3]3) [2] -- Int(0.0) --> [3], [1] starts being filled4) and so on...

Using the typical perlin noise octaves (3 will do) will further enhance the visuals.

Alternativly, you could implement a Perlin noise function in HLSL. I've been thinking about that too, but never got around to doing it since the approach above worked fine for me. Nevertheless, this article describes a noise function that seems simple enough (details in article near "Making your noise functions"):

  function IntNoise(32-bit integer: x)			     x = (x<<13) ^ x;    return ( 1.0 - ( (x * (x * x * 15731 + 789221) + 1376312589) & 7fffffff) / 1073741824.0);      end IntNoise function

Whatever you decide to do, the look of plasma will probably also benefit from decent interpolation techniques and the exponentional cut-off which is commonly used for clouds. The article describes these thorougly, so it proved a must-read for noise stuff for me [smile]