ISL Questions

[CTZn]

How do I use pos for bump mapping ?

Can't do eval(vec3 pos) real.

[OnoSendai]

Hi CTZn,
I'll add a posWS() function soon that will return the world space position when available.

[CTZn]

You seem to imply that I could use posOS() already, wich makes sense for wooden objects. I'll have to try this, thanks !

[CTZn]

You seem to imply [...]

Ah no you didn't

[CTZn]

Is it possible to compute the maximum range returned by a fbm ? At iteration 1 it is ~ [-0.83 : 0.83], at it8 absolute is above 1.1. It's not easy to get everything into the 0:1 range since min and max are variant, depending (mostly) on the iterations. I guess I need to know a (simplified) ratio that adds to intensity with iterations.

Could I also have a working example for the print statement ? I thought I would print the fbm values as they go evaluated but couldn't figure it out. Thanks.

[CTZn]

Practically I'd need a fbm within the 0:1 range to apply curves on it.

[OnoSendai]

Hi CTZn,
Base Perlin noise returns values in the range [-1, 1].
FBM with one octave is just perlin noise.
FBM with infinite octaves should lie in the range [-2, 2], but most values will be near the zero than near the extreme values I guess.

Printing should work, the only problem is that it prints to stdout. Try running the console indigo

[CTZn]

The problem with extreme values is when they produce clipping, my concern is vertical

I'll play more with the normalise functions eventually, maybe that's just what I want.

Thank you for the tip Sir !

[Borgleader]

If I may add a question to this thread, is it possible to emulate normal map functionnality through ISL?

[CTZn]

Perhaps someone like galinette could make use of a normal map data while dealing with bump mapping, but otherwise I think emulation would require the writing of certain parameters that are not exposed in Indigo despite a request (would allow NPR shading).

I am not sure on that topic though, I may be mixing facts... for the matter I am bumping your question

[galinette]

It is mathematically possible do derivate the normal map from a height map, by calculating the gradient vector of the height map. That's actually what renderers do.

However, it's not possible to compute a height map from a normal map in general : arbitrary normal map has a non-zero rotational vector, which means it cannot be derivated from a height map by a gradient.

If the normal map has a uniform zero rotational vector, it's theoretically possible to compute a height map, but this require numerical integration which would be a non-local calculation... not possible in a simple shading language, and always computationally intensive.

If your normal map is a texture (not a shader), there are maybe tools that could compute a height map from it (provided the fact that they will need to remove the rotational component of it if non-zero, thus modifying the appearance of the material)

Try this : http://www.crazybump.com/

Etienne

[Borgleader]

Thanks I'll try that