[smeggy]

You could try sss with very small clipmap to simulate the effect of net curtains, that might give you a decent compromise.

[Mihai]

It's for a flower, the flesh eating one if anyone remembers it, it has "tentacles" that are pretty thick but they are only slightly colored, they look almost transparent.

edit:

Like in this pic


You see how some of the tentacles look almost transparent against a bright background?

[smeggy]

Oooo, col subject matter. I'd loveto see this one pulled off

[Aldaryn]

I think this is something what you're looking for

Exactly!
Thanks, I had no time this afternoon here to do further testing.

[Mihai]

More examples:

this white thin curtain:


or this one:


you get the picture




I wonder if in these cases it's not more practical to use a really fine clipmap texture? Wouldn't that increase the noise a lot?

[Aldaryn]

Well, the little strands in the curtains are pretty dense, but scatter the light, this is where cast shadows come from, and there is gap between the strands, this is where transparency comes from.

So, to mimic life, a clipmapped sss surface would be the best, along with a robust AA engine.

[Mihai]

ok, this image is a good example of what i mean, transparent rice paper:



Here I don't think we could use a dense clipmap, you see how it has varying transparency? It has virtually the same thickness but different density, thus making a blurrier or clearer transparency.

[smeggy]

This is a perfect example of where a true Alpha, not a clip-map would be really useful. Combined with SSS you could really do some cool stuff.

[Thomas An.]

Mihai,

I think your example with the curtains is 100% clip-map application.

In the real curtains light passes through tiny holes in between the fiber pattern.

Another thing. Do you have to use SSS ? From what you describe I think a thin dielectric sheet with medium UV rougness should do. No ?

[Aldaryn]

Thomas, I dont think shadows would appear on the dielectric surface.

Heres my test for a clipmepped sss plane:



20min, the first few passes were totally correct, after the blur was applied, much detail lost. Strange that the spheres seen through the plane get brighter and brighter as time passes. More samples lead to an odd solution?

With a denser clipmap its even worse.

[smeggy]

clipmaps really aren't suited to something so fine as lace curtains or sheer fabric, it just doesn't work.

[Aldaryn]

Real alpha map support would solve this problem. Or the chance to set the material density correctly for scattering.

Btw, this is the shader effect I would love to see done accurately by MaxWell:


!!Not MaxWell images!! sorry, but this is an easy way to show, what I was thinking of.

This is just a teapot in some SSS jelly.

[Mihai]

Mihai,

Another thing. Do you have to use SSS ? From what you describe I think a thin dielectric sheet with medium UV rougness should do. No ?

I'm making a test now with clipmaps, I'll see what happens. But for other things, like the flower image, we really need some density control. As soon as you deal with rays entering an object, it's natural to have a control over how dense that object is. Which is what I thought the scattering param would do actually.

Increase scatter param=more molecules for the rays to bounce off of=more chaotic scattering=more blur.

[Thomas An.]

With clipmaps, I would suggest you use the error diffusion method and adjust the size of the grain:
http://www.maxwellrender.com/forum/view ... 6099#26099
I know it works becasue I did some tests with curtains before (embroidered and plain) . If you apply a color map simultaneously then you get colored curtains.

I was getting an example ready for you (with curtains) but I found a bug now with the new beta and all. Not sure if it is a Rhinoll or Maxwell related.

But for other things, like the flower image, we really need some density control

True !

Also, in your shoes I would also test a dielectric with a 0.95-0.99 absorbance. Not sure if it works right...

To me SSS and dielctrics are (should be) inherently related. So this seperation of concepts confuses my idealistically.