[req] maps for transparency and for sss color...

[Kosmokrator]

Ono i dont know if is in your todo list this features but its very cool and usefull to have this stuff in indigo! thnx

[OnoSendai]

Well, i've thought a little about it.
The main problem is how to keep it physically realistic.
For example, imagine a thin object with a image map determining the absorption coefficients. How will the coefficients inside the object be effected by the map?
etc...

[Kosmokrator]

if i understand well with uvw map cords.....as i can think about its easy to assign diferent color absorption coefficients by uv mapping...i dont know how dificult can be to code this but if maxwell have this feature its possible at least!

[OnoSendai]

Let me explain some more...
The problem is how to determine the correct value for the absorption (or the throughput along a given path) at a given point inside an object, given the maps across the surface of the object.
Some kind of depth-based weighting scheme would probably be needed.

[Kosmokrator]

hhmmm... now i understand.....but i think need more research about this...thnx Ono...search about...worth it!

[oodmb]

this is definitly possible. i have seen something like it at least with pbrt(i think) with a multicolored sss dragon. this might be easier with a procedural texture however. possibly if you could implement a way to turn an image into a vector based image like inkscape

[CTZn]

Yes but projecting textures onto a volume is just ugly, you are ending with spacially linear or concentric strokes of values. You could use two or more textures, but 3d procedurals are the key I would say !!!

[oodmb]

well, if no method currently exists, i am sure there are enough scientific readers of these forums that we can colaborate with and develop a method. if not, well, i'm sure somebody will figure out a way. in the mean time i had the idea of allowing a surface texture that changes the incoming and outgoing color of the ray or either one of these. the effect would be that the color would bleed or be blurry, but thats not too bad.

[CTZn]

Sorry I'm not a scientific but...

Are you talking about a system based on 2d textures ? By interpolating 3 textures one can recreate a true 3d texture, so IMHO everything is about 3d textures (also called procedurals, but not all procedurals are volumetric textures), at the end.

A 3d noise function is something mathematically very simple I think, at least the most common kind of noises. At every moment the renderer asks for, they can provide a value in space. Such data can be used to define... well, what you want: absorbtion, ior, cauchy etc.

That's one of the reason why I'm requesting for dynamic parameters write access in the API... yes it's official

Your idea about blending colors along a ray is feasible I think, and I'm quite curious of the result cause I never saw that; it may create nice soft volumetric effects. But I think that the most difficult aspect of coding a true 3d noise function will be knowing the coding language

[IanT]

It would be trivial to simulate uniformly-coloured glass with a transparent, textured coating painted on the outside (simply multiply the entry/exit radiance) and, if viewed from the outside, would probably look accurate. I'm guessing many trinkets or ornaments like this exist in real life.

For glass objects with internally varying absorptivity, I'd agree with Ono ... using an image to base it on would be tough. (I'd also guess that no objects like this exist in real life either). More realistic would be using a colour-banded perlin noise volumetric texture, perhaps with a bit of turbulence added, and march the ray through to determine how much light is absorbed. I might be wrong but I think that POV-Ray is already capable of this.

Ian.

[oodmb]

For glass objects with internally varying absorptivity, I'd agree with Ono ... using an image to base it on would be tough. (I'd also guess that no objects like this exist in real life either).

have you ever realy looked at jade? or for that matter any other sss material? the idea that any of them are truly uniform is rediculous. Jade or any other sss stone is filled with veins. skin too is filled with veins, although of a completely different sense. translucent wood has rings.

as for simulating it by having a painted coat on the outside, yah it would be a good and easy solution, but it is by no means accurite. the paint on the outside of the trinkets you are thinking of is actualy another layer of varying SSS.

now that i think about it however, one method of doing this, although not realy a great solution, would be to have the absorbtion and scatter properties linked to each ray, so that a ray that enters one part of the texture behaves with those appropreate throught that medium.

[IanT]

have you ever realy looked at jade? or for that matter any other sss material? the idea that any of them are truly uniform is rediculous

Umm, I think my post was a bit misleading and/or badly written ... what I was trying to say was that volumetric procedurals would be much more suited to "painting" the inside of an SSS shape than UVs (because real-world objects' internal characteristics rarely have any relationship to the pattern that happens to exist at the surface ... the surface pattern will correspond to wherever "man" decided to cut or polish it ).

Ian.