View Full Version : multi-layered skin model

02 February 2006, 08:11 AM
I read about a multi-layered skin model from down address
I understood kinda what he tought it
but the problem is i can't guess at all about how to make it
so.. i wish you guys read that paper
and tell me about it.. a little

anyway in last of the paper there is 7 layer images and a perfact image..
how did he make it? with his own program? maya or xsi?
and 7 layers....
how to use that for last image?
in compositing? or put those layer in some shader and render?
i have no idea what i'm thinking now..

so,, me please

wish you have good day

03 March 2006, 01:55 AM
I know Max8 Has A Sub Surface Scattering texture type built into it... but otherwise i would guess you can use an extremely complex set of falloff maps? Hmm... just my 10 cents.

03 March 2006, 09:53 AM
This paper is basically an advancement on jensen's original method in his paper, "A Practical Model for Subsurface Light Transport". To be honest, the differences don't really seem to be worth the extra complexity, and most of the realism in the images he shows comes from the fact that he's using a high-resolution model and texture scan. The renders were most likely done with jensen's own renderer, Dali.

The seperate component images are inputs to the shader, which then calculates the skin colour based on those parameters. The method is detailed in the paper if you can fight your way through jensen's rather opaque writing style.

For a more practical multi-layered shader, the misss_fast_skin shader is a good bet.

03 March 2006, 10:39 PM
I donīt think that you should think about a special package like Maya or XSI.
Try to think of a Software package as two parts. There is a part where you manipulate things on a low level of abstraktion. That means things are very close to the real world. You create geometry, tweak lights, or connect nodes to manipulate a flow of information.
Thatīs what a GUI is for.

The other part is the internal mechanic of the software, how things are calculated, itīs based on a high level of abstraction, symbols, mathematics. To understand the paper you have to go to that level of abstraction. The reason why software is rarely mentioned in papers is, that itīs not the most important thing.

More important than he modeling or animation package is -in my opinion- the renderer. A good render package is usually well documented, and it should enable a more or less advanced user to transfer the theoretical model described in a paper to a practical model. My favorite in this case would be renderman, beceause there is a lot of code avaible on the net, and itīs extremly good documented. You should read some of the SIGGRAPH course material on renderman, especially the newer ones. There is a lot about Subsurface Scattering and Skin Shading, most of it on the background of real feature-films like Hollow Man, or Finding Nemo. I think this will give you a completely different perspektive.

As far as I know the other package used in feature-film productions is Mental Ray. I heard that V-Ray is used more often, but I donīt believe it yet.

Anyway, I recently read some stuff about the Lafortune-Shader. Itīs not as complex as a Subsurface-Shader, but Itīs capable of simulating some aspekts of skin, and a lot faster. The advantage of Lafortune is that you can feed it with measured reflection data, wich is also aviable on the web. A friend showed me some tests, and It looked jawdropping considering the render time.

CGTalk Moderation
03 March 2006, 10:39 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.