Sign in to follow this  
lipsryme

Specular model for human skin

Recommended Posts

lipsryme    1522

I know a few samples that are using the kelemen-szirmay kalos brdf that uses a precomputed beckmann distribution texture (e.g. nvidia skin rendering sample & more recently Jimenez SSSS sample).

I've read a bit through their original paper describing the specular model but I'm not seeing anything to the like of "use this when rendering skin"...

Does anyone know why this model is used for realistic skin rendering ?

 

As far as I can tell their brdf is just a beckmann NDF + the usual schlick fresnel + their geometry term approximation of cook torrance.

I'm already using a custom BRDF model that is trying to be physically correct (GGX distribution + Schlick fresnel + the kelemen szirmay-kalos geometry approximation described in hoffman's physically based shading 2010 paper (1 / LdotH²), which is somehow different to the one described in the KSK brdf paper. 

 

My question now is if there's any benefit to using their brdf over mine as a realistic model for human skin ?

Edited by lipsryme

Share this post


Link to post
Share on other sites
Styves    1792

The main idea behind the paper's use of the Kelemen-Szirmay BRDF is physically based specular. Since you already have a pretty good physically based BRDF, I don't think there's any need to use the KSK model for skin (especially if you're using deferred). A PBR model is all you need. smile.png

 

The only reason behind it is that skin has a lower specularity when facing it directly, but at grazing angles it gets much brighter. Basically, fresnel is very important to achieving realistic skin.

Edited by Styves

Share this post


Link to post
Share on other sites
lipsryme    1522

Alright thanks for the input. That's what I was wondering about. It's quite a coincidence though that both skin rendering samples use the same brdf for specular reflections. I guess Jimenez also just looked at the nvidia sample rolleyes.gif ...

Share this post


Link to post
Share on other sites
Styves    1792

Jimenez's tech is all based on the nvidia stuff IIRC. It originally started with doing the convolutions in screen space. Eventually he managed to simplify it to a 2-pass approximation by applying the colors to the weights of the blur. When you really think about it, it's nothing all that special. :P

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this