SSS Weird Results/Colours

[abvfx]

I posted this on the official forums and got no luck, still be playing with it and yet still no solution. I thought i would give it a shot here.

This model has true displacement on, polys are rendered as subdivision and diffuse intensity is set to 0.2 with a colour of white. The point cloud generated is set to about 1000 points.

The lower eyelids render really dark and certain points read as green along with an area below his neckline that looks like it has lost specular or sss ?

Little confused here . Maybe when colour textures are in place it will be different, right now im just testing.

Any help as always much is appreciated.

[talashonak]

Hello phrenzy84!

I have similiar issues with SSS when the model not closed on the bottom. You need to use polycap node with triangulate option switched on.

So, here results of such operation:

Render before:

After applying node:

About dark and green areas - try to play with lights, attenuation density and scattering phase. On the model above I have density at 20 and phase at 0.1, environment light and two area lights.

Edited May 29, 2012 by talashonak

[abvfx]

As always there is usually a simple solution. If only i didnt delete the body when generating the Point Cloud and Rendering i wouldnt have this problem. hehe oh well.

As for the colour problems I am playing more with the density & scattering options, and feel you just have to find a happy balance. but i will try your settings out. They are decent results. I thought you had to make sure scattering phase was set at zero to allow both back and forward scattering together but i can clearly see both in that render. Great job.

Thanks so much.

-andrew

[old school]

When using the fast sss single scattering method, the sss shading model causes Mantra to construct what you would call a volume representation of your object. This is why not closing the base of your head caused the artefact as the geometry wasn't closed and the internal volume representation wasn't constructed properly for the open neck region.

This is similar to adding the vm_volumeuniform "Uniform Volume" property to a shader which then renders the incoming geometry as a volume after displacements are run on the surface. See the example file in case this is new to you as it was added in Houdini12.

See the attached file: render_time_volumes_with_Volume_Density_property.hip

Knowing this, you can do funky stuff with your skin and single fast sss. I don't know if this is useful but you can offset your skin inward with a Peak SOP and then clean up the self-intersections (PolyDoctor SOP? or manually smooth with Sculpt SOP) and merge together and render with sss. What you will find is that the volume constructed to perform the sss will be that thin shell between the two surfaces. Again I haven't found a use yet but that's the way it works if both surfaces are manifold (water tight).

As for the sss parameters exposed on the Mantra Surface shader, the Subsurface Color is the starting color of the ray as it penetrates the surface, not the diffuse surface color as that is what diffuse color is for. If this is a bit of skin, it could be a reddish color. The Attenuation Color is the color that the ray will tint towards as it penetrates the volume representation of your skin. The deeper it penetrates, the more it tints. This means that the Attenuation Density determines how quickly the Attenuation Color is reached. If your Subsurface color is beige skin color but the Attenuation Color is red, then looking at each of red, green and blue channels individually, you may see some odd tints in the sss output.

To play it safe, you can set the attenuation color to a white to a warm white and then dial in the Attenuation Color, set it more to a purplish type of color always previewing the render with the appropriate extra export image planes in ipr.

It is interesting to add a point geometry override to the Attenuation Density inside the Surface Model VOP (unlocked of course) and then paint a varying value to the surface that represents the thickness of the effect or the distance to bone or whatever material under the skin would block all scattered light. Obviously this could be represented as a texture map but what better way than to displace the head geometry to simulate sub-dermal depth.

Along with modifying the Attenuation Density, creating texture maps for Attenuation Color with the supplied Attenuation Map are important (not done it my file).

See the attached example file as a hipnc file as I can not find credit for the head model.

older_head.hipnc

[abvfx]

Thanks for the info old school. Really helpful stuff.

I wish i made the model watertight now , I made the inside of the mouth, throat, tongue and teeth as a separate model so i could use it on other characters.

I will try to play with some of the 'funky stuff' you mention. I think painting the Attenuation Density as you mentioned would be essential, to account for fingers and ears.

Big thanks for a step in the right direction.

edit://

Quick Update, ive unhidden the main model, its watertight from the neck down, and polycapped the eyes and mouth cavities (with triangulation on). This is with the default mantra surface shader.

Still dont understand why the nose is weird, the nostrils are watertight.

Edited May 30, 2012 by phrenzy84

[talashonak]

So, if you closed them , may be these areas are too thin for your scatter radius (attenuation density)? Also try to make render without subdivision of surface.

[abvfx]

I think i have reached a balance (with the help for some temp textures), but i still need to work on keeping textures sharp. (They are being gamma corrected if that is important).

Edited June 5, 2012 by phrenzy84

[talashonak]

Very nice result! Only single SSS or multi? What settings and light setup?

[abvfx]

Very nice result! Only single SSS or multi? What settings and light setup?

I feel you need both Single and Multi scattering to be effective. As for the settings, i will post them but it is nothing crazy. I did turn the Single intensity down, something around 0.6-8. Normal front, back lighting setup, with a hdr for additional support.

I will probably do a wip because its gonna take some time to get everything working right, with the different maps (diff, subscat, attn, bump, disp, reflect, gloss), eye shaders, and im doing the hair with fur.