Progressive path tracing and white materials

[PixelJockey]

I've got something odd happening with white materials using progressive path tracing. There are four boxes with gray, red, light gray and white Vray materials applied. The ground plane is also white. The white box in the foreground appears like it's self-illumniated when using PPT. There are no shadow details. Even if I lower the sky value from 255 to 128 or 50 there are still no shadow details. There are no lights in the scene - default lights are disabled. Only using skylight for illumination. Is this the way it's supposed to work?

Scene file (Max 7.1)

Wireframe


Rendered with irradience map / QMC


Rendered with Light cache / progressive path trace / sky = 255


Rendered with Light cache / progressive path trace / sky = 128


Rendered with Light cache / progressive path trace / sky = 50

[Paul Oblomov]

Make the white value 252

[vlado]

Never use perfectly white materials when using the light cache (with or without ppt). Perfectly white materials bounce away all the light that falls on them; since shadows are formed by light being absorbed by materials, they won't show up on the perfectly white cube.

In the Irradiance map + GMC GI case, the number of light bounces is limited. That means that a part of the GI solution is missing, and you get shadows.

The light cache (and ppt therefore) allow for infinite light bounces and produce the accurate GI solution for scene - and the accurate solution has no shadows.

Best regards,
Vlado

[cpnichols]

oh wow... that makes sense.... but it is a bit freaky.

[timmatron]

Phew! Saved me the trouble of posting that question. I started a new rendering yesterday where the IR/QMC looked great, but the light cache was overbright & washed out. Hopefully this explains the problem I was having...

[Frances]

Does this have anything to do with reflectance values of real-world materials? From my experience with radiosity, values that are too high make for unrealistic renders.

[Gijs]

I think you can either do two things: lower the colors to realistic values, or use a secondary bounce multiplier below 1 as some sort of 'global absorb setting'
IMO the last option is the easiest, because then you don't have to take care of it anymore in your materials.

[Geddart]

Perfectly white materials bounce away all the light that falls on them; since shadows are formed by light being absorbed by materials, they won't show up on the perfectly white cube.

I think this leads to a question here. Is it really realistic to have the "reflectance" of the material only calculated by it's rgb color?
Yes Frances, i think this has to do with reflectance values of real-world materials. In Lightscape you had these nice % graphs with Surface Reflectance values for different kind of materials. New white paint = 75%, New snow = 85% and New concrete = 40% Surface Reflectance, that kind of Stuff. I think it was not only based on color but also on the actual Surface Structure. (well, not the complex version...) The nice thing in Lightscape was, that the final Solution of the light calculation came out incredibly realistic with simple preset Values. The Bad thing that made the Workflow tedious though, was tweaking your Radiosity Mesh, well...
Hey PixelJockey, you can try to mimic this Workflow in Vray when you start to adjust the GI Multiplers although this can be tricky sometimes. (Especially when you turn in the wrong direction. ) I don't know exactly what Vray does with the Multipler Input, (Vlado?) i haven't tested it with Lightcache yet. Could one just use 0.x Values instead of Reflectance %?. IMO It's strange that not more People asked about it already, especially cause Vray is used by many Architects and former Lightscape Users.
Maybe the Way Vray Handles it right now is not yet optimal? (maybe reflectance directly in the Vray Material?)
Maybe i missed something?

regards