This study is an experiment with physically correct lighting, both direct and indirect. We have two lanterns (one slightly out of scene) illuminating an entirely enclosed room.
In the first pass, I used standard Maya blinn, lambert, and ansiotropic shaders with mental ray (MR) physical lights. Light intensity, falloff, photon intensity, and other parameters were varied endlessly without success with global illumination and final gathering on.
In the second pass, mental ray DGS material and photon shaders (physically correct in the amount of light they reflect) were attached to every Maya shader. Frankly, it was s a lot of work to do this, but the hope was to stabilize the light in the room, and indeed it did. Perhaps the DGS material shaders were not needed, but the DGS photon shaders definitely were.
In the third pass, the struggle was to control the shadows and its here I was not successful. Mental Ray area lights are the correct solution to softening the shadows, but render times were estimated at 60+ hours using 10x10 sampling. Since significant experimentation would be needed to find optimum area light size and sampling size (least the shadows be over-softened), I opted for point lights and a 10 hour render (light radius 0.2, shadow rays 7). Hardly satisfactory given the shadows you see now.
Shadows were also lightened using the mib_shadow_transparency shader on the lantern cloth. There is a dubious tradeoff between lighter shadows and a brighter room given the lantern cloth is transparent and lets light into the room. If there was time, shadow depth maps (as opposed to raytraced shadows) would have been a good avenue to explore.
The only bump network for DGS shaders is found on the tatami mats. Since these are nurb surfaces, its a matter of setting the correct select and selspaces value on the mib_texture_vector node. To avoid using the DGS specular channel for bump mapping, I used the mib_color_interpolate work around described in the v6 docs.
The glow from the fire in the hibachi was a last minute addition (an incandescent plane inside the hibachi), and no time was allotted for improvement.
The robe was done in Maya cloth and allowed to fall to the floor.
Some nurbs cubes (wood surfaces in the center) were micro-beveled using the round tool on the 12 edges of the cube.
The texture on the katana blade are projected images of a real blade, with DGS glossy adjusted for shininess.
The brightness of the lantern is disconcerting, but this could not be helped. In real life our eyes can compensate when starring directly at the lantern, here we cannot.