Material Null, Glass Homogeneous Test
Posted: Mon Sep 02, 2019 5:34 pm
After I got currently stuck with orange juice in a glass I required a few new inputs since material Glass is still miraculous for me.
First, the easy test: What happens if one uses material Null as inter-mediator? (Please open the image to be able to read the texts.) There are openCL hybrid (large images) and CPU BiDir (small images) renders. Labeled yellow cuboids overlap (right) with both glass cuboids (top, bottom). Yellow cuboids overlap with cuboids made from material Null (top, bottom) the latter cuboids overlap each with one glass cuboid.
As you can see in openCL renders there is not much change in appearance of yellow cuboids glass intersection. In contrast CPU BiDir renders appear differently, i.e. yellow cuboid intersection with clear glass cuboid (top) appears as if made from glass if inter-mediator Null is used.
Let us keep in mind what is observable
(a) Looking top-down through Glass cuboid (ior crown glass)
- Null, top is saturated yellow
- Matte, top is dark almost black
- Glass, top is dark and
- Glass, top is dark yellow with yellow border because of inter-mediator Null
(b) Looking top-down through Glass (ior water)
- Null, top is saturated yellow with a black, horizontal bar in the middle
- Matte, top is dark almost black
- Glass, top is dark and
- Glass with inter-mediator Null, top is medium dark yellow (views into yellow cuboid) with thin yellow border (Null material)
(c) Looking from front side to bottom through Glass cuboid (ior crown glass)
- Glass, clear view into yellow cuboid and
- Glass with inter-mediator Null, very clear view into yellow cuboid, view is brighter than without inter-mediator
At least one question rises:
Why are yellow cuboids' tops dark, see (a) and (b), especially with inter-mediator material Null too
Paths: Glass top - Glass bottom - [Null] - Material m, i.e. SS[N]m | [] in some cases
m=Glass S, Matte D | S Specular, D Diffuse, N Null
How cautious did you look at the figures in the pictures? Did you notice the difference in convergence per sample for Sobol and Metropolis?
First, the easy test: What happens if one uses material Null as inter-mediator? (Please open the image to be able to read the texts.) There are openCL hybrid (large images) and CPU BiDir (small images) renders. Labeled yellow cuboids overlap (right) with both glass cuboids (top, bottom). Yellow cuboids overlap with cuboids made from material Null (top, bottom) the latter cuboids overlap each with one glass cuboid.
As you can see in openCL renders there is not much change in appearance of yellow cuboids glass intersection. In contrast CPU BiDir renders appear differently, i.e. yellow cuboid intersection with clear glass cuboid (top) appears as if made from glass if inter-mediator Null is used.
Let us keep in mind what is observable
(a) Looking top-down through Glass cuboid (ior crown glass)
- Null, top is saturated yellow
- Matte, top is dark almost black
- Glass, top is dark and
- Glass, top is dark yellow with yellow border because of inter-mediator Null
(b) Looking top-down through Glass (ior water)
- Null, top is saturated yellow with a black, horizontal bar in the middle
- Matte, top is dark almost black
- Glass, top is dark and
- Glass with inter-mediator Null, top is medium dark yellow (views into yellow cuboid) with thin yellow border (Null material)
(c) Looking from front side to bottom through Glass cuboid (ior crown glass)
- Glass, clear view into yellow cuboid and
- Glass with inter-mediator Null, very clear view into yellow cuboid, view is brighter than without inter-mediator
At least one question rises:
Why are yellow cuboids' tops dark, see (a) and (b), especially with inter-mediator material Null too
Paths: Glass top - Glass bottom - [Null] - Material m, i.e. SS[N]m | [] in some cases
m=Glass S, Matte D | S Specular, D Diffuse, N Null
How cautious did you look at the figures in the pictures? Did you notice the difference in convergence per sample for Sobol and Metropolis?