I am modelling a 3DP-printed object made of transparent plastic in Radiance. I want to obtain a BSDF description using genBSDF to compare to a data-driven one.
To do this, I created a geometrical model in Rhino where I modele the layer-by-layer resolution and exported it to Radiance. Then, I apply a dielectric material to it and generate the BSDF using genBSDF and the Klems resolution.
I made different attempts at tuning the genBSDF parameters (-c, -ab, -ad,…) and the dimensions of the sample material; however, I always obtained the same strange results:
- the BSDF is not symmetrical, though the geometry is symmetrical
- Simulation and experimental results start to deviate significantly for high incoming light elevation angles (e.g. >50°) where my simulated BSDF shows transmission values greater than 100%.
Does anyone have experience with similar modelling challenges?
Thanks in advance,
If you have modeled the material as pure dielectric with no absorption, then it is not physical and rounding errors during Monte Carlo sampling could lead to slightly over 100% transmission. I recommend you enter some non-zero absorption coefficients corresponding to your substrate if you can find such information.
As for the asymmetrical distribution, you should share your genBSDF command options. It is probably just a sampling accuracy issue, though it may be that the type of output itself (i.e., Klems or tensor tree) is responsible.
It also occurs to me that you need to make sure your surfaces are oriented correctly for the dielectric material. Depending on the software used to create the model, this can be tricky.
You can check surface orientation by using the Radiance “glow” material, which will have a constant color on one side and be black on the back side. If any black surfaces are visible from the outside of your construction, these are probably reversed normals. You can try using rvu to visualize, or create several renderings from all sides using rpict.