No Difference in Illuminance Using Different Tree Models in Rtrace

Hello everyone,

I am a new Radiance user, and I am currently conducting daylight simulations to analyze the impact of trees on illuminance levels in an indoor setting. However, I am facing an issue where the illuminance values remain nearly the same when using two different tree modeling approaches, despite both having the same material properties.

Here are the details of my workflow:

Model & Scene: The study involves comparing two tree representations— simple sphere-based trees and voxel-based trees using photogrammetry —to assess their influence on illuminance values at various points. The trees are modeled with the same material properties (e.g., reflectance, transmittance).
Radiance Parameters: I have used -I -h -ab 7 -ad 2048 -aa 0.1 -ar 128 -as 32 for the simulations.
Illuminance Results: The measured field data and the corresponding simulated values using both tree models in the figure.

Issue: I expected a noticeable difference in the illuminance values between the two tree models due to their structural differences, but the results are quite similar. I am unsure whether this is due to how Radiance handles tree geometry, my material definitions, or potential issues in my simulation settings.
Could anyone provide insights into why this might be happening and suggest possible troubleshooting steps? Any help would be greatly appreciated!

Thank you!

Best,
kodi

Hi Kodi,

Welcome to the Radiance forum!

It is possible that there really is not much influence of the trees on your results. Or, it’s possible that your rendering parameters are insufficient to tease apart what difference exists.

One of my favorite debugging methods for this type of calculation is to generate views using a hemispherical fisheye view from the perspective of your measurement points. For example, you can use rvu for a point at XYZ=(10,5,2) looking up in the Z-direction with:

rvu [rendering parameters] -vth -vp 10 5 2 -vd 0 0 1 -vu 0 1 0 -vh 180 -vv 180 [octree]

This tells you what the illuminance point sees, and allows you to visually judge how much the trees outside matter.

Following that, if you still think there should be a significant difference, you can try increasing your parameters bit by bit. For example, I would think a smaller value for -aa (say 0.05) might give you a more accurate result.

Best,
-Greg

P.S. Make sure the other parameters for rvu match those of rpict by following the parameter list with “-defaults” in each case, since the default values are set very differently. Or, use the rad tool to set parameters for you, since it ensures that they match between interactive and batch rendering.

1 Like