Reflected sun spot is missing from fisheye image in relatively lower resolution

In the scene shown below, each of the 4x12 reflective surfaces (hypothetical PV panels) was titled purposefully so that the sun at a given time point is reflected right towards the same viewpoint.

  1. However, quite a lot of the reflected sun spots were missing from several surfaces when rendered in the resolution of 800x800:

  2. Some of the missing spots were “recovered” when the resolution was increased to 1000x1000, with a few still missing:

  3. They were all shown correctly only when the resolution was increased to 2000x2000:

  4. Singling out the surface which failed to show the sun reflection in step 1&2 didn’t seem to work, either, when rendered by 800x800 and 1000x1000:

If surface luminance levels are the key references for visual comfort analysis, such as glare probability evaluation using Evalglare which prefers an image of 800x800, the missing of the bright spots may lead to significant underestimation of the brightness across the field of view, not to mention annual analysis which may rely on efficient generation of large amount of images with relatively smaller image size, isn’t it?

I suspect the Radiance material properties for the reflective surfaces and Radiance parameters might be factors of this issue, but I haven’t test it yet.

I’d appreciate your advice if you have encountered similar issue before.

Thanks. (710.3 KB)

Does setting -ps 1 in rpict change anything ?


Thank you, @sarith

Yes, setting “-ps 1” for rpict solves this issue.

Fixing the image size at 800, decreasing “-ps” from the default 4 to 1 increased the simulation time by about 9%, which is definitely affordable considering that the sun reflection is correctly shown. This is the case even when the image size is reduced to 500x500 as shown below, and the simulation time is reduced by 9%.

1 Like

Just out of curiosity, what is the glare mitigation strategy here in case the panels are placed where they are shown?

This reminds me of a paper by Alstan in which a similar problem at a flight control tower was discussed. Here it is. jakubiec,reinhart_assessing-disability-glare-potential.pdf (2.1 MB)

Thanks for sharing, @devang Yes, Alstan told me his PhD research is related to impact of PV induced glare around airport. Will read his paper you shared.

@sarith I’m exploring how to model PV glare accurately. Other than Alstan’s approach, I’m reading on the SGHAT approach developed by Sandia Lab which refers to retinal irradiance as the criteria for glare evaluation. Key references by HO et al. are listed here, and the key paper is this one. Mitigation strategies will be explored in parallel, hopefully…

Currently, I’m exploring the following: 1) apply BSDF file for typical PV module product in Radiance, if its available so as to simulate the reflectance of PV panel accurately as indicated by Brotas, and 2) calculate retinal irradiance through Radiance.

Your advice are most welcome.

On the issue of using BSDFs for modeling PV panels, you might want to subdivide the panels. The Klems patches act funny if the BSDFs are assigned to large polygons.

just curious, how do we show the sun spot with Honeybee 0.063 and 1.4.0 ?
For example, in HB 1.4.0, I have tried to set the sky type as type 0, i.e. +s, sunny with sun. However, there is no sun spot even my viewpoint towards the solar position.