First of all, findglare and glareindx are fairly old programs in the Radiance suite. Since their publication, an entire glare metric (DGP) and tool for calculating DGP (evalglare) have emerged and been added to the Radiance distribution. DGP is more focused on daylight glare (hence the 'D' in DGP), so perhaps you're going more of an electric lighting glare study, but wanted to make this point.

Assuming you are indeed looking at an electric lighting glare evaluation, I'd say based on these warnings you have an issue with the input HDR image, your chosen area of interest (angles in the -ga option), or both.

The person best qualified to answer these questions is Greg Ward but I'm not sure he monitors this forum as frequently as he (and the rest of the Radiance community) monitors the Radance Discourse forums here: https://discourse.radiance-online.org/

Maybe try reposting your issue there...

There is a bug in the dgi calculation of glarendx when sources are behind the viewer. Raphael Compagnon wrote this particular formula, and I am not sure what is wrong with it. I will try writing to him about it again. The 24% missing samples has to do with an insufficient view as Rob points out.

I also recommend looking into the newer evalglare program.

P.S. to the above. I have checked in a fix to the glarendx program, which will need to be recompiled from source until the distributions update. You can find & download it from the CVS repository https://www.radiance-online.org/cgi-bin/viewcvs.cgi/ray/src/util/glarendx.c?view=log

Yes, I got something similar! Does anyone know what's happening? Is it because I am using Wndows 10?

-60.000000  1.#QNAN0
-50.000000  1.#QNAN0
-40.000000  1.#QNAN0
-30.000000  1.#QNAN0
-20.000000  1.#QNAN0
-10.000000  1.#QNAN0
0.000000    1.#QNAN0
10.000000   1.#QNAN0
20.000000   1.#QNAN0
30.000000   1.#QNAN0
40.000000   1.#QNAN0
50.000000   1.#QNAN0
60.000000   1.#QNAN0