I have been running a simulation in Radiance for a room with both an artificial lighting system and daylight, using a consistent set of calculation points. I used the following rtace command for lighting simulation
The .oct files are l7_whole.octl7_daylight.oct, and l7_led.oct.
I noticed that the results of the full simulation ( including LED and daylight) are consistently about 55 units lower than the sum of the results from separate simulations (one with only LED and one with only daylight).
Are there any adjustments to the rtrace command that might alleviate this difference?
55 units out of how many? What is the percentage difference?
The likely issue is inadequate settings with -aa 0. While -aa 0 is a fine choice, you should alter other default values at the same time. I suggest trying:
Using the new rtrace -pY option, you won’t need the rcalc filter on the output, as it accomplishes the same thing. (The “-ov” option is the default, so I left it off.)
Regarding the 55 units – they represent the average illuminance difference computed with the same .pts file for the different simulations mentioned above.
Can you let me know where I can find the manpage for the new rtrace -pY option? I installed the Radiance 6.0a (2025-03-14), but I haven’t been able to locate the updated documentation for rtrace.
It looks like the online manual page is still on the last official release. The new manpages are included with the distribution in the ray/doc/man/man1/ directory. You can copy these pages to a standard man page location such as /usr/local/man/man1, put the /full/path/ray/doc/man folder in your MANPATH environment variable, or use them directly with:
With the same points grid (points_whole.pts), the average illuminance of the full simulation ( including LED and daylight) is consistently about 55 units lower than the sum of the results from separate simulations (one with only LED and one with only daylight).
I also adjusted the values of -lw :
when the value increased from 1e-4 to 1, the difference increased;
When the value decreased from 1e-4to1e-20`, the difference stabilized around 55.
It seems that the discrepancy remains as a fixed bias.
here is a fixed bias.
Would you have any insights on what might be causing this issue?
I would still like to know what fraction of the average 55 represents. Is it 1%? Is it 10%? Is it more? Knowing the difference in values doesn’t tell me much.
The only other suggestion I can offer is to set “-dt 0” as this can affect the result in scenes with many light sources. Since you have daylight in one scene and not in the other, the scene with daylight could be under- or over-estimating the contribution from your LED lights.
Sorry for the incomplete reply. The 55 represents 2% of the average. The full simulation result is always 2% lower than the sum of separate simulations.
Sorry for the delayed reply, there is an error in the computer I am running Radiance. I did try it again with the rtrace -dt 0 option, but the result is the same. My concern stems from consistently observing about a 2% lower value. However, as you say, a 2% difference is not that large and can be considered a bias.
2% is not a lot in this sort of calculation, but it would be nice to understand it. As you say, it appears to be a consistent bias. The -as calculation is known to be biased, so you could try switching that off (-as 0) as a last attempt to resolve the difference.
Thank you so much for your suggestion. I truly appreciate it.
I set the -as 0, and the result is the same. Maybe because I didn’t have enough experiments; I only ran it 20 times, and currently, the computer I am running Radaince has some other errors. Once I have updates I will show it here.
My only other suggestion is that you upload the two versions of your scene, somewhere we can get to them, along with your points file and your commands and results. This will let me reproduce the differences you are seeing and hopefully find out their origin.