Three Phase Matrix Method using irradiance

radiance-general
1

In understanding Subramaniam paper “Daylighting simulations with radiance using matrix-based methods”, and his accompanying document (https://www.radiance-online.org/learning/tutorials/matrix-based-methods), I have a question related to rmtxop:

  • My gendaylitmtx parameters includes -O1 for output so should be total solar radiance rather than visible radiance
  • In rflixmtx parameters includes -I+ for irradiance contributions are reported

Therefore in the rmtxop step,
rmtxop -fa -c 1 1 1 /res.dat | rmtxop -t - | getinfo - > radiation.tab
To convert into a single column matrix should the -c parameter be -c 1 1 1 or -c 0.33 0.33 0.33?
My thoughts are -c 1 1 1 would be no scaling.

Some advice on the -c parameter would be appreciated when using irradiance rather than illuminance.

Regards
Terrance McMinn

2

Hi Terrance,

I feel like this question has been answered in another context, but I don’t remember where. Anyway, it is a bit confusing but the latest version of Radiance makes it easy with the “-c A” option, which is equivalent to “-c 0.33333 0.33333 0.33333” for an RGB input. You can check out the rmtxop man page for details. Your command can in fact be shortened to:

rmtxop -fa -c A res.dat -t | getinfo - > radiation.tab

Placing the -t at the end applies the transpose following any other operations, but since the only other operation is channel averaging, which isn’t affected by the transpose, the -t option could have also come before the input file name.

I hope this helps.
-Greg

P.S. Sorry for the slow response - the discourse server was down and I am traveling.

3

Hello Greg,

Many thanks for your response.

In summary then, even when using irradiance input, you still need to use the -c 0.33333 0.33333 0.33333.

The 1/3rd factor in each of the RGB channels to nominally equal an overall factor of 1.

Regards
Terrance

4

Yes, you are always trying to come up with some kind of weighted average. If there’s a conversion factor as well (e.g., 179 lumens/watt) it’s best to think of that separately from the spectral averaging. RGB weights chosen usually arrive at photopic sensitivity, since that is the most commonly-used metric. There are of course others, such as scoptic and melanopic, which would use slightly different coefficients for R, G, and B. Spectral simulations with more than 3 channels still take the same approach, but with a longer list of weights.

Cheers,
-Greg

5

Thank you Greg for clearing this up.

Regards
Terrance