2

Radiance options for direct sunlight calculation

I'm using rcontrib to calculate sunlight hours for a grid of test points. Each sun position is created as a light source modifier and then I use rcontrib -M modifiers.sun [options] *.oct < *.pts > *.dc to calculate the daylight coefficient. I tried to use rigorous radiance options such as -ab 0 -aa 0 -ad 65536 -lw 1.52e-5 -dt 0 -dc 1 but radiance is still missing ~3% of suns in the scene. I also tried turning on and off -u.

This image shows one of the cases. Black spheres show the place with inaccurate values. I draw sun vectors from one of the test points and in this case Radiance is missing the vector which is parallel to the wall.

sunlight hours

You can read more here.

Mostapha Roudsari's avatar
1.1k
Mostapha Roudsari
asked 2016-04-28 16:12:56 -0500
__AmirRoth__'s avatar
4.4k
__AmirRoth__
updated 2017-04-26 14:05:51 -0500
edit flag offensive 0 remove flag close merge delete

Comments

add a comment see more comments

1 Answer

4

It looks like the errors occur on the boundary between shaded and unshaded. The default for rcontrib is -dj 1 which sends a ray to a random location within the angle subtended by the sun, but you are probably comparing this to rays sent to the center of the sun. Using -dj 0 will hopefully bring the two into agreement.

Andyrew's avatar
790
Andyrew
answered 2016-04-28 19:56:20 -0500
edit flag offensive 0 remove flag delete link

Comments

@Andyrew thank you for your help. I wish I could do more than +1. It would have taken me forever to figure this out. Changing dj to 0 did it. My follow up question is does the rest of options make sense in the way that I have it set up? Anything that I should have considered? For instance I never really noticed dj until now. Anything else that I'm missing here? Thanks in advance.

Mostapha Roudsari's avatar Mostapha Roudsari (2016-04-28 20:47:21 -0500) edit
1

You might not have noticed dj because the default is 0 for everything except rcontrib and rfluxmtx. I mainly use it when I want to render realistic shadows, but it requires substantial over sampling.

The most important parameters for this type of simulation are dc 1 and dt 0. I've been told that setting both is redundant, but I always use both anyway. I can't think of any other important parameters. I don't know if ad aa and lw have any role when ab is 0.

Andyrew's avatar Andyrew (2016-04-29 09:51:07 -0500) edit

Thanks Andy! I feel a little bit better now. Now I can finalize the parameters for the recipe.

Mostapha Roudsari's avatar Mostapha Roudsari (2016-04-29 11:22:55 -0500) edit
add a comment see more comments