Need some help tuning Radiance parameters

[delrocco]

EDIT: found this post, but it’s not helping me

Radiance Parameters for Daylight Analysis

HI! I am setting the radiance parameters for a daylight analysis of a simple shoe box with an exterior ground. While setting the radiance parameters for the simulation I noticed that some parameters are somehow offset for the notion I had of theirs maximum. But since I guess that @sarith set it, I think that somehow I am missing something. Since my scene has a maximum dimension of 30 meters and I set the -aa 0.1, the -ar 128 is more than enough for this case. Because I need accuracy but I do …

I’m doing an imageBasedSimulation + runDaylightAnalysis to compute a render for a small office:

I have this problem (spotty light):

RADParameters _quality 0 doesn’t have this problem.
RADParameters _quality 1 or 2 (with other parameters the same) does have this problem.

I started by modifying the parameters as per Radiance suggestions:
http://radsite.lbl.gov/radiance/refer/Notes/rpict_options.html

I’ve tried so many parameter permutations that I’m losing my mind!
I cranked _ad_ to 10000 and _as_ by at least half that, but same problem.

Current “not highest, just trying to get rid of this problem” parameters are:
_quality = 1
pt = 0.1
ds = 0.25
aa = 0.15
pj = 0.9
dt = 0.25
ad = 512
dj = 0.5
dp = 256
lw = 0.01
xScale = 2
yScale = 2
ar = 128
as = 256
dc = 0.5
av = 0
lr = 6
ps = 4
st = 0.5
sj = 0.7
dr = 1
ab = 8

PS: What is the -c parameter talked about on this page?

Ladybug Tools | Radiance Image Parameters

I can’t seem to find that documented elsewhere.
Should it be passed into additionalIP_?
What should the ranges be?

EDIT: from these docs posted by @sarith , is -c the pcomb parameter?

rtcontrib_lesson.pdf

http://radsite.lbl.gov/radiance/pcomb.1.html

[delrocco]

Now I’m not overriding any parameters at all.

If I set RADParameters _quality to 0, it succeeds…

pt = 0.15
ds = 0.5
aa = 0.25
pj = 0.6
dt = 0.5
ad = 512
dj = 0
dp = 64
lw = 0.05
xScale = 1
yScale = 1
ar = 16
as = 128
dc = 0.25
av = 0
lr = 4
ps = 8
st = 0.85
sj = 0.3
dr = 0
ab = 2

But if I set RADParameters _quality to 1, the simulation fails…

pt = 0.1
ds = 0.25
aa = 0.2
pj = 0.9
dt = 0.25
ad = 2048
dj = 0.5
dp = 256
lw = 0.01
xScale = 2
yScale = 2
ar = 64
as = 2048
dc = 0.5
av = 0
lr = 6
ps = 4
st = 0.5
sj = 0.7
dr = 1
ab = 3

[delrocco]

_simulationType_ 0 (illuminance) does work with _quality set to 1:

But not luminance.

[sarith]

Hi Joe, can you try -ab 3 -aa 0.05 -ar 1024 -ad 1024 -ps 1 -x 300 -y 300 in additional parameters and see if it improves anything? If it does, you can crank up the resolution -x 1000 -y 1000 and set -ps 2 in the next run.
If it doesn’t change anything, I’d probably have to look into the model to figure out what is going on. The discussion on rcontrib is not relevant to what you are doing (your simulations are not pure Monte Carlo).

(PS: I wont be able to look into this at least tomorrow afternoon)

[delrocco]

No problem! Thx very much for the reply!

I’ll try both now and reply my results…

[delrocco]

Ok… those parameters did keep it from failing during a luminance _simulationType_ this time. However, the results were roughly the same (spotty light problem).

I figured it must be something with my geometry, since some of it was just imported and not optimized. So I stripped the office of nearly everything, except the shell, glazing, and outside ground surface. But looks like I still have the same problem with default RADParameters and _quality of 1.

Here is all the information I can think to give to debug it. Let me know if there is anything else I can help with.

USA_FL_Orlando.Executive.AP.722053_TMY3.epw
Climate Based Sky
7/1/2017 @ 0800
Ladybug files for this office

Putting the mullions back in between the glazing panes produces the same problem:

[delrocco]

I should also note that using those extra additionalP_ produces the same problem:
-ab 3 -aa 0.05 -ar 1024 -ad 1024 -ps 1 -x 300 -y 300

I am by no means an expert at Radiance tool chain (obviously), so this probably doesn’t mean much, but I noticed differences in the cmd line parameters in the first rpict call vs the 8 separate core calls. Here are captures of the cmd line parameters when it kicked off with default RADParameters + _quality 1 + those additionalP_ above.

First cmd:

Core cmds:

[sarith]

Hi @delrocco,

@delrocco

I am by no means an expert at Radiance tool chain (obviously), so this probably doesn’t mean much, but I noticed differences in the cmd line parameters in the first rpict call vs the 8 separate core calls. Here are captures of the cmd line parameters when it kicked off with default RADParameters + _quality 1 + those additionalP_ above.

The first calc preps up the simulation by creating an ambient file. So, thats fine. More details here. And if you want to get further indepth info, see this (do a text search for “overture”).

With regards to the spotty light thing, the issue is that your surfaces are specular (when they shouldnt be).

If you set the values for “spec” to 0 in your plastic materials (I guess its grass?), it will work out fine.
mod plastic id
0
0
5 red green blue spec rough

[delrocco]

So… looks like it’s the materials I used. Either the way I’m using them or they are too reflective. If I remove the setRADMaterials component where I specifiy materials, the results are fine. If I change the order of setRADMaterials and addHBGlz it doesn’t make a different; the problem still exists. So it is likely the material of the wall is too reflective…

[delrocco]

Yea that was it… what a waste of time lol I was using that color picker and perhaps didn’t notice when I copied out the results.

Thanks so much for your help @sarith !