# gensky -u not uniform?

Hi Greg,

I tried to generate a hemisphere with a completely uniform luminance distribution and wanted to use gensky for this, since it offers the option -u (uniform sky). But it turns out that the distribution is not uniform.
Am I doing something wrong?

gensky 6 12 21 -b 55.86592178771 -u

Gives a zenith-luminance of 10000 cd/m2 (was the goal) and close to the horizon around 6500 cd/m2.
Also without the -b option the distribution is not uniform.
Any idea?

Thx!
Jan

···

--
Dr.-Ing. Jan Wienold
Ecole Polytechnique Fédérale de Lausanne (EPFL)
EPFL ENAC IA LIPID

http://people.epfl.ch/jan.wienold
LE 1 111 (Office)
Phone +41 21 69 30849

Hi Jan,

Ah, a trip down memory lane -- I think I first came across this in 2002. Gensky applies a mixing function around the horizon to `blend' the brightness of the sky and whatever might be below, e.g. usually the glowing ground but it's still applied if there's nothing below the horizon. I guess this was for purely visual appearance. A small change to skybright.cal should get rid of the blending at the horizon:

Change:

skybr = wmean((Dz+1.01)^10,
select(A1, sunnysky, cloudysky, unifsky, intersky),
(Dz+1.01)^-10, A3);

to:

skybr = if( Dz, select(A1, sunnysky, cloudysky, unifsky, intersky), A3);

That should fix it.

I didn't want to alter how skybright.cal is used, so I created another file (imaginatively) called skybright_nomix.cal. At the time, I'm sure Greg must have nudged me in the right direction.

Cheers
John

John Mardaljevic PhD FSLL FIBPSA
Professor of Building Daylight Modelling
School of Architecture, Building & Civil Engineering
Loughborough University
Loughborough
Leicestershire
LE11 3TU, UK

Tel: +44 1509 222630 (Direct)
Tel: +44 1509 228529 (Secretary)

[email protected]

http://www.lboro.ac.uk/departments/abce/staff/john-mardaljevic

Personal daylighting website:
http://climate-based-daylighting.com

Hi Jan,

John is on spot with the reason for the non-uniformity near the horizon as well as the "fix." For the record, the blending is not just for visual appearance; it's there to avoid sampling artifacts caused by unnatural discontinuities. This can result in noisy as well as erroneous output.

You should still get a uniform sky using "-u" without "-b", but for your given date and time, the sun is likely below the horizon, so the sky would naturally be zero.

Cheers,
-Greg

···

From: John Mardaljevic <[email protected]>
Date: November 19, 2017 5:25:18 AM PST

Hi Jan,

Ah, a trip down memory lane -- I think I first came across this in 2002. Gensky applies a mixing function around the horizon to `blend' the brightness of the sky and whatever might be below, e.g. usually the glowing ground but it's still applied if there's nothing below the horizon. I guess this was for purely visual appearance. A small change to skybright.cal should get rid of the blending at the horizon:

Change:

skybr = wmean((Dz+1.01)^10,
select(A1, sunnysky, cloudysky, unifsky, intersky),
(Dz+1.01)^-10, A3);

to:

skybr = if( Dz, select(A1, sunnysky, cloudysky, unifsky, intersky), A3);

That should fix it.

I didn't want to alter how skybright.cal is used, so I created another file (imaginatively) called skybright_nomix.cal. At the time, I'm sure Greg must have nudged me in the right direction.

Cheers
John

+++++++

From: Jan Wienold <[email protected]>
Date: November 19, 2017 4:22:40 AM PST

Hi Greg,

I tried to generate a hemisphere with a completely uniform luminance distribution and wanted to use gensky for this, since it offers the option -u (uniform sky). But it turns out that the distribution is not uniform.
Am I doing something wrong?

gensky 6 12 21 -b 55.86592178771 -u

Gives a zenith-luminance of 10000 cd/m2 (was the goal) and close to the horizon around 6500 cd/m2.
Also without the -b option the distribution is not uniform.
Any idea?

Thx!
Jan

PS. Jan reminded me (off list) that, of course, if you *only* want a uniform luminance hemisphere, then there is no need to use gensky. However, should you want an *exact* representation of any of the CIE types (i.e. without any blending of the luminance at the horizon), then the suggested fix will work.

By the way, for quantitative work, say, simulating illuminance values using CIE skies created by gensky, this luminance blending at the horizon will almost certainly *not* have any significant effect for anything but weird scenarios. Thought I'd better mention that

Back in 2002, I was setting up a `weird scenario': mapping the CIE clear sky pattern onto a hemisphere of finite size to predict the effect of parallax in sky simulator domes. Since some of the sensor points were nudging right up against the (finite extent) horizon, the effect of the blending was noticeable. That (rather arcane) study is described in a 2004 paper:

http://journals.sagepub.com/doi/abs/10.1191/1365782802li055oa

Cheers
John

"For the record, the blending is not just for visual appearance"

I should have guessed

oops... obviously there was an update of my email client and the reply-button is not automatically replying to the discussion group any more...
of course I wanted to share it to all.
below is what I wrote.

cheers
Jan

···

Hi John,

thanks! That explains it! I was a bit confused that uniform was not uniform, but it makes sense not to have a drastic change between sky and ground.

The easiest way for a uniform hemisphere is of course not using skybright.cal at all and use directly 10000/179 for the three channels of the glow material... I realized this while walking outside under a nice mixed winter sky.

thanks again!
cheers

Jan

On 11/19/2017 07:33 PM, John Mardaljevic wrote:

PS. Jan reminded me (off list) that, of course, if you *only* want a uniform luminance hemisphere, then there is no need to use gensky. However, should you want an *exact* representation of any of the CIE types (i.e. without any blending of the luminance at the horizon), then the suggested fix will work.

By the way, for quantitative work, say, simulating illuminance values using CIE skies created by gensky, this luminance blending at the horizon will almost certainly *not* have any significant effect for anything but weird scenarios. Thought I'd better mention that

Back in 2002, I was setting up a `weird scenario': mapping the CIE clear sky pattern onto a hemisphere of finite size to predict the effect of parallax in sky simulator domes. Since some of the sensor points were nudging right up against the (finite extent) horizon, the effect of the blending was noticeable. That (rather arcane) study is described in a 2004 paper:

http://journals.sagepub.com/doi/abs/10.1191/1365782802li055oa

Cheers
John

_______________________________________________