Integrated photon mapping and material roughness

Hello experts,

playing around with the new integrated photon mapping I met the following strage effect.

First I am using the old version (Radiance 3.7 and Photon map patch 4.3.1) and this very simple geometry (sorry for the german notes):

# LED als IES-Verteilung

# strahlt in Richtung der x-Achse

!xform LED.rad | xform -ry -90 -t -0.2 0 0

# Umlenkspiegel als Metal

void metal spiegel

0

0

5

0.9 0.9 0.9 1 0

!gensurf spiegel mir ?-0.25+s*0.5? ?-0.25+t*0.5? ?0? 1 1 | xform -ry -135

# Testflaeche

void plastic weiss

0

0

5

0.8 0.8 0.8 0 0

!gensurf weiss test ?-4+s*8? ?-4+t*8? ?-2? 1 1

A caustic map is generated like this: mkpmap -apc map2 1000000 -dp 50000

And based on that rad file:

# Rad Input File created by trad Son Aug 2 11:57:29 CEST 2015

DETAIL= Medium

EYESEP= 1

INDIRECT= 1

OCTREE= Dir_Pmap.oct

PENUMBRAS= True

PICTURE= Dir_Pmap

QUALITY= Medium

RESOLUTION= 1024

UP= Z

VARIABILITY= High

ZONE= Interior -4 4 -4 4 -2 0.5

render= -apcb map2 100 2000

scene= System_Pmap.rad

view= g1 -vtl -vp 0 0 -0.5 -vd 0 0 -1 -vu 0 1 0 -vh 8.5 -vv 8.5

I get this output: https://dl.dropboxusercontent.com/u/27495200/HGF.jpg

If I change the roughness for the "spiegel" material to 0.1, I get this output: https://dl.dropboxusercontent.com/u/27495200/RoughF.jpg

which seem to be quite Ok for me.

Now I change to the radiance HEAD from the 4th July. I use the same geometry and the following rad file:

# Rad Input File created by trad Son Aug 2 11:59:05 CEST 2015

DETAIL= Medium

INDIRECT= 0

OCTREE= Pmap1.oct

PCMAP= map3 1000000 50

PENUMBRAS= True

PICTURE= Pmap1rough

QUALITY= Medium

RESOLUTION= 1024

UP= Z

VARIABILITY= High

ZONE= Interior -4 4 -4 4 -2 0.5

mkpmap= -dp 50000

scene= System_Pmap.rad

view= g1 -vtl -vp 0 0 -0.5 -vd 0 0 -1 -vu 0 1 0 -vh 8.5 -vv 8.5

With the roughness of 0.0 I get this output: https://dl.dropboxusercontent.com/u/27495200/HGnF.jpg

Which, compared to a run with mirror material, seems to be even better than the old result.

But when I change to the roughness 0.1 I get this output: https://dl.dropboxusercontent.com/u/27495200/RoughnF.jpg

I already tried differend setting for the bandwidth parameter but the were no fundamental chages.

May thanks for any ideas

Martin Klingler

Hi Martin,

this is indeed weird, and resembles irradiance cache extrapolation artefacts, even though I understand ambient rays aren't even traced with INDIRECT=0, nor do they have to be if there are no other diffuse surfaces, so none of that should be going on.

Is the observed pattern (which resembles a Fresnel lens) from the LED's IES file? That would explain why the "jaggies" are also visible on the left edge of the mirror, where no photons reside. What worries me more is that the
lighting levels are too high.

The code has changed substantially in the course of this year, so it's hard to pinpoint the root cause offhand. I'll try reproducing this with a "point" source to check what's up. Failing that, could you send me the LED definition with IES file for testing?

Thanks & best regards,

--Roland

ยทยทยท

On Sun, 02 Aug 2015 12:21:22 +0200, Martin Klingler <[email protected]> wrote:

Hello experts,

playing around with the new integrated photon mapping I met the following strage effect.

Now I change to the radiance HEAD from the 4th July. I use the same geometry and the following rad file:

But when I change to the roughness 0.1 I get this output: https://dl.dropboxusercontent.com/u/27495200/RoughnF.jpg

I already tried differend setting for the bandwidth parameter but the were no fundamental chages.

--
Dr. Roland Schregle
Senior Research Associate

Lucerne University of Applied Sciences and Arts
School of Engineering and Architecture
CC Envelopes and Solar Energy (EASE)
www.hslu.ch/ccease

T direct: +41 41 349 36 26 (New)
[email protected]

Technikumstrasse 21, CH-6048 Horw (Correspondence)
Ebenaustrasse 20, CH-6048 Horw (Office)
T +41 41 349 33 11, F +41 41 349 39 60