mkillum usage

[Marija_Velickovic]

Hi,

I've been thinking about mkllum usage and parameters for mkillum
calculation. I've explored mailing list and my private archives and also
RWR book, and come to some conclusions, but also some questions.

All notes below are for few types of daylight simulations, where windows
with/without fenestration are (not)treated with mkillum.

···

########

*Luminance images calculation*
Based on RWR page 351
“For image generation, the conversion of a window to a light source using
the mkillum utility can significantly speed up the production of smooth
renderings. The technique works well as long as the total number of
secondary light sources is kept reasonably small.”

“For those rare occasions when images need to be rendered using high value
for -ab, it may be best to avoid using secondary sources and rely on
ambient calculation.”

Classical usage of these ideas is usage of mkillum either manually or
through rad program, for window surfaces.
Mkillum parameters:

   - -ab 2 is usually sufficient if building exterior is not too complex.
   If illum surfaces should seal some complex shading system (Venetian blinds)
   this parameter should be >2.
   - other parameters - default values can be used because they are
   sufficiently high
   - -aa 0 can be used when mkillum is used for Venetian blinds output
   calculation
   - -av r g b is set to:
      - if glazing is between building exterior and interior - av should
      be set to *“ground ambient value*” obtained from gensky command.
      - if glazing is between exterior and interior for upper floors in the
      building - use the same value as for bottom floors
      - if glazing is between two interior areas - either use 0 0 0 (which
      will always underestimate values) or use the compamb script with interior
      views to come up with a good setting

*Question:* If outside environment is heavily obstructed, as it is usually
the case in urban areas, how should we determine -av value for mkillum
calculation? I think that simple usage of 'ground ambient value' for -av
would overestimate luminance inside of the building.
Rad parameters:
Beside setting of ambient parameters it is good idea to experiment a little
and find nice set of -d* parameters.
Example of parameters: -ab 3 -av 0 0 0 -ad 1024 -as 256 -aa 0.1 -ar 600 -ps
1 -ds 0.05 -dj 1 -dt 0.03 -dc 0.9 -af tmp/floor1_illum.amb

###
*Illuminance images calculation*
Based on RWR page 351:
“For illuminance calculations, where -ab >2 is usually essential, the
preprocessing of windows to secondary light sources is generally not
recommended.”

Illuminance calculations are needed for DF or Autonomy calculations.

*DF calculations*
Generally this approach is used in dayfact script, where illuminance on a
workplane is calculated with rtrace -I option, and rather high parameters
are recommended eg: -ab 5 -av 0 0 0 -ad 2048 -as 512 -aa 0.1 -ar 600 -af
tmp/amb_file.amb

*Question:* Why it is not recommended to use mkillum for illuminance
calculations?
I've tried to include mkillum for DF calculation on workplane, and set
parameters with similar manner like for luminance images - and internal
illuminances were much higher then without mkillum (with same rtrace
parameters for the image).

So I need some answer if for illuminace calc approach with mkillum gives
wrong results, or parameters should be set differently.

*Autonomy calculations*
These are also illuminace calculations, but since numerous time moments are
included, mkillum doesn't have any sense to be used.
For this we can use 3-phase method, or Daysim approach - and calculate with
high ambient parameters.

###

So if we say it all in short: what is bothering me the most is -av setting
for mkillum, because "ground ambient values" seem to high for real project
cases where there is complex environement outside of the building.

Any info about list members' experience with mkillum and daylighting is
welcome.

Thanks in advance,
Marija

[Greg_Ward]

Hi Marija,

From: Marija Velickovic <[email protected]>
Date: February 21, 2012 5:17:07 AM PST

Hi,

I've been thinking about mkllum usage and parameters for mkillum calculation. I've explored mailing list and my private archives and also RWR book, and come to some conclusions, but also some questions.

All notes below are for few types of daylight simulations, where windows with/without fenestration are (not)treated with mkillum.

########

Luminance images calculation
Based on RWR page 351
“For image generation, the conversion of a window to a light source using the mkillum utility can significantly speed up the production of smooth renderings. The technique works well as long as the total number of secondary light sources is kept reasonably small.”

“For those rare occasions when images need to be rendered using high value for -ab, it may be best to avoid using secondary sources and rely on ambient calculation.”

Classical usage of these ideas is usage of mkillum either manually or through rad program, for window surfaces.
Mkillum parameters:
-ab 2 is usually sufficient if building exterior is not too complex. If illum surfaces should seal some complex shading system (Venetian blinds) this parameter should be >2.
other parameters - default values can be used because they are sufficiently high
-aa 0 can be used when mkillum is used for Venetian blinds output calculation
-av r g b is set to:
if glazing is between building exterior and interior - av should be set to “ground ambient value” obtained from gensky command.
if glazing is between exterior and interior for upper floors in the building - use the same value as for bottom floors
if glazing is between two interior areas - either use 0 0 0 (which will always underestimate values) or use the compamb script with interior views to come up with a good setting

Thanks -- this is a really nice summary of what we know. It would be great to include in a FAQ if we ever get our act together...

Question: If outside environment is heavily obstructed, as it is usually the case in urban areas, how should we determine -av value for mkillum calculation? I think that simple usage of 'ground ambient value' for -av would overestimate luminance inside of the building.

You are probably right that it should be set lower. You can always set it to zero (the default) and accept that you will underestimate values slightly. Otherwise, you can render a fisheye image just outside the window setting "-ab 0 -dv-" and compute the average pixel value with:

  pvalue -h -H -o -df fisheye.hdr | total -if3 -m

That's pretty close to what compamb does.

Rad parameters:
Beside setting of ambient parameters it is good idea to experiment a little and find nice set of -d* parameters.
Example of parameters: -ab 3 -av 0 0 0 -ad 1024 -as 256 -aa 0.1 -ar 600 -ps 1 -ds 0.05 -dj 1 -dt 0.03 -dc 0.9 -af tmp/floor1_illum.amb

In daylighting situations, the calculation shouldn't be very sensitive to the -d* settings. You can set -dt 0 if you're worried about it, though.

###
Illuminance images calculation
Based on RWR page 351:
“For illuminance calculations, where -ab >2 is usually essential, the preprocessing of windows to secondary light sources is generally not recommended.”

Illuminance calculations are needed for DF or Autonomy calculations.

DF calculations
Generally this approach is used in dayfact script, where illuminance on a workplane is calculated with rtrace -I option, and rather high parameters are recommended eg: -ab 5 -av 0 0 0 -ad 2048 -as 512 -aa 0.1 -ar 600 -af tmp/amb_file.amb

Question: Why it is not recommended to use mkillum for illuminance calculations?
I've tried to include mkillum for DF calculation on workplane, and set parameters with similar manner like for luminance images - and internal illuminances were much higher then without mkillum (with same rtrace parameters for the image).

So I need some answer if for illuminace calc approach with mkillum gives wrong results, or parameters should be set differently.

It's just that mkillum doesn't pay off if you are only calculating a small number of points. The only reason to use it is to reduce the error when you cannot afford a difficult indirect calculation over a million or so pixels. If you are only computing a few hundred illuminance values to begin with, you are better off simply cranking up the indirect calculation parameters and letting rtrace deal with the window complexity.

Autonomy calculations
These are also illuminace calculations, but since numerous time moments are included, mkillum doesn't have any sense to be used.
For this we can use 3-phase method, or Daysim approach - and calculate with high ambient parameters.

###

So if we say it all in short: what is bothering me the most is -av setting for mkillum, because "ground ambient values" seem to high for real project cases where there is complex environement outside of the building.

Any info about list members' experience with mkillum and daylighting is welcome.

Thanks in advance,
Marija

Thanks for your input, always appreciated.

-Greg

[Marija_Velickovic]

Hi Greg,

After years of work with Radiance, I can always learn something new

Related to FAQ, is Radiance Wiki still active, or is there some other
knowledge database related to Radiance (beside list members

*Question:* If outside environment is heavily obstructed, as it is
usually the case in urban areas, how should we determine -av value for
mkillum calculation? I think that simple usage of 'ground ambient value'
for -av would overestimate luminance inside of the building.

You are probably right that it should be set lower. You can always set it
to zero (the default) and accept that you will underestimate values
slightly. Otherwise, you can render a fisheye image just outside the
window setting "-ab 0 -dv-" and compute the average pixel value with:

pvalue -h -H -o -df fisheye.hdr | total -if3 -m

That's pretty close to what compamb does.

Interesting method, I'll surely try it.

So ambient value i get from this calculation I use -av parameters for
mkillum?

Rad parameters:

Beside setting of ambient parameters it is good idea to experiment a
little and find nice set of -d* parameters.
Example of parameters: -ab 3 -av 0 0 0 -ad 1024 -as 256 -aa 0.1 -ar 600
-ps 1 -ds 0.05 -dj 1 -dt 0.03 -dc 0.9 -af tmp/floor1_illum.amb

In daylighting situations, the calculation shouldn't be very sensitive to
the -d* settings. You can set -dt 0 if you're worried about it, though.

For nice luminance images of daylight spaces, I usually experiment with
these -d* parameters.
Combination of -d* parameters in example above (although too high and
computationally expensive) gives really nice images in most cases.

*Question:* Why it is not recommended to use mkillum for illuminance

calculations?

It's just that mkillum doesn't pay off if you are only calculating a small
number of points. The only reason to use it is to reduce the error when
you cannot afford a difficult indirect calculation over a million or so
pixels. If you are only computing a few hundred illuminance values to
begin with, you are better off simply cranking up the indirect calculation
parameters and letting rtrace deal with the window complexity.

We usually calculate illuminance images over complete workplane and set
rather high ambient parameters ab>=5, and calculations of an image take
several hours.
Maybe mkillum in such process can make calculation time shorter, but again
we should determine correct ambient value previous to mkillum running. Also
if project has big number of windows, calculation of mkillum would take
additional time - so no real time saving in this approach.

As a conclusion: For illuminance calculations better use high ambient
parameters and avoid mkillum.

Thanks again,
Marija.