Heavily different sensor values with different (rcontrib?) versions

Stefan,

The only major change to rcontrib I know of was the addition of extra
details in the header, but you're not writing the header so that can't be
the problem. The only thing I can think of is that there are some
differences in shell tools between BSD and Linux, so perhaps test seq and
tr independently to make sure they are doing what you expect. Other than
that I don't know what could be happening.

Now that you're using a recent version of Radiance instead of the for loop
you could use a single inline rmtxop command like this:

cat $sensors | rcontrib -I -fo @scene.opt -ab 4 -ad 100000 -lw 0.000004 -y
$numsensors -e MF:1 -f reinhart.cal \
    -b rbin -bn Nrbins -m sky_glow -w scene.oct | rmtxop -c 0.265 0.67 0.065
- > $sensors_vec

just make sure you define $numsensors as your number of sensors.

As for settings to consider, if you're interested in specular reflections
you probably want to set -st 0. And you'll probably need to reduce -lw
further to something lower than your facade reflectance divided by ad
setting. For example, if your facade VLT is 5% it would have to be less
than 0.05/100000=0.0000005 or 5e-7 (you currently have 4e-6).

Best,
Andy

···

On Mon, May 23, 2016 at 5:32 AM, Stefan Bohren <[email protected] > wrote:

Dear Radiance community

I am doing my master thesis about the relevance of reflections on solar
fassades, for what I want to make some simulations with radiance, and also
testing it with a little live experiment.

After having worked myself through most radiance tutorials and the book
(very helpful), my prof helped me to put a script together, which in the
end should be able to return radiation values at several sensor points, and
opt on different ambient settings, surfaces and sensor point locations.

And now where we are stuck atm: His results with the script are reasonable
(on OSX 10.9, likely with radiance 4.1), but if I use exactly the same
script and files (on Ubuntu 16 with newest radiance), I get values that are
far away from realistic (three to ten times too high radiation values, and
slightly different to physically impossible rgb-tripplets), without any
error messages. I searched through the archive, but didn't find a similar
issue.

As the results of gendaylit -> genskyvec are identical, I think the
problems are starting with rcontrib, which we use to produce the sensor
radiation readings for each sky patch:

  cat $sensors |rcontrib -h -I -fo @scene.opt -ab 4 -ad 100000 -lw
0.000004 -e MF:1 -f reinhart.cal \
    -b rbin -bn Nrbins -o results/p%04d.dat -m sky_glow -w scene.oct

and are getting worse where we set them all together:

  for i in $(seq -f "%04g" 0 145); do
    cat results/p${i}.dat |rcalc -e '$1=($1*0.265+$2*0.67+$3*0.065)' |tr
'\n' '\t' >> $sensors_vec
    echo >> $sensors_vec
  done

As the results are fine with the older version, and not the other way
around, we'd really be happy if you helped us on how to adapt the script to
the current radiance version - or told us if we are doing something
completely wrong!

Looking ahead this issue, we also are wondering which will be the
important ambient settings to optimise, in the end it will be a large
(several buildings) outdoor setting, and I'd be glad if you could hint me
into the right direction with that as well!

Thank you very much,
Stefan

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

Dear Andy

Thank you very much for your answer and help, especially also for the hints about the abbreviation with rmtxop and the settings.

I finally found the issue, with another old script that not just produced bad results, but refused to work. And it actually is a header-issue, but not because of rcontrib: It looks like genskyvec didn't use to write a header, but nowadays it does. So as soon as I told it not to write a header, I got reasonable results too :slight_smile:

Cheers,
Stefan

···

________________________________
Von: Andy McNeil [[email protected]]
Gesendet: Mittwoch, 25. Mai 2016 01:49
An: Radiance general discussion
Betreff: Re: [Radiance-general] Heavily different sensor values with different (rcontrib?) versions

Stefan,

The only major change to rcontrib I know of was the addition of extra details in the header, but you're not writing the header so that can't be the problem. The only thing I can think of is that there are some differences in shell tools between BSD and Linux, so perhaps test seq and tr independently to make sure they are doing what you expect. Other than that I don't know what could be happening.

Now that you're using a recent version of Radiance instead of the for loop you could use a single inline rmtxop command like this:

cat $sensors | rcontrib -I -fo @scene.opt -ab 4 -ad 100000 -lw 0.000004 -y $numsensors -e MF:1 -f reinhart.cal \
    -b rbin -bn Nrbins -m sky_glow -w scene.oct | rmtxop -c 0.265 0.67 0.065 - > $sensors_vec

just make sure you define $numsensors as your number of sensors.

As for settings to consider, if you're interested in specular reflections you probably want to set -st 0. And you'll probably need to reduce -lw further to something lower than your facade reflectance divided by ad setting. For example, if your facade VLT is 5% it would have to be less than 0.05/100000=0.0000005 or 5e-7 (you currently have 4e-6).

Best,
Andy

On Mon, May 23, 2016 at 5:32 AM, Stefan Bohren <[email protected]<redir.aspx?REF=8RQxChPI4tTaam-L0w9llHpKX9_axdFoZ1HokCx4bXnJ4a2esITTCAFtYWlsdG86c3RlZmFuLmJvaHJlbkBzdHVkLnVuaWJhcy5jaA..>> wrote:
Dear Radiance community

I am doing my master thesis about the relevance of reflections on solar fassades, for what I want to make some simulations with radiance, and also testing it with a little live experiment.

After having worked myself through most radiance tutorials and the book (very helpful), my prof helped me to put a script together, which in the end should be able to return radiation values at several sensor points, and opt on different ambient settings, surfaces and sensor point locations.

And now where we are stuck atm: His results with the script are reasonable (on OSX 10.9, likely with radiance 4.1), but if I use exactly the same script and files (on Ubuntu 16 with newest radiance), I get values that are far away from realistic (three to ten times too high radiation values, and slightly different to physically impossible rgb-tripplets), without any error messages. I searched through the archive, but didn't find a similar issue.

As the results of gendaylit -> genskyvec are identical, I think the problems are starting with rcontrib, which we use to produce the sensor radiation readings for each sky patch:

  cat $sensors |rcontrib -h -I -fo @scene.opt -ab 4 -ad 100000 -lw 0.000004 -e MF:1 -f reinhart.cal \
    -b rbin -bn Nrbins -o results/p%04d.dat -m sky_glow -w scene.oct

and are getting worse where we set them all together:

  for i in $(seq -f "%04g" 0 145); do
    cat results/p${i}.dat |rcalc -e '$1=($1*0.265+$2*0.67+$3*0.065)' |tr '\n' '\t' >> $sensors_vec
    echo >> $sensors_vec
  done

As the results are fine with the older version, and not the other way around, we'd really be happy if you helped us on how to adapt the script to the current radiance version - or told us if we are doing something completely wrong!

Looking ahead this issue, we also are wondering which will be the important ambient settings to optimise, in the end it will be a large (several buildings) outdoor setting, and I'd be glad if you could hint me into the right direction with that as well!

Thank you very much,
Stefan

_______________________________________________
Radiance-general mailing list
[email protected]<redir.aspx?REF=okBVF0SQVmfq9p1DI6dq-4Je4Vgc6ytOtkp6IIMl0uDJ4a2esITTCAFtYWlsdG86UmFkaWFuY2UtZ2VuZXJhbEByYWRpYW5jZS1vbmxpbmUub3Jn>
http://www.radiance-online.org/mailman/listinfo/radiance-general<redir.aspx?REF=Y44plK10NT3zQLV-SAX6Zm48ARY3K-YHEL3mkvSn0ZvJ4a2esITTCAFodHRwOi8vd3d3LnJhZGlhbmNlLW9ubGluZS5vcmcvbWFpbG1hbi9saXN0aW5mby9yYWRpYW5jZS1nZW5lcmFs>