Hello everyone,
I’ve been using gendaylit and rtrace to simulate scenes illuminated by daylight. Given the rise of hyperspectral rendering, I’d like to adjust gendaylit to handle a spectral sun and sky defined by specific spectral direct and diffuse irradiances.
I am aware of other spectral sky generators, such as gensdaymtx and genssky. However, gensdaymtx uses the Reinhart sky subdivision, and genssky relies on a built-in atmospheric extinction model, which does not allow me to input my specific spectral irradiances directly.
I’ve attempted to compile a modified version of the gendaylit.c file to replace the existing gendaylit.exe without success. This compilation step has been my primary challenge. I have reviewed relevant forum posts and documentation but am still unclear on the proper procedure to accomplish this.
I am a newborn baby when it comes to this, so do you know an easy way to compile such .c files.
Kind regards,
Alex
Hi Alex,
Your question is more for @Taoning_Wang who created the spectral models for genssky and gensdaymtx. I believe he is still working on direct inputs for sun/sky radiance values for these, which will probably be integrated rather than spectral values, anyway.
The simplest approach would not be to modify gendaylit or any of the other tools, but to apply a “spectrum” modifier to the sun and sky material chains equal to your per-wavelength measurements divided by the corresponding direct and diffuse spectral irradiance wavelengths you get from gendaylit or genssky.
You will need to run “rtrace -h- -I+ -co+ -cs 20” on the sky description by itself and the sun by itself, passing it a single point directed upwards in each case. This should give you the spectral denominators to use against your measured(?) spectra, and from these create a “spectrum” modifier to apply to sun and sky appropriately.
If you use gendaylit as a starting point, then your sky will have the same spectrum everywhere. If you use genssky, then you will get some spectral variation over the sky, though there is no guarantee these variations will make sense if you weren’t pretty close to the desired spectral sky distribution in the first place.
Best,
-Greg
If you have directly measured spectral radiance, you can use it directly with specdata.
void specdata sky_spect
4 noop sky_spect.dat . "Acos(Dz)/DEGREE" "atan(-Dx, -Dy)/DEGREE"
0
0
sky_spect glow skyglow
0
0
4 1 1 1 0
skyglow source skydome
0
0
4 0 0 1 180
Your spect.dat file shoud look something like this. It’s a 3-dimensional data file describing the direction of your spectral sample point in theta and phi, which goes from 0 to 90 and 0 to 360 at 3 deg interval. The last dimension is your wavelength. here i have 20 wavelength samples. the data values increment from wavelength to phi to theta.
3
0 90 30
0 360 120
0 19 20
0.2
0.8
...
Thank you for the fast response.
@Greg_Ward
My plan was to apply the modifier but that would require adjusting the sky.rad file that is eventually passed onto oconv.
I’ve created a script in Python that replicates some of the functionalities of gendaylit and prints a .rad file containing the Radiance sky description. It looks like this:
void light solar
0
0
3 1.237e+07 1.237e+07 1.237e+07
solar source sun
0
0
4 -0.440714 0.373752 0.816138 0.533000
void brightfunc skyfunc
2 skybright perezlum.cal
0
10 -4.627e-25 6.185e+01 1.253062 3.840531 -293.856228 10.528083 -19.053431
-0.440714 0.373752 0.816138
skyfunc glow sky_glow
0
0
4 1 1 1 0
sky_glow source sky
0
0
4 0 0 1 180
skyfunc glow ground_glow
0
0
4 1 1 1 0
ground_glow source ground
0
0
4 0 0 -1 180
void plastic RGB_ground
0
0
5 0.510 0.265 0.074 0 0
RGB_ground ring groundplane
0
0
8
0 0 -.01
0 0 1
0 100
However, when I try to run rtrace I receive a warning: no light sources found.
If I can make this work, I should then be able to apply the spectrum modifier.
@Taoning_Wang
What I have is the spectral irradiance (not radiance), so I probably cannot use specdata. Unless, I’ve misunderstood something.
Any tips would be highly appreciated!
Hi Alex,
There is some mistake in your running of oconv or rtrace, as the file you provided above is valid and contains a light source. I get no warnings and it behaves as expected with rtrace.
Regarding the spectral modifier, it is easy enough to specify a average irradiance over the spectrum to gendaylit then compute the ratio between your spectral irradiance and this average, putting the result into a .dat file using “specfile” or directly into a “spectrum” primitive. Inserting the modifier into your scene description is as simple as putting it ahead of your sky file and running the gendaylit output through sed:
gendaylit [arguments] | sed ‘s/^void brightfunc skyfunc$/my_spectrum brightfunc skyfunc/’
Good luck!
-Greg
You are absolute right! I had not passed the new sky.rad file to the oconv command leading to the error of missing light source 
.
I also miscalculated the relative luminance of each sky element leading to a weird skyfunc. Here is the correct one:
void brightfunc skyfunc
2 skybright perezlum.cal
0
10 5.156e+01 1.383e+02 -0.966875 -0.175090 20.918068 -5.595514 1.326504
0.440714 -0.373752 0.816138
I’ll post again once I’ve implemented the spectral part.
Many thanks!
Oh yes!
Thank you both once again,
Alex
1 Like
