Glitchy illuminance image

Hello everyone,

I’m trying to use Radiance to see whether a camera, in situation with just one light source, would receive enough light to see anything and what efficiency there is between energy consumed by the light, and energy received by the camera.

The scene is very simple, a kind of torchlight illuminates an object, and some of the reflected light comes back to the camera which is close to the light source.

What I’ve tried is creating a fake camera, with genbox, and making an illuminance false-color image of it (the light is a bit behind the camera so it doesn’t light it directly).This seems to be working more or less, however I’ve sometimes found things I cannot explain. The image below shows one such example, as well as the scene in Blender, in case my description wasn’t clear.

I can clearly see a separation on the fake camera. Can someone explain what’s going on?
Additionally, if someone has ideas on how to achieve my goal, I am open to suggestions.

Thanks,
Evan

Hi Evan,

I am not sure I understand what is going on from your description, but it looks like you may be seeing the result of a too-crude irradiance cache calculation. What happens if you render with -aa 0?

Without knowing the dimensions of your scene, the geometry of the camera, or any of your rendering methods or options, it is difficult to discern the problem.

Cheers,
-Greg

Thanks for your answer

I wasn’t sure if including everything would be too much text, which is why I didn’t write all the details.

I played around with the parameters before, but never as far as the max since I used an old computer. Using just -aa 0 made the fake camera look like TV static, but I trusted your insight, increased the other parameters as well and let it run a while and that seems to have done the trick, there are no more weird separations.

It’s going to take a while to render the whole final scene, but I know what to do now, so thanks a lot for your help!

Evan

The -aa 0 suggestion was just to determine the source of the problem. Glad you were able to confirm it. There are settings with -aa > 0 that will not have this artifact, but we would need to know more about your scene.

Well if you can do better than I’d be glad to share the details.

I first export the object from Blender to an .obj file. Suzanne, Blender’s mascot, for this test. And it is centered at coordinates 0, 0, 0 and uses the ‘subject’ plastic material defined below

File scene.rad

materials

void light Lamp
0
0
3 100 100 100

void plastic subject
0
0
5 .9 .9 .9 .0 .2

void plastic cameraSensor
0
0
5 0 0 0 0 0

light source

Lamp sphere torchlight
0
0
4 0 -5 0 0.1

fake camera that receives reflected light

!genbox cameraSensor fakeCamera 0.2 0.01 0.2 | xform -t 0.5 -4 0

Then I do these commands to generate the image

obj2rad monkey.obj > monkey.rad
oconv scene.rad monkey.rad > scene.oct

rpict -vp 0.5 -3.0 0.0 -vd 0.0 -1.0 0.0 -vu 0.0 0.0 1.0 -ab 3 -aa 0.00 -ar 0 -ad 4096 -as 1024 -i -x 512 -y 512 -t 1 scene.oct> scene_i.hdr

falsecolor -ip scene_i.hdr -l Lux -log 1 -n 10 -s 1 > Lux_scene.hdr

ra_bmp Lux_scene.hdr > Lux_scene.bmp

And here is the result:

If you have another suggestion, or notice something weird in my method, let me know.

Thanks,
Evan