[BUG?]Cylinder-Emitter-Cautics-Weirdness

[manitwo]

Hi

I was trying to render the WinOsi testscene(here)
There is something wrong with the caustics in that scene.
To figure the problem out i reduced the scene to the following testscene:
http://rapidshare.com/files/15246854/cy ... e.zip.html (indigo_xml + an .obj file)

Maxwell:




Indigo:




As you can see there are some strange lines in the indigo pic which aren't occur in the maxwell one.
Unfortunatly my brain is too small and i'm not able to figure out where these lines come from.
I'm not sure but i think its a bug.

Scene Info:
Cam is faceing downwards.
cylinder has a mirror material.
ground_plane is grey.
two emitting objects inside the cylinder.

Btw: This might also be a good scene for testing fireflies.

[Zom-B]

both of your emitters are at the exact same position,
just try to push one of them 0.0001 mm away from the other...
hope this helps...

[manitwo]

fuck ... THIS is embarrassing
i already tested that but didn't put them far enough away from each other.

[CTZn]

Well, that's still usefull to know

[manitwo]

both of your emitters are at the exact same position,
just try to push one of them 0.0001 mm away from the other...
hope this helps...

hmm ... hope i embarrass myself not again but these lines also occur when i delete one light. has to be something else
Same scene with one light:
http://rapidshare.com/files/15257053/cy ... 2.zip.html



any help would be greatly appreciated!

[manitwo]

[Kram1032]

hmmm... looks, as if indigo doesn't use the full range of degree... seems to be around the 90° value... after this, it works fine again.

Let's post a request in this thread too: double sided faces with normals pointing in both directions. I.e. a tag called "<double_sided>true</double_sided>
Then, there wont be intersections 'cause of two faces spotted similary but pointing in different directions.

[IanT]

Based on the fact that a square emitter sends out rays in a cosine-weighted distribution, I'd expect to see a couple of dark bands like this ... maybe the extra bands are due to reflections inside the reflective cylinder?

Just a guess...

Ian.

[Kram1032]

It's not a squareemitter, but a square shaped meshemitter...
Or did you mean this?

[manitwo]

Based on the fact that a square emitter sends out rays in a cosine-weighted distribution, I'd expect to see a couple of dark bands like this ... maybe the extra bands are due to reflections inside the reflective cylinder?

Just a guess...

Ian.

hmm ... as i said this scene is only a modified part of the winosi testscene. (one of the lamps)
Those shadows in the indigo render of the winosi test are the same as the shadows in this scene.
As far as i seen indigo is the only renderer which shows this shadows - so i think they shouldn't be there.

If i find out how to render that without these shadows i'll make a kickass render of the winosi-test.

[IanT]

It's not a squareemitter, but a square shaped meshemitter...
Or did you mean this?

Yes, sorry ... I should have said "thin emitter"

Ian.

[manitwo]

It's not a squareemitter, but a square shaped meshemitter...
Or did you mean this?

Yes, sorry ... I should have said "thin emitter"

Ian.

Tried with an plane-like-cube - no success.

[manitwo]

hope i embarrass myself not again but ...

exactly what i did
I modeled a new cylinder and it works, altough i don't know whats wrong with the original cylinder and also
dont know why this poblem doesnt occur in e.g. maxwell. (maybe a problem with the c4dexporter - but
i am carefully with my statements now)
Anyway ... appologies to everyone who read this crappy thread.

[Kram1032]

very very weird...

[jurasek]

call them!



greetz,
jur