Indigo Renderer Forum

I am currently trying to render a glass piece with stone pebbles inside of it. The problem is the pebbles that are seen through the glass have a shiny quality even though they are using a Diffuse shader. The pebbles that are not seen through the glass appear as they should. It seems like the pebbles are somehow changing their IOR to that of the glass making them really shiny (seemingly).

Indigo 2.4.13, Blender 2.49, Blendigo 2.4 Stable (R1 and R2)

I tried putting different objects with and without any shaders applied to them, put objects inside and behind the glass, but anything viewed through the glass gets this shiny effect. I have tried different glass materials other than my own from the material database and the same effect still occurs.

Here are some of the things that I have tried to no avail:
I ensured that no objects (pebbles) were touching the glass
Adjusted every setting in the glass material that I could (IOR of the medium, uniform (0 - 5), regular
I tried changing the material from Spec to Glossy transparency
I tried using internal materials instead of .igm's from the material editor
I tried older versions of Indigo (2.2)
I tried PT, Bi, MLT +Bi, MLT + PT
I tried changing the environment scale
I went to the eye doctor for an eye exam
I went to a neurologist for a brain exam, still waiting for the results.
I tried turning on and off Aperture Diffraction
I changed the material for the pebble to phone and oren-nayer and adjusted the IOR and smoothness.
I inverted the normals of the pebbles.

Any help would be appreciated I will try to provide as much detail as possible, I will upload the pigs and blend files as well.

I rendered one image at 20% of it's original resolution and I also render a backlit version where the problem isn't as noticeable at full resolution.

UPDATE:

I went ahead and rendered a 100% res version with the problem clearly visible.

Here are the PIGS file:

http://www.mediafire.com/?nt2eejfzvzhojtc

Here are the Blend and Packed Materials

http://www.mediafire.com/?j7bx0jj1d9drm5n

An update:

I have figured out that the issue at hand only occurs when the light is angled so that it goes through the glass first and then touches an object behind the glass.

I have uploaded two comparisons, the scene is a simple sheet of glass and two balls, the glass has the "tinted green glass" material applied to it from the material database, and the two spheres have nothing applied to them at all which should default them to a diffuse white. One of the spheres is positioned behind the glass the other is not, I also placed another sphere in the scene with a glass material applied to it.

On the first picture I have the light positioned directly overhead, and on the second I have it at its original angle passing through the glass first and then striking the sphere behind the glass.

It should be noted that the same effect occurs with a Sun light as well as a mesh light, also the atmospheric light does not cause the problem to occur.

Hi Kristoff,
I think what you are seeing is MLT sampling artifacts, which are more obvious in the pebbles in glass scenario due to the tricky lighting situation, which we like to call the 'sunlight through glass' problem.
Just let the image converge and the artifacts should go away.
Alternatively, you might want to try the Indigo 2.6.1 beta, which might help in this situation.

Hi OnoSendai,

Thanks for the help, it seems you are right it is just taking an extra ordinary amount of time to clear the image and it looks worse before it looks better. I have also figured out that the problem only occurs where the caustics (shadow area) of the glass lands. Perhaps if I reduce the thickness of the glass it may not be such a problem.

I have decided to take this opportunity to compare Architectural Glass Acceleration in 2.6.1. The image below has been rendered for 14h and has 17k SPP, I will post one with the use of glass acceleration as soon as it is done.

Also note that the pebbles are slightly offset to ensure that they were not touching the glass.

Ok so I just let the same render run overnight with glass acceleration, and the problem is completely cleared up in 10h with 11k spp (glass acceleration) vs not entirely cleared up with 14h and 17k spp (no glass acceleration).

Are the pebbles a little darker on the glass-accelerated render or it is just me?

It seems like the noise in the pebbles texture/shader varies.

Yes Pibuz, I believe it is related with convergence; the first images have a strong bright noise by lack of convergence >> more convergence has reduced the bright noise >> looks darker.