[gmenzel]

I get strange reflections when rendering this simple scene:



The material of the ground plane is a simple gray BSDF with a reflective layer added on top. Here the normal map used:



The scene renders as follows:



Notice the overly bright secondary reflelection of the emitter above. Where is this coming from? I have added a simple reflective sphere in the bottom left to show that there really is nothing more in this scene, that could cause this reflection.

Anyone have a clue?

(3DS Max 2012 x64, Maxwell 2.6.10.0 / 2.6.21)

[gmenzel]

Sorry for my impatience, but I'm really stuck because of this problem. If maybe someone of the staff could look into this matter (and confirm that it is indeed a bug and not something I'm doing wrong), I would greatly appreciate it!

[bograt]

Looks very odd, I am afraid I cannot help all that much but have you tried different normal maps, slight variations in glossiness, using a variety of glossy materials, moving the light, removing the light above to see if there is another invisible one etc..
all of these things can shed light on the problem, I would also suggest uploading your scene (possibly for other max versions as well) so people can have a look at it..

[gmenzel]

I built this scene specifically to demonstrate the problem I have run into. The symptom is actually really easy to recreate. All you need is a normal map with very steep normal information, like this simple sphere map has (attached to my initial post).

This reflection is definitely caused by the emitter plane above. Changing the color of the emitter, changes the color of the reflection. Moving the emitter below the ground plane, eliminates all reflections.

I have deliberately not provided my scene, because I want to be sure that it is not something specific to my setup or configuration. The scene consists of only two planes and a camera and should be easily reproducible within a couple of minutes.

If someone is able to confirm this behavior, then this would mean that there is something wrong in general with the way Maxwell calculates normal vectors from normal maps with steep angles.

[Fernando Tella]

Try to change the way the normal map is read (flip x, flipY or wide options in the map picker). Maybe it works.

[gmenzel]

I think I may have track down the cause. Have a look at the following comparison rendering:



The arrow marks the point after which the calculated normal vector starts to point downwards, below the ground plane. Renderers usually limit the outgoing vector to the polygon plane. Maxwell however handles such cases in a different way and seems to flip the outgoing vector along the polygon plane, resulting in the vector pointing upwards and therefore causing 'wrong' reflections.

The 'wide' parameter does not change this behavior.

[tom]

Could you export MXS and email me with the map? Thank you. (tom at nextlimit dot com)

[gmenzel]

Could you export MXS and email me with the map? Thank you. (tom at nextlimit dot com)

Done.

[tom]

Thanks! We're going to check it...

[gmenzel]

Thanks! We're going to check it...

Any news on this, Tom? Can I hope that this issue will be resolved with the next update?

[tom]

Fixed in the release

[numerobis]

that was fast... nice!

[gmenzel]

Fixed in the release

Great! You mean the next release?

[tom]

Yes.