strange behavior of Architectual Mat (MIA) at low light angles

[nielskretschmann]

Hi, just recently i noticed a rather odd feature of the MIA material.

when hit by light at low angles it appears darker compared to a lambert shader or the new Mila shader (or a Vray shader or any other shader i tested).

 

To see what i mean, you find a table with several images under this link:

 

http://www.roomservice3d.com/files/Post_160721_MiaShader.jpg

 

all materials are pure white and diffuse only, there is only one light in the scene lighting the plane (Lambert) and the two boxes (MIA + MILA) at different angles. test was done with linear color workflow.

 

at a light angle of 20° everything looks as expected with all 3 materials showing the same color. at lower angles of 10°,5° and 1° you notice how the mia shader gets darker compared to the other two. At 10° it is 70% percent as bright as the others (in linear color space), at 5° it is 32% and at a light angle of 1° only 7%.

 

 

 

what is up with that? is this behavior of the mia shader known by you (i found nothing with google)? is it a bug? is it a feature? if it is a feature WHY?

Edited July 21, 2016 by nielskretschmann

[Francisco Penaloza]

MIA and MILA are new shaders compared to Lamber, these new shader have a better falloff. Phisical correct as claimed by Mental ray developers. Lamber does not.

 

The shading you are displaying are from viewport, those difference I can't explain. Did you try to put each cube side by side, instead one in front of the other? this way the light will touch both at the same angle.

What software are you testing this?

Softimage?

[nielskretschmann]

i know that the lambert shading model is an relatively old concept. still the modern mila shader shows exactly the same result as a lambert shader if its roughness is zero. if the roughness parameter is higher it gets brighter than a lambert shader at low light angles and not darker.

 

as the light in my example is directional (parallel) it has the same angle to any surface.

 

the software is indeed softimage but should not matter since rendering is the job of mental ray. if its a preview or a real render should not matter either. Still i could have made some error and my findings could be wrong. It would be nice if somebody could do a simelar test to verify the result.

Edited July 21, 2016 by nielskretschmann