Mantra Surface Model equivalent of Glossiness(Vray), Roughness(arnold)

[Erik_JE]

As the title suggests. What would be the equivalent of Glossiness(Vray), Roughness(arnold) in the Mantra Surface Model and how does it map to them.

 

Changing the Reflection/Refraction Angles gives a similar result as adjusting Glossiness/Roughness materials but it's hard to get it right as it goes between 0 to 90 degrees and the gloss and rough attributes goes between 0 and 1.

 

Links:

http://help.chaosgroup.com/vray/help/150SP1/examples_material.htm#reflection_glossiness

http://help.chaosgroup.com/vray/help/150SP1/examples_material.htm#refraction_glossiness

 

https://support.solidangle.com/display/AFMUG/Specular

https://support.solidangle.com/display/AFMUG/Refraction

 

Someone on odforce somewhere posted that:

rough = sin(radians(angle))

 

But we could not get that to work.

 

Cheers,

Erik

[anim]

without being able to test and compare I would assume that there are 2 straightforward mappings which may make sense

1.  angle = 90*rough                                =>  rough = angle/90

2.  angle = degrees(arcsin(rough))        =>  rough = sin(radians(angle))

 

but it may highly depend on lighting model you are using as well

if you have the ability to compare then it's quite easy to find out the math behind (if exists), but be sure to compare the same lighting models

[Erik_JE]

We tried to use Phong everywhere.

 

1. This was the first I tried but it gave very different result. 0.2 roughness in arnold would be 18 degrees in mantra reflection angle but that was a lot more blurry. About 5 degrees gave a similar result.

2. 0.2 roughness if calculating angle would give ~11 degrees which also is a lot blurrier.

 

Maybe it don't map. Who knows

 

Mapping between arnold and vray seems to be:

roughness = 1 - glossiness

So 0.8 glossiness in vray gives the same result as 0.2 roughness in arnold.

 

 

It's funny how exactly the same arnold and mantra looks on diffuse surfaces.

[anim]

the only thing I can suggest is try to render out few sample images of roughness of 0 .1 .2 .3 ... .9 1 (or smaller steps)

then try to match each of them visually in Mantra and plot the values it returns into the angle/roughness graph

then see if it can be simplified with some function or not

[symek]

Sebkaine had studied similar topic.

 

 

 

 

https://support.solidangle.com/display/AFMUG/Specular

https://support.solidangle.com/display/AFMUG/Refraction

 

 

Why?! Why, this documentation is so good!?

Edited April 10, 2014 by symek

[abvfx]

It makes me a little sad.

I know we all want great new features etc but this is some thing currently in Houd could do some some house cleaning, i would sacrifice a new feature set for a release so things like these could get filled out. /rant

[eetu]

Some PBR shading models use the square of the roughness parameter inside, to get a more linear-feeling control to the user. This might be another source of perceived discrepancy.

 

Physically-Based Shading at Disney:

 

 

For roughness, we found that mapping alpha = roughness^2 results in a more perceptually linear change in the roughness. Without this remapping, very small and non-intuitive values were required for matching shiny materials.

Edited April 12, 2014 by eetu

[magneto]

Sebkaine had studied similar topic.

 

 

 

 

 

Why?! Why, this documentation is so good!?

 

Wow that documentation is insane. I mean SESI docs are also very good and very long but that Arnold one seems like it has a professional writer doing these or something

[Guest tar]

Some PBR shading models use the square of the roughness parameter inside, to get a more linear-feeling control to the user. This might be another source of perceived discrepancy.

 

Physically-Based Shading at Disney:

 

Yup:

https://support.solidangle.com/display/ARP/General+FAQ#GeneralFAQ-Whatisthere-parameterizationofspecular_roughnessinthestandardshader,andwhyisitnon-linear?

[Erik_JE]

Some PBR shading models use the square of the roughness parameter inside, to get a more linear-feeling control to the user. This might be another source of perceived discrepancy.

 

Physically-Based Shading at Disney:

 

Yeah it is probably something like this.

Did some quick tests as anim suggested with trying to replicate vray glossiness with mantra reflection angle and got something like:

 

4 degrees

0,2 gloss

 

20 degrees

0,5 gloss

 

70 degrees

0,8 gloss

 

Which indeed seems very exponential.

 

Tested with:

glossiness = sqrt(angle) / sqrt(90)

 

Calculating the angle like that the values would be:

0.2 ~ 3.6

0.5 ~ 22.5

0.8 ~ 57.6

 

Which is pretty close.

Edited April 22, 2014 by Erik_JE

[Erik_JE]

Did some rendering tests as well and this seems to work good against both vray and arnold for specular reflections. Refractions however look very different.

Will investigate some more.

[Guest tar]

Wow that documentation is insane. I mean SESI docs are also very good and very long but that Arnold one seems like it has a professional writer doing these or something

 

I'd imagine that as Solid Angle does rendering only, that it's docs will have to be good for rendering.  Mantra is a part of Houdini.