unexpected behaviour of material's "max reflection depth" settings

[mts]

Hi there,
I am currently tweaking some render settings and found some unexpected behaviour regarding the max depth value under a vray material's reflection settings.

My assumption always has been as follows:

when changing the max reflection depth of a material, say "steel", it only has an impact on the objects assigned with the material "steel". No other objects are influenced.
steel's max reflection depth = 1: an object assigned with the material "steel" only reflects the surrounding objects
steel's max reflection depth = 2: an object assigned with the material "steel" reflects surrounding objects PLUS the (first) reflections in these objects...
and so on...

Anyways, when tweaking some renderings it just ocurred to me that Vray seems to behave differently. Please have a look at the attached image. When changin the max reflection depth on the steel material, it has an impact on the material made of plastic. Has my assumption been wrong or does Vray behave in a strange way here? If this is the way it is supposed to be it makes tweaking renderings way harder in my opinion as you never know what objects you will influene when setting up a certain material.



I am grateful for any explanations.
Thanks,
Matthias



V-Ray for Maya version 2.10.01, revision 18083 from Dec 11 2011

[vlado]

The "max. reflection depth" for a material determines when the material will stop tracing reflections; this is usually noticeable in parts of the image where the material itself is reflected in other materials.

Best regards,
Vlado

[mts]

Thanks Vlado - I already assumed that I was wrong when realizing that Mental Ray behaves the same way a Vray does. Seems like I have to rethink what I wanted to do.

[mts]

Vlado, just out of curiosity - would you mind shedding some light on how the actual algorithm basically works then?
We are debating with some people here, but everyone always thought that it is working as follows:

Eye Ray hits some "steel" object. Steel has the max depth set to value "x" so for each reflection ray x-1 secondary reflection rays are sent out (assuming there are only reflective objects in the scene). The more secondary reflection rays there are, the more likely it is that more reflected light arrives at the point the eye ray has hit.

Apparently it is not like that but we have no idea what other way it could be. It would be great if you could give us a hint.
Thanks!

[vlado]

It works like this: some ray hits a material (the ray path may have already gone through one or several other reflections/refractions etc); the material needs to shoot a new reflection ray, but it first checks if the number of reflection rays on the path so far is beyond or equal to the specified max depth value. If there are already more reflection rays on the path, it does not shoot the new reflection ray and instead returns the reflection exit color. Otherwise, a new reflection rays is spawned and the path tracing continues.

So, in short, the number of reflection ray depth is not determined by the first material that the camera sees; instead each material along a ray path determines if it needs to trace reflections based on the number of reflections that have been traced before it.

Best regards,
Vlado

[pechart]

So to say its more an absolute then a relative value. I always thought about it like mts. Maybe you could add an "relative" option? Because the way it gets handled by now is a little bit hard to handle, i think. Just an idea.

[vlado]

Because the way it gets handled by now is a little bit hard to handle, i think.

Well, the other way round might be just as confusing depending on what is the particular situations. Most raytracers work in the same way as V-Ray, so it might be a better idea to stick to what is known.

Best regards,
Vlado

[pechart]

What I mean is, if you want to tune up your scene you usually not know where the ray comes from. You just know the actual position, in this case the current material. So you want to cut the depth related to that specific position. At least this is how I think about it.
How are other guys here thinking about that?

[mts]

Thanks Vlado, your description clarifies a lot. I see it the same way as pechart though. If the reflection depth was controlled by the first material the camera sees I'd be able to say "I want Material X to have realistic reflections and Material B does not have to behave that realistically", which would make great sense to me.
The way it is now I am only able to say "I want Material X to be more likely to be reflected in other materials and Material Y does not have to be as likely to be reflected." I'm sure there is an advantage to this, too, as you say most raytracers work this way, Vlado. I just don't see it at the moment.

Anyhow, my question has been answered and I finally understand what's going on.