[gmenzel]

And that shouldn't happen with additives, at least since V3. You should be able to use as many additive layers as you wish now, without your object starting to glow or look overly bright. Your test is puzzling, and checking with an SSS equivalent, I get more what I expect....

As I understand it, Maxwell V3 only clips the total energy of stacked additive layers that exceeds the allowed absolute range. In my example, the energy stays well within the allowed range on every bounce but builds up from bounce to bounce. On each reflection bounce, a little bit of energy is added.

[tom]

Dear Everybody,

You have a master mistake assuming layers as sheets physically stacked on each other. They are really *not* onion-skin. Technically speaking, stacked layers is *not* different than BSDF blending. BSDF blending uses weights and normalizes the result. Layer system offers you opacity maps which helps you to *visually* stack groups of BSDFs. The computation is just like managing the windows on your screen. You bring a window to front or send to back or minimize. Have you ever seen stacking many windows on your desktop caused a thickness on your display? No. So, the term *stacking* is virtual.

Imagine you have a pure red BSDF and a pure green BSDF. Now, if you weight them 50% each, you will have a 50% yellow output using the formula BSDF1 * weight + BSDF2 * weight. Please try it yourself. So it means, 2 weight-blended BSDF can only be at 50% of total expected brightness.

Now, try doing the same under separate layers. Make 2 layers each containing 1 BSDF and make one of them pure red and the other pure green. Make the upper layer Additive and render. You will have a pure yellow output at 100% (not more). You might wonder what would happen if both of your BSDF were pure red. Give it a try and you will see the result will not be 200% red unlike in the past and it will not exceed 100% red. Because, Maxwell checks and avoids reflections exceeding 100%. To make sure, create numerous layers contatining a full white BSDF and blend all of them additively.

Briefly saying, the terms Normal or Additive are all about the computation. They simply have no physical meaning and they all come from the maths in code. So, it's a miserable effort wondering if there are additive reflections in real world or not. Because, there is no such thing called "additive reflection" or "normal reflection" in Maxwell either. You simply compare normalization to addition and that's all. Apples and oranges...

Suggestions done by several users including our friends may vary but, of course there is only one truth. You can use both methods making your materials without worrying about physical correctness. Both methods are safe and they are not exceeding 100% reflectance. Apart from this fact, I'd like to check gmenzel's scene if he shares.

Thank you

[Fernando Tella]

Very clarifying example Tom.

In gmenzel's scene I think the brighter part are just reflected caustics from the mirror part of the material.

[gmenzel]

You have a master mistake assuming layers as sheets physically stacked on each other. They are really *not* onion-skin.

That’s actually how I see it too. I did not at all want to imply that I assume a 1:1 relation between MXM layers and real-world layered materials, like coatings. It is exactly the computation aspects that caught my attention. By ‘additive reflection’ I was referring to the commonly used approach to produce glossy materials. In this context, the gloss-layer represents an actual physical material layer. My question was in regards to certain phenomena that can occur when using such materials, namely the problem of light amplification as demonstrated.

Apart from this fact, I'd like to check gmenzel's scene if he shares.

He certainly will. I didn't save the torus scene, but I made another one that shows more clearly what is reflected.



Scene: http://s000.tinyupload.com/index.php?fi ... 6042250425

[tom]

Thanks for the scene! Meanwhile, I've made torus myself and we started checking it already. Seems to be an unexpected leak obviously and I just hope it's not a limitation of the additive-fix. Soon to let you know and thanks for the heads-up!

[gmenzel]

In gmenzel's scene I think the brighter part are just reflected caustics from the mirror part of the material.

Hm, interesting. Deactivating caustics has indeed an effect. I wonder what is going on here.

But even with caustics deactivated, the result still shows amplification.

[tom]

Btw, did you upload wrong scene? It's not rendering these balls.
-Edit: Nevermind, I found the camera was not in its place...

[eric nixon]

Tom, I'm confused, you haven't mentioned a transition between methods, surely in the real world there are situations where a coated surface gets worn down and the reflections become duller. In those cases should we use a gradient weight-map between two layers -(one additive and one normalized) ? Or do we just reduce the weight of a single additive layer as it gets rougher or duller?

I mean how do we treat soft-highlights say around roughness 40, should they be blended in additive mode also? Is there a cutoff?

I need to understand this in real world terms, because thats how I design materials.

Looking at gmenzel's render I wonder if using roughness > 0 would change the result, that was an issue with MW3.0, which may still be problematic.

[eric nixon]

Just noticed a bug with additive spec layer @ roughness 0

pic shows rough0 vs rough2

It still seems that the question 'when to use additive spec and when not?' has not been answered, Is the answer to always use additive mode whatever the surface is made of, and that topcoats/laquers are irrelevant?

[Mihai]

I thought I did

As for when additive, when not, I think always additive when trying to mimic a topcoat/laquer on top of something else. Because in the real world, say you have a yellow plastic with and without a topcoat. Would the version with the topcoat have a duller and darker looking yellow? I'd guess no. So the topcoat never really takes away from the saturation of the "texture" in my opinion. Duller vs stronger, that's just a matter of finding the right nD, having a good roughness map etc. That's where good references taken in controlled conditions come in handy.

I'm really not seeing a situation when I would want to use normal blending for plastics or shiny wood with a top laquer, etc...Also I'm not sure what you mean by how to transition between these two blending, it's not necessary really. If you have a coating that is faded away in parts, well use a layer opacity mask for the top additive layer, combined with maybe a roughness map for it.

[Mihai]

Just noticed a bug with additive spec layer @ roughness 0

The bottom glass BSDF is non zero right? Because I'm seeing strange things also if the bottom glass is left at 0 and I set the top additive to any roughness. If none of them is set to zero it works fine.

[eric nixon]

Yes, bottom glass is roughness 2.

I'm really not seeing a situation when I would want to use normal blending for plastics or shiny wood with a top laquer, etc...

So your saying it is only for topcoats, (and plastics are a special case even if they're rough, because they're actually transmissive and using SSS for plastics usually is overkill, but this way we can brighten the appearance without sss)

But in that case what actually defines a topcoat? I assume greasy fingerprints count, and also water on a non-porous surface.....

If something is semi-coated is it ok to just reduce the spec layer weight?

[Mihai]

I would say using SSS for plastics is not overkill - it's a necessity for realism but in some cases with very thick solid looking plastics you can get away with a non SSS plastic and it will render faster.

but this way we can brighten the appearance without sss

Maybe this is just semantics, but not "brighten the appearance". Rather - not make the underlying color lose its saturation because of blending it with a shiny BSDF which by itself is like a shiny dark "metal".

[seghier]

i always read this material physicaly or incorrect
i don't know how they know ! if the material is mixed from different materials
if i draw something i draw what i see i don't need to know physic or chimic
if i create car model i don't need how the motor worked
materials like glass metals ... or any simple material ; anyone can say it correct or incorrect but with complex material i don't think is simple to say that and how many layers we need to create it .
this is my personal opinion .

[Hervé]

at the end it is up to the user... the rule is... there are no rules...