All posts related to V3
#385342
Well some of us want actual answers, but the NL boys seem determined to avoid the question (when to use additive spec?) which is very suspicious. Perhaps they dont know and are too insecure to admit they dont know. Or is there a flaw in the maxwell material model they dont want to talk about?

Tom states that 'everybody' is confused about what a layer actually is, I didnt notice anyone making that mistake in this thread. He then dissapears from the thread.... even though hes the only one who has the answer.

Mihai why do you choose to ignore direct questions, and focus on side issues, its pure distraction. Ill ask you again how do you define a top-coat?, can we have an example of something that is nearly a top-coat?
#385343
Sorry for droning on about this! But seeing that so many of you seem to have no qualms about the additive approach to coatings, I wonder if there may be an error in my reasoning. So I’ll state my case again.

There are two different approaches to glossy coatings in Maxwell:
  • Simple blending of BSDFs/Layers: A high-roughness BSDF (i.e. the base) is mixed with a low-roughness BSDF (i.e. the coating) with a specific mix-ratio. This case computes as follows:
    Code: Select all
    (BSDF[base] * weight[base] + BSDF[coating] * weight[coating]) / (weight[base] + weight[coating])
    The result is always properly normalized, i.e. the total result will never exceed the higher of the individual BSDF solution.
  • Additive blending of Layers: A layer containing a high-roughness BSDF (i.e. the base) is added to a layer containing a low-roughness BSDF (i.e. the coating). This will compute as follows:
    Code: Select all
    Layer[base] * weight[base] + Layer[coating] * weight[coating]
    There is no normalization here, as the individual solutions are just added up. Unless at least one of the two is zero or the sum of all weights is less than 1.0, the total result will always exceed both individual solutions. This means more light might be reflected than physically possible, leading to the aforementioned amplification problem.
The formulas are of course not meant literally and are only intended to illustrate how these two approaches relate different layers/BSDFs to each other. Please let me know if either of these is incorrect.
tom wrote:You can use both methods making your materials without worrying about physical correctness. Both methods are safe and they are not exceeding 100% reflectance.
This is only true in regards to computational problems that might arise from reflectance exceeding the allowed range. I’m not saying this is what Tom meant, but it does not mean that every material will always behave in a manner that is physically plausible or even possible. So I don’t think it is good advice to recommend to not worry, when it is not at all hard to produce physically incorrect setups.
Mihai wrote:Would the version with the topcoat have a duller and darker looking yellow? I'd guess no.
Yes, it most certainly would! Think about it, all light that is reflected by the coating will never even reach the yellow base underneath it. So, naturally, the yellow will be dimmed. Normal blending would be the way to go here. If the result doesn’t ‘feel’ right, then the reason is either a missing SSS component, that would brighten the base, or the coating is set up as too reflective.

I need to make a strategic decision for one approach or the other. So I thought about this issue quite a lot and did some experiments over the last couple days. I have a tentative conclusion, but I'll wait for Tom's results.
#385344
eric nixon wrote:Well some of us want actual answers, but the NL boys seem determined to avoid the question (when to use additive spec?) which is very suspicious. Perhaps they dont know and are too insecure to admit they dont know.
Relax your sphincter, Mr Sunshine. I don't know the ins and outs of light transport algorithms. I just said what in my mind seems logical and confirmed by my eyes when I create a material. So it's intuitive for me. You're the one who starts making all sorts of weird connections and assumptions and then asks for validation.
#385345
gmenzel wrote:
Mihai wrote:Would the version with the topcoat have a duller and darker looking yellow? I'd guess no.
Yes, it most certainly would! Think about it, all light that is reflected by the coating will never even reach the yellow base underneath it. So, naturally, the yellow will be dimmed. Normal blending would be the way to go here. If the result doesn’t ‘feel’ right, then the reason is either a missing SSS component, that would brighten the base, or the coating is set up as too reflective.
.
It would depend on the viewing angle wouldn't it? And also the incidence angle of the light falling on that coating. If it doesn't fall at an angle, or a very shallow one, most of it will get to the yellow base, despite the coating (it's very transparent after all) then bounce back out of through the topcoat and into our eyes. Of course as the viewing angle increases we see mostly the reflected light from the topcoat and less of the color underneath. This happens of course also for the SSS. So no, it would not look duller. It may in fact even make the color seem more vivid because of higher contrast in real life. Why is it that a print that's coated with a high gloss coating makes the colors pop more?

As far as when to use the different blending methods, I just think of it as normal blending is like blending two paints in a bucket, forceably you will get some mix of the two and not one or the either can be equally "pure" as before. And additive blending is for mimicking a clear top coat on top of something else. I may be a bit interested what goes on behind the scenes but as long as this intuition is confirmed visually, I'm happy. This is apart from the scene you posted where apparently there are some leaks with additive blending which shouldn't be there (although if you keep the nD within reasonable limits, as you should anyway for the sake of realism, this problem is largely minimized compared to V2.x).
#385346
Mihai wrote:
gmenzel wrote:It would depend on the viewing angle wouldn't it? And also the incidence angle of the light falling on that coating. If it doesn't fall at an angle, or a very shallow one, most of it will get to the yellow base, despite the coating (it's very transparent after all) then bounce back out of through the topcoat and into our eyes. Of course as the viewing angle increases we see mostly the reflected light from the topcoat and less of the color underneath. This happens of course also for the SSS. So no, it would not look duller. It may in fact even make the color seem more vivid because of higher contrast in real life. Why is it that a print that's coated with a high gloss coating makes the colors pop more?
Of course the ratio of light reflected vs. light absorbed varies depending of several factors. This doesn't change the fact though, that light that is reflected by the coating can not also be reflected by the whatever is underneath the coating.

An example: A coated surface is hit with light of 100% intensity. Depending on the angle of incident, the coating will reflect, say, 10% of this light and absorb the other 90%. Only these 90% (a little bit less in practice, actually) will make it to the base layer. There, let's say, 50% is absorbed and the rest reflected. Let's ignore any further interaction of this light with the coating layer on its way 'back out'. The total reflectance of the surface works out as 10% from the coating plus 45% from the base layer, so 55% of the light all in all. I don't see how it could be any way else.

That a coated print seems to make colors 'pop', in my opinion has more to do with the greater local contrast. That is also why glossy screens look so much better next to anti-reflex-coated ones.

If additive blending in Maxwell works as I think it does, then the amplification effect is not just some leak or other computational bug. Rather, it is exactly what I would expect to happen. I don't think this is something that can be 'fixed'.
#385347
For context: a standard glass for a picture frame absorbs/reflects roughly 7% of the light... which means a print behind glass is 7% darker than it would be without glass. This is true regardless of whether regular or non-glare glass is used (there are special types of anti-reflective coated glass that will only absorb/reflect roughly 3% of the light).

Also FWIW, the reason a color looks more saturated when coated is due to reduced roughness... causing the light to be less scattered.
#385348
Relax your sphincter, Mr Sunshine.
Some of us are trying to learn here, and your just getting in the way. If you dont know much, stop pretending that you do. Remember because you work for NL, most forum readers will assume you are somewhat knowledgeable and may make the mistake of listening to your 'advices'. I can remember when I was a noob in 2006 and your advice (presented with a scientist avatar) cost me several weeks of confusion, until I realised you are a quite illogical, and best ignored.
#385360
Eric, stop being nasty and people won't be nasty back to you. Don't talk shit to people/about people and they won't talk shit to you. I have no freaking clue what you are referring to....in 2006. Something I said in 2006??? 9 years ago??? What the hell are you talking about??? Do you still remember? If I were to make a list of all the confusing conclusions you arrive at, which you post "spontaneously" as fact (while many times drunk as admitted by yourself), and then change your mind, and then rechange it, I could fill a whole forum section. Have another pint of Smirnoff and get the hell off my back.
Of course the ratio of light reflected vs. light absorbed varies depending of several factors. This doesn't change the fact though, that light that is reflected by the coating can not also be reflected by the whatever is underneath the coating.
I didn't say that it could. In real life, light which falls at 0 angle or close to it to that topcoat, practically all of it will reach the surface underneath, practically 0 of it will be simply reflected only by the topcoat. This means that the colors underneath will be reflected back at almost the same intensity as if nothing transparent and reflective was on top of it, do you agree? While as the angle of incidence increases, of course less light will be transmitted through and we will see more and more of the topcoats specular reflections. This effect, I can mimic the best with additive blending, it's not possible with normal blending. If I mix a bright red with a shiny BSDF, that bright red will get very dark and dull *all over the material* when using normal blending, indifferent of the angle of incidence of the light, or my viewing angle.
Code: Select all
Layer[base] * weight[base] + Layer[coating] * weight[coating]
There is no normalization here, as the individual solutions are just added up. Unless at least one of the two is zero or the sum of all weights is less than 1.0, the total result will always exceed both individual solutions. This means more light might be reflected than physically possible, leading to the aforementioned amplification problem.
I hardly think that's all what happens behind the scenes. The proof would be that you can stack 50 additive materials if you wish, on top of a very bright base, and it won't start to glow or emit light. I think what you have shown in this thread can be a bug or maybe a limitation when using 0 roughness on the topcoat. Try for example to make a material with several additive layers, with roughness 1 or two. Furthermore this bug is exaggerated if you use a very high nD, but you wouldn't do that anyway if you want a realistic topcoating which wouldn't have an nD higher than around 1.55.
#385361
eric nixon wrote:NL boys seem determined to avoid the question (when to use additive spec?) which is very suspicious. Perhaps they dont know and are too insecure to admit they dont know. Or is there a flaw in the maxwell material model they dont want to talk about?
Eric, I am not disappearing or NL staff is not running away from your questions at all. It's just we all have a life and cannot spend hours replying you over and over for hours here. Your uqestion is when "to use" and when "not to use" additives, right? This is very easy to explain and one doesn't have to be a scientist to tell it to you. Here in my previous answer I have already explained it but, maybe you need to hear it in a different way. Let's say you have a full yellow ball and now you want it shiny (here shiny means polished and not like chalk). You simply have just 2 ways to go. One of them is blending this BSDF to a mirror BSDF and end up with *reduced* yellowness. Deal?.. OK, and the other one is adding the mirror BSDF in a separate layer on it and keep rendering the shiny reflections without losing the full yellow base. The answer is; up to you! :)
eric nixon wrote:Tom states that 'everybody' is confused about what a layer actually is, I didnt notice anyone making that mistake in this thread. He then dissapears from the thread.... even though hes the only one who has the answer.
Some people did and still doing so, I must warn. "Additive Blending" is just a mathematical way of summing up reflections and it has *nothing* to do with physical top coats or lacquer layers etc. They are *coatings* and we already have a coating component. "Additive Blending" is nothing else than a computation method.
#385364
tom wrote:"Additive Blending" is just a mathematical way of summing up reflections and it has *nothing* to do with physical top coats or lacquer layers etc. They are *coatings* and we already have a coating component. "Additive Blending" is nothing else than a computation method.
Thanks, Tom! I appreciate the definitive statement.

I have yet to do more thorough experiments with Maxwells coating option but a quick test seems surprisingly promising for my purposes. Is there something like an official guideline on when/how to use coatings and what the drawbacks might be, if there are any? The documentation is a bit sparse here.

I figured out how to indirectly map the opacity of a coating. But there seems to be no direct way to map roughness, right (other than via a noise normal map)?
#386174
I have question here cos it seems to be related, about additive fix.
With additive layer enabled whole bottom layer is dimmed if I'm not wrong. But shouldn't it be similar to that example from Corona renderer:
https://corona-renderer.com/forum/index ... l#msg31355
...that it works like with falloff map controlled by IOR?
#386196
The "dimming" you are reporting may be due to the Roughness setting, hopefully this video clarifies the usage of Additive blending for Clear Coats a bit more... I do show how the roughness can contribute haziness to the final additive mix and how to get around it.



This is an excerpt from a longer upcoming tutorial series, so it assumes the viewer already has a fair bit of fundamental knowledge about making Maxwell materials -- this video will be free along with many others when the series is released.
#386219
At the moment, the only way to produce physically correct coatings is the (unfortunately, somewhat limited) coating-function that Maxwell provides. Using additive layers should only be seen as a work-around, as it will produce artifacts and unnatural results in certain situations.
  • 1
  • 2
  • 3
  • 4
  • 5
  • 7
Will there be a Maxwell Render 6 ?

Let's be realistic. What's left of NL is only milk[…]