[Hybaj]

By high ND you probably mean atleast 1000 because there's no way I see any darkening at the edges of aluminium (pure and 8 types of alloy), steel, silver, gold, bronze, copper, titan etc. So it's really not only silver.

he fresnel effect eventually disappears by the roughness. More rough means less ref90 influence.

Yes it comes from the way that the whole roughness effect is acheived via the most commonly used technique, which actually hits the brick wall of physical-reality so in order to make it real you have to develop some kind of a workaround. So yes it's physicaly incorrect. Do we agree on this or not?

+Edit:


Also the glossy glass-like materials suffer too (only one renderer as far as I know does compensation for this). Check the dark edge

Actually the only reason why i posted this topic is to see how many people recognize the "problem" (not a tragic problem really). I never really had any hopes of hearing anything constructive from anyone from the dev team or beta-team

[hyltom]

Hybaj, could you expalin you scene setup?
In your last image, i can see a gradient in the background, so this make me think that it tends to something dark...this could explain the black edge around the sphere.

Today i have made many test with some metallic sphere and i couldn't reproduce your problem...even with some roughness. The ND set to 1000 give me a perfect reflection of the background (means no black edge around the sphere).

[Hybaj]

Nope.. that's just vignetting (I was turning it off in some previous pictures.. but got lazy) I'll pm you with the scene.. you pm me with your scene (but before that render it again and turn down the exposition/iso a bit so we'll be sure it's not the intensity of light that burns the slight fade into total whiteness - because that's why you probably don't see the fade)

But speaking of time efficiency.. why bother here's the setup



And just to make sure that people won't burn me at the stake, i'll put in a picture with a perfect flat background (simple sky dome this time, the last scene setup really did a sliiiight gradient but there was no place on the object that was as dark as the darkest color on the edge of the ball)

[KurtS]

but with roughness=0 and nd=1000 - there is no problem...?

[Hybaj]

The only problem with ND is my philosophical beef with it. I really hate it. I somehow don't see the reason for it to exist inside a renderer.. definitely in a different form... but that's something this whole thread wasn't inteded for

Now the problem with roughness and reflectivity is somewhere between the method used for "roughing" up the surface, weird angles, and fresnel law. It's a problem only solvable by some coding. Glossy transparent materials look damn good after the fix - well they just look real without any abstract edge darkening (the "low" ND dark edges also look very abstract)

[hyltom]

Hybaj,

I have done many others test to try to find how you get these results and now i can reproduce it. However i think your setting to tend to these results are not accurate and realistic. You will see at the end of my battery of test in my next post that there is a solution to your problem.

[hyltom]

Skydome used for lighting (please for all these test don't pay attention to the maxwell material preview...the picture was not updated...forget to do that)

Test 1











Test 2











So on the test 2 it's obvious we can see the dark edge when there is some roughness applied to the material or when the Nd is too low. But i think this material setting is really not representative of a metal....anyway we got the black edge.

[hyltom]

Now the solution:

Your mistake come from the roughness setting. With maxwell we have to think realism to get the best result, so my question is: how a material is rough?....because it surface get some bump, scratch is it? so instead of using the roughness setting in maxwell, we have to apply some bump.



With a better looking metal, the solution give this:

[Hybaj]

Yes I know about this "workaround" but it can create sampling problems in different type of scenes where even the default glossy parameter is quite slow (btw.. the ND you have used to avoid the edge darkening looks not like any metal I know.. just a comment ) anyway thanks for pics

Your mistake come from the roughness setting.

Next time please refrain from telling users that it's their mistake when they use a parameter that should be working well in such an advanced version of the renderer like 1.7 should be

anyway we got the black edge.

And it's been with us for a looong time

[hyltom]

Yes I know about this "workaround"

Great, ... thanks to make me waste my time! If you knew there was a solution, why not mentioned it from the beginning. I wouldn't have think about helping you and let you express your anger against NL team for a problem that is not really one.

EDIT: No ok, it's a problem, but there is a solution. So no need to make a big deal of this case.

[Hybaj]

Hyltom lol just chill down Why should I be introducing a quite innefective solution?? Innefective solution is not a solution. It's something like a half-truth - worse than a lie.

And I'm not angry toward the NL team at all (Maxwell is pretty cool now). That's merely just your opinion or "feeling". If I was angry about this situation I would post a thread some years ago when I first saw it being discussed elsewhere. But I already did mention it in a little number of my previous posts but it never got any attention. And now that I had a scene that clearly showed the effect and even my customer pointed that out, I thought it was time to raise the public awareness of this problem

Actually there was a situation in this thread when my hero Tom (Not really.. he's cool) chose a very diplomatic answer and I was sad. It's like my heart was broken. I guess i'll play a slow polka rhytm today in the memmory of the "less diplomatic" Tom that I know from this forum for a long time (not personally of course).

So you see i'm a complete angel

[hyltom]

Personally, i was not aware of this problem until you point it out few days ago and i come to check this forum almost everyday since 2 years. So maybe the old thread was not as explicit as this one.

Anyway this new thread as open my eyes to this matter, and i can tell you that from now i will change the way i make my rough metal.

[Hybaj]

Great you see? It had atleast a small positive effect. That's my goal

[hyltom]

And why do you call this an ineffective solution? For me it appear to be the most realistic parameter for a rough metal.

[Hybaj]

I'll explain.

Do you realize how most people are touchy when it comes to speed in Maxwell?? How they always cry for optimizations??

Now i've been using very fine bump maps for ages (I put alot of effort into realistic bump and roughness maps.. that nicely breaks the synthethism that the "pure" renderer materials have) and I know alot of situations where the noise clears very very slowly in comparison to the roughness parameter. So it's really only about speed which is ussualy one of the most important factors for alot of people. Your ussual scene will deal with the speed quickly because you have alot of light comming from relatively large emitters. But what about other people?? Actually they're not bothered by this whole phenomenon at all - questioning the dark edge has been avoided quite effectively. So this whole thread is not a question of life and death.

And the word ineffective comes from my experience with the marketing of maxwell. Realistic, physically based... well.. the most commonly used roughness parameter still doesn't have the "needed" compensations