PBR - Rendering difference between PBR and Blinn-Phong on a matt non metallic material
expected
Gael Streeter
I do not know if it is "normal" but I noticed a troublesome difference of rendering between a PBR and a Blinn-Phong matt non metallic material which is very annoying for the heads and bodies business notably.
My use case is a mesh body set in Blinn-Phong and a mesh head set in PBR. In both it is a rough, not metallic skin material (a texture and a normal).
Here is the result : https://gyazo.com/d1cb0ad0e52cc5bcfa632188f24c95dc
As you can see the head does not have the same color as the body....
To compare here is the result when both are set in Blinn-Phong : https://gyazo.com/0e5b4151a9bee5382b89e45a7509ad55
This brings the compatibility between the various heads and bodies problematic...
Log In
Dan Linden
expected
Dan Linden
Hi Gael.
I do see the difference in the screenshots you provided but this is as close as we can make it while using a different lighting model. This is expected behavior.
Dan Linden
under review
Allegory Malaprop
They aren't the same, nor are they intended to ever be. That's going to be something the body part makers (or whatever other items that might straddle the two) are going to have to address.
Gael Streeter
Allegory Malaprop Yes indeed ! And that customers will have to deal with too... If you have an "old" Blinn-Phong head and a "new" PBR body for example...
Ceri Quixote
I would consider this expected behavior - different materials will produce different results. As PBR is adopted, heads and bodies will need to use the same type of material to avoid differences from lighting - this is no different than what happens now when one body part is using materials and the other isn't (or is using significantly different settings).
Gael Streeter
Ceri Quixote The problem is that the customer can not "fix" the issue if the body (or the head) does not give the two options : BP or PBR...
Ceri Quixote
Gael Streeter Agreed, that could be painful - but that doesn't mean it's fixable when two completely different lighting systems are in play.
I tried duplicating your results by splitting a sphere in half and using PBR on one half and BP on the other with the same normal map and no other materials. I'm maybe 80% sure I can see the difference - but it's really subtle. https://gyazo.com/feed73a8c60c6bb5f036eaed62517d55
That's with just a normal map though, which you'd almost never use by itself. If you add Specular (BP) or tweak the Roughness at any level (PBR) the visuals become radically different and there's no way to get them match. https://gyazo.com/eda74a831ccd5d0f7597868f43ebdbed
Particularly with skin, a certain level of subtle gloss is desirable and should be considered normal. At that point, getting body parts with two different material systems to match up basically involves disabling materials on both of them - which is what is done now.
I am really excited about the potential of PBR skins on mesh bodies. However, issues like this which exist because we're trying to get parts from two separate creators to work together generally require the creators to agree on a standard first. There's going to be a transition period as the new systems get implemented across brands.
As a plus, since PBR doesn't allow disabling the rendering of materials people will actually be able to see what they're working with - the rush of issues I anticipate seeing is from people who normally run around with ALM turned off and have no clue what their current material settings look like.
Ceri Quixote
As a note on the two images I shared. On the second image the BP side has an even all-over gloss, while on the PBR side the intensity of the glossy effect conforms to the shape of the sphere. This is due to PBR using a more realistic lighting model. We are probably seeing the same difference in lighting in the first image - it's just incredibly subtle without a gloss effect.
Gael Streeter
Ceri Quixote By default, I set the skin material as mat at the neck level to avoid any issue with the body transition (expecting the body does the same). For a full shiny or metallic skin material the solution would be to do both head and body by the same creator...
But here my concern is not about different shiny materials set on the head and on the body. My concern is about the fact that even without any shininess or metaliness the neck separation is visible between Blinn-Phong and PBR....
Ceri Quixote
Gael Streeter Right, but even without shininess you're still using materials so the expectation is that it will interact differently with light - because what materials do is interact with light. Glossiness makes the problem obvious, but even without glossiness it is relevant that the rules governing light are different. PBR materials conform better to the shape of the object than BP materials do. PBR materials will react differently to EEP. These are innate qualities - we are literally changing the rules of physics. The solution is to use the same rules on both sides of the equation.
Neck seams have been an issue for as long as we've had mesh heads and bodies made by different creators - easily 10+ years. Entire sub industries have grown up around fixing it. The problem was actually solved about 2 years back when Lelutka promoted their neck fix. The neck fix isn't a technical solution - it's a set standard that any creator can use. Trying to get two completely different lighting systems to match across a mesh border seamlessly is fairly equivalent to trying to get normals to match based on guesswork.
Gael Streeter
Ceri Quixote Yes I follow this Neck Standard for all my heads as most of the other heads and bodies creators now do. That is how we have now no visible neck seam in Blinn-Phong ! That is not the subject.
My concern is that with the coming of the PBR material we will have to face again visible neck seams and the creators and customers will have to deal with it during a short/long period of transition.
Ceri Quixote
Gael Streeter Which I understand. But it is a transition period and it will resolve itself. These are fundamentally different lighting models and a very limited case scenario where there is even an issue. There is also an established solution (don't use parts with different material systems) that will continue to work through the transition period. Not using materials still works. Both pieces using BP still works. There is a transition caused by some creators adopting PBR sooner than others - but that doesn't invalidate the first two options.
Ceri Quixote
Looking at an actual solution. We are attempting equivalency and the problem exists because the equation on one side is radically different from the equation on the other. As implemented neither system is currently capable of modeling the other.
glTF 2.0 has an extension called KHR_materials_specular which allows specular to be rendered within the PBR model. With that extension it might allow rendering legacy BP content using PBR. That would have both sides using the same model and should solve the problem.
I know LL is working on PBR extensions for IO and transmission - I don't know if the specular extension is in the pipeline. It would be a nice feature addition above and beyond this one specific use case, but I'm not sure if it's even possible to implement it in time for it to be relevant to this scenario - the transition period might end before they could get to it.
Cooter Coorara
Geez. To be honest, I see no difference. But take into account that I'm a guy so...
Gael Streeter
Cooter Coorara lol There is one ! ;)