PBR Smoothness bleeding issue

McDev

Hello, not sure if this issue has a name already but I suffer from this quite often and I would like to know if it can be avoided.
The issue i talk about occurs for any PBR material which has different Smoothness/Roughtness values on it.

When there is an area on a texture which has high smoothness next to an area which has low smoothness there will be artifacts in the transitions.
I sometimes can void this by using the specular workflow and instead of having variable smoothness I use variable specular values.
Anyway this doesn't always work as this example below shows.

I see great examples of others with similar scenes but I am unable to achieve a proper look. Am I missing something?

McDev

I gave Substance Designer a try as I had the idea to manipulate the mask differently for each channel and I think that this can solve many issues. I am only not happy with this as there is no wetness on the dark tinted bricks. If I would increase the area of smoothness here I would end up with wide edges as on the post above.


Yet as this image shows it is definitely not working for all situations. Here I solved it by using Specular instead of Smoothness for the definition of the rough areas of a marble surface.

gnoop

While you first puddle example looks actually pretty realistic  and real life puddles often have such shiny halo ,   looking a bit differently although,   it 's definitely a problem with PBR, especially metalness approach in certain situations.      Too contrast roughness pixel values always makes such filtering halo and often looks totally wrong on distant mips.  

  I use extra Specmask/speclevel   channel in textures together with less contrast roughness for things like you marble example.      I believe U4 has speclevel input too.   

radiancef0rge

Water is not metallic, it should not be having this issue. 
Darken the albedo, raise the roughness, and flatten the normal.
The Specular value should not change. 

gnoop

radiancef0rge said:

Water is not metallic, it should not be having this issue. 

The halo issue is not because of metal vs non metal , it's because of too contrast  100% black to  100% white  transition in metalness channel .  Do same kind of contrast in roughness channel and you get similar halo too in both PBR approaches.  

radiancef0rge

 Do same kind of contrast in roughness channel and you get similar halo too in both PBR approaches.  

? I mean I guess its possible where the texture mips or has filtering there;s a small halo. But what your saying as fact - is incorrect.

Here is a plane with a 100% black to 100% white roughness texture. No feathering. 

@McDev Post your texture flats. 

McDev

radiancef0rge said:

Water is not metallic, it should not be having this issue. 
Darken the albedo, raise the roughness, and flatten the normal.
The Specular value should not change. 

It is about hte Roughness/Smoothness contrast. In a normal situation you have at least one pixel of 50% blending between two values. As I learned in my second post it is helpful to play around with the mask and make it highly contrasted.

I don't have the maps here right now, but my water has flat normals, high smoothness (low rougness) vs low smoothness of the bricks. Specular or Metal doesn't make an issue. The only thing is that the specular becomes wide while the values blend. I wonder if a proper linear workflow could make things better.

gnoop

Here is the screen from Substance Designer  showing contrast roughness making a halo with metallic PBR and nothing is metall at all.

Have no idea  radiancef0rge  how you get rid of it.    Some special shader?