What does the Specular channel in UE4 actually do.
polycounter lvl 12
I am doing some experimenting with materials in UE4, and it is straight forward enough to understand, but to me it feels like their metalness workflow is limiting compared to how Marmoset does it. In Marmoset set I have full control over the exact specular value I want within my texture, but in UE4 I only have control over my specular if it is metal. This feels very restricting as there are some technically non metal materials that require a higher specular value then the default non metal one UE4 is assigning. What is also really strange is they have a specular input but it doesn't seem to do anything at all, so what is the point of it?
I would like to see what you guys think.
I would like to see what you guys think.
Replies
In UE4 when you set metalness to 1 it will use the value/color in your base color as the specular, but when you set it to 0 it sets a default value that you have no control over. It has a specular input, but it doesn't seem to have an effect on anything not matter what kind of inputs I use.
but for non-metals it give more control over the specular independently of the diffuse value
for example, you cant do this in Marmoset I think while keeping the same diffuse.
Also on top of that this process is wasteful because it requires an extra texture to be authored, instead of a standard spec map by itself.
So it is just a tweak you can add to non metallic objects to boost the spec or tone it down if you want. However, the workflow just seems like a bad idea to me, and kind of restricting. What do you guys think?
The metallic value is for saying what type of material it is and what shader model it's using.
I'm simplifying but a non-metalic material will have glancing angle reflections that are the same color as the light source and based on the IOR of the material. A metallic material will react differently to light than most organic materials and it will tend to have color to its reflections (for example gold has gold reflections, it absorbs certain wavelengths).
So you have two different shader types that you're switching between (0 for most things 1 for metals). The roughness is a value that tries to describe the look of the surface if the surface isn't perfectly polished (imagine tiny scratches in the surface, or a texture to it).
The specular value is currently a tweak value to add extra control to the non-metallics and it can be used to tone down the reflections. You might be making skin for example and not want to make the surface too rough and lose the definition of the highlights but still want to reduce the overall strength of them to match a reference. In that case you could use the specular value to bring the highlights in a bit but keep them from being too dull.
I'd expect IOR values at some point and maybe they'd get rid of the specular tweak then, but if it helps when trying to match a reference, there are so many lurking variables as it is with surface details and such just do what it takes to get close.
So then we will get no control over our specular values on non metallic surfaces? That just doesn't seem like a good idea from an artist point of view, because it feels like I am being needlessly restricted.
"control" is the key term here. In PBR - you don't get 100% control over the way specular works. The math handles a bunch of it in a much more realistic fashion; preventing you from making materials that aren't physically plausible, giving better results that weren't 100% possible before, and taking some control away from you.
Keep in mind if you made something super rough and non-metallic it would effectively look like it had very little reflection. You have some control, it's just different than before.
In addition - the entire reason this specular value exists is to push or pull that PBR result in a way that you deem artistically fit, outside of what may be physically correct. So they have given you even more artistic control than most PBR implementations give.
The point being - you should always START with something based in reality, and only push it outside of that realm when you want to achieve something specific artistically. This is the philosophy used by many already - including Disney and their CG films. In the past we instead (as an industry) made a lot of mistakes and assumptions that led to wildly unrealistic results, and most of the time it wasn't on purpose.
Clearly there is an argument for both having more control from a pure PBR approach, or more accessibility/intuitive asset creation from a metalness approach. The benefits of both could be argued pretty heavily, and it probably depends a lot on the artist and their personal opinion.
It's fine to disagree with the ideas behind UE4's approach - but that's different than saying it's a completely "unnecessary" approach. Clearly they felt pretty strongly that it was necessary, and it wasn't an arbitrary decision.
I can see why they did it, but having a choice would have been nice, just like Marmoset does. You have the option of doing the metalness workflow or the specular map workflow. I feel like giving artist more choices is the better approach.