[Official] Specular vs Metallic workflow

it says right in the release notes, there were changes to the standard shader. So if you authored textures before, they are going to look different now.

at this point it looks very close to Marmoset Toolbag 2. Which IMO is a good thing.

I have not tried the Metal/Rough workflow yet, so I cant speak for how that one looks. But spec/gloss still works fine I think they just changed it.

it may be an issue with substances? did you try baking them? substances have been a constant source of pain during the unity 5 beta, i would just bake them out for now..

 

Can someone clarify this for me? I use SD4 and substance painter so I am a bit confused.

Metallic goes into the R channel
roughness goes into the alpha
roughness has to be inverted from what it is in SD4 as well?

Also, please get rid of the term smoothness, nobody uses that word. Its either roughness or glossiness. There is no reason to use the word smoothness.

Honestly I would prefer to not have to pack roughness and metallic together...

 

The tooltip on the Metallic input says: Metallic (R) and Smoothness (A), while the tooltip of the Specular input in spec/gloss shader says: Specular (RGB) and Smoothness (A), which means either that:
1. Metallic workflow uses glossiness instead of roughness, which is confusing
2. Metallic workflow has an error in the tooltip, which is confusing

Honestly I would prefer Unity to not support metallic workflow (as it's the limiting workflow of the two), so they could focus on keeping the spec/gloss shader as close as possible to Marmoset Toolbag. I've fiddled a bit with both Quixel Suite and Substance Painter, and I don't see any point in the metallic workflow at all. Both these tools allow you to modify materials 4 maps at a time, and I don't see how it's easier to set your metallic property to white and base-color to yellow for gold, than it is to set diffuse to black and specular to yellow. You have the same number of 4 maps that you need to set for both metal/rough and spec/gloss. And unless you're a total noob that heard the word PBR an hour ago and have no basic understanding about the values that go into each maps, metallness workflow is not in any way easier to use. It's also as easy to screw up the values in metallness workflow as it is in spec/gloss.

 

If they've done their metallic shader the way I would expect (just transforms inputs into specular color and uses spec color internally) then they don't need to drop metallic to "focus on specular" at all (they just work on specular, and the metallic bit doesn't need to change at all).

 

A channel combiner/splitter/packer would be a useful feature/tool (anything on the Asset Store?).

As would picking layers/groups from Photoshop files...

And making FilterForge presets directly in the Editor....

and.... I'll shut up now.

 

I really hope they add real roughness not just smoothness. i hate gloss. i'm a CGI guy so roughness is where all the numbers in my head are. also i would like to see more control over the textures once added. like a levels control for roughness. most programs use roughness differently. also the metallic version needs a specular offset. once thats added you will be able to get the same range of values that spec can. currently it doesn't support wet surfaces which sucks. also can we get distance based gloss/roughness at some point. rather than the basic reflection blur. it should be sharper at contact points as roughness is a cone.

bitching aside. great work guys! also i'm looking forward to the reflection cards that my friend found in the docs. =3
absolutely loving what i've used of unity 5(been using a friends computer to play with things).

 

Ok so here is a problem that is showing up. I am having to adjust my roughness in photoshop just for unity which is unacceptable. Also my Toolbag shots look nothing like my unity shots (aren't they supposed to be very close?)

First here is the roughness map in Toolbag 2

Now here is the same roughness in Unity (specularity is messed up)



Here is the roughness map after it has been adjusted in photoshop (character look less greasy).


PBR is supposed to be more "scientific". I should not have to adjust roughness maps to make the specular look right in unity.

What gives?

 

Something about roughness in b13 is screwed up, it looks totally wrong. I'm seeing it on my end as well - I've got the sci-fi helmet from the Quixel site. I can plug into Lux and Skyshop both and it looks perfect in either one, but when I import it into b13 it looks way too shiny.

 

If you all file bug reports, with examples, it would really help our developers to tweak it and make it better!

 

Bug filed. Case #649997

The example is a Test scene which has the sci fi helmet example from the Quixel website. As mentioned, it looks great when imported into a 4.x project with Lux or Skyshop, but looks totally wrong in Unity (I did try inverting roughness/smoothness by the way, it didn't help)

 

Ahh, I am suppose to use a rough/gloss map in the alpha of my spec map? ok... Is this discussed somewhere? It wasn't mentioned in the video tutorial about the standard shader...

(d'oh, nevermind. I wondered where that gloss slider went... *note to self: read the little hint windows)

 

You know that everybody with the link is able to access your whole bug- and asset store log history, right?

 

Huh, didn't realize it included asset store history too.
Guess I'll just include the case number and leave out the link.

 

in addition to that article, the artifacts that sometimes pop up in metallic workflow are mentioned here as well:
http://www.marmoset.co/toolbag/learn/pbr-conversion

both the author of quixel's smart material library, and Mr. Joe Earthquake Wilson at marmoset say these artifacts can be an issue with the metallic workflow...

 

Is this still going on? Make a layer over which either of specular / metallic you choose that lets you choose common real everyday things in your house and outside to create shaders with. It's not like the variable for a specular / metallic workflow can't be exposed underneath that layer. Unity is supposed to be about democratizing game / app development to those that don't have degrees in physics or mathematics. Sometimes the speed of light is so slow.

 

Not really feasible in the core product. Not like you could have a "chair" preset. A chair is made up of multiple different materials. You could maybe have a wood preset, but it would have no way of matching up w/ your unwrap. Something like this is best left to your content creation tool, several of which have presets like you're asking for. Check out Substance Designer, Substance Painter, or Quixel's DDO. All of those have presets you can use to get realistic values onto a map that Unity will accept.

 

I'm pretty new to PBR. I personally like the metallic/roughness model better... it's just easier to set up. I wish it was actually Metal/Roughness instead of Metal/Smoothness though to match how it's done in Substance designer. Easy enough to just invert the alpha in Substance, but it's an extra step that makes it a tad more difficult to preview what I'm going to get while designing a material.
It's great to have both types of workflows though... it seems as though spec/gloss might give better control and possibly a little better quality for certain types of things.

 

theres no quality difference between metallic/rough as long as it has a specular offset. only thing would be certain extreme unrealistic materials. metallic rough also follows the laws of physics more as all materials can be broken into conductive/non-conductive. the mix area between is for when you have things like dirt on metal. for more reflective non-metal surfaces you can use the specular offset.

 

hello, to get albedo calibrated I m using this tool : https://colortarget.wordpress.com/
it gives pretty good results

 

Sorry for bringing this up again, but I've just read the changelog to Substance Designer 4.6, and there was a thing that caught my attention, and maybe Wes can comment on it in this thread.
* [3D View] Add specular level channel for dielectrics in metal/rough PBR shader
I haven't had the chance to check out what that means yet. Does this mean you'll be able to assign specular values other than 0.04 to dielectrics using a separate map? If that's the case, isn't it easier and more efficient to just use spec/gloss workflow?

 

Hi,

Yes. This option allows you to control the F0 for dielectrics in for the metal/rough workflow. It is the same as the Specular channel that UE4 uses. It is mapped 0.0 - 1.0 where 0.0 = 0 and 1.0 = 0.08. This means that 0.5 (default) corresponds to 0.04 or 4% F0 for common dielectrics.

We have opened specularLevel as a texture sampler on the PBR metal/rough shader. You can utilize a texture to control this channel.

Cheers,

wes

 

Thanks for a quick reply and it's good to hear you're giving more control over this values!

Is 0.8 a typo? Or is there something I'm missing?

 

Oh, sorry. It was a typo. I fixed my post. It should be 0.08. Sorry about that.

Cheers,

Wes

 

In the latest release is the metalness shader behaving properly yet?

 

I don't think the specular shader is. All my textures are super glittery in the b17 version of the specular setup shader but were fine in b15 and before.

 

So I should be seeing roughly the same rendering in Marmoset Toolbag2 and in engine in Unity 5 for the metalness workflow?

I'm currently using Quixel DDO, things look great in previewer and then look wrong in Unity. I'm starting to belive that Quixels Unity 5 Metalness preset is either generating the maps wrong or at least sending the wrong data to the different channels (rgba).

 

1. If you're using DDO, why do you use the matalness workflow?
2. IMHO Unity needs to stick as close as possible to Marmoset Toolbag2, as that's the more or less industry's standard. If every authoring tool has fine exporters to Marmoset, then we're all set, and it's that authoring tool's job to have the Marmoset calibration as close as possible to Marmoset.

 

I'll try switching to specular and see how things go.

 

Hi,
We have released "The Comprehensive PBR Guide," which covers the theory of PBR as well as practical guidelines for application. It is divided into 2 volumes.

Volume One: The theory of physically-based shading and rendering

Volume Two: Practical guidelines and workflows for creating PBR textures. *Coming Soon.

Download the Free guide from our site
http://www.allegorithmic.com/pbr-guide

Discuss PBR theory and workflows in our official support forums.
http://forum.allegorithmic.com/index.php/board,27.0.html

Cheers,
Wes

 

Great resource Wes. Your first link is broken, should be? http://www.allegorithmic.com/pbr-guide

 

Oh, yes. I had a typo when copy/paste. Thanks for the heads up! I updated the post.

Cheers,

Wes

 

@wesm It seems like the Unity shader is constantly changing. Are you goings going to be updating the Unity exporter in Substance Painter to match the up to date shader?

 

Hi,
Yes, we are watching the changes closely. We will make sure that it matches for the public release of U5. We are working on keeping up with shader changes. I think we will not make any adjustments to SP's export until the U5 shaders are finalized.

Cheers,

Wes

 

@wesm Thanks, that's good to hear!

 

Nice PDF @wesm ! Very well written. Good to see somebody like Allegorithmic trying to make sense of all of it

 

Thanks Wes. It's because of support like this, both your involvement on these forums as well as the resources you guys put out, that motivated me a few weeks ago to switch to Allegorithmic products entirely.

I haven't looked back.

Quixel's products, especially their megascans, were neat. But their notion of customer service is terrible. Diverting all customer support to a Polycount forum that they sometimes read and sometimes respond to based on the thread is complete nonsense in the 21st century.

Any way, thanks again!

 

At this point, is there any chance of keeping both 'flavours' (Specular and metallic) on the final release? Or are you going to discard one of them anyway?

Thanks

 

As they have both now, why would they want to discard one of them?

 

That's what I think, Jashan. But maybe they are decided to keep just a single one... To keep things as simple as possible.

Edit: Unity will keep both, as stated before in this very thread. Sorry for asking.

 

I thought I was the only one.. Agreed, quixel is a joke.

Buy yes, it's great to see more pbr reference charts from unity and this paper from allegorithmic is interesting as well. Thanks!

 

I clearly favor specular one.

If we are going for a physical model, then let's do it properly and use actual physics. The refractive index is a complex number, each RGB channel has its own complex refractive index, so must be represented by two real numbers. That means RGB diffuse and RGB specular. Not RGB-something-unclear.

There are useful materials that are between perfect dielectric and perfect conductor (gemstones, I need those), and the metallic workflow does manage them well; for these you need the specular model. Otherwise it's all guess-work again.