[Official] Specular vs Metallic workflow

Hello,
With the introduction of physically based rendering in Unity 5, we need a workflow and way to specify materials that makes sense.

Currently there are two main approaches, one based on specularity and the other on how metallic something appears.

We would like to learn more about the current and planned tools and workflows to author and create physically based materials for use with Unity. Some interesting info for us is:

What type of game(s) are you trying to make?
Do you use on "scanned" data?
Do you hand author textures?
Which tools do you use?
Do you have a preference for specular or metallic or is this all new to you?

Thanks!
Brett

(@Aras edit: we'll try to have both metallic and specular Standard shaders in 5.0 beta soon)

 

We're not trying, but making a stylised pixar ish title that's graphically heavy.

We roll our own "PBR" shaders an awful lot so with that in mind:

We don't currently use scanned data, we think it's a red herring as it doesn't really get you far for a lot of games unless you want to limit yourself to scanned data, so it's fairly essential to support artists.

Yes - but as a feature of Unity, it would be nice if Unity would be able to take all of the textures our tools (like Painter) spits out (AO, Gloss, Specular, Roughness, Diffuse, Base Colour, Normal, Height, Metallic) as separate textures, and have Unity combine / optimise them down into new textures - non destructively optimising.

In short, Unity could potentially use any shader design it wanted to, but create new textures from unoptimised source textures to fill the inputs that suit it's shader, and any optimisations it would require such as baking to a particular channel like alpha.

Failing this then it would be a requirement to support both Substance Painter and UE4 workflows.

We do not use the Metallic PBR style. We have a sort of specular approach where brighter pixels offer less roughness (or more shininess/reflectivity). But that's mostly because it's artist friendly, and we create our own shaders.

So far as tools for texturing/materials? We are currently using Substance Painter a lot alongside Photoshop for hand painting + texture overlays, and we use bake information (AO etc).

 

I don't have a real preference myself. Specular has the advantage of a little more flexibility for non-realistic materials, but you can always use a custom shader for that if needed. Metalic has the advantage of being a little more widely used in my experience. My main concern is that there's a clean workflow for authoring them and getting them into Unity as authored. I'd love it if Unity and/or Allegorithmic would release shaders for Substance Designer/Substance Painter that match Unity's rendering (either Metallic or Specular) so that we could get an accurate idea of what it's going to look like while authoring. Metallic seems to potentially allow for using one less texture. If you guys can find a way to leverage that to give better performance or a more flexible standard shader, then go that route please. A set of tools to convert textures of the unused type to the correct format would be awesome.

We don't use scanned data currently, but I am looking forward to trying out Megascans when it's released.

We currently use Substance Designer/Substance Painter for a lot of stuff, so a well supported path from those tools would be awesome.

 

+1 for this. I actually use ShaderMap only for combining different maps in the color channels of a texture.

I've found that in Unity 5 Specular could be better interpreted as "Reflectivity" and provides a very intuitive and straightforward method for defining how much reflection the material should have. To me, it's much easier and flexible to use than the metallic map, which is typically black or white.

(Pictures borrowed from the Marmoset PBR page)



In addition, there are are are some useful material charts with reflectivity (specular) values that can be used directly. In the chart below "microsurface" refers to the glossiness value:



So, in sumary:

Games with vehicles: car paint, car interiors, roards, buildings, outdoor sceneries...

I use charts like the above one that give indicative values for common materials. Values are later fine tunned for each specific material.

Not myself, but I want to have some definitive guidelines I could send to 3D artists so they could author the textures properly. The charts in the Marmoset PBR page are the best reference I have so far.

Allegoritmic's Bitmap2Material and ShaderMap 2.
Adobe Fireworks for authoring reference maps (i.e. checkerboards, gradients, etc.)

Having compared how metallic and specular maps typically work, I feel that specular map (I'd better name it "reflectivity") is much more straightforward to use and easy to understand. I think it's far more useful than metallic map when it comes to modeling the fictitious materials typically used in games.

 

Thanks for asking the community about this!

I do have a slight preference for metallic, because you can pack it tighter into the memory, plus, it's harder to make unrealistic materials. I also prefer it because Unreal 4 has gone this route. (Although they seem to have some sort of hybrid)

However, I also want to stress the importance of being flexible as a game engine. Unity should definitely support both.

 

What type of game(s) are you trying to make?
I am working on a audio/visual experience utilizing the Oculus Rift. Visually the app will be a mix of hand painted and more realistic textures.

Do you use on "scanned" data?
No and we don't plan to either.

Do you hand author textures?
Yes we do.

Which tools do you use?
We are using Substance Designer and Painter, Bitmap2Material and Photoshop to do texture work.

Do you have a preference for specular or metallic or is this all new to you?
We feel very strong about a Metallic approach as it simply feels more intuitive and natural to work with. Substance Painter is our main tool and it's wonderful to work with. We would hope to bring our textures into Unity(5) and have it look the same using the new Standard Shader - which is currently a challenge and requires some modification/workaround on our end. The new Standard Shader isn't really fitting our needs.

Note: I wish the new "Standard" shader would just offer to allow for both workflows. It seems a bit odd to wanting to settle for just one.


Thomas

 

Note we are planning to support both, but we might only support one initially due to time and testing constraints.

 

Different types of games including games that have a more realistic look, a more cartoonish look and a more "stylized" look (glowy stuff with really intense colors like this):

​

For more photorealistic games / applications (game based learning), that's certainly an important option to have for us. It's not something we have a lot of experience with, yet, though.

You mean with Photoshop or the likes? Or with tools like Substance Painter? Either way, it's yes ;-) ... but with Photoshop, just for really simple stuff like simple tiles ... and also doing some corrections to textures baked or created with other tools.

Substance Painter, Substance Designer, Photoshop.

At the moment I prefer specular because that's what I've used with the Unity betas so far ... when using Substance Painter, I can export whatever the engine supports (but I agree with Thomas: so far, I wasn't able to reproduce the results from Substance Painter in Unity easily even when using the Unity export options ... and that would be something that should work easily).

 

I am on the same page as Thomas. Developing for Rift, mobile and desktop platforms. Different types/styles of games require different workflows. That being said the Metallic route is the one that matches best the tools I use. Substance Designer, Substance Painter.

Gary

 

Another vote for some kind of texture packer utility here, even if its just a script available as part of a standard asset bundle.

I can't answer all the questions properly at this stage as the team I'm likely to be working with hasn't fully formed yet. But it's looking likely that I've sold the art guy on PBR and they got very excited about Substance Designer and Substance Painter.

We have the additional complication of being unsure if our title will be done in Unity 4 or 5 yet, for example due to unknowns about how swiftly certain 3rd party unity asset store items that are key to our project will be made Unity 5 compatible. So lately I've been trying to come up with a workflow for the artists that allows me to hedge our bets and fall-back on a 3rd party Unity 4 PBR shader set if something doesn't work out with Unity 5 for us in time. The fact all the previous Unity 5 announcements, videos etc have shown a specular workflow/shader slots has made this quest something of a challenge due to the somewhat prolific support for metallic in tools & shaders to date, though I reckon it will turn out ok for us one way or another. So of course I was very interested to see you mention that Unity will support both eventually, although I suspect that progress and news on that front won't come quickly enough to strongly influence the decision I will have to make in the coming weeks.

 

On a related note, is it possible to confirm that I got the following information right (from watching past unity videos) and that it is still the plan for Unity 5 standard shader, at least in the first version?

The alpha channel of the specular texture is used to store smoothness map. If there isn't one, you get a slider to control overall smoothness.
If opaque, cut-out or transparent mode of shader is used, info comes from alpha channel of diffuse texture.

Cheers.

 

I started using substance painter and substance designer about 2 weeks ago. Since then i cant really think of going on without them. Creating materials (on hand crafted basis) based on the metallic workflow and using substance painter to paint the models is definitely the workflow of the future. It is fast, predictable and intuitive, saves tons of time and delivers excellent results. Going with the old specular stuff... not with me.

 

For the sake of clarity the specular stuff mentioned in this thread isn't the 'old specular stuff' but rather the PBR spec/gloss approach, which also seems to have several other names just to ensure that this stuff remains slightly harder than it should be to learn.

Tools like Substance Painter should be able to handle the Unity 5 PBR approach in a way that doesn't totally spoil the workflow you've already been trying in recent weeks. Although its always possible I got something wrong or there is some devilish detail I haven't considered.

 

Hi,

Substance Designer and Substance Painter both support Spec/Gloss and Metal/Rough PBR workflows.

Substance uses the Disney (Principled) GGX BRDF. The BRDF is what is most important. Unity 5, I believe, is also GGX (Not 100% sure on this) as is UE4. Marmoset Toolbag also now has a GGX option for reflectivity. The map types themselves come down to workflow.
Of course, there still can be deviations between applications (using GGX or not) depending on how the BRDF is implemented because with PBR, there isn't a specific standard per se. For example, a shader implementation could have different tweaks to the BRDF for specific purposes. However, GGX is becoming the most widely adopted method and map types come down to personal preference or studio choice.

We have seen metal/rough to be the dominant workflow. On a personal note, I much prefer metal/rough. The metal/rough workflow is less prone to user error as there is not a specific control over the F0 (fresnel 0 degree) reflection value in the maps themselves. Some engines such as UE4, provide a specular option to override the default 0.04% reflectance value, but this is for dielectric materials only and reserved for custom material exceptions.

Substance Designer has a Metal/Rough and Spec/Gloss PBR Shader. When you create a substance, you can choose to work in with spec/gloss. The Spec/Gloss PBR shader is still utilizing the GGX BRDF, but the workflow for creating the maps, which is spec/gloss in this case, is different. It's different only in how the artist creates the maps. The results are identical in metal/rough and spec/gloss if the maps are authored correctly. You can create a material using spec/gloss or metal/rough and the result should look the same in either metal/rough or spec/gloss PBR shader. This is because of the GGX BRDF.

Our Spec/Gloss workflow within Substance Designer works very well with the Unity 5 Standard Shader. So, if you are working with Unity 5, you would choose the PBR Spec/Gloss shader and use spec/gloss outputs.

With Substance Painter 1.0, the PBR shader is currently Metal/Rough. However, we have an option when exporting to convert the metal/rough maps to spec/gloss. We currently have a Unity 5 export preset that handles this. The preset also places the gloss in the alpha of the specular for the Unity 5 standard shader. We will also implement our Spec/Gloss shader directly into Substance Painter as well in a future update.

In Substance, the PBR workflow uses the following map types.

Metal / Rough
Base Color (albedo)
Metallic
Roughness
Normal
Height

Specular / Gloss
Diffuse
Specular
Glossiness
Normal
Height

Cheers,

Wes

 

Could you clarify exactly what you mean by metallic and specular? because specular could mean different things to different people. I suspect some people don't know what's meant by Unity specular vs classic specular (old school) shading.

 

Thanks for taking the time to ask the community, its a very important question.

What type of game(s) are you trying to make?
Most of my client work is based on creating unique experiences for tablets, web and physical kiosk installations. As such I rarely make a 'type' of game/project, they vary considerably in scope, style and art direction, but one unifying aspect is that we always try to make them look as good as possible for the budget/time.

For this reason I've been really pushing Physically Based Rendering for the last few months and bought Marmoset Skyshop to offer it to my clients. It will push up overall costs due to the additional assets (texture work) but ultimately if they want their projects to stay up to the current state of the art then that's what they are going to have to pay for.

In terms of actual projects I'll use this on, it will range from product promotion (e.g. vehicles to watches), sport games (mainly on people with a nice skin shader too), and anything in-between such as The Queen Elizabeth Prize for Engineering Trophy Designer should I ever need to update it.


Do you use on "scanned" data?
Not directly but I am using Quixel Suite and their Mega Scans which do.


Do you hand author textures?
Will depend on project requirements and objectives, though overall the aim is to use something like Quixel Suite where possible to speed up production. Hand authored will be for special cases and touching up, where I believe that a metalness work flow will pay dividends as its easier to understand what effect changing the maps will have.


Which tools do you use?
Marmoset Skyshop
Quixel Suite
Substance Designer - will be looking into this soon.
xNormal

One related aspect is I keep meaning to ask or double check if Unity is using mikktspace for its Tangent calculations (like xNormal does) as it solves a number of issues and ensures that you'll see the same results across different renderers/engines. It was developed by Morten S. Mikkelsen during his master's thesis and is free for anyone to use. It is contained in the two standalone files mikktspace.h and mikktspace.c.


Do you have a preference for specular or metallic or is this all new to you?
I voted for metallic , though actually supporting both would be best to give people the option.

I would rather metallic as I feel in terms of producing assets by hand its far more understandable what the results of the values you are applying will be. It also helps to differentiates itself from traditional spec/gloss methods to avoid artists falling back into their old methods, where they would often fake specular highlights.

This is a great webpage (http://artisaverb.info/PBT.html) that explains the differences and provides the arguments for using metalness maps instead of specular

 

Rocking Summary @wesm !!

As the fellow who's been providing a Metal-Rough PBR solution for Unity for the past year or so, and training studios on bringing PBR to their projects, I wanted to chime in on this.

I yell a lot about how I _personally_ enjoy the Metal Rough workflow the most, but I'm going to put that aside for a moment and make a couple observations about PBR in Unity, and how it sits within the general spectrum of the parts of the community I've had dealings with, and then propose some questions here. These are just my observations, some of them might be over-generalizations, so take from this all what you will.

1. High-level operators (ie. folks who are versed in AAA/AA-level authoring pipelines seem to be preferring the Metal-Rough workflow, but I wouldn't call it a blow-out preference. 70/30 if I had to guess.

2. Folks doing split-development between mobile and desktop are not trying to use the same maps from what I've seen. Instead they're using things like Substance Designer to Decimate their full-detail PBR map-stacks down to Mobile-oriented Diffuse + BlackWhiteSpec + Normal sets., and using a legacy shader. Its fairly obvious why. The mid-range mobile GPU (that one needs to target for deep user penetration) blows goats for quarters, and the trade-off of fancier shaders vs. more content/vfx/level-size/battery-life simply _doesn't_ work out well for better shaders.

3. I don't use raw scan data. No client, working partner, etc. to my knowledge has used raw scan data, except in the creation of tech demos that show off the use of raw scan data. At best, I've seen folks pipe it into a more complex texturing pipeline where its being used as an intermediary, rarely ever as a final output. Usually just the normal map and an extracted occlusion map end up being truly useful. Scan data just ends up being waaaay too noisy for most applications (imho).

4. And this is the big one. You're marketing a _fundamentally_ pro-only feature as though it wasn't. It is. PBR without linear lighting, hdr, and postfx is borderline useless. I was asked this (awkwardly) at the end of my Unite talk by @QuantumTheory . It's just as much of an awkward issue now. Every indie user asks this when I give a talk on PBR, and this will continue to be an issue throughout the U5 release cycle.

So this provokes the following questions for me.

Are you configuring this system towards the Unity-Pro licencees, or are you configuring it for Free Users?
-- It is important to understand that a configuration built for one user-base/skill level/production scale will almost certainly be mal-adapted for the other. The notion that you can design one shader for all cases is just silly.

What's the impetus to try to create this magical shader that 'just works' on mobile?
-- No-one I work with is going to use that vs. a hand-tuned set of their own to scrape every last bit of perf out of weak devices. If you drop the decimate-to-mobile issue, the difficulty of putting a metal-rough-dropdown in the standard shader drops to near-zero.

So, to close, I think whichever PBR workflow you offer as default is borderline irrelevant. Any set of shaders shipped with the engine is going to cover at best, a minority of user-needs. Hand-tuned high-performance shaders only cover a minority of user-needs. Visual scripting/node based setups only cover a minority of user-needs. Hell, even PBR shaders as a whole, given the market-space of Unity's usage, cover a minority of user-needs. There is no one ideal solution in this space. Trying to design something that fits all needs risks doing them all in a mediocre fashion.

I would say the most important thing here is to make sure that whatever system is in place from Enlighten to the new Deferred Pipeline is _flexible_ enough for your community to use it with whatever shader-based solution they require. Given how the 4.x cycle has been defined by incredible community work extending/modifying/augmenting/replacing pieces of the draw pipeline, focusing on opening up more of the renderer to things like custom light types, custom light falloff, uncapped light intensity, etc. would be so much more impactful for the v5 release cycle than how the standard shader is configured.

/endrant

 

Another +1 on the 'packing to optimized' aspect as well as being a user of Substance Designer, Substance Painter, B2M and Photoshop here.

I normally prefer metallic too as its rather 'easy' to grasp when you get a hang for it.
Specularity could probably work but then it requires decoupled specularity and reflectivity and roughness so I get the control over the light reaction AND the reflectivity, not one or the other.

 

Hey - thanks for this thread

What type of game(s) are you trying to make?
Most of the time (90%) I'm unfortunately not making games of my own but rather just author for others (employed and freelance). For myself I like the more arcad-ish and classic games. Things you can do with few people.

Do you use on "scanned" data?
Not so far ...no. Didn't need to.

Do you hand author textures?
Yes. Most of the time. And even there there's asplit between hand authoring as in painting/cartoonish and using photos as a basis.

Which tools do you use?
Currently I'm using mostly Substance Designer & Painter, Photoshop and ZBrush Polypainting.

Do you have a preference for specular or metallic or is this all new to you?
I currently strongly prefer the metallic approach. I never really cared about the specular workflow after I understood the metallic workflow, any more. It just feels very logical and natural.

 

What type of game(s) are you trying to make?
My studio is working on PC, IOS, Android and will eventually be considering porting to Console. Our original release was for mobile but because we dont want to do a straight port we are rebooting our graphics to leverage PBR. Our goal is to have a great looking game for PC and High End Mobile devices. With that said because we authored our content in a reasonable way we can then crush the maps down into a single diffuse map for mid to low range mobile devices for performance/Memory gains. I have found through early prototyping this is most easily achieved through PBR"Matallic" pipeline

Do you use on "scanned" data?
Our studio only uses scan data if its been provided through Quixel and Substance Designer default presets

Do you hand author textures?
Yes. We currently author textures by hand using Quixel and Substance Designer/Painter

Which tools do you use?
We are using Substance Designer and Painter, Bitmap2Material and Photoshop Quixel to do texture work.

Do you have a preference for specular or metallic or is this all new to you?
I Strongly suggest you use the Metallic pipeline. Its easy to train and author content using this pipeline because once you have the correct value in your diffuse map you don't have to worry if the dielectrics or metals are reacting correctly. Your metallic map defines this for you just it just works. Spec/Gloss approach makes you have to get the correct highlight color through the Spec, then you have to define the correct base color in the Diffuse and its a constant back and forth. This is devastating because everyone has a different idea of how real world materials look and react. Remove the guess work out and simply allow for both shader models within Unity. In my studio we would be using Metallic for 99% of our content and the 1% would be for edge case where Spec/Gloss is more appropriate

Finally I will invoke xenius If you
4. And this is the big one. You're marketing a _fundamentally_ pro-only feature as though it wasn't. It is. PBR without linear lighting, hdr, and postfx is borderline useless. I was asked this (awkwardly) at the end of my Unite talk by @@QuantumTheory . It's just as much of an awkward issue now. Every indie user asks this when I give a talk on PBR, and this will continue to be an issue throughout the U5 release cycle.


Note: I strongly suggest if and when Metallic is supported to allow artist to create a single texture they can then imbed into a single maps R,G,B and A channel to save memory. I have heard people from Unity mention that if you use a PBR pipeline "it is expected you have no memory budget" Well that's simply a horrible way to think about building content in a production setting. There will ALWAYS be memory concerns and if you support a range of devices that have different hardware specs. Allow artist/developers to dictate what is best for there product and support a configuration like. (R) Metal, (G) AO, (B) Spec and (A) for Roughness or allow them to decide that on there own

 

Thanks for asking!

What type of game(s) are you trying to make?
No games, architectural visualization.

Do you use on "scanned" data?
Yes, depending on the material.

Do you hand author textures?
Yes, for higher quality.

Which tools do you use?
Substance Designer and Substance Painter.

Do you have a preference for specular or metallic or is this all new to you?
I prefer the metallic workflow. It is becoming some kind of standard in the industry and feels more "natural".

 

What type of game(s) are you trying to make?
3D first-person. Going for realistic texturing/lighting with semi-retro geometry.

Do you use on "scanned" data?
No, not currently.

Do you hand author textures?
Yes.

Which tools do you use?
Learning Substance Designer.

Do you have a preference for specular or metallic or is this all new to you?
Metallic for sure. It just makes more sense to me in terms of how you think about it and create textures. I'm still trying to figure out the Spec-Gloss model compared to Metallic, but I understood Metallic right away.

 

What type of game(s) are you trying to make?
FPS / Space flight shooter VR title

Do you use on "scanned" data?
No. Tried, seemed more of a waste of time if you're not targetting ultra-realistic.

Do you hand author textures?
Yes.

Which tools do you use?
ZBrush, Photoshop.

Do you have a preference for specular or metallic or is this all new to you?
Specular, but don't care either way which is supported at first. It seems easier to port old art to specular style, but Metallic is probably better in the long run.

 

Do you have a preference for specular or metallic or is this all new to you?

Metallic, it just make sense to me. Support Metallic first then add specular

 

What type of game(s) are you trying to make?
That question is strange indeed. But anyway, I am creating Environments and Architecture in Unity.

Do you use on "scanned" data?
No.

Do you hand author textures?
Yes.

Which tools do you use?
Substance Designer, Substance Painter

Do you have a preference for specular or metallic or is this all new to you?
I prefer metallic/roughness: it just seems more intuitive and faster to author. And I guess, I am not the only one who got used to it by the amazing Allegorithmic tools.
But I agree with the other posters, that both shading models should be implemented right from the beginning.

 

Seriously, someone said that at Unity? Well, whoever said it probably didn't think about the fact there's less texture fetches. It's not just less ram it's also faster.

 

What type of game(s) are you trying to make?
I'm working on a fighting game.

Do you use on "scanned" data?
Nope.

Do you hand author textures?
Yes. I prefer this method because I can create surfaces that would be near impossible for me to capture.

Which tools do you use?

ZBrush, Modo, Substance Designer/Painter, xNormal

Do you have a preference for specular or metallic or is this all new to you?

Since I have moved my games shading to PBR using Alloy, I definitely prefer Metallic since it is a bit easier for me to create realistic surfaces and with tools like Substance Painter, it stops the guesswork of trying to tweak Specular and Gloss Maps.

 

And I'm not hating on spec/gloss, but I do own a polarization filter and a Macbeth chart

 

A realistic multiplayer naval fps game. Art style and direction is realistic.

Yes, Quixel megatextures.

Yes, depending on the asset.

Quixel Suite, Substance Painter, Photoshop, Marmoset, Knald

Currently we are using the Specular workflow. We do know how to do Metallic. As for preference, why not have both options available? It would be easy to implement if other game art tools are doing it.

 

Do both, but you really must make sure that linear lighting and HDR are available in the free version and relatively unplagued by bugs to keep the visual quality of the engine close enough to being on par with your competitors. Otherwise there's really no point in using the standard shader, as commonly used ways to define various materials won't work correctly in gamma-space.

I don't rely on scanned data yet for the most part, as Quixel hasn't made most megascans publically available and the scans that come with Substance Painter are lackluster in comparison. I generally find that I can get better material definition for non-uniform materials like sedimentary rock and wood by authoring custom, tweakable substances in Substance Designer.

 

Anyone here still Beast? Or is everyone prepped to use Enlighten?

 

Beast is a beast ... Enlighten is enlightening. So I'm glad they're getting rid of Beast in Unity 5. But I'd say that's a little off-topic.

 

Definitely Metalic.
My impression is that Specular is only good in that it is a much simpler transition to the workflow from traditional non-PBR.
But Metallic is actually far more logical and intuitive, and easier to get specific desired results.

 

What type of game(s) are you trying to make?
A micromanagment strategy game with addition to build cities.

Do you use on "scanned" data?
No, We don't use them.

Do you hand author textures?
Yes, we make all our assets inhouse.

Which tools do you use?
Photoshop, Marmoset,

Do you have a preference for specular or metallic or is this all new to you?
We are using simple specular workflow. Curently we don't need advance materials for simple objects. But in the future we would like to have both options.

 

What type of game(s) are you trying to make?
all kinds

Do you use on "scanned" data?
no

Do you hand author textures?
yes

Which tools do you use?
Allegorithmic Substance Designer
Allegorithmic Substance Painter
Photoshop

Do you have a preference for specular or metallic or is this all new to you?
strongly for metallic as it has a clear production path and it's authoring process behaves a lot more like the physical properties of the material you want to create

 

If you're only going for one, how about you choose spec/gloss, then create the texture packer that others have requested in this thread, and in the texture packer, allow inputting metal/rough textures, which are then converted to spec/gloss for the shader? AFAICT converting metal/rough to spec/gloss is straight forward, the other way around is not.

 

What type of game(s) are you trying to make?
Horror modular environment.

Do you use on "scanned" data?
Nope.

Do you hand author textures?
Yes, I dont have access to procedural texturing tools yet.

Which tools do you use?
Photoshop, Materialing , Xnormal, Marmoset Toolbag 2.

Do you have a preference for specular or metallic or is this all new to you?
I prefer the Specular workflow, because for me it's easier to find proper material values on the net, and because I've already built a small materials library using Spec maps, and I would love to be able to use them again on unity 5.
If you guys decide to go for the specular approach, please consider not packing the roughness in the specTex alpha, because generally Roughness values should be in Linear Space, while Specular should be in Srgb. You could however pack it with AO/Heightmap in a separate texture, so we can bypass srgb sampling for this particular texture.

 

Do you use on "scanned" data?
Nope

Do you hand author textures?
yup

Which tools do you use?
Photoshop...maybe substance designer.

Do you have a preference for specular or metallic or is this all new to you?
Please please please give us control of reflectivity for all surfaces. Having worked with both workflows I find the specular workflow to make more sense, and gives much better control and surface variation to non-metals.

Metalness workflow is weak, and makes any non-metallic surface look like plastic poop.

I really hope that everybody voting here actually has an understanding of both workflows, and isn't just choosing metalness, because thats what UE4 does.....

 

I really hope that you have valid examples to back up your statement. Having used both I find the Metallic workflow to make more sense and is in fact easy to understand/logical. Below are some pieces I created using the Metallic workflow (using Substance Painter/Designer). Please point out to me where exactly my non-metal parts look weak and poop as you state. My decision to pick metallic has nothing to do with Epics approach in UE4. It's merely based on my experiments and tests and lastly real world examples.

https://dl.dropboxusercontent.com/u/449179/The Raven/OilLamp_Painter_Final.mp4

https://dl.dropboxusercontent.com/u/449179/The Raven/RR_HallwayTable_Final.mp4

https://dl.dropboxusercontent.com/u/449179/The Raven/RR_Door_01.mp4

https://dl.dropboxusercontent.com/u/449179/The Raven/RR_RoundTable_Final.mp4


There are many benefits in using metallic. The obvious one is that you clearly define which parts are metallic and which aren't. It doesn't get any simpler than that.

Thomas Pasieka

 

Another I've found, is that you have only one colormap, which makes it easier to preview in your 3D modelling app of choice, to adjust UVs and things like that.

 

@Gskol: really? plastic poop..? anywho..

I will say this. Metal-Rough as a workflow is markedly inferior to use IF one does not have a map input for specular intensity/f0 (can be black and white, should map to 0-0.08 f0). If this isn't present, nonmetals all just have a constant f0 of 0.04, and thus all look they're made of the same material. The early UE4 demos tended to look like this. But they fixed it. We have the channel in Alloy. Allegorithmic still needs to add it to the substance set (come oooooon guys, I've been asking for a year now).

So yeah, Unity, unless you want the Unreal plastic sheen to your metal rough, have a black-and-white spec intensity for dielectrics channel (that internally defaults the value to 0.04 when a map/input isn't present).

For everyone else, a handy reference (and why this is important):
Specularity at zero Metallic (0.00-0.08 f0)
Dielectrics Color Parameter
Air 0 0
Water, Ice, Snow 64 0.25
Milk 70 0.275
Skin 89 0.35
Glass 128 0.5
Quartz 147 0.575
Polystyrene, Plastics 159 0.625
Calcite, Rocks Minerals 191 0.75
Ruby 245 0.9625
Alumina, Gems 255 1

 

High quality AAA stuff at home, crazy blow-your-mind stuff at work; something about Manhattan.

Though the results are nice, I'll likely never use scanned data unless it's my own, and I don't have the means to do it. I prefer to iterate and create my materials from calibrated photography.

Exclusively.

Photoshop, ZBrush, Mudbox, World Machine, Substance Painter/Designer.

I find Metallic to be more straightforward.

 

I use Substance Designer / Painter.

Metalic / Rough is easier to learn & work with, and it also requires less texture data. The metallic map just makes sense. Is easier to convert from metal/rough to spec/gloss.

Spec Gloss kinda pissed me off at first. I have been playing with Unity 5 beta lately and trying to get stuff to look good with the standard shader. I have authored a number of PBR substances in Metal/Rough. First I tried using the "Metal/Rough to Spec/Gloss conversion" node in substance designer - which looks horrible (uses a default specular value that isn't always right, gloss seems off) in the standard shader. So I have to go in and tone down the glossiness, and tweak specular value sometimes. I'm not having much fun with spec/gloss. After working with spec/gloss a little, I feel like you DO have more control and can make a wider range of materials. But it just takes more work, is annoying, and I'd rather not do it. I think a *lot* of people are going to be confused by Spec/Gloss if they pick up Unity 5 without reading up on PBR, whereas with metal/rough it's pretty clear you need different maps.

So my vote is for Metal/Rough, but bottom line I'm also fine with the Spec/Gloss shader. I am however going to continue making materials using metal/rough, then use the conversion node and tweak the specular and glossiness.

Looks like everyone is voting for Metal/Rough, huh? I guess you guys should have done this poll before deciding on Spec/Gloss for the standard shader

 

If there's one thing I've learned over the years making shaders and tools which don't conform to normality is:

Sometimes a simpler workflow or shader will yield much better graphical results, because it's understood and tweaked to the limits. You can't maximise the potential of anything that takes too much work or fiddling around, because you're working on using it rather than using it for work.

It's one of the reasons I suggested Unity have a small utility or tool path where you plug in a lot of separate textures, and it generates the desired result non destructively for the specular approach, which makes it tool-neutral and easier to understand. Possibly an idea for asset store if someone wants to bother.

 

+ bump

 

Either workflow is fine for me.

I setup a dropdown selector in my Substances to change between workflows inside of Unity:
- Metallic/Roughness
- Spec/gloss
- Alloy
- Lux

The main downfall currently with doing this is that I need to check "Generate all outputs" in the Unity Inspector for my substance, hence all possible outputs calculate and display in the unity Inspector even if not required.

Therefore I would rather see better Unity Inspector support for Substances ie:
- Respect the "visibleIf" instruction for outputs from Substance Designer for cleaner Inspector results.

 

What I don't understand is the decision to make ONE shader to rule them all. It doesn't make much sense. Why can't we have 1 Spec, 1 Metallic and 1 that works for mobile instead of trying to make this Ueber/Standard Shader that tries to cover everything? Could somebody answer that? What is the logic behind this?

 

@ARtfx You should prob. ask Allegorithmic for that, considering they're in charge of the substance integration (to my knowledge). I agree that it's desperately needed.

@Thomas Pasieka Big ol bag of YUP!

 

@xenius, yup I have spoken to both about this. Just added to the SD forum feature requests http://forum.allegorithmic.com/index.php/topic,2969.0.html

Hopefully others will add their support for it.

 

Man I wish I could get more pbr values easily with data sheets. I could probably hit up a fabric store and take pictures with my polarized lens and match what I see