[SOLVED] ShaderFX - Stingray Viewport PBS Not Rendering Correctly

eCstatic

For a while now I've been trying to figure out a way to get a PBR shader working inside my Max viewport. Whenever I use the ShaderFX Stingray shader however, my roughness map always becomes splotchy and pixelated.

 I've read that the stingray shader doesn't support grayscale roughness maps so I've went with the ORM (Occlusion, Roughness, Metallic) packed map approach but I'm still getting similar results. All my maps are set to linear encoding, and my gamma/lut is turned off in the viewport. Is this even fixable?





What is causing my roughness map to be rendered this way, and how can I remedy this?

bitinn

Like I said, post your BaseColor/Normal/RMA texture node settings.

For example, Normal map should definitely not use Linear, you should use Normal map encoding.

And to avoid any doubts on textures, post them as well.

(I am using Maya so I can only make suggestion in terms of ShaderFX)

eCstatic

Thanks for helping me look into this I've played around with it some more and I still cant get it to work. I'm starting to think it might be inherently broken.

eCstatic

The effect is even more apparent when using the default cubemap

bitinn

The maps you uploaded are different from the one you are showing.

But I think you should try changing the Standard Base's Normal to "Tangent Space". At least I managed to produce a similar weird rendering using "World Space"

Good luck.

World Space



Tangent Space

eCstatic

Whoops, I uploaded the wrong maps here are the correct ones with the teapot fbx also.

I tried tangent space it just flips the environment map, the roughness is still as pixely and splotchy as ever.

poopipe

StingrayPBS is not the same as Substance in terms of specular calculations and also has some sort of curve applied that affects output  values.
  
What I see in your initial images is consistent with what I've seen myself during testing. 

It's not broken as such, just different. 

bitinn

@poopipe I think @eCstatic could have done better. This is it in my Maya viewport. Using the same airport environment we created at my thread.

bitinn

I don't see any obvious mistakes in your material so I am suggesting you remove all other control nodes. And create just the base node like in my old thread.

That way you have a lot less variable and can tune each node one parameter at a time to figure out what might be wrong.

EDIT: a few things I would try after creating the base nodes.

- BaseColor: sRGB Map encoding.
- Standard Base: IBL encoding use Linear (to match Substance).
- If all else fails, you might want to open a support ticket to Autodesk (I figure out mine with their support).

poopipe

You're absolutely right - I was attempting to add to what you'd already said but I'd only been awake ten minutes so I phrased it badly. 

eCstatic

Ok, can't get more basic than this



The result is messy still. @poopipe I don't think this is normal consistent behavior for the shader because I can see much better and different results in bittin's examples


Someone had a similar issue to mine and the answer reply from an Autodesk employee was that the issue was in the material code.
http://polycount.com/discussion/comment/2379869#Comment_2379869

poopipe

I've just checked out what happens in max 2017 and maya 2018 using this graph setup (identical in maya)

My material is in world space because i use the tangent to world node - mainly out of habit.  i didn't change any settings etc.

max 2017 first: (had to make a new teapot as it didnt like your FBX for some reason)

and maya 2018


This is what happens if you tell it the ORM texture is srgb - you get the same thing in Maya:



That does look suspiciously like what you're getting in your viewport .  

This should get you to where bitinn is - it doesn't resolve the differences between stingray and substance. 

If you're not getting this then something is broken. 

you could try adding a power node (0.45) to un-gamma correct the image if you think there's an issue with encoding

marks

why people are using linear textures in games. Facepalm. This isn't vfx industry. You're not helping yourselves here.

poopipe

Because they're a far better representation of actual values than srgb( ie 0.5 is actually 0.5) and you don't get crushing of values at either end of the spectrum

marks

That's pretty misleading - there's a lot more to it than that, I wrote a ton of info about this in an old thread in this forum about linear vs srgb textures and I've just spent 20mins looking for it, but I can't find it. Sorry. 

eCstatic

poopipe said:

max 2017 first: (had to make a new teapot as it didnt like your FBX for some reason)

Thanks for the suggestions, they didnt fix my issue unfortunately. In this screenshot you posted, it looks like you are getting the same issue as me with your roughness map being drawn pixelated and splotchy as if its being mip mapped. I'm not seeing it in your Maya screengrab which makes me think that this issue is exclusive to Max.

poopipe

Substance doesn't mip so that may contribute, also the brdf is completely different from what's used in stingray and tends to be more contrasty.  What I saw looked about as close as I think you'll get without building a Shader of your own.

I'd publish mine but I built it at work so I wouldn't be allowed. 

marks said:

That's pretty misleading - there's a lot more to it than that, I wrote a ton of info about this in an old thread in this forum about linear vs srgb textures and I've just spent 20mins looking for it, but I can't find it. Sorry. 

It's a gross simplification I'll grant but it's not untrue. 

To answer the original question in practical terms - it's what the shaders expect and it's what the authoring tools produce (WRT the grayscale "maths" images at least . 

Why is it a good idea?
You'll know this but... if your image is stored with a gamma curve on it then information is squashed at the ends and basically lost. If it's stored as linear then that information can be converted to gamma space without loss.
Working linear also makes all the maths in your shader cheaper. 

Overall, it's higher quality and cheaper..

eCstatic

poopipe said:

Substance doesn't mip so that may contribute, also the brdf is completely different from what's used in stingray and tends to be more contrasty.  What I saw looked about as close as I think you'll get without building a Shader of your own.

I'd publish mine but I built it at work so I wouldn't be allowed. 

You got it! Thanks! The BRDF was at fault here, the one I'm using now makes my material looks 100 times more cleaner thanks again.

This is the BRDF map I'm using if anyone is curious, it works great.






Max and UE4 comparison

shaderfx

nice work @eCstatic
I've linked your solution on shaderfx.com

SebKaine

hi guys sorry for bringing out the dead, but i would be curious to know how we can generate a brdf LUT for stinray PBS shaders in Maya.
i am having a similar issue and i am sure that i could solve this, if i just understand how to create my own LUT.

Cheers
E

Eric Chadwick

This might help you
http://brdflab.sourceforge.net/

And this
https://alastaira.wordpress.com/2013/11/26/lighting-models-and-brdf-maps/

SebKaine

Thanks for your help Eric,
BRDF map looks to be a generic game concept. i only know BRDF throw standard precalc renderer. The fact that they approximate this with a Map in games is new for me.

In the case of game development, it’s often not necessary to strive for physically-accurate BRDF models of how a surface reacts to light. Instead, it’s sufficient to aim for something that “looks” right. And that’s where BRDF maps come in.

A BRDF map is a two-dimensional texture ... the BRDF map represents different parameters on each of its two axes :
- the incoming light direction
- and the viewing direction 

This is also interesting :  
https://www.imgtec.com/blog/physically-based-rendering-with-powervr-part-2/
https://x3dom.org/pbr/pbr2016.pdf
https://www.programmersought.com/article/8883369575/
https://learnopengl.com/PBR/IBL/Specular-IBL
https://github.com/HectorMF/BRDFGenerator
https://docs.imgtec.com/PBR_with_IBL_for_PVR/topics/Assets/pbr_ibl_the_bidirectional_reflectance_distribution_function__brdf__look_up_table.html

I think the initial concept of using a 2D Map to approximate BRDF is coming from UE4 and this paper :
https://de45xmedrsdbp.cloudfront.net/Resources/files/2013SiggraphPresentationsNotes-26915738.pdf

Cheers
E

LeoBounce

https://polycount.com/discussion/comment/2601719#Comment_2601719

Thank you very much for this! Even after years 3ds max still have broken build-in BRDF texture (pixelated texture which produce artifacts), and your sulution solved the problem! (in my case I also edited orientation)