pbr textures PBR - Separate AO Layer | Multiple Material Layering | Unique Uv Channels per face

[Tyiler Scarborough]

Hey Everyone,  

I wanted to explain and share my use case for Material Layering and Multiple UV mapping  and how this could help increase performance while also adding fidelity to our environments.

Let me explain the best I can.

Multiple Material Layering with Unique UV mapping Channel (Baked on mesh on all mesh not just avatars)

What?  Similar to Photoshop or Gimp, you would be able to layer one image on top of another. 
Why?  To allow a defined AO map that doesn't tile
Why not use AO map which is in the red channel of the ORM?  Its not that good, it's not strong enough and with a custom AO layered on top will give a clean result + Roughness control will block the light reflection in the AO shaded areas. 

Main Benefits of this?

• To allow tileable textures but retain a strong AO to give objects a MASSIVE increase quality
• Lowers multiple textures downloads because of Ao being required to stay within it's 0-1 uv space
• Most likely speeds up performance due to not loading as many textures, creates overall better result.
  • You do add a few Textures for the AO, But it doesn't need to be a 1024; And using tileable textures instead will balance out in our favor
• Allows for material changes on mesh without rebake/repaint/re UV
• Allows Users to Customize their homes/products with add-on Material kits? or their own materials?

I've provided examples, in which the one labeled "AO shell" has a duplicate mesh on top of the base mesh to simulate a layering effect, but doubles my land impact.

 

FULL RESOLUTION IMAGES (please allow google photos to fully load image, they're close to 4k images) : https://photos.app.goo.gl/kPM1VmvqVfXs9SH47

 

See in world : secondlife://Aditi/secondlife/Materials1/65/57/23

 

Edited May 9, 2023 by Tyiler Scarborough

[ChinRey]

I've been thinking along the same lines for a while. I'm not quite convinced it will offer improved performance since it will require client side baking but it offers several other advantages as your pictures show so well.

 

11 hours ago, Tyiler Scarborough said:

Why not use AO map which is in the red channel of the ORM?  Its not that good, it's not strong enough and with a custom AO layered on top will give a clean result + Roughness control will block the light reflection in the AO shaded areas

Also, and perhaps most important, because an efficient AO map on a tiled texture require different resolution and repeat rate than the other maps.

[Rick Nightingale]

I would love a non-tiled AO layer! Could have done with that on the thing I just finished building last night, and was actually wishing we could have it at the time.

[GreyLasagna]

Been hoping for something like this since I started making stuff so this would be such an improvement, especially for buildings and interiors. I hate to split my models in so many UVs just to get decent looking textures. I've used simple planes with AO shapes with transparency to fake the bake but this increase the LI considerably, depending on the building complexity of course.

[Jabadahut50]

Wow... the quality difference between those two pictures really shows a dramatic difference. I support this idea.

[ChinRey]

29 minutes ago, Jabadahut50 said:

Wow... the quality difference between those two pictures really shows a dramatic difference. I support this idea.

Yeah but before we get too excited, remember that this is just wishful thinking; there's  no realistic chance LL will implement the idea. Still, it would be itneresting if somebody filed a formal feature request if only to see how they responded.

[arton Rotaru]

19 hours ago, ChinRey said:

Still, it would be itneresting if somebody filed a formal feature request if only to see how they responded.

Tyiler did that actually. It's status has been set to Accepted, but no further response as of yet.
https://jira.secondlife.com/browse/BUG-233843

[ChinRey]

On 5/25/2023 at 8:00 PM, arton Rotaru said:

Tyiler did that actually. It's status has been set to Accepted, but no further response as of yet.
https://jira.secondlife.com/browse/BUG-233843

This can be interesting.

The reason why I don't believe anything will come out of it is not the usual one. It's because LL seems to be committed to the glTF standard now and it may be difficult for them to implement features not covered by it.

[Jabadahut50]

10 hours ago, ChinRey said:

This can be interesting.

The reason why I don't believe anything will come out of it is not the usual one. It's because LL seems to be committed to the glTF standard now and it may be difficult for them to implement features not covered by it.

The GLTF standard is mostly about asset portability. How your engine interprets said assets is up to you... to an extent... I think it does define some specific requirements TBF, so It is still up in the air, but separating texture channels and manipulating them is trivial nowadays, so I'd imagine this wouldn't be too hard. I suppose we'll wait and see.

[Extrude Ragu]

there was talk on the sl discord from one of the developers who was thinking of implementing something along these lines, but it sounded kinda different like he wanted the viewer to generate a special uv/ao map for the whole object and have it be unique to that object rather than be part of the material asset

[Jabadahut50]

On 6/7/2023 at 3:04 PM, Extrude Ragu said:

there was talk on the sl discord from one of the developers who was thinking of implementing something along these lines, but it sounded kinda different like he wanted the viewer to generate a special uv/ao map for the whole object and have it be unique to that object rather than be part of the material asset

Sounds like they're talking about on the fly AO baking or maybe on upload AO map generation of some kind. It'd be a nice feature to be able to generate for a mesh on upload and then let people tick on option on a mesh during building so that you can have the mesh override the one in a material, so materials can be more portable between objects. So long as it's not the only method, and we can still make the ones in the GLTF at our discretion to maintain artistic vision.

[Tyiler Scarborough]

So, What I was wanting to see is more of a secondary layer to put a image into ...

You can think of the second "layer",  I specifically would NOT want "on the fly AO", but rather a defined AO image that Icreated and imported.  

Also You can think of this idea of adding drips from window ledges, rust stains, edge grunge, graffiti... Anything you can imagine... 

So one layer would be tiled.... so i can repeat it over and over to use a tillable material, then i could add a Grunge/AO map that is defined in a 0-1 UVspace... to add a extra detail.   How you can do this now with duplicate geometry.. and i feel it will also be that way going forward..

 

I just figured before a release is done, we should try to add as many features now and fix them before adding them later/never >_>

 

[OptimoMaximo]

14 hours ago, Tyiler Scarborough said:

I just figured before a release is done, we should try to add as many features now and fix them before adding them later/never >_>

 

Yeah, and never be listened by LL and whoever is working on such feature, at least not until "it's too late to add/change" what they've got and the request be dismissed, with the promise of "in future iterations" that will never come. Exactly like it happened with path finding, animesh, EEP, etc etc...

[ChinRey]

2 hours ago, OptimoMaximo said:

Yeah, and never be listened by LL and whoever is working on such feature, at least not until "it's too late to add/change" what they've got and the request be dismissed, with the promise of "in future iterations" that will never come. Exactly like it happened with path finding, animesh, EEP, etc etc...

And mesh and MOP ... and the Marketplace.

It's still worth trying though, you never know. I heard LL brought in some new developers with game developing experience recently at that may help.

Edited June 12, 2023 by ChinRey