Spec gloss vs metal rough

[emackey]

This adds the "two water bottles" side-by-side comparison of spec/gloss and metal/rough as a new sample model in its own right. New labels were added, and they too are part of the test.

/cc #181 and CesiumGS/cesium#7006

/cc @bghgary @OmarShehata @abwood

[emackey]

See the README

[lexaknyazev]

"License Information" part of the readme contains only authors, not usage / distribution license.

[emackey]

Yes, this is because it is derived from the Microsoft WaterBottle, which similarly lists authors but not a license. @bghgary or @sbtron can a real license be added to WaterBottle and friends? Preferably one that permits derivative works! I recommend CC-BY 4.0.

[bghgary]

I'll let Patrick Ryan answer this in more detail. I think we wanted to avoid putting any license because it still caused restrictions even with the most permissive one. We wanted this to be completely open for use. @PatrickRyanMS what do you think?

[lexaknyazev]

Lack of license usually means that nothing could be done with the model without explicit permission.
To give it away completely, the typical way is to declare it "public domain" (if applicable) or to use something like CC0.

[PatrickRyanMS]

I agree with @bghgary that we want these assets publicly distributable and open for anyone to use, remix, share, etc. The issue with even applying a CC-BY 4.0 license is that since we work for Microsoft and these contributions are part of our work for the company, I don't know that we can just assign a license, even CC-BY, without getting our legal department to sign off on us doing that. This is a question that we will need to dig a little further on before we can attach any legal language to the assets, but we do want them to be public domain.

[emackey]

@bghgary @PatrickRyanMS Thanks, and please do follow up on this. @lexaknyazev is correct that legally we shouldn't be doing anything with the Microsoft-supplied models, not even including them for distribution with the sample models, without a license.

[cx20]

@emackey I added this model to gltf-test.

I think the current status is as follows.
https://github.com/cx20/gltf-test#extension-test-models

[emackey]

@cx20 Thanks. I didn't realize the spec/gloss ecosystem was in such bad shape. Cesium support is coming soon. Some of the others appear to be really close, like ThreeJS and Hilo3D.

[emackey]

@bghgary @PatrickRyanMS How should we proceed with this one? Did you reach out to legal, or is there any other option?

[bghgary]

@sbtron is going over this with legal which takes some time. @sbtron, any status on this?

[emackey]

OK, in the master branch, Microsoft has added CC0 licenses to its sample models (thanks so much @bghgary and @sbtron!). I've copied that license here to this new model, since it's just a re-packaging of two different versions of the existing water bottle model side-by-side.

This should be good to merge now.

[bghgary]

I don't think it's appropriate to say "donated by Microsoft for glTF testing" here?

[emackey]

Removed.

[bghgary]

Add two spaces at the end of the line to put the text on a separate line.

[emackey]

Fixed. I had copy/pasted this, but I have my copy of VSCode configured to remove trailing spaces automatically. Sad that markdown places significance on such things.

[cx20]

@emackey @donmccurdy I have doubts about the display result of Three.js.
The Metallic Roughness model reflects the environment map, but the Specular Glossiness model does not seem to reflect the environment map. Is this a correct display result?

Three.js + SpecGlossVsMetalRough.gltf result:

[donmccurdy]

Thanks again for these side-by-side comparisons. ❤️

It's entirely possible the spec/gloss result is incorrect in three.js, I'll file a bug.

EDIT: Filed mrdoob/three.js#15195.