Gltf 2.0

[moneimne]

Continuation of #4808.

To do:

• Upgrade all test models in specs/data
• Write tests for morph targets
• Remove gltf-sample-models from this branch, use simpler test models instead
• Investigate hemispheric lighting some more
• support for sparse accessors
• include some glTF 1.0 models in the Apps/SampleData folder for easy visual testing in Sandcastle
• Update CHANGES.md
• Update LICENSE.md

[lilleyse]

Added some fixes for broken tilesets. Will investigate the new "destroy" related test failures next, as well as the Classification Demo.

[lilleyse]

A snapshot of the performance of going around the BIM tileset. The results seem pretty reasonable, about as much time in upgrading gltf 1.0 to 2.0 as initializing the b3dm.

[pjcozzi]

Profile is probably fine, but it would be better to run without the readPixels bottleneck.

[lilleyse]

Performance numbers without readPixels. Looks fine.

[pjcozzi]

Thanks, I agree, looks OK. Later we could make the 1.0 -> 2.0 path a bit faster if we need, but would rather focus on 2.0.

[emackey]

Some kind of Unicode fail happening in this file.

[emackey]

I fixed this in my branch, PR: moneimne#3

[lilleyse]

I think we are all set now! All the 1.0 and 2.0 sample models have been thoroughly tested, as well as a variety of 3D Tiles tilesets. It's hard to catch everything, so definitely let us know if you have issues loading models or tilesets. Also for 2.0 models check out the glTF-validator to aid with testing.

There are some known issues here:

• Mixing morph targets in materials crashes for glTF 2.0 models #5695
• Multiple animations in a glTF 2.0 model causes an error #5694

Thanks to everyone who worked on this: @lasalvavida for laying the foundation and keeping up with the 1.0-2.0 changes from the start, @emackey for the procedural PBR reflection and merge assistance, @moneimne for writing the PBR shaders and bringing everything to production.