WIP: add GTLFReader

[daker]

Context

This is a draft PR, a starting point to learn what's missing to render the GLTF correctly

Results

• What versions of glTF format should we support? (both or just 2.0 ?)
• Load glTF (.gltf) with separate resources: .bin (geometry, animation, skins) and .jpg or .png image files
• Load glTF (.gltf) with embedded resources (as Data URIs)
• Load glTF geometry compressed with the Draco library
• Load binary glTF (.glb) using the binary container format
• Handle animations loading
• Handle skins loading
• Handle lights loading
• Extensions support? what is the minimal set of extensions to support?
• ORM Texture support in vtk.js ([Feature] Add for support for combined ORM Texture #2961)

Changes

• Documentation and TypeScript definitions were updated to match those changes

PR and Code Checklist

• semantic-release commit messages
• Run npm run reformat to have correctly formatted code

Testing

• This change adds or fixes unit tests
• Tested environment:
  • vtk.js:
  • OS:
  • Browser:

[supriome]

Hi @daker , Any progress there?

[jourdain]

Hi @daker,

I may start looking into it. Do you have a feel of how much more work needs to go into it?
Does your code mostly work but fails in some use cases? Basically I just want to have a feel on what kind of help you are looking for knowing that I'm looking for a glb loader...

Do you think we can reformat/rebase and merge it in its current state, or would it be better to work on the same branch until X and Y are solved? If so, knowing what X and Y are would be great.

Thanks for your hard work pushing vtk.js forward,

Seb

[daker]

I have updated the description with a list of questions/features to be addressed. Basically the loader is able to parse the gltf with simple embedded resources like the box or colored box.

As you can see on the images, there is something missing in the code to render those boxes correctly, @finetjul told me that faces/cells do not seem to be made of the right point ids, i am not good on this part.

#210 (comment)

My idea is to have a simple SceneLoader that will act like this :
Scene[$sceneId] -> Nodes -> Node -> Actors -> Actor (load texture and apply materials)-> Add the actor to the renderer

I would suggest to work on a branch, once we have a basic SceneLoader with basic parsing and rendering then we can merge and iterate over with adding animations and extensions depending on what we have decided.

[jourdain]

Sounds good, thanks for the update.

[daker]

@jourdain I am back working on this, i made some very good progress, no animation, lights , cameras, skins, morph or extensions so far

The ORM textures are slow to display since they are computed(using canvas) by extracting the channels from a the combined texture, this should be done by the GPU like in vtk c++ by providing a setORMTexture (FYI @martinken)

screen-capture.2.webm

[jourdain]

@finetjul @tbirdso @floryst @sankhesh @jspanchu