Feature Request..

[digitaledge]

Hi,
this is actually a really interesting converter and the client side conversion makes it even more useful than gltf. Really a great work you did on this.

Is it maybe possible to add 2 more converter options for the custom binary format?
It would be great if these could be loaded as a standard three.js model, keeping all hierarchy of meshes (like with the three collada loader). I did manage to load skinned models as normal three.skinned mesh, but the merged geometry and hierachy is an issue with our project.

[crobi]

A format that preserves the hierarchy is not the focus of this converter - I wanted to have a tool that produces model files, not scene files. So it's unlikely that I will add this as an option myself.

If you are willing to invest some time however, it should be possible to modify this tool to do what you want. This project consists of three main classes: the loader, the converter, and the exporter. You can use the loader and converter, but have to write your own exporter:

• First, use the loader to load and parse the COLLADA file
• Then, use the converter to process the loaded data. In particular, have a look at the singleGeometry and enableExtractGeometry options. Without these options, the converter produces a scene graph with geometries attached to scene graph nodes.
• The data preprocessed without the above mentioned options will not be compatible with the built in exporter, so you'd have to modify the exporter or write your own. This requires some work, but shouldn't be too hard - the converter does all the heavy work and produces geometry data stored in a format more or less identical to that of THREE.BufferGeometry.
• I have no idea how such a scene export would interact with the animation export code

[rchadwic]

We're looking into this as well. This importer really seems to do a much better job than the original THREE.js importer. Maybe we could modify the code to behave exactly as it does for skinned meshes, and preserve the hierarchy for non-skinned mesh. We could even just prompt the user to tell us which. Either way, I'm hoping this library will solve our import woes. Looks great so far.