Long story short, I am basically forced to use the ColladaLoader to load my animated models into three.
The issue I'm having is that a model loaded with ColladaLoader use about ten times as much memory as a model loaded with ColladaLoader2, and so I would very much like to go with ColladaLoader2, but it lacks animation support.
So basically I am going with the old ColladaLoader for now, but I would be immensely grateful if someone could add animation support to the new one. I tried adding it myself, but in the end it proved too complicated for my brain to handle.
I will trying to add animation support to ColladaLoader2. Because the collada spec is very sophisticated in this context, this will definitely take some time. I'll report any progress here...
A basic support of skeletal animation is now implemented in ColladaLoader2. Because the loader does not implement the whole (comprehensive) Collada spec, you might have models that won't work properly yet. If you encounter problems, just open a new issue.
How about basic pos/rot/scale animations that doesn't use a skeleton?
That's still missing. But it should be a little bit easier now to implement this kind of animations because a lot of necessary code is already written.
Against this backdrop, i think it's fair enough to reopen the issue.
All right, thanks.
@donmccurdy I guess GLTFLoader supports the loading of arbitrary keyframe animations, right? Something like the animation of the position/rotation/scale of meshes like in this example:
https://threejs.org/examples/#webgl_animation_keyframes_json
The new ColladaLoader only supports skeletal animations since it's absolutely horrible to implement this according to the spec. I'd like to close this issue and refer to GLTFLoader instead.
GLTFLoader supports arbitrary pos/rot/scale keyframes with or without skinnning, yes. The COLLADA2GLTF 2.0 branch should preserve them correctly on conversion, as well.