This adds Vector3.getRotationFromQuaternion and changes Object3D.applyMatrix to use Matrix4.decompose.

Also added SceneUtils.attach / detach and cleaned up unused vector parameter in Geometry.applyMatrix.

Renamed Vector3.setRotationFromMatrix() to Vector3.getRotationFromMatrix().
Added Vector3.getScaleFromMatrix().

As usual, it's up to caller to cache the value if it's not used right away.

Had to use Matrix4's internal matrix to avoid circular cross-dependency between Vector3 and Matrix4 (added extra internal matrix m3 for this not used anywhere in Matrix4 itself to minimize chances of messing up something).

Not sure it's really correct, it's kinda hard to tell.

Also made stress test more fancy and heavy. It's interesting how antialiasing increases performance difference for more complex materials.

Also some shaders refactoring.

Still todo: doubleSided handling for reflections / refractions.

As per discussion in #1324

This fixes geometry tessellation example (before TextGeometry by default was having UVs defined for just a subset of faces).

Also trying to make a bit nicer UVs for ExtrudeGeometry / TextGeometry.

Is as it was before, but I implemented the event dispatching in a different way (checking the distance of the object position between frames). Also renamed the 'update' event to 'change'.

controls.addEventListener( 'change', render );

Right now only misc_camera_trackball.html uses this.

At least till we don't have this solved somehow better I think we should keep the old one. New one makes everything using it feeling broken, it's distracting.

Moved new event-driven `TrackballControls` into `TrackballControlsExperimental`. For the moment used just in "misc_camera_trackball.html" example.

Now we can do things like this:

controls = new THREE.TrackballControls( camera );
controls.addEventListener( 'update', render );

Sadly, some examples are slightly broken until we have better pipeline for assets load complete event dispatching.

Seems horse model was wrong, it had one extra frame at the end (almost identical to the first frame), which caused our animation playback code to be incorrect in order to accommodate for this.

Side-effect was that two frame animations were completely broken.

Bah.

Also updated build script to fill in missing dependency (ImageUtils => DataTexture).

As per discussion in #1256

Also improved error reporting in JSONLoader (getting DEPRECATED errors for incorrect URLs was really confusing).

Default computation of morph normals for all models with morphs was taking unnecessary time and memory.

Also made it less resource intensive to recompute morph normals.

It seems to work, though not sure yet how to handle some things. Also for the moment it's implemented just for Lambert material.

As per discussion in #1207