3d camera orientation widget works only when a frag file is loaded

[JackConradGray]

In the app I am developing I use lightweight 3JS objects to represent abstractions that need to be manipulated graphically. These need to respond to user actions without any hesitation or slowdown.

The 3d camera orientation widget makes the view change when a frag file is loaded but not without it. Is there a way to keep that widget active without a frag being loaded so I can use it with my non-frag 3D geometry?

Alternatively if there could be a way of wrapping the 3JS objects so they appear to the platform to be Fragments objects, then there would be no need ever for implementing parallel operations that are already functioning for fragments--as long as that would not noticeably degrade performance. This request duplicates one in issue #35.

[HoyosJuan]

Hey @JackConradGray! Good news on this one too.

If you're adding your Three.js objects straight to world.scene.three, the gizmo already frames them correctly, no extra work needed there.

If you're on the deferred postproduction pipeline, though, you've probably already run into this: a plain mesh added directly to world.scene.three is invisible under DEFERRED mode, since its single-pass G-buffer capture only draws materials carrying the engine's own internal marker (set on frag/BIM surface materials only). The way around that is to draw your Three.js content in its own scene and render it through renderer.postproduction.addOcclusionPass(...), which runs after the deferred composite and skips the G-buffer requirement entirely.

The catch is that scene is separate from world.scene.three, so the gizmo had no way to know it exists, that's the actual bug you ran into. Fixed now: top-viewer has a new extraScenes property, pass it whatever scenes you're rendering that way and the gizmo will include them when framing:

<top-viewer .extraScenes=${[yourOcclusionScene]}></top-viewer>

On your wrapper idea: that already exists, and it's not a workaround, it's a first-class part of the Fragments Edit API. You can take any BIM element from a loaded model, edit it as a native Three.js object (geometry, position, whatever you need), and it stays a real fragment, participating in everything fragments already do (undo/redo, visibility, measurement, etc). It also works the other way: build plain Three.js geometry and convert it into a fragment at runtime. Either direction, no parallel machinery to maintain on your side.

Two tutorials worth a look:

• https://docs.thatopen.com/Tutorials/Fragments/Fragments/FragmentsModels/EditApi
• https://docs.thatopen.com/Tutorials/Fragments/Fragments/FragmentsModels/Rebars

The Rebars one in particular is a good example of building custom geometry from scratch and folding it into the fragment model, probably close to what you'd want for your space-planning objects.

Let us know how it goes!