Wasn't VRML support in browsers usually plugins and not native?
https://www.cnet.com/tech/services-and-software/silicon-grap...
I would create the files in a txt editor and then explore the world-space created via browser.
There was an "object" view which was basically a bunch of still images of an object from different angles. The cursor moved a virtual camera that just displayed the still frame for the appropriate angle.
The "panorama" view was a panoramic image mapped to view angles from a central point. Images could be cubic or spherically mapped IIRC.
Both types supported hyperlinks. So clicking a linked region could load another scene which could have yet more links to more scenes.
The Star Trek Interactive Technical Manual was a famous use of QTVR. It enabled visual fidelity a real-time 3D rendered of the era couldn't deliver. It was made by actually photographing the Star Trek production sets.
Classic MacOS had pretty well-integrated support for QuickDraw 3D Metafiles, they could be dragged and dropped, you could embed 3D objects into a word processing document or presentation or something and manipulate and explore them in there.
Kind of mirrors what Apple is doing with USDZ these days (you can send a USDZ model in iMessage and the recipient can manipulate it right there in chat)