[drdeadringer]

In a summer tech program for pre-college high school tech-interested//inclined students, my group of three created four different VRML world-spaces. We each made our own, and I made the "gateway" world-space where you would enter a Parthenon-like structure with three hyperlink objects inside which you could click on to travel to each of the other places.

Each place had a similar hyperlink object to get back to the Parthenon.

I enjoyed it. I probably still have the VRML 2.0 book around somewhere.

[giantrobot]

VRML Sourcebook [0]. I did a similar little project in VRML. I took the general organization of my homepage and VRMLized it. Authoring it was such a pain in the ass.

I loved the concepts behind VRML, and still do. Unfortunately all of the implementations around it were terrible.

When it was in vogue PCs weren't powerful enough to handle anything but the most trivial models. Dial-up was entirely insufficient to deliver anything but the most trivial models as well.

[0] https://archive.org/details/vrml20sourcebook00ames

[jandrese]

I remember the VRML player having some weird skewmorphic navigation widget that was incredibly clunky to use. You had to use the mouse to turn virtual wheels to adjust your speed or something, all laid out on a horizontal tube thing at the bottom of the screen.

And this was years after Doom and other 3D shooters introduced wasd movement.

VRMLs biggest problem was that it was just way ahead of its time. Everybody who tried it agreed that 90s PC hardware just wasn't ready for VR, and dialup modems were never going to be adequate. Unfortunately almost everything that has come up to replace it has been proprietary and usually monetized. We've lost that sense of "anybody in the community can be a creator" that defined the early web. It was too hard to monetize.

[drdeadringer]

> I loved the concepts behind VRML, and still do. Unfortunately all of the implementations around it were terrible.

Being ~18 years away and therefore memory is skewed, may I ask regarding how the implementations were terrible? Honest question.

[giantrobot]

On the playback/viewing side many VRML viewers were browser plugins. The idea being you might navigate from a web page to a VRML space or embed a 3D object in a web page. These were (as I remember) very sluggish and crash prone which took down the whole browser. Even dedicated VRML browsers weren't all that great.

I don't remember any without awful skeuomorphic buttons and view borders. Their navigation UIs also tended to be terrible. Most I ever tried aped 3D modeling apps using the mouse (a workstation mouse with three buttons) to fly through a scene. But with the typical PC mouse having two buttons (and a Mac with one) you had to use modifier keys to switch between fly and "look". VRML supported world gravity (intensity and a normal) so a viewer program could just navigate like a FPS game but I don't remember any that did.

Another issue I remember but may have just been my PC, because WRL files don't embed any linked content they've got to go fetch every resource referenced like on a web page. On my systems at least this led to a lot of pop-in as I moved through scenes, even locally. On dialup it's was a thousand times worse.

So modulo actual hardware performance, or lack thereof, you had poorly performing software with very alien feeling UI. While that is more than 20 years of time between me trying VRML and now those are things that stand out to me. Maybe none of these problems existed if you were blasting around on an SGI O2 workstation or knew just the right software to use. I had neither so I just stumbled around playing with stuff.