[causality0]

There's just not enough uniformity of content. There are museum collections and specialist websites, but there's no "WikiSimulate" that aims to be a universal compendium of different objects and their functions. The deepest modeling resources are behind paywalls because it simply isn't practical to gamify every object in the world for free. For example, I can purchase World of Guns or Car Mechanic Simulator, but I don't think the author realized how fragmented and piecemeal the effort to digitally represent the real world would be even decades later.

[Animats]

> There's just not enough uniformity of content.

There's much more uniformity than there used to be. Creation tools now import and export glTF. That's a neutral format and is becoming the industry standard. Until recently, people were trying to move content around in proprietary formats such as .obj and .fbx.

There's NVidia Omniverse, which is a way to link together multiple tools on the same model in real time.

Portable behavior is tougher, although the USD people are working on that. The coming thing there is Unreal Engine's "verse", but it's too soon to see how that will work out.

A big problem now is importing some high-detail model and getting terrible performance. Level of detail generation needs more automation.

3D creation tools are hard to use. It's a really hard user interface problem.

[causality0]

Frankly I'm surprised there hasn't been a 3D scanning revolution the same way there's been a 3D printing revolution.

[Animats]

It's not that useful, except for artworks. The copy has less precision than the scan, which has less precision than the original object, which has less precision than its design. If you scan an auto part, you're unlikely to get a usable duplicate auto part.

[causality0]

I feel like AI could help with this by intelligently asking questions and refining the design. "Is this a flat surface? What is the radius of this curve? What is the exact distance between these two edges?"