Hacker News Reader

Could probably be (semi?)automated to run on 3d models of places that doesn't exist. Even ai-built 3d models.

The paper they’re basing this off already does this.

What's the point? You already have the 3d models. If you want an interactive video just use the 3d models.