Figured this vid I just found is worth posting here. This is seriously impressive. In the comments, the guy mentions he did it using some software & a Gopro while riding a bike around the neighborhood. Maybe Asobo could hire him and buy him a motorcycle instead! Imagine the distance he could cover!! lol
Either way, it really gets the imagination going! The future is gonna be cool!
That’s quite impressive indeed.
It’s the next generation of Ortho4XP. Crowd sourced photogrammetry.
I think until the technology matures where it can accurately model each objects perfectly, I prefer a generic Auto-generated buildings and trees compared to this.
The potential is there, but I don’t think this is currently at an acceptable quality at the moment.
You need multiple angles, a few more passes, and imaging from the ground and some from the air or space it’ll be solid.
We already have air/space (top down) photogammetry in the sim via satellite imaging thanks to Bing and Asobo. But if we had some low altitude or ground based imaging data (like google street view) then maybe we can merge that data with the top-down imaging data to create a more comprehensive world with less errors. However I dont think Bing has “street view” (too lazy to check).
I dunno… maybe use the photogrammetry as a first cut pass to create the outline of the 3D model. And have someone, somewhere to actually polish the 3D building objects accordingly. It would still need a lot of data to make, though.
Big data… wew, at one side is a really powerful tool to have… in the other side, a very scary tool to have.
Eventually drones will be collecting the visual data. Drones deployed by google, amazon, private companies specifically for purpose.
Every commercial drone has an HD camera. Delivery drones will need street levell data. That imagery will be collected and stored somewhere and since it has wider value it will be marketed
Google already does this now, you’ve probably seen a Google car with a multiple camera turret driving around your neigborhood, sometimes it’s self-driven (with a bored safety driver sitting behind the wheel). The problem isn’t so much gathering the data, it’s optimizing it for streaming delivery and striking a balance on fidelity versus data size.
This is a technology that in its infancy atm. The fact is works so well in Google and Bing Maps (and MSFS as a byproduct) is already pretty impressive.
It’s still fraught with issues atm. But in time as the core data improves, AI gets better, and client-side processing power increases, it will only get better.
MSFS just gives us a taste of what this kind of tech will be able to do in the future. This stuff will seem pretty quaint 20 years from now.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.