Tokyo point cloud: Continuous/fast travel between all regions, as well as imported some geometry through the CityGML data. Working on integrating Mapillary street images to improve the building sides. (Using my own Cloudflare R2 bucket and CF Cache to deliver data)

Travelling beneath the ground in forested areas is very cool. you can see how well lidar penetrates the leaves and generates good coverage of the ground.

Yeah this dataset is really great, even pre-colorized. I love the way the trees look too with the lighting baked in :). Some areas are tone mapped a little differently.

Unlock up to 2x faster real-time data processing on Ververica Cloud, surpassing Apache Flink®

Upscale + add-in randomized human avatars when camera is focused on small enough coordinates + plug in randomized roleplaying personalities = you just created a new universe

I wanna try rendering large environment point clouds in WebGPU! Any tips for openly sourcing the LiDAR data?

Both the point cloud and CityGML data are open and can be downloaded from Tokyo's 3D viewer: Explore Map Data > Point Cloud Data > LP Point Cloud Download. You can download by tile in their GUI, or inspect your network requests to see geojson files which include file lists per region.

so good for my mental health

Insanely cool to fly around in!

r2 is great!

Yeah, them not charging for egress bandwidth is a major game changer. This would be very expensive to host with S3

Do you have semantic labels for the objects?

The data has things preclassified to ground or not, the building geo should help with extracting relevant points for each building. At the moment I don’t have building internals or things like door/window. I would also like to extract the cars from the streets and do some in painting.