Hi everyone,
We’re rethinking how we tile 3D geospatial data, and we’d like your input early.
Today we run a separate, specialized tiler for each data type — terrain, imagery, photogrammetry, point clouds. As that list grows, our approach has shifted from “how do we build a tiler for this new data type?” to “how do we combine the best of our algorithms to produce the most efficient 3D Tilesets?” Our answer: composable tiling pipelines: reusable, configurable components you chain together in a node graph, so a single pipeline can handle heterogeneous data end to end, scale across compute nodes, and reuse work instead of duplicating it.
We’ve written up the high-level vision on the blog, covering the architecture, some examples, a mockup of a potential graph editor, and a sample JSON config.
We’d love your feedback:
- What capabilities would be most valuable to you in a composable tiling pipeline?
- What formats, transformations, or integrations would support your workflows?
- What does your tiling workflow look like today, and what improvements would be most impactful?
Feel free to respond directly to this post. Your feedback will directly shape our roadmap!
— Jake