With the submission earlier today, OpenNeRF has been added as an officially supported third party method for users.
Similar to LERF, OpenNeRF looks at providing semantic based awareness within a capture.
OpenNeRF was shown recently at ICLR 2024. Now that the conference is past, the authors have opened the pathway for people to begin using it directly in Nerfstudio. A full description of OpenNeRF can be found on Nerfstudio’s site.
The command for OpenNeRF is quite simple, with it using: ns-train opennerf. There is not yet any indication how much VRAM OpenNeRF takes to run.
The foundation that Nerfstudio has created makes it easy for authors to open their methods up to the public to try them out. If you are a researcher or just want to implement a method, Nerfstudio has provided documentation on how to do so.
OpenNeRF comes with an MIT License on top of Nerfstudio’s Apache 2.0 license, making it easy for people to build ontop of and use commercially.
New to Gaussian Splatting? Start here
