Key points
- LoFT is designed to reconstruct video geometry.
- The paper, source code and model weights have not yet been released.
- The available results are reported by Horizon Robotics' own team.
Context
LoFT is positioned as a model for reconstructing video geometry. For developers and technical decision-makers, the announcement is not yet a ready-to-use release because its supporting materials remain pending and its results are team-reported.
What this means for you
Not yet happened
For you, LoFT is currently relevant mainly for tracking future evaluations of video-geometry workflows, since the paper, code and weights needed for practical testing are still pending.
Who benefits: Developers evaluating video-geometry pipelines can watch LoFT for later assessment, but no one benefits from an immediately available implementation yet.