weiyithu/NerfingMVS
NerfingMVS uses neural radiance fields optimized with depth priors to reconstruct 3D scenes from multiple indoor views.
Not currently ranked — collecting fresh signals.
star history
NerfingMVS is a guided optimization approach for neural radiance fields focused on indoor multi-view stereo reconstruction. The method trains depth priors using learned estimators and leverages these priors to guide NeRF optimization for better geometric accuracy. It combines traditional structure-from-motion (COLMAP) with deep learning to produce high-quality 3D reconstructions from multiple RGB images.
Frequently asked
- What is weiyithu/NerfingMVS?
- NerfingMVS uses neural radiance fields optimized with depth priors to reconstruct 3D scenes from multiple indoor views.
- Is NerfingMVS open source?
- Yes — weiyithu/NerfingMVS is open source, released under the MIT license.
- What language is NerfingMVS written in?
- weiyithu/NerfingMVS is primarily written in Python.
- How popular is NerfingMVS?
- weiyithu/NerfingMVS has 436 stars on GitHub.
- Where can I find NerfingMVS?
- weiyithu/NerfingMVS is on GitHub at https://github.com/weiyithu/NerfingMVS.