Supplementary Results for

👻 Nerfbusters 🧹

Removing Ghostly Artifacts from Casually Captured NeRFs

Frederik Warburg o 👻
Ethan Weber o 👻
Matthew Tancik 👻
Aleksander Hołyński 👻🧹
Angjoo Kanazawa 👻

🧹Google Research           👻 UC Berkeley

oDenotes equal contribution
paper
code-and-data
code-and-data
Paper
Code and data
Nerfstudio page

Nerfbusters Dataset

We show training and evaluation videos from the 12 scenes of the Nerfbusters dataset. We hope this casual captured dataset that bridges the gap between current evaluation practice and the users requirements, will refocus the community's efforts.

Nerfbusters Results

We compare Nerfbusters with various baselines. Click the baselines buttons and swipe left and right to see the results. Our Nerfbusters method appears on the right. You can also use the pause button and select different methods for the same capture.