Home Teaser Abstract Video Links More Examples BibTex Acknowledgments ☰

Base-NeRF ExtraNeRF(Ours)
Base-NeRF ExtraNeRF(Ours)

Abstract

We propose ExtraNeRF, a novel method for extrapolating the range of views handled by a Neural Radiance Field (NeRF). Our main idea is to leverage NeRFs to model scene-specific, fine-grained details, while capitalizing on diffusion models to extrapolate beyond our observed data. A key ingredient is to track visibility to determine what portions of the scene have not been observed, and focus on reconstructing those regions consistently with diffusion models. Our primary contributions include a visibility-aware diffusion-based inpainting module that is fine-tuned on the input imagery, yielding an initial NeRF with moderate quality (often blurry) inpainted regions, followed by a second diffusion model trained on the input imagery to consistently enhance, notably sharpen, the inpainted imagery from the first pass. We demonstrate high-quality results, extrapolating beyond a small number of (typically six or fewer) input views, effectively outpainting the NeRF as well as inpainting newly disoccluded regions inside the original viewing volume. We compare with related work both quantitatively and qualitatively and show significant gains over prior art.

Supplementary Video

More Examples


BibTex

@misc{shih2024extranerf,
    title={ExtraNeRF: Visibility-Aware View Extrapolation of Neural Radiance Fields with Diffusion Models}, 
    author={Meng-Li Shih and Wei-Chiu Ma and Lorenzo Boyice and Aleksander Holynski and Forrester Cole and Brian L. Curless and Janne Kontkanen},
    year={2024},
    eprint={2406.06133},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

Acknowledgments

This work was supported by the UW Reality Lab, Meta, Google, OPPO, and Amazon. We extend our gratitude to Yu-Hsuan Yeh for her assistance in refining the pipeline figure.