Saved in:
Bibliographic Details
Main Authors: Xing, Yan, Wang, Pan, Liu, Ligang, Li, Daolun, Zhang, Li
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.14586
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916137893101568
author Xing, Yan
Wang, Pan
Liu, Ligang
Li, Daolun
Zhang, Li
author_facet Xing, Yan
Wang, Pan
Liu, Ligang
Li, Daolun
Zhang, Li
contents We present a novel framework, called FrameNeRF, designed to apply off-the-shelf fast high-fidelity NeRF models with fast training speed and high rendering quality for few-shot novel view synthesis tasks. The training stability of fast high-fidelity models is typically constrained to dense views, making them unsuitable for few-shot novel view synthesis tasks. To address this limitation, we utilize a regularization model as a data generator to produce dense views from sparse inputs, facilitating subsequent training of fast high-fidelity models. Since these dense views are pseudo ground truth generated by the regularization model, original sparse images are then used to fine-tune the fast high-fidelity model. This process helps the model learn realistic details and correct artifacts introduced in earlier stages. By leveraging an off-the-shelf regularization model and a fast high-fidelity model, our approach achieves state-of-the-art performance across various benchmark datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2402_14586
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle FrameNeRF: A Simple and Efficient Framework for Few-shot Novel View Synthesis
Xing, Yan
Wang, Pan
Liu, Ligang
Li, Daolun
Zhang, Li
Computer Vision and Pattern Recognition
Graphics
We present a novel framework, called FrameNeRF, designed to apply off-the-shelf fast high-fidelity NeRF models with fast training speed and high rendering quality for few-shot novel view synthesis tasks. The training stability of fast high-fidelity models is typically constrained to dense views, making them unsuitable for few-shot novel view synthesis tasks. To address this limitation, we utilize a regularization model as a data generator to produce dense views from sparse inputs, facilitating subsequent training of fast high-fidelity models. Since these dense views are pseudo ground truth generated by the regularization model, original sparse images are then used to fine-tune the fast high-fidelity model. This process helps the model learn realistic details and correct artifacts introduced in earlier stages. By leveraging an off-the-shelf regularization model and a fast high-fidelity model, our approach achieves state-of-the-art performance across various benchmark datasets.
title FrameNeRF: A Simple and Efficient Framework for Few-shot Novel View Synthesis
topic Computer Vision and Pattern Recognition
Graphics
url https://arxiv.org/abs/2402.14586