Transforming Radiance Field with Lipschitz Network for Photorealistic 3D Scene Stylization

TL;DR

This paper introduces LipRF, a novel framework that uses Lipschitz networks to achieve photorealistic and consistent 3D scene stylization by transforming NeRF appearance representations, improving view consistency and stylization quality.

Contribution

The paper proposes a new Lipschitz mapping approach to enhance 3D scene stylization with NeRFs, ensuring view consistency and photorealism, which was challenging with previous methods.

Findings

LipRF achieves high-quality 3D stylization results.

The framework maintains cross-view consistency in stylized scenes.

LipRF demonstrates robustness in object appearance editing.

Abstract

Recent advances in 3D scene representation and novel view synthesis have witnessed the rise of Neural Radiance Fields (NeRFs). Nevertheless, it is not trivial to exploit NeRF for the photorealistic 3D scene stylization task, which aims to generate visually consistent and photorealistic stylized scenes from novel views. Simply coupling NeRF with photorealistic style transfer (PST) will result in cross-view inconsistency and degradation of stylized view syntheses. Through a thorough analysis, we demonstrate that this non-trivial task can be simplified in a new light: When transforming the appearance representation of a pre-trained NeRF with Lipschitz mapping, the consistency and photorealism across source views will be seamlessly encoded into the syntheses. That motivates us to build a concise and flexible learning framework namely LipRF, which upgrades arbitrary 2D PST methods with Lipschitz mapping tailored for the 3D scene. Technically, LipRF first pre-trains a radiance field to reconstruct the 3D scene, and then emulates the style on each view by 2D PST as the prior to learn a Lipschitz network to stylize the pre-trained appearance. In view of that Lipschitz condition highly impacts the expressivity of the neural network, we devise an adaptive regularization to balance the reconstruction and stylization. A gradual gradient aggregation strategy is further introduced to optimize LipRF in a cost-efficient manner. We conduct extensive experiments to show the high quality and robust performance of LipRF on both photorealistic 3D stylization and object appearance editing.