S2RF: Semantically Stylized Radiance Fields

TL;DR

S2RF introduces a novel method for stylizing 3D scenes by transferring styles from arbitrary images, using a nearest neighborhood loss to ensure multi-view consistency and detailed style transfer.

Contribution

The paper presents a new approach for 3D scene stylization that enables flexible style transfer with multi-view consistency using a neighborhood-based loss.

Findings

Effective style transfer in 3D scenes from arbitrary images

Preserves multi-view consistency and detailed styles

Allows customizable scene stylization from different viewpoints

Abstract

We present our method for transferring style from any arbitrary image(s) to object(s) within a 3D scene. Our primary objective is to offer more control in 3D scene stylization, facilitating the creation of customizable and stylized scene images from arbitrary viewpoints. To achieve this, we propose a novel approach that incorporates nearest neighborhood-based loss, allowing for flexible 3D scene reconstruction while effectively capturing intricate style details and ensuring multi-view consistency.