Large Point-to-Gaussian Model for Image-to-3D Generation

TL;DR

This paper introduces a novel Point-to-Gaussian model that leverages initial point clouds from 3D diffusion models to improve image-to-3D asset generation, achieving state-of-the-art results.

Contribution

The paper proposes a new Point-to-Gaussian approach with an APP block for better image-to-3D generation, integrating point cloud priors and advanced feature fusion techniques.

Findings

Achieves state-of-the-art performance on GSO and Objaverse datasets.

Significantly improves the quality and speed of image-to-3D generation.

Effectively utilizes initial point clouds to facilitate Gaussian parameter prediction.

Abstract

Recently, image-to-3D approaches have significantly advanced the generation quality and speed of 3D assets based on large reconstruction models, particularly 3D Gaussian reconstruction models. Existing large 3D Gaussian models directly map 2D image to 3D Gaussian parameters, while regressing 2D image to 3D Gaussian representations is challenging without 3D priors. In this paper, we propose a large Point-to-Gaussian model, that inputs the initial point cloud produced from large 3D diffusion model conditional on 2D image to generate the Gaussian parameters, for image-to-3D generation. The point cloud provides initial 3D geometry prior for Gaussian generation, thus significantly facilitating image-to-3D Generation. Moreover, we present the \textbf{A}ttention mechanism, \textbf{P}rojection mechanism, and \textbf{P}oint feature extractor, dubbed as \textbf{APP} block, for fusing the image features with point cloud features. The qualitative and quantitative experiments extensively demonstrate the effectiveness of the proposed approach on GSO and Objaverse datasets, and show the proposed method achieves state-of-the-art performance.