DiffSplat: Repurposing Image Diffusion Models for Scalable Gaussian Splat Generation
This addresses the challenge of scalable 3D content creation for applications like gaming or VR, though it builds incrementally on existing diffusion and 3D generation methods.
The paper tackles the problem of generating consistent 3D content from text or images by introducing DiffSplat, a framework that uses image diffusion models to produce 3D Gaussian splats, achieving superior results in text- and image-conditioned generation tasks.
Recent advancements in 3D content generation from text or a single image struggle with limited high-quality 3D datasets and inconsistency from 2D multi-view generation. We introduce DiffSplat, a novel 3D generative framework that natively generates 3D Gaussian splats by taming large-scale text-to-image diffusion models. It differs from previous 3D generative models by effectively utilizing web-scale 2D priors while maintaining 3D consistency in a unified model. To bootstrap the training, a lightweight reconstruction model is proposed to instantly produce multi-view Gaussian splat grids for scalable dataset curation. In conjunction with the regular diffusion loss on these grids, a 3D rendering loss is introduced to facilitate 3D coherence across arbitrary views. The compatibility with image diffusion models enables seamless adaptions of numerous techniques for image generation to the 3D realm. Extensive experiments reveal the superiority of DiffSplat in text- and image-conditioned generation tasks and downstream applications. Thorough ablation studies validate the efficacy of each critical design choice and provide insights into the underlying mechanism.