# DiffSplat: Repurposing Image Diffusion Models for Scalable Gaussian Splat Generation

Source: [DiffSplat: Repurposing Image Diffusion Models for Scalable Gaussian Splat Generation](<https://www.sota2.com/research/paper/68fa76b61e209131cea6ea4c>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-27T06:45:57.319Z

Authors: Chenguo Lin, Panwang Pan, Bangbang Yang, Zeming Li, Yadong Mu

Published: 2025-01-28T07:38:59.000Z

## Abstract

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.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image-conditioned 3D Generation on GSO](<https://www.sota2.com/research/sota/image-conditioned-3d-generation-on-gso.md>): Image-conditioned 3D Generation; GSO; PSNR: 19.59.

- [Text-conditioned 3D asset generation on G-Objaverse (evaluation)](<https://www.sota2.com/research/sota/text-conditioned-3d-asset-generation-on-g-objaverse-evaluation.md>): Text-conditioned 3D asset generation; G-Objaverse (evaluation); IS: 13.75.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2501.16764>)

[PDF](<https://arxiv.org/pdf/2501.16764.pdf>)