Customizing 360-Degree Panoramas through Text-to-Image Diffusion Models

TL;DR

This paper introduces a method for customizing 360-degree panoramas using text-to-image diffusion models, addressing global geometry and seamless stitching through a specialized dataset, fine-tuning, and a novel StitchDiffusion technique.

Contribution

It presents a new approach combining dataset curation, model fine-tuning with LoRA, and StitchDiffusion to generate seamless, high-quality 360-degree panoramas from text prompts.

Findings

Effective generation of seamless 360-degree panoramas.

Model generalizes well to unseen scenes.

StitchDiffusion improves continuity between image edges.

Abstract

Personalized text-to-image (T2I) synthesis based on diffusion models has attracted significant attention in recent research. However, existing methods primarily concentrate on customizing subjects or styles, neglecting the exploration of global geometry. In this study, we propose an approach that focuses on the customization of 360-degree panoramas, which inherently possess global geometric properties, using a T2I diffusion model. To achieve this, we curate a paired image-text dataset specifically designed for the task and subsequently employ it to fine-tune a pre-trained T2I diffusion model with LoRA. Nevertheless, the fine-tuned model alone does not ensure the continuity between the leftmost and rightmost sides of the synthesized images, a crucial characteristic of 360-degree panoramas. To address this issue, we propose a method called StitchDiffusion. Specifically, we perform pre-denoising operations twice at each time step of the denoising process on the stitch block consisting of the leftmost and rightmost image regions. Furthermore, a global cropping is adopted to synthesize seamless 360-degree panoramas. Experimental results demonstrate the effectiveness of our customized model combined with the proposed StitchDiffusion in generating high-quality 360-degree panoramic images. Moreover, our customized model exhibits exceptional generalization ability in producing scenes unseen in the fine-tuning dataset. Code is available at https://github.com/littlewhitesea/StitchDiffusion.