SpheRoPE: Zero-Shot Optimization-Free 360 Panorama Generation with Spherical RoPE

Hugging Face Daily Papers Papers
zero-shot optimization-free 360-panorama spherical-rope diffusion-transformers image-generation

Summary

This paper introduces SpheRoPE, a zero-shot and optimization-free framework that injects spherical priors into pre-trained diffusion transformers to generate 360 panoramic images and videos, overcoming topological constraints without retraining.

We present a zero-shot, training-free and optimization-free framework for generating 360 panoramic images and videos by directly injecting spherical priors into pre-trained diffusion transformers. Existing methods either rely on costly fine-tuning on scarce panoramic data that limits generalization, or leverage multi-step optimization that incurs prohibitive inference latency. We observe that contemporary generative models natively exhibit some panoramic priors from large-scale training. However, these emergent capabilities are insufficient, as the models fundamentally fail to satisfy the rigorous topological constraints imposed by equirectangular projection (ERP). We introduce a zero-shot and optimization-free approach that resolves these constraints at inference time. Spherical RoPE replaces standard rotary position embeddings: low-frequency channels are re-parameterized as 3D Cartesian coordinates to natively encode the spherical manifold, while high-frequency channels are harmonically quantized to enforce exact periodicity. Coupled with complementary Semantic Distortion classifier-free guidance (CFG) that explicitly steers geometry, we avoid retraining and inherit the full creative breadth of state-of-the-art models. Our approach generalizes across diverse backbones and 360 generation modalities. We demonstrate this across text-to-panorama using Flux.1, Flux.2, and LTX-Video backbones, achieving competitive performance against baselines, all while remaining training-free. Project page: https://orhir.github.io/SpheRoPE
Original Article
View Cached Full Text

Cached at: 07/01/26, 11:45 PM

Paper page - SpheRoPE: Zero-Shot Optimization-Free 360 Panorama Generation with Spherical RoPE

Source: https://huggingface.co/papers/2606.32033

Abstract

A novel zero-shot framework injects spherical priors into pre-trained diffusion transformers for 360 panoramic generation, using spherical RoPE and semantic distortion guidance to overcome topological constraints without training or optimization.

We present azero-shot,training-freeandoptimization-freeframework for generating 360 panoramic images and videos by directly injectingspherical priorsinto pre-traineddiffusion transformers. Existing methods either rely on costly fine-tuning on scarce panoramic data that limits generalization, or leverage multi-step optimization that incurs prohibitive inference latency. We observe that contemporary generative models natively exhibit some panoramic priors from large-scale training. However, these emergent capabilities are insufficient, as the models fundamentally fail to satisfy the rigorous topological constraints imposed byequirectangular projection(ERP). We introduce azero-shotandoptimization-freeapproach that resolves these constraints at inference time.Spherical RoPEreplaces standardrotary position embeddings: low-frequency channels are re-parameterized as3D Cartesian coordinatesto natively encode the spherical manifold, while high-frequency channels are harmonically quantized to enforce exactperiodicity. Coupled with complementarySemantic Distortionclassifier-free guidance(CFG) that explicitly steers geometry, we avoid retraining and inherit the full creative breadth of state-of-the-art models. Our approach generalizes across diverse backbones and 360 generation modalities. We demonstrate this across text-to-panorama using Flux.1, Flux.2, and LTX-Video backbones, achieving competitive performance against baselines, all while remainingtraining-free. Project page: https://orhir.github.io/SpheRoPE

View arXiv pageView PDFProject pageGitHub6Add to collection

Get this paper in your agent:

hf papers read 2606\.32033

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.32033 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.32033 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.32033 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

SP^3: Spherical Priors for Plug-and-Play Restoration

Hugging Face Daily Papers

This paper introduces SP³, a method using Spherical Encoder priors for Plug-and-Play image restoration, achieving perceptual quality comparable to zero-shot diffusion priors while being 3–630× faster across tasks.

PanoWorld: Real-World Panoramic Generation

Hugging Face Daily Papers

PanoWorld proposes a method for long-range memory in panoramic world models using rotation-equivariant representations, with a three-stage training pipeline and a new large-scale dataset World360. The model outperforms alternatives by a large margin.