UltraGen: Efficient Ultra-High-Resolution Image Generation with Hierarchical Local Attention
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911505413308416 |
|---|---|
| author | Zhang, Yuyao Tai, Yu-Wing |
| author_facet | Zhang, Yuyao Tai, Yu-Wing |
| contents | Ultra-high-resolution text-to-image generation is increasingly vital for applications requiring fine-grained textures and global structural fidelity, yet state-of-the-art text-to-image diffusion models such as FLUX and SD3 remain confined to sub 2MP (< $1K\times2K$) resolutions due to the quadratic complexity of attention mechanisms and the scarcity of high-quality high-resolution training data. We present \textbf{\ourwork}, a novel framework that introduces hierarchical local attention with low-resolution global guidance, enabling efficient, scalable, and semantically coherent image synthesis at ultra-high resolutions. Specifically, high-resolution latents are divided into hardware aligned fixed-size local windows to reduce attention complexity from quadratic to near-linear, while a low-resolution latent equipped with scaled positional embeddings injects global semantics as an anchor. A lightweight LoRA adaptation bridges global and local pathways during denoising, ensuring consistency across structure and detail. To maximize inference efficiency and achieve scalable ultra-high-resolution generation, we repermute token sequence in window-first order, so that the GPU-friendly dense local blocks in attention calculation equals to the fixed-size local window in 2D regardless of resolution. Together~\ourwork~reliably scales the pretrained model to resolutions higher than $8K$ with more than $10\times$ speed up and significantly lower memory usage. Extensive experiments demonstrate that~\ourwork~achieves superior quality while maintaining computational efficiency, establishing a practical paradigm for advancing ultra-high-resolution image generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_16325 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | UltraGen: Efficient Ultra-High-Resolution Image Generation with Hierarchical Local Attention Zhang, Yuyao Tai, Yu-Wing Computer Vision and Pattern Recognition Ultra-high-resolution text-to-image generation is increasingly vital for applications requiring fine-grained textures and global structural fidelity, yet state-of-the-art text-to-image diffusion models such as FLUX and SD3 remain confined to sub 2MP (< $1K\times2K$) resolutions due to the quadratic complexity of attention mechanisms and the scarcity of high-quality high-resolution training data. We present \textbf{\ourwork}, a novel framework that introduces hierarchical local attention with low-resolution global guidance, enabling efficient, scalable, and semantically coherent image synthesis at ultra-high resolutions. Specifically, high-resolution latents are divided into hardware aligned fixed-size local windows to reduce attention complexity from quadratic to near-linear, while a low-resolution latent equipped with scaled positional embeddings injects global semantics as an anchor. A lightweight LoRA adaptation bridges global and local pathways during denoising, ensuring consistency across structure and detail. To maximize inference efficiency and achieve scalable ultra-high-resolution generation, we repermute token sequence in window-first order, so that the GPU-friendly dense local blocks in attention calculation equals to the fixed-size local window in 2D regardless of resolution. Together~\ourwork~reliably scales the pretrained model to resolutions higher than $8K$ with more than $10\times$ speed up and significantly lower memory usage. Extensive experiments demonstrate that~\ourwork~achieves superior quality while maintaining computational efficiency, establishing a practical paradigm for advancing ultra-high-resolution image generation. |
| title | UltraGen: Efficient Ultra-High-Resolution Image Generation with Hierarchical Local Attention |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.16325 |