ResDiT: Evoking the Intrinsic Resolution Scalability in Diffusion Transformers
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914175999016960 |
|---|---|
| author | Ma, Yiyang Zhou, Feng Yin, Xuedan Cao, Pu Dang, Yonghao Yin, Jianqin |
| author_facet | Ma, Yiyang Zhou, Feng Yin, Xuedan Cao, Pu Dang, Yonghao Yin, Jianqin |
| contents | Leveraging pre-trained Diffusion Transformers (DiTs) for high-resolution (HR) image synthesis often leads to spatial layout collapse and degraded texture fidelity. Prior work mitigates these issues with complex pipelines that first perform a base-resolution (i.e., training-resolution) denoising process to guide HR generation. We instead explore the intrinsic generative mechanisms of DiTs and propose ResDiT, a training-free method that scales resolution efficiently. We identify the core factor governing spatial layout, position embeddings (PEs), and show that the original PEs encode incorrect positional information when extrapolated to HR, which triggers layout collapse. To address this, we introduce a PE scaling technique that rectifies positional encoding under resolution changes. To further remedy low-fidelity details, we develop a local-enhancement mechanism grounded in base-resolution local attention. We design a patch-level fusion module that aggregates global and local cues, together with a Gaussian-weighted splicing strategy that eliminates grid artifacts. Comprehensive evaluations demonstrate that ResDiT consistently delivers high-fidelity, high-resolution image synthesis and integrates seamlessly with downstream tasks, including spatially controlled generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_01426 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ResDiT: Evoking the Intrinsic Resolution Scalability in Diffusion Transformers Ma, Yiyang Zhou, Feng Yin, Xuedan Cao, Pu Dang, Yonghao Yin, Jianqin Computer Vision and Pattern Recognition Leveraging pre-trained Diffusion Transformers (DiTs) for high-resolution (HR) image synthesis often leads to spatial layout collapse and degraded texture fidelity. Prior work mitigates these issues with complex pipelines that first perform a base-resolution (i.e., training-resolution) denoising process to guide HR generation. We instead explore the intrinsic generative mechanisms of DiTs and propose ResDiT, a training-free method that scales resolution efficiently. We identify the core factor governing spatial layout, position embeddings (PEs), and show that the original PEs encode incorrect positional information when extrapolated to HR, which triggers layout collapse. To address this, we introduce a PE scaling technique that rectifies positional encoding under resolution changes. To further remedy low-fidelity details, we develop a local-enhancement mechanism grounded in base-resolution local attention. We design a patch-level fusion module that aggregates global and local cues, together with a Gaussian-weighted splicing strategy that eliminates grid artifacts. Comprehensive evaluations demonstrate that ResDiT consistently delivers high-fidelity, high-resolution image synthesis and integrates seamlessly with downstream tasks, including spatially controlled generation. |
| title | ResDiT: Evoking the Intrinsic Resolution Scalability in Diffusion Transformers |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2512.01426 |