Research map / Image generation / Autoregressive and latent generation
HierEdit: Region-Aware Hierarchical Diffusion for Efficient High-Resolution Editing
Yuyao Zhang, Alexander Huang-Menders, Yu-Wing Tai
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add local window attention (Flash Sparse kernel) (ablation) | HierEdit · Acceleration ablation (1K, 50% edited) | speed (s): 4.98 -> 2.34 | Improved |
| add feature caching on top of local window attention (ablation) | HierEdit · Acceleration ablation (1K, 50% edited) | speed (s): 9.32 -> 4.98 | Improved |
| add joint token integration on top of LWA and feature caching (ablation) | HierEdit · Acceleration ablation (1K, 50% edited) | speed (s): 29.12 -> 9.32 | Improved |
| add all three acceleration components (LWA, FC, TI) (ablation) | HierEdit · Acceleration ablation (1K, 50% edited) | speed (s): 29.12 -> 2.34 | Improved |
| add local window attention (ablation) | HierEdit · Acceleration ablation (1K, 50% edited) | CLIP-I: 0.922 -> 0.931 | Improved |
| add local window attention (ablation) | HierEdit · Acceleration ablation (1K, 50% edited) | PSNR: 18.84 -> 19.01 | Improved |
| add LWA, FC and TI to the full model (ablation) | HierEdit · Acceleration ablation (1K, 50% edited) | PSNR: 19.03 -> 19.01 | No clear effect |
| use HierEdit instead of FLUX-Fill for text-guided inpainting at 1K | FLUX-Fill · Text-guided inpainting (1K) | latency (s): 21.4s -> 6.97s | Improved |
| use HierEdit instead of FLUX-Fill for text-guided inpainting at 1K | FLUX-Fill · Text-guided inpainting (1K) | FID: 56.1 -> 39.5 | Improved |
| use HierEdit instead of ACE++ for text-guided inpainting | ACE++ · Text-guided inpainting (1K) | FID: 37.2 -> 39.5 | Hurt |
| use HierEdit instead of ACE++ for image-guided inpainting | ACE++ · Image-guided inpainting (1K) | latency (s): 52.9s -> 7.84s | Improved |
| use HierEdit instead of OminiControl2 at 4K with 25% edited | OminiControl2 · Inpainting speed (4K, 25% edited) | inference time (s): 155s -> 91.4s | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- OminiControl2* — OminiControl: Minimal and Universal Control for Diffusion Transformer (ICCV 2025) · better on 2 of 3 shared results
- EasyControl — EasyControl: Adding Efficient and Flexible Control for Diffusion Transformer (ICCV 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- EditBridge: Towards Faithful and Efficient Ultra-High-Resolution Image Editing (NeurIPS 2026 · similarity 0.91)
- High-Fidelity Diffusion-Based Image Editing (AAAI 2024 · similarity 0.89)
- Enabling Instructional Image Editing with In-Context Generation in Large Scale Diffusion Transformer (NeurIPS 2025 · similarity 0.89)
- Towards Efficient Diffusion-Based Image Editing with Instant Attention Masks (AAAI 2024 · similarity 0.89)
- h-Edit: Effective and Flexible Diffusion-Based Editing via Doob's h-Transform (CVPR 2025 · similarity 0.89)
- Accelerating Text-to-Image Editing via Cache-Enabled Sparse Diffusion Inference (AAAI 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
