Research map / Image generation / Text-to-image diffusion
Uni-ControlNet: All-in-One Control to Text-to-Image Diffusion Models
Shihao Zhao, Dongdong Chen, Yen-Chun Chen, Jianmin Bao, Shaozhe Hao, Lu Yuan, Kwan-Yee K. Wong
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 |
|---|---|---|---|
| Use shared multi-scale local condition injection rather than resized direct SPADE injection (ablation) | Uni-ControlNet · Canny condition | FID: 25.89 -> 18.24 | Improved |
| Use multi-scale local condition injection rather than input-layer-only injection (ablation) | Uni-ControlNet · Canny condition | FID: 22.22 -> 18.24 | Improved |
| Inject global condition into the main diffusion model as well as the control adapter (ablation) | Uni-ControlNet · Content condition | FID: 27.06 -> 24.86 | Improved |
| Fine-tune local and global adapters separately instead of jointly from scratch (ablation) | Uni-ControlNet · Canny condition | FID: 21.21 -> 18.24 | Improved |
| Further jointly fine-tune after separate adapter training (ablation) | Uni-ControlNet · Canny condition | FID: 18.24 -> 18.80 | No clear effect |
| Use Uni-ControlNet rather than ControlNet for single-condition generation | Stable Diffusion controllable generation · COCO2017 val, HED | FID: 26.59 -> 17.86 | Improved |
| Use Uni-ControlNet rather than the strongest reported baseline | Stable Diffusion controllable generation · COCO2017 val, Sketch | FID: 18.83 -> 20.11 | Hurt |
| Use Uni-ControlNet rather than the strongest reported baseline | Stable Diffusion controllable generation · COCO2017 val, Pose | mAP: 0.5283 -> 0.2164 | Hurt |
| Use Uni-ControlNet rather than the strongest reported baseline | Stable Diffusion controllable generation · COCO2017 val, Depth | MSE: 87.57 -> 91.05 | Hurt |
| Use Uni-ControlNet rather than the strongest reported baseline | Stable Diffusion controllable generation · COCO2017 val, Segmentation | mIoU: 0.4431 -> 0.3160 | Hurt |
| Use Uni-ControlNet rather than ControlNet | Stable Diffusion controllable generation · COCO2017 val, MLSD | SSIM: 0.7455 -> 0.6773 | Hurt |
| Use Uni-ControlNet rather than ControlNet | Stable Diffusion controllable generation · COCO2017 val, Canny | SSIM: 0.4828 -> 0.4911 | Improved |
Later papers that compared with this method
15 later accepted papers report Uni-ControlNet as a baseline in their results tables; 11 of them report a better number on most of the results they share with it.
- MTVDiff: Multimodal Conditional Latent Diffusion for Enhanced Thermal-to-Visible Face Translation (ECCV 2026 · better on 5 of 5 shared results)
- TrioPose: Native Triple-Stream Diffusion Transformers for Pose-Guided Text-to-Image Generation (NeurIPS 2026 · better on 1 of 1 shared result)
- Semantic Alignment for Pose-Invariant Identity Preserving Diffusion (CVPR 2026 · better on 0 of 1 shared result)
- Simplifying Control Mechanism in Text-to-Image Diffusion Models (AAAI 2025 · better on 4 of 4 shared results)
- FoundHand: Large-Scale Domain-Specific Learning for Controllable Hand Image Generation (CVPR 2025 · better on 4 of 4 shared results)
- Universal Few-shot Spatial Control for Diffusion Models (NeurIPS 2025 · better on 0 of 3 shared results)
- Subjective Camera 1.0: Bridging Human Cognition and Visual Reconstruction through Sequence-Aware Sketch-Guided Diffusion (ICCV 2025 · better on 2 of 2 shared results)
- UniCombine: Unified Multi-Conditional Combination with Diffusion Transformer (ICCV 2025 · better on 2 of 2 shared results)
7 more not listed.
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
- Simplifying Control Mechanism in Text-to-Image Diffusion Models (AAAI 2025 · similarity 0.92)
- CtrLoRA: An Extensible and Efficient Framework for Controllable Image Generation (ICLR 2025 · similarity 0.90)
- Local Conditional Controlling for Text-to-Image Diffusion Models (AAAI 2025 · similarity 0.89)
- Universal Few-shot Spatial Control for Diffusion Models (NeurIPS 2025 · similarity 0.89)
- Compositional Text-to-Image Synthesis with Attention Map Control of Diffusion Models (AAAI 2024 · similarity 0.89)
- T2I-Adapter: Learning Adapters to Dig Out More Controllable Ability for Text-to-Image Diffusion Models (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.
