Research map / Image generation / Text-to-image diffusion
UniControl: A Unified Diffusion Model for Controllable Visual Generation In the Wild
Can Qin, Shu Zhang, Ning Yu, Yihao Feng, Xinyi Yang, Yingbo Zhou, Huan Wang, Juan Carlos Niebles, Caiming Xiong, Silvio Savarese, Stefano Ermon, Yun Fu and 1 more
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 MoE-style adapter (ablation) | UniControl · six-task avg | FID: 28.7 -> 25.7 | Improved |
| add task-aware HyperNet (on top of MoE adapter) (ablation) | UniControl · six-task avg | FID: 25.7 -> 24.0 | Improved |
| add TaskHyperNet (ablation) | UniControl · Depth-to-image | FID: 23.7 -> 21.3 | Improved |
| unify nine tasks in one model instead of single-task ControlNets | UniControl · Depth-to-image | FID: 25.5 -> 21.3 | Improved |
| unify nine tasks in one model instead of single-task ControlNets | UniControl · HED-to-image | FID: 25.1 -> 23.6 | Improved |
| unify nine tasks in one model instead of single-task ControlNets | UniControl · Canny-to-image | FID: 22.7 -> 22.9 | No clear effect |
| unify nine tasks in one model | UniControl · Segmentation perceptual distance | image perceptual distance: 0.693 -> 0.693 | No clear effect |
| unify nine tasks in one model | UniControl · Normal perceptual distance | image perceptual distance: 0.778 -> 0.623 | Improved |
| unify nine tasks in one model | UniControl · Pose perceptual distance | image perceptual distance: 0.747 -> 0.741 | Improved |
| unify nine tasks in one model | UniControl · model size | parameters: 4.32B -> 1.44B | Improved |
Later papers that compared with this method
7 later accepted papers report UniControl as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- DivControl: Knowledge Diversion for Controllable Image Generation (AAAI 2026 · better on 4 of 4 shared results)
- SketchEvo: Leveraging Drawing Dynamics for Enhanced Image Synthesis (ICLR 2026 · better on 2 of 2 shared results)
- CoLoGen: Progressive Learning of Concept-Localization Duality for Unified Image Generation (CVPR 2026 · better on 1 of 1 shared result)
- UniCombine: Unified Multi-Conditional Combination with Diffusion Transformer (ICCV 2025 · better on 2 of 2 shared results)
- ControlNet-XS: Rethinking the Control of Text-to-Image Diffusion Models as Feedback-Control Systems (ECCV 2024 · better on 2 of 3 shared results)
- AnyControl: Create Your Artwork with Versatile Control on Text-to-Image Generation (ECCV 2024 · better on 2 of 2 shared results)
- ControlNet++: Improving Conditional Controls with Efficient Consistency Feedback (ECCV 2024 · better on 1 of 1 shared result)
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
- VersaGen: Unleashing Versatile Visual Control for Text-to-Image Synthesis (AAAI 2025 · similarity 0.90)
- UNIC-Adapter: Unified Image-instruction Adapter with Multi-modal Transformer for Image Generation (CVPR 2025 · similarity 0.89)
- FlexControl: Computation-Aware Conditional Control with Differentiable Router for Text-to-Image Generation (ICML 2025 · similarity 0.88)
- OmniGen: Unified Image Generation (CVPR 2025 · similarity 0.88)
- Imagine Before You Draw: Visual Prompt Engineering for Image Generation (NeurIPS 2026 · similarity 0.88)
- CtrLoRA: An Extensible and Efficient Framework for Controllable Image Generation (ICLR 2025 · 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.
