Research map / Image generation / Text-to-image diffusion
Cross-ControlNet: Training-Free Fusion of Multiple Conditions for Text-to-Image Generation
Xiang Liu, Junjun Jiang, Wei Han, Kui Jiang, Xianming Liu
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 PixFusion over MaxFusion fusion (ablation) | Cross-ControlNet · Conflicting conditions (Pose, Seg) | mIoU-P: 0.3528 -> 0.3610 | Improved |
| add ChannelFusion in final layer (ablation) | Cross-ControlNet · Conflicting conditions (Pose, Seg) | mIoU-P: 0.3610 -> 0.3648 | Improved |
| add KV-Injection (ablation) | Cross-ControlNet · Conflicting conditions (Pose, Seg) | mIoU-P: 0.3648 -> 0.3719 | Improved |
| apply ChannelFusion at shallow layers instead of final layer (ablation) | Cross-ControlNet · Conflicting conditions (Pose, Seg) | mIoU-P: 0.3648 -> 0.3447 | Hurt |
| start KV-Injection at step 5 instead of step 0 (ablation) | Cross-ControlNet · Conflicting conditions (Pose, Seg) | MSE-S: 0.2003 -> 0.1983 | Improved |
| start KV-Injection late (step 20) (ablation) | Cross-ControlNet · Conflicting conditions (Pose, Seg) | CLIP score: 0.2978 -> 0.2952 | Hurt |
| increase Gaussian kernel std from 0.8 to 3.0 (ablation) | Cross-ControlNet · Conflicting conditions (Pose, Seg) | mIoU-P: 0.3718 -> 0.3667 | Hurt |
| set threshold delta to 0 (naive averaging) (ablation) | Cross-ControlNet · Qualitative | pose faithfulness | Hurt |
| use Cross-ControlNet instead of MaxFusion on complementary conditions | Cross-ControlNet · Complementary (Hed, Seg) | SSIM: 0.2382 -> 0.2404 | Improved |
| use Cross-ControlNet vs EasyControl on FLUX | Cross-ControlNet · Complementary (Canny, Depth), FLUX.1-dev | SSIM: 0.4129 -> 0.3735 | Hurt |
| use Cross-ControlNet vs MaxFusion latency | Cross-ControlNet · Inference time | time (s): 12.932 -> 14.759 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- AnyControl — AnyControl: Create Your Artwork with Versatile Control on Text-to-Image Generation (ECCV 2024) · better on 8 of 8 shared results
- MaxFusion — MaxFusion: Plug&Play Multi-Modal Generation in Text-to-Image Diffusion Models (ECCV 2024) · better on 7 of 8 shared results
- EasyControl — EasyControl: Adding Efficient and Flexible Control for Diffusion Transformer (ICCV 2025) · better on 0 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
- 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 0.91)
- FlexControl: Computation-Aware Conditional Control with Differentiable Router for Text-to-Image Generation (ICML 2025 · similarity 0.89)
- Universal Few-shot Spatial Control for Diffusion Models (NeurIPS 2025 · similarity 0.89)
- Local Conditional Controlling for Text-to-Image Diffusion Models (AAAI 2025 · similarity 0.89)
- R&B: Region and Boundary Aware Zero-shot Grounded Text-to-image Generation (ICLR 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.
