Research map / Image generation / Text-to-image diffusion
IntrinsicControlNet: Cross-distribution Image Generation with Real and Unreal
Jiayuan Lu, Rengan Xie, Zixuan Xie, Zhizhen Wu, Dianbing Xi, Qi Ye, Rui Wang, Hujun Bao, Yuchi Huo
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 IntrinsicControlNet instead of RGB↔X, Multi-ControlNet, Ctrl-X | IntrinsicControlNet · All scenes | FID: 42.90 -> 26.40 (vs RGB↔X) | Improved |
| Use IntrinsicControlNet instead of RGB↔X on Walkway scene | IntrinsicControlNet · Walkway scene | LPIPS: 0.177 (RGB↔X) vs 0.210 (Ours) | Hurt |
| Train on both synthetic and real data instead of real data only (ablation) | IntrinsicControlNet · Ablation | LPIPS: 0.5141 -> 0.3826 | Improved |
| Train on both real and synthetic data instead of synthetic data only (ablation) | IntrinsicControlNet · Ablation | CLIP score: 0.2933 -> 0.3137 | Improved |
| Add separate material and geometry branches (ablation) | IntrinsicControlNet · Ablation | LPIPS: 0.4181 -> 0.3826 | Improved |
| Add realistic discriminator (ablation) | IntrinsicControlNet · Ablation | LPIPS: 0.4403 -> 0.3826 | Improved |
| Add lighting (environment map) condition (ablation) | IntrinsicControlNet · Ablation | LPIPS: 0.4150 -> 0.3826 | Improved |
| Add synthetic data alongside real data on DINOv2 similarity (ablation) | IntrinsicControlNet · Ablation | DINOv2 similarity: 0.9535 -> 0.9796 | Improved |
| Train synthetic-only versus full on LPIPS (ablation) | IntrinsicControlNet · Ablation | LPIPS: 0.3862 -> 0.3826 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Ctrl-X — Ctrl-X: Controlling Structure and Appearance for Text-To-Image Generation Without Guidance (NeurIPS 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
- Generative Photography: Scene-Consistent Camera Control for Realistic Text-to-Image Synthesis (CVPR 2025 · similarity 0.89)
- DreamControl: Control-Based Text-to-3D Generation with 3D Self-Prior (CVPR 2024 · similarity 0.87)
- FlexControl: Computation-Aware Conditional Control with Differentiable Router for Text-to-Image Generation (ICML 2025 · similarity 0.86)
- LightIt: Illumination Modeling and Control for Diffusion Models (CVPR 2024 · similarity 0.86)
- CtrLoRA: An Extensible and Efficient Framework for Controllable Image Generation (ICLR 2025 · similarity 0.86)
- FashionR2R: Texture-preserving Rendered-to-Real Image Translation with Diffusion Models (NeurIPS 2024 · similarity 0.86)
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.
