Research map / Image generation / Text-to-image diffusion
Concept Sliders: LoRA Adaptors for Precise Control in Diffusion Models
Rohit Gandikota, Joanna Materzynska, Tingrui Zhou, Antonio Torralba, David Bau
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 disentanglement objective with preservation concepts (ablation) | Concept Sliders (age) · Age slider, 2500 images | Interference: 0.36 -> 0.10 | Improved |
| add disentanglement objective (ablation) | Concept Sliders (age) · Age slider, 2500 images | LPIPS: 0.17 -> 0.06 | Improved |
| add low-rank (LoRA) constraint (ablation) | Concept Sliders (age) · Age slider, 2500 images | Interference: 0.19 -> 0.10 | Improved |
| add low-rank constraint (ablation) | Concept Sliders (age) · Age slider, 2500 images | LPIPS: 0.23 -> 0.06 | Improved |
| add disentanglement or low-rank constraint (ablation) | Concept Sliders (age) · Age slider, 2500 images | Delta CLIP: 3.39 -> 3.93 (disentanglement); 3.18 -> 3.93 (low rank) | Improved |
| use slider instead of Prompt2Prompt for age edit | textual concept editing · Age slider, SDXL | Interference: 0.33 -> 0.10 | Improved |
| use slider instead of Prompt2Prompt on Rusty concept | textual concept editing · Rusty slider, SDXL | Delta CLIP: 7.67 -> 7.60 | No clear effect |
| use slider instead of Textual Inversion and Custom Diffusion for eye size | visual concept learning · Eye size slider | Delta_eye: 0.97 and 0.81 -> 1.75 | Improved |
| apply fix-hands slider to SDXL | SDXL · Hand distortion user study | distorted hands (%): 62% -> 22% | Improved |
| apply repair slider to SDXL | SDXL · Realism A/B study | preference for slider image: 80.39% | Improved |
| use SDEdit-style inference (LoRA off for first t steps) | Concept Sliders · Slider scale sweep, 2500 images | LPIPS and CLIP | Mixed |
| apply repair slider direction measured by FID | Repair slider · Realism | FID | Hurt |
Later papers that compared with this method
6 later accepted papers report Concept Sliders as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.
- HEART: Hyperspherical Embedding Alignment via Kent-Representation Traversal in Diffusion Models (NeurIPS 2026 · better on 5 of 6 shared results)
- Learn Once, Edit Anywhere: Visual Direction Transfer for Diffusion Models (ECCV 2026 · better on 5 of 5 shared results)
- Kontinuous Kontext: Continuous Strength Control for Instruction-based Image Editing (CVPR 2026 · better on 2 of 2 shared results)
- Continuous, Subject-Specific Attribute Control in T2I Models by Identifying Semantic Directions (CVPR 2025 · better on 3 of 3 shared results)
- SliderSpace: Decomposing the Visual Capabilities of Diffusion Models (ICCV 2025 · better on 1 of 1 shared result)
- Interpreting the Weight Space of Customized Diffusion Models (NeurIPS 2024 · better on 9 of 10 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
- The Hidden Language of Diffusion Models (ICLR 2024 · similarity 0.88)
- LoRAShop: Training-Free Multi-Concept Image Generation and Editing with Rectified Flow Transformers (NeurIPS 2025 · similarity 0.88)
- Concept Lancet: Image Editing with Compositional Representation Transplant (CVPR 2025 · similarity 0.87)
- Concept Replacer: Replacing Sensitive Concepts in Diffusion Models via Precision Localization (CVPR 2025 · similarity 0.87)
- LoRACLR: Contrastive Adaptation for Customization of Diffusion Models (CVPR 2025 · similarity 0.87)
- Time-Varying LoRA: Towards Effective Cross-Domain Fine-Tuning of Diffusion Models (NeurIPS 2024 · similarity 0.87)
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.
