Research map / Image generation / Text-to-image diffusion
Instruction-based Image Editing with Planning, Reasoning, and Generation
Liya Ji, Chenyang Qi, Qifeng Chen
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 CoT planning on top of region reasoning and hint-guided editing (ablation) | Multimodal CoT editing framework · HQEdit-Abstract (user study) | Abstract Score: 24.80% -> 29.41% | Improved |
| add CoT planning on top of region reasoning and hint-guided editing (ablation) | Multimodal CoT editing framework · HQEdit-Abstract (user study) | Editing Quality: 28.64% -> 27.10% | Hurt |
| add CoT planning (ablation) | Multimodal CoT editing framework · MagicBrush test | Total Score: 0.5881 -> 0.5904 | Improved |
| add CoT planning (ablation) | Multimodal CoT editing framework · MagicBrush test | CLIP-T (Global): 0.3026 -> 0.2995 | Hurt |
| train an MLLM editing-region reasoner instead of using pretrained LISA (ablation) | Hints generation network · MagicBrush dev | CLIP-I: 0.9081 -> 0.9219 | Improved |
| use ground-truth masks instead of predicted masks at test time (ablation) | Hints-guided editing network · MagicBrush dev | CLIP-I: 0.9219 -> 0.9277 | Improved |
| augment training data with generative fill (ratio 0.5 versus none) (ablation) | Hints-guided editing network · MagicBrush dev | CLIP-T (Local): 0.2771 -> 0.2835 | Mixed |
| increase hint-related classifier-free guidance scale from 1.0 to 2.0 (ablation) | Hints-guided editing network · MagicBrush dev | CLIP-I vs CLIP-T | Mixed |
| use CoT planning with Flux-based editing model | Flux editing with ControlNet · qualitative | alignment with input | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- HQEdit — HQ-Edit: A High-Quality Dataset for Instruction-based Image Editing (ICLR 2025) · better on 1 of 1 shared result
- HIVE — HIVE: Harnessing Human Feedback for Instructional Visual Editing (CVPR 2024) · better on 1 of 1 shared result
- InstructPix2Pix — InstructPix2Pix: Learning To Follow Image Editing Instructions (CVPR 2023) · 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
- Reasoning to Edit: Hypothetical Instruction-Based Image Editing with Visual Reasoning (ICML 2026 · similarity 0.91)
- SmartEdit: Exploring Complex Instruction-based Image Editing with Multimodal Large Language Models (CVPR 2024 · similarity 0.91)
- ReasonEdit: Towards Reasoning-Enhanced Image Editing Models (CVPR 2026 · similarity 0.91)
- InsightEdit: Towards Better Instruction Following for Image Editing (CVPR 2025 · similarity 0.90)
- Guiding Instruction-based Image Editing via Multimodal Large Language Models (ICLR 2024 · similarity 0.90)
- Meta-CoT: Enhancing Granularity and Generalization in Image Editing (CVPR 2026 · similarity 0.90)
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.
