Research map / Vision-language and multimodal / Multimodal LLMs
Unlocking Complex Image Editing via Natively Interleaved Visual Textual CoT with Deep Confidence Reasoning
Zhentao Zou, Zhengrong Yue, Kunpeng Du, Binglei Bao, Hanting Li, Haizhen Xie, Guozheng Xu, Yue Zhou, jie hu, Xue Jiang, Xinghao 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 interleaved text-image CoT to text-only CoT (ablation) | BAGEL · MagicBrush test | L1: 0.067 -> 0.058 | Improved |
| add MMDC deep confidence reasoning (width 5) on top of interleaved CoT (ablation) | MURE · MagicBrush test | L1: 0.058 -> 0.049 | Improved |
| add MMDC to interleaved CoT (ablation) | MURE · Emu Edit test | DINO: 0.836 -> 0.897 | Improved |
| add MMDC to interleaved CoT (ablation) | MURE · Emu Edit test | CLIP-Out: 0.303 -> 0.301 | No clear effect |
| increase MMDC search width from 1 to 5 (ablation) | MURE · Emu Edit test | CLIP-I: 0.880 -> 0.920 | Improved |
| increase MMDC search width from 1 to 3 (ablation) | MURE · MagicBrush test | L1: 0.058 -> 0.052 | Improved |
| use interleaved CoT instead of text CoT on challenging MagicBrush subset (ablation) | MURE w/o MMDC · MagicBrush challenging subset | L1: 0.081 -> 0.063 | Improved |
| add MMDC on challenging MagicBrush subset (ablation) | MURE · MagicBrush challenging subset | L1: 0.063 -> 0.052 | Improved |
| use interleaved CoT instead of text CoT on SmartEdit spatial reasoning (ablation) | MURE w/o MMDC · SmartEdit Spatial Reasoning | PSNR: 23.823 -> 25.230 | Improved |
| add MMDC on SmartEdit spatial reasoning (ablation) | MURE · SmartEdit Spatial Reasoning | PSNR: 25.230 -> 25.610 | Improved |
| force interleaved CoT inference for object size change | MURE · MagicBrush test Obj.Size Chg. | L1: 0.0499 -> 0.0479 | Improved |
| replace human scoring with Qwen-VL reward for MMDC on background change | MURE + MMDC · MagicBrush (121 cases) Bg. Chg. | L1: 0.0585 -> 0.0581 | 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.
- SmartEdit-13B — SmartEdit: Exploring Complex Instruction-based Image Editing with Multimodal Large Language Models (CVPR 2024) · better on 2 of 2 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
- TVI-CoT: Text-Visual Interleaved Chain-of-Thought Reasoning for Multimodal Understanding (ICML 2026 · similarity 0.90)
- Meta-CoT: Enhancing Granularity and Generalization in Image Editing (CVPR 2026 · similarity 0.90)
- GIR-Bench: Versatile Benchmark for Generating Images with Reasoning (ICLR 2026 · similarity 0.89)
- Envisioning Beyond the Pixels: Benchmarking Reasoning-Informed Visual Editing (NeurIPS 2025 · similarity 0.89)
- ReasonEdit: Towards Reasoning-Enhanced Image Editing Models (CVPR 2026 · similarity 0.89)
- Reasoning to Edit: Hypothetical Instruction-Based Image Editing with Visual Reasoning (ICML 2026 · 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.
