Multi-Turn Image Editing

Follow this topic Ask about this topic

Latest papers

14 papers3 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

TailSteer: Trajectory-Tail Steering for Drift-Resistant Multi-turn Image Editing
Nov 4, 2026Ruikang Li, Zixuan Chen, Jiarui Wu +6Multi-Turn Image Editing Diffusion-Based Image Editing
MMLab, CUHK · Shanghai Artificial Intelligence Laboratory · SenseTime Research
MT Forcing: Towards Stable and Consistent Multi-Turn Image Editing with Autoregressive Diffusion
Oct 18, 2026Houyuan Chen, Fanzhou Wang, Xiangyu Fan +10Multi-Turn Image Editing Diffusion Model Distillation
SenseTime Research · HKUST MMLab · Stanford University +1
MTEdit: Packed Trajectories Beat Rollouts for Multi-Turn Image Editing
Oct 16, 2026Houyuan Chen, Fanzhou Wang, Xiangyu Fan +8Multi-Turn Image Editing Image Editing
SenseTime Research · HKUST MMLab · Stanford University +1
ACEdit: Stable Multi-Turn Image Editing via Asymmetric Conditioning
Oct 15, 2026Sara Dorfman, Nir Goren, Gal Barak +2Multi-Turn Image Editing Image Editing
Tel Aviv University
On-Policy Self-Distillation for Multi-Turn Image Editing
Sep 28, 2026Liangbing Zhao, Le Zhuo, Mohamed ElhoseinyText-Guided Image Editing Diffusion Model Distillation
KAUST · Krea AI
RegionCache: Semantic-Aware Region Reuse for Efficient Multi-Turn Image Generation
Aug 30, 2026Peizheng Li, Xin Ai, Hanyuan Liu +2Multi-Turn Image Editing Diffusion Model Caching
School of Computer Science and Engineering, Northeastern University, China · School of Computer Science
What to Edit Next: Visually Aligned Image-Editing Follow-Up Suggestions in Conversational Systems
Aug 3, 2026Zhijing Zhang, Jinpeng Yu, Xin Song +6Multimodal Recommendation Image Editing
1Qwen Business Unit of Alibaba · 2Southeast University
Making Implicit Preservation Intent Explicit in Conversational Image Editing
Jul 8, 2026Soomin Han, Jihyung Ahn, Bumsoo Kim +1Image Editing Evaluation Image Editing
Sogang University · Korea University · Chung-Ang University
An Extensive Benchmark for Single-round and Multi-round Instruction-based Image Editing
Jun 14, 2026Yiwei Ma, Ke Ye, Weihuang Lin +4Image Editing Evaluation Text-Guided Image Editing
Key Laboratory of Multimedia Trusted Perception and Efficient Computing, Ministry of Education of China, Xiamen University, 361005, Fujian, P.R. China. · National University of Singapore, 119077, Singapore.
AnchorEdit: Maintaining Temporal Consistency in Multi-turn Image Editing via Causal Memory
Jun 10, 2026Hang Xu, Xiaoxiao Ma, Guohui Zhang +7Multi-Turn Image Editing Image Editing
University of Science and Technology of China · JD Explore Academy
Edit-R2: Context-Aware Reinforcement Learning for Multi-Turn Image Editing
Jun 4, 2026Yuxiao Ye, Haoran He, Fangyuan Kong +4Reinforcement Learning Image Editing Evaluation
Hong Kong University of Science and Technology · Kuaishou Technology
MT-EditFlow: Reinforcement Learning for Multi-Turn Image Editing with Flow Matching
Jun 1, 2026Jiahui Huang, Yasi Zhang, Tianyu Chen +6Multi-Turn Image Editing Multimodal Reward Modeling
Apple · University of California, Los Angeles · University of Texas at Austin +1
Why Do DiT Editors Drift? Plug-and-Play Low Frequency Alignment in VAE Latent Space
May 7, 2026Xiaoce Wang, Sifan Zhou, Kaifei Wang +4Diffusion Transformer Multi-Turn Image Editing
♠Tsinghua University · ♡Carnegie Mellon University · ♢Peking University +3
IMAGAgent: Orchestrating Multi-Turn Image Editing via Constraint-Aware Planning and Reflection
Feb 12, 2026Fei Shen, Chengyu Xie, Lihong Wang +5Multi-Turn Image Editing Image Editing
National University of Singapore · Nanjing University of Science and Technology · Nanjing Forestry University

Nov 4, 2026J/K move · Enter open · S save

TailSteer: Trajectory-Tail Steering for Drift-Resistant Multi-turn Image Editing

Ruikang Li, Zixuan Chen, Jiarui Wu, Mingde Yao+5

MMLab, CUHK · Shanghai Artificial Intelligence Laboratory · SenseTime Research

Multi-Turn Image Editing Diffusion-Based Image Editing Diffusion Model Fine-Tuning

Open paper AskPDF arXiv

Strong single-turn editing performance does not guarantee reliable multi-turn interaction: as edited outputs are repeatedly reused, small unintended changes can accumulate into severe visual drift. This long-term drift is often mitigated through training on multi-turn data or cross-turn rollouts, potentially altering pretrained priors, or through specialized inference-time mechanisms that are difficult to scale to foundation models. We introduce TailSteer, a final-stage training framework that improves multi-turn robustness while retaining the generative and editing capabilities acquired through earlier training. TailSteer follows a simple principle: keep early denoising dynamics close to those of the base model and concentrate preservation corrections near the trajectory tail, where fine-grained deviations can be suppressed. These corrections are learned from the editor's own on-policy denoising rollouts, using unchanged source content as a preservation reference without requiring curated ground-truth editing pairs. Experiments across the FLUX, SenseNova, and Qwen families show that TailSteer substantially reduces cumulative drift over 15-20 editing turns while largely preserving base-model performance on extensive text-to-image and single-image editing benchmarks. These results demonstrate that long-horizon editing robustness can be unlocked in foundation models after TailSteer, rather than by relearning their existing capabilities.