UniEdit-Flow
Unleashing Inversion and Editing
in the Era of Flow Models

Guanlong Jiao1,3  Biqing Huang1  Kuan-Chieh Wang2  Renjie Liao3 

1Tsinghua University  2Snap Inc.  3The University of British Columbia 


Paper Code (Coming Soon) BibTex

UniEdit-Flow for image inversion and editing. Our approach proposes a highly accurate and efficient, model-agnostic, training and tuning-free sampling strategy for flow models to tackle image inversion and editing problems. Cluttered scenes are difficult for inversion and reconstruction, leading to failure results on various methods. Our Uni-Inv achieves exact reconstruction even in such complex situations (1st line). Furthermore, existing flow editing always maintain undesirable affects, out region-aware sampling-based Uni-Edit showcases excellent performance for both editing and background preservation (2nd line).


Home Compare: Reconstruction Compare: Editing Gallery

Quantitative Comparison


Quantitative results for inversion and reconstruction on the Conceptual Captions validation dataset. For Stable Diffusion 3 (SD3), we keep each method's NFE close to 100, which means we set sampling step to 50 for once-forward methods (i.e., Euler, FireFlow, and Ours) and to 25 for twice-forward methods (i.e., Heun and RF-Solver). For FLUX, we keep NFE close to 60 (i.e., 30 for once-forward methods and 15 steps for twice-forward methods). We adopt the official implementations of baselines for FLUX, and reimplement their methods for SD3. The best and second best results are bolded and underlined, respectively. Cells are highlighted from worse to better.

Quantitative results for inversion and reconstruction of our Uni-Inv based on Heun method with Flow models and DDIM with Diffusion models. We set the step to 50 for SDXL, 25 for SD3, and 15 for FLUX models to conduct the experiments. The best results are bolded.


Qualitative Comparison





BibTeX

@misc{jiao2025unieditflowunleashinginversionediting,
    title={UniEdit-Flow: Unleashing Inversion and Editing in the Era of Flow Models}, 
    author={Guanlong Jiao and Biqing Huang and Kuan-Chieh Wang and Renjie Liao},
    year={2025},
    eprint={2504.13109},
    archivePrefix={arXiv},
    primaryClass={cs.CV},
    url={https://arxiv.org/abs/2504.13109}, 
}