@misc{indiciae2c2ec30733ab,
  title = {VeloEdit: Training-Free Consistent and Continuous Instruction-Based Image Editing via Velocity Field Decomposition},
  author = {Zongqing Li and Zhihui Liu and Yujie Xie and Shansiyuan Wu and Hongshen Lv and Songzhi Su},
  year = {2026},
  url = {https://arxiv.org/abs/2603.13388},
  note = {Source identifier: 2603.13388}
}