I am current a Tenure-Track Assistant Professor at Ecole Polytechnique, IP Paris. Prior to that, I was a postdoc researcher at @VISTA Team, LIX, Ecole Polytechnique, working with Vicky Kalogeiton. I finished my Ph.D. at Univ Rennes 1, Inria Rennes and IRISA advised by Marc Christie and Eric Marchand. My main research interests lie in image, video or other content generation, and I am also very interested in computational cinematography.

Xi Wang wearing a steampunk top hat
Xi Wang, formal portrait
Oct 27, 2026 MUSE: Manipulating Unified Framework for Synthesizing Emotions in Images via Test-Time Optimization is accepted at IEEE Transactions on Multimedia.
Sep 26, 2026 FakeParts: a New Family of AI-Generated DeepFakes is accepted at the NeurIPS 2026 Evaluations & Datasets Track.
Aug 1, 2026 I received an ANR JCJC grant (Young Researchers program) for the project REEL-WORLD: World Models for Cinematic Tasks, as PI.
Mar 18, 2026 Three papers accepted at ECCV 2026: VERTIGO, VIGOR, and What Images Cannot Say.
Mar 10, 2026 Long Story Short: Story-level Video Understanding from 20K Short Films is accepted at IJCV.
Jan 26, 2026 Two papers accepted at ICLR 2026: Pulp Motion, and Soft-Di[M]O.

Recent and selected publications

all publications →
  1. DualHeads: Reward-tilted few-step diffusion made simple
    Yuanzhi Zhu, Eleftherios Tsonis, Simo Alami Chehboune, Xi Wang, Stéphane Lathuilière, and Vicky Kalogeiton
    arXiv preprint arXiv:2610.28858, 2026
  2. When Left Is Right? Camera-Motion Understanding beyond Screen
    Chong Cheng, Ran Zhang, Xinyuan Li, Alan Zhao, Hao Wang, and Xi Wang
    arXiv preprint arXiv:2610.17242, 2026
  3. MUSE: Manipulating Unified Framework for Synthesizing Emotions in Images via Test-Time Optimization
    Yingjie Xia, Xi Wang, Jinglei Shi, Vicky Kalogeiton, and Jian Yang
    IEEE Transactions on Multimedia (TMM), 2026
  4. FakeParts: a New Family of AI-Generated DeepFakes
    Ziyi Liu, Firas Gabetni, Awais Hussain Sani, Xi Wang, Soobash Daiboo, Gaëtan Brison, Gianni Franchi, and Vicky Kalogeiton
    In Advances in Neural Information Processing Systems (NeurIPS), Evaluations and Datasets Track, 2026
  5. VERTIGO: Visual Preference Optimization for Cinematic Camera Trajectory Generation
    Mengtian Li, Yuwei Lu, Feifei Li, Chenqi Gan, Zhifeng Xie, and Xi Wang
    In European Conference on Computer Vision (ECCV), 2026
  6. VIGOR: VIdeo Geometry-Oriented Reward for Temporal Generative Alignment
    Tengjiao Yin, Jinglei Shi, Heng Guo, and Xi Wang
    In European Conference on Computer Vision (ECCV), 2026
  7. What Images Cannot Say: Language-Guided Olfactory Representation Learning
    In European Conference on Computer Vision (ECCV), 2026
  8. Long Story Short: Story-level Video Understanding from 20K Short Films
    International Journal of Computer Vision (IJCV), 2026
  9. Pulp Motion: Framing-aware multimodal camera and human motion generation
    In International Conference on Learning Representations (ICLR), 2026
  10. Soft-Di[M]O: Improving One-Step Discrete Image Generation with Soft Embeddings
    In International Conference on Learning Representations (ICLR), 2026
  11. Di[M]O: Distilling Masked Diffusion Models into One-step Generator
    In International Conference on Computer Vision (ICCV), 2025
  12. AKiRa: Augmentation Kit on Rays for optical video generation
    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025
  13. Analysis of Classifier-Free Guidance Weight Schedulers
    Transactions on Machine Learning Research, 2024

Collaborators

Nefeli AndreouMarie-Paule CaniBaoquan ChenChong ChengMarc ChristieRobin CourantNicolas DufourVictoria Fernández AbrevayaGianni FranchiRidouane GhermiHeng GuoHongda JiangVicky KalogeitonIvan LaptevStéphane LathuilièreMengtian LiLibin LiuZiyi LiuDavid LoiseauxEric MarchandDavid PicardJinglei ShiEleftherios TsonisYingjie XiaJian YangTengjiao YinRan ZhangYuanzhi Zhu