We propose LCUDiff, a stable one-step framework for human body restoration that upgrades latent diffusion from the 4-channel latent space to a 16-channel latent space. The method uses channel splitting distillation and prior-preserving adaptation to improve fidelity while keeping one-step efficiency, and further introduces a decoder router for per-sample decoder routing under diverse restoration conditions.
@article{gong2026lcudiff,title={{LCUDiff}: Latent Capacity Upgrade Diffusion for Faithful Human Body Restoration},status={preprint},author={Gong, Jue and Zhou, Zihan and Wang, Jingkai and Li, Shu and Liu, Libo and Lan, Jianliang and Zhang, Yulun},journal={arXiv preprint arXiv:2602.04406},year={2026},}
Light Up Your Face introduces a physically grounded dataset and one-step diffusion model for controllable face fill-light enhancement. It uses a 160K-pair dataset with 6D area-disk lighting control and physics-aware conditioning to improve face illumination while preserving identity and background consistency.
@inproceedings{gong2026lightup,title={{Light Up Your Face}: A Physically Consistent Dataset and Diffusion Model for Face Fill-Light Enhancement},status={accepted},author={Gong, Jue and Zhou, Zihan and Wang, Jingkai and Liu, Xiaohong and Zhang, Yulun and Yang, Xiaokang},booktitle={Proceedings of the International Conference on Machine Learning},year={2026},}
This organizer paper reports the NTIRE 2026 Image Super-Resolution (x4) challenge at CVPRW. It benchmarks bicubic x4 super-resolution across restoration and perceptual tracks, summarizes the datasets, metrics, results, and participant methods, and provides a unified view of recent progress in image super-resolution.
@inproceedings{chen2026ntire_sr_challenge,title={The Fourth Challenge on Image Super-Resolution (x4) at {NTIRE} 2026: Benchmark Results and Method Overview},status={accepted},author={Chen, Zheng and Liu, Kai and Wang, Jingkai and Yan, Xianglong and Li, Jianze and Zhang, Ziqing and Gong, Jue and Li, Jiatong and Sun, Lei and Liu, Xiaoyang and Timofte, Radu and Zhang, Yulun and others},booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},year={2026},}
SODiff improves JPEG artifact removal with semantic-oriented guidance for a pre-trained one-step diffusion model. It introduces a semantic-aligned image prompt extractor and a quality factor-aware time predictor to better handle varying compression levels.
@inproceedings{yang2026sodiff,title={{SODiff}: Semantic-Oriented Diffusion Model for {JPEG} Compression Artifacts Removal},status={accepted},author={Yang, Tingyu and Gong, Jue and Guo, Jinpei and Li, Wenbo and Guo, Yong and Zhang, Yulun},booktitle={The AAAI Conference on Artificial Intelligence},year={2026},}
HAODiff targets practical human-centric restoration under mixed degradation and human motion blur. It introduces triple-branch dual-prompt guidance and a dedicated MPII-Test benchmark, enabling stronger robustness and better visual quality in a single diffusion step.
@article{gong2025haodiff,title={{HAODiff}: Human-Aware One-Step Diffusion via Dual-Prompt Guidance},status={accepted},author={Gong, Jue and Yang, Tingyu and Wang, Jingkai and Chen, Zheng and Liu, Xing and Gu, Hong and Zhang, Yulun and Yang, Xiaokang},journal={Advances in Neural Information Processing Systems},year={2025},}
This work introduces the PERSONA benchmark for human body restoration and proposes OSDHuman, a one-step diffusion model guided by a high-fidelity image embedder. The method improves restoration quality while addressing the lack of large-scale, task-specific datasets for human-centric restoration.
@inproceedings{gong2025osdhuman,title={Human Body Restoration with One-Step Diffusion Model and a New Benchmark},status={accepted},author={Gong, Jue and Wang, Jingkai and Chen, Zheng and Liu, Xing and Gu, Hong and Zhang, Yulun and Yang, Xiaokang},booktitle={Proceedings of the International Conference on Machine Learning},year={2025},}
This organizer paper reviews the NTIRE 2025 Real-World Face Restoration challenge at CVPRW. The challenge focuses on producing realistic face restoration results while preserving identity, using no-reference IQA metrics, FID, and AdaFace-based identity validation to rank participating methods.
@inproceedings{chen2025ntire_face_challenge,title={{NTIRE} 2025 Challenge on Real-World Face Restoration: Methods and Results},status={published},author={Chen, Zheng and Wang, Jingkai and Liu, Kai and Gong, Jue and Sun, Lei and Wu, Zongwei and Timofte, Radu and Zhang, Yulun and others},booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},year={2025},}
This organizer paper presents the NTIRE 2025 Image Super-Resolution (x4) challenge at CVPRW. It covers the bicubic x4 benchmark setting, restoration and perceptual tracks, evaluation protocol, final rankings, and method summaries from participating teams.
@inproceedings{chen2025ntire_sr_challenge,title={{NTIRE} 2025 Challenge on Image Super-Resolution (x4): Methods and Results},status={published},author={Chen, Zheng and Liu, Kai and Gong, Jue and Wang, Jingkai and Sun, Lei and Wu, Zongwei and Timofte, Radu and Zhang, Yulun and others},booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},year={2025},}
OSDFace is a one-step diffusion model for face restoration that combines a visual representation embedder, identity-aware supervision, and GAN-based guidance. It improves realism and identity consistency while keeping inference efficient.
@article{wang2025osdface,title={{OSDFace}: One-Step Diffusion Model for Face Restoration},status={accepted},author={Wang, Jingkai and Gong, Jue and Zhang, Lin and Chen, Zheng and Liu, Xing and Gu, Hong and Liu, Yutong and Zhang, Yulun and Yang, Xiaokang},journal={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},year={2025},}