{"ID":2921207,"CreatedAt":"2026-06-02T02:42:49.606572591Z","UpdatedAt":"2026-06-04T00:54:56.190393508Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2606.01651","arxiv_id":"2606.01651","title":"Restoring Initial Noise Sensitivity in Text-to-Image Distillation via Geometric Alignment","abstract":"Generative distillation significantly accelerates text-to-image (T2I) generation by compressing multi-step trajectories into few-step student models while preserving perceptual quality. However, existing methods primarily optimize efficiency and output fidelity, often neglecting critical properties of the original trajectory. In this work, we identify a key missing property: sensitivity to initial noise, whose degradation impairs downstream control methods relying on noise-based optimization and manipulation. We trace this issue to standard distillation objectives that enforce pointwise output alignment, inadvertently flattening the input-output landscape and suppressing the teacher's local geometric structure. To address this, we propose Geometry-Aware Distillation (GAD), a sensitivity-preserving framework that aligns the local functional behavior of teacher and student models. Specifically, GAD matches Jacobian-vector products with respect to input noise, enabling the student to reproduce the teacher's differential response to perturbations. Extensive experiments across multiple T2I paradigms and noise-driven control tasks demonstrate that GAD significantly restores sensitivity and improves diversity while maintaining high visual fidelity. Code is available at https://github.com/Hannah1102/GAD.","short_abstract":"Generative distillation significantly accelerates text-to-image (T2I) generation by compressing multi-step trajectories into few-step student models while preserving perceptual quality. However, existing methods primarily optimize efficiency and output fidelity, often neglecting critical properties of the original traj...","url_abs":"https://arxiv.org/abs/2606.01651","url_pdf":"https://arxiv.org/pdf/2606.01651v1","authors":"[\"Huayang Huang\",\"Ruoyu Wang\",\"Jinhui Zhao\",\"Wei Deng\",\"Daiguo Zhou\",\"Jian Luan\",\"Yu Wu\",\"Ye Zhu\"]","published":"2026-06-01T03:58:47Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[]","has_code":false,"code_links":[{"ID":612573,"CreatedAt":"2026-06-02T02:42:49.606572591Z","UpdatedAt":"2026-06-02T02:42:49.606572591Z","DeletedAt":null,"paper_id":2921207,"paper_url":"https://arxiv.org/abs/2606.01651","paper_title":"Restoring Initial Noise Sensitivity in Text-to-Image Distillation via Geometric Alignment","repo_url":"https://github.com/Hannah1102/GAD","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}