Restoring Initial Noise Sensitivity in Text-to-Image Distillation via Geometric Alignment
Overview & Synopsis
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.
hal.science)
← / → for pages • + / - for zoom • 0 to fit width
Powered by Mozilla PDF.js v3.11
Document Information
| Document Title | Restoring Initial Noise Sensitivity in Text-to-Image Distillation via Geometric Alignment |
|---|---|
| Author / Organization | Huayang Huang, Ruoyu Wang, Jinhui Zhao, Wei Deng, Daiguo Zhou, Jian Luan, Yu Wu, Ye Zhu |
| Subject & Discipline | Multidisciplinary Academic Open Research |
| Categories | Science & Mathematics |
| Publication Year | 2026 |
| License / Rights | Open Access / Creative Commons |
| Resource Verification | Listed PDF (Verification Pending) |
| Original Host Website |
HAL Open Science
(hal.science)
|
| File Format | application/pdf (Portable Document Format) |
| Language | EN |
| Catalog ID | DOC-111370 |
| Original Source Link | Visit Source Page on hal.science » |