Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Pu, Yifan, Han, Yizeng, Tang, Zhiwei, Tang, Jiasheng, Wang, Fan, Zhuang, Bohan, Huang, Gao
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2512.13006
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917146733314048
author Pu, Yifan
Han, Yizeng
Tang, Zhiwei
Tang, Jiasheng
Wang, Fan
Zhuang, Bohan
Huang, Gao
author_facet Pu, Yifan
Han, Yizeng
Tang, Zhiwei
Tang, Jiasheng
Wang, Fan
Zhuang, Bohan
Huang, Gao
contents Diffusion distillation has dramatically accelerated class-conditional image synthesis, but its applicability to open-ended text-to-image (T2I) generation is still unclear. We present the first systematic study that adapts and compares state-of-the-art distillation techniques on a strong T2I teacher model, FLUX.1-lite. By casting existing methods into a unified framework, we identify the key obstacles that arise when moving from discrete class labels to free-form language prompts. Beyond a thorough methodological analysis, we offer practical guidelines on input scaling, network architecture, and hyperparameters, accompanied by an open-source implementation and pretrained student models. Our findings establish a solid foundation for deploying fast, high-fidelity, and resource-efficient diffusion generators in real-world T2I applications. Code is available on github.com/alibaba-damo-academy/T2I-Distill.
format Preprint
id arxiv_https___arxiv_org_abs_2512_13006
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Few-Step Distillation for Text-to-Image Generation: A Practical Guide
Pu, Yifan
Han, Yizeng
Tang, Zhiwei
Tang, Jiasheng
Wang, Fan
Zhuang, Bohan
Huang, Gao
Computer Vision and Pattern Recognition
Diffusion distillation has dramatically accelerated class-conditional image synthesis, but its applicability to open-ended text-to-image (T2I) generation is still unclear. We present the first systematic study that adapts and compares state-of-the-art distillation techniques on a strong T2I teacher model, FLUX.1-lite. By casting existing methods into a unified framework, we identify the key obstacles that arise when moving from discrete class labels to free-form language prompts. Beyond a thorough methodological analysis, we offer practical guidelines on input scaling, network architecture, and hyperparameters, accompanied by an open-source implementation and pretrained student models. Our findings establish a solid foundation for deploying fast, high-fidelity, and resource-efficient diffusion generators in real-world T2I applications. Code is available on github.com/alibaba-damo-academy/T2I-Distill.
title Few-Step Distillation for Text-to-Image Generation: A Practical Guide
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.13006