Enregistré dans:
Détails bibliographiques
Auteurs principaux: Cheng, Shihan, Kulkarni, Nilesh, Hyde, David, Smirnov, Dmitriy
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2511.17844
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866908946769379328
author Cheng, Shihan
Kulkarni, Nilesh
Hyde, David
Smirnov, Dmitriy
author_facet Cheng, Shihan
Kulkarni, Nilesh
Hyde, David
Smirnov, Dmitriy
contents Fine-tuning large-scale text-to-video diffusion models to add new generative controls, such as those over physical camera parameters (e.g., shutter speed or aperture), typically requires vast, high-fidelity datasets that are difficult to acquire. In this work, we propose a data-efficient fine-tuning strategy that learns these controls from sparse, low-quality synthetic data. We show that not only does fine-tuning on such simple data enable the desired controls, it actually yields superior results to models fine-tuned on photorealistic "real" data. Beyond demonstrating these results, we provide a framework that justifies this phenomenon both intuitively and quantitatively.
format Preprint
id arxiv_https___arxiv_org_abs_2511_17844
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Less is More: Data-Efficient Adaptation for Controllable Text-to-Video Generation
Cheng, Shihan
Kulkarni, Nilesh
Hyde, David
Smirnov, Dmitriy
Computer Vision and Pattern Recognition
Artificial Intelligence
68U05
I.3.3; I.5.4
Fine-tuning large-scale text-to-video diffusion models to add new generative controls, such as those over physical camera parameters (e.g., shutter speed or aperture), typically requires vast, high-fidelity datasets that are difficult to acquire. In this work, we propose a data-efficient fine-tuning strategy that learns these controls from sparse, low-quality synthetic data. We show that not only does fine-tuning on such simple data enable the desired controls, it actually yields superior results to models fine-tuned on photorealistic "real" data. Beyond demonstrating these results, we provide a framework that justifies this phenomenon both intuitively and quantitatively.
title Less is More: Data-Efficient Adaptation for Controllable Text-to-Video Generation
topic Computer Vision and Pattern Recognition
Artificial Intelligence
68U05
I.3.3; I.5.4
url https://arxiv.org/abs/2511.17844