Tune-A-Video: One-Shot Tuning of Image Diffusion Models for Text-to-Video Generation

TLDR

Text‑to‑video generation typically relies on large video datasets, making the approach computationally expensive despite promising results. This study introduces a one‑shot video tuning framework that requires only a single text‑video pair. The method adapts pre‑trained text‑to‑image diffusion models with a spatio‑temporal attention module and efficient one‑shot tuning, using DDIM inversion for inference guidance. Experiments show that the adapted models can generate verb‑representative still images, maintain content consistency across multiple frames, and achieve strong performance across diverse applications.

Abstract

To replicate the success of text-to-image (T2I) generation, recent works employ large-scale video datasets to train a text-to-video (T2V) generator. Despite their promising results, such paradigm is computationally expensive. In this work, we propose a new T2V generation setting—One-Shot Video Tuning, where only one text-video pair is presented. Our model is built on state-of-the-art T2I diffusion models pre-trained on massive image data. We make two key observations: 1) T2I models can generate still images that represent verb terms; 2) extending T2I models to generate multiple images concurrently exhibits surprisingly good content consistency. To further learn continuous motion, we introduce Tune-A-Video, which involves a tailored spatio-temporal attention mechanism and an efficient one-shot tuning strategy. At inference, we employ DDIM inversion to provide structure guidance for sampling. Extensive qualitative and numerical experiments demonstrate the remarkable ability of our method across various applications.