Few-Shot-Based Modular Image-to-Video Adapter for Diffusion Models
This addresses the challenge of efficient motion control in image-to-video generation for users, though it is incremental as it builds on existing diffusion models.
The paper tackles the problem of image animation with diffusion models, which struggle with motion generalization and prompt compliance due to limited training data, by proposing MIVA, a lightweight modular adapter that enables precise motion control with just ~10 training samples while matching or exceeding the quality of models trained on larger datasets.
Diffusion models (DMs) have recently achieved impressive photorealism in image and video generation. However, their application to image animation remains limited, even when trained on large-scale datasets. Two primary challenges contribute to this: the high dimensionality of video signals leads to a scarcity of training data, causing DMs to favor memorization over prompt compliance when generating motion; moreover, DMs struggle to generalize to novel motion patterns not present in the training set, and fine-tuning them to learn such patterns, especially using limited training data, is still under-explored. To address these limitations, we propose Modular Image-to-Video Adapter (MIVA), a lightweight sub-network attachable to a pre-trained DM, each designed to capture a single motion pattern and scalable via parallelization. MIVAs can be efficiently trained on approximately ten samples using a single consumer-grade GPU. At inference time, users can specify motion by selecting one or multiple MIVAs, eliminating the need for prompt engineering. Extensive experiments demonstrate that MIVA enables more precise motion control while maintaining, or even surpassing, the generation quality of models trained on significantly larger datasets.