CustomCrafter: Customized Video Generation with Preserving Motion and Concept Composition Abilities

2024-08-23

arXiv.org

AI Summary

Current video generation models trained on a single subject image struggle to simultaneously achieve motion diversity and compositional generalization of subject concepts, while requiring frequent retraining with new video data—entailing high interactive overhead. To address this, we propose a fine-tuning-free, text-image jointly driven personalized video generation framework. Our method introduces a plug-and-play lightweight subject learning module, coupled with a dynamic weighted video sampling strategy that preserves motion priors in early denoising steps and emphasizes appearance reconstruction in later steps. We further incorporate parameter-efficient Video Diffusion Model (VDM) updates, phased weight scheduling, and cross-modal concept alignment. Extensive experiments demonstrate significant improvements over state-of-the-art methods across multiple benchmarks, achieving superior motion naturalness, subject fidelity, and flexibility in compositional concept generation. The source code is publicly available.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Machine Learning: Multimodal Learning

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Customized video generation aims to generate high-quality videos guided by text prompts and subject's reference images. However, since it is only trained on static images, the fine-tuning process of subject learning disrupts abilities of video diffusion models (VDMs) to combine concepts and generate motions. To restore these abilities, some methods use additional video similar to the prompt to fine-tune or guide the model. This requires frequent changes of guiding videos and even re-tuning of the model when generating different motions, which is very inconvenient for users. In this paper, we propose CustomCrafter, a novel framework that preserves the model's motion generation and conceptual combination abilities without additional video and fine-tuning to recovery. For preserving conceptual combination ability, we design a plug-and-play module to update few parameters in VDMs, enhancing the model's ability to capture the appearance details and the ability of concept combinations for new subjects. For motion generation, we observed that VDMs tend to restore the motion of video in the early stage of denoising, while focusing on the recovery of subject details in the later stage. Therefore, we propose Dynamic Weighted Video Sampling Strategy. Using the pluggability of our subject learning modules, we reduce the impact of this module on motion generation in the early stage of denoising, preserving the ability to generate motion of VDMs. In the later stage of denoising, we restore this module to repair the appearance details of the specified subject, thereby ensuring the fidelity of the subject's appearance. Experimental results show that our method has a significant improvement compared to previous methods. Code is available at https://github.com/WuTao-CS/CustomCrafter

Problem

• Video Generation
• Action Diversity
• Training Efficiency

Innovation

• CustomCrafter
• Plug-and-Play Module
• Dynamic Weighted Video Sampling