A Comprehensive Ecosystem for Open-Domain Customized Video Generation
This work provides a large-scale dataset and efficient model for open-domain customized video generation, addressing the lack of diverse identity-specific data and enabling broader real-world applications.
The authors introduce PexelsCustom-1M, the first million-scale dataset for identity-preserving video generation, and propose CustoMDiT, a parameter-efficient framework that achieves state-of-the-art performance with only 8% additional parameters. They also construct OpenCustom, a benchmark with over 1,000 categories, to address the limitations of existing benchmarks.
Recent progress in video generation has shown impressive visual synthesis capabilities. However, open-domain customized video generation remains limited by the lack of large-scale, annotated datasets capturing diverse identity-specific attributes. To address this, we introduce PexelsCustom-1M, the first publicly available million-scale dataset for identity-preserving video generation, containing one million curated <identity, text, video> triplets across 8,000+ categories. Leveraging this, we propose CustoMDiT, a parameter-efficient framework that adapts a pretrained multimodal Diffusion Transformer into a customized video generator with only 8% additional learnable parameters. Our method surpasses prior state-of-the-art. However, benchmarks such as DreamBooth cover only 100 classes, which is insufficient for real-world applications. To overcome this, we construct OpenCustom, a new benchmark with 1,000+ categories, created via cross-dataset knowledge fusion from ImageNet and MS-COCO. Extensive experiments confirm the advantages of both our dataset and model. We will open-source the entire ecosystem--including dataset, pipeline, benchmark, and implementations--to support further research.