InstantFamily: Masked Attention for Zero-shot Multi-ID Image Generation

arXiv:2404.19427v1 Announce Type: new
Abstract: In the field of personalized image generation, the ability to create images preserving concepts has significantly improved. Creating an image that naturally integrates multiple concepts in a cohesive and visually appealing composition can indeed be challenging. This paper introduces "InstantFamily," an approach that employs a novel masked cross-attention mechanism and a multimodal embedding stack to achieve zero-shot multi-ID image generation. Our method effectively preserves ID as it utilizes global and local features from a pre-trained …

abstract arxiv attention concepts create cs.cv image image generation images indeed multiple novel paper personalized type zero-shot