FaceChain-MMID: Generating highly identity-consistent realistic portraits via dividing & merging multi-modal representations

Open Webpage

Chao Xu, Fei Wang, Cheng Yu, Baigui Sun, Jian Zhao

Published: 01 Jan 2025, Last Modified: 15 Oct 2025Pattern Recognit. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•We utilize facial image, masks, prompts for identity, and design a dividing strategy to express them.•We propose merging strategy including designed training pairs, networks, and loss functions to achieve multi-modal fusion.•Experiments show superior resutls than SOTA.