A Dual-Stream Diffusion Transformer for High-Fidelity Multimodal Face Generation - View it on GitHub