DreamActor-M1 is a human animation framework based on Diffusion Transformer (DiT), designed to achieve fine-grained overall controllability, multi-scale adaptability, and long-term temporal consistency. The model, through blending guidance, can generate high-expressiveness and realistic human videos suitable for various scenarios from portrait to full-body animation. Its main advantages lie in its high fidelity and identity preservation, bringing new possibilities to human behavior animation.