MusePose is an image-to-video generation framework developed by Tencent Music Entertainment's Lyra Lab, designed to generate virtual character videos through pose control signals. It is the last building block of the Muse open-source series, alongside MuseV and MuseTalk, aiming to push the community towards the vision of generating virtual characters with full-body motion and interaction capabilities. Based on diffusion models and pose guidance, MusePose can generate dance videos of characters from reference images, and the results surpass the quality of almost all open-source models on the same topic.