PhysDreamer is a novel approach that endows static 3D objects with interactive dynamics by leveraging the object dynamics priors learned by video generation models. The key idea is to distill these priors to estimate the physical material properties of static 3D objects, enabling realistic interactions with novel stimuli. The method uses 3D Gaussians to represent objects, a neural field to model the physical material field, and the differentiable Material Point Method (MPM) to simulate 3D dynamics. By optimizing the material field and initial velocity field to match a synthesized video reference, PhysDreamer can generate realistic object responses to external forces or agent manipulations. The effectiveness of PhysDreamer is demonstrated through experiments on diverse elastic objects and a user study, showing superior motion realism compared to state-of-the-art methods. PhysDreamer enhances the realism and interactivity of virtual environments, paving the way for more engaging and lifelike simulations.PhysDreamer is a novel approach that endows static 3D objects with interactive dynamics by leveraging the object dynamics priors learned by video generation models. The key idea is to distill these priors to estimate the physical material properties of static 3D objects, enabling realistic interactions with novel stimuli. The method uses 3D Gaussians to represent objects, a neural field to model the physical material field, and the differentiable Material Point Method (MPM) to simulate 3D dynamics. By optimizing the material field and initial velocity field to match a synthesized video reference, PhysDreamer can generate realistic object responses to external forces or agent manipulations. The effectiveness of PhysDreamer is demonstrated through experiments on diverse elastic objects and a user study, showing superior motion realism compared to state-of-the-art methods. PhysDreamer enhances the realism and interactivity of virtual environments, paving the way for more engaging and lifelike simulations.