Abstract
Creating hand-drawn animation sequences is labor-intensive and demandsprofessional expertise. We introduce PhysAnimator, a novel approach forgenerating physically plausible meanwhile anime-stylized animation from staticanime illustrations. Our method seamlessly integrates physics-based simulationswith data-driven generative models to produce dynamic and visually compellinganimations. To capture the fluidity and exaggeration characteristic of anime,we perform image-space deformable body simulations on extracted meshgeometries. We enhance artistic control by introducing customizable energystrokes and incorporating rigging point support, enabling the creation oftailored animation effects such as wind interactions. Finally, we extract andwarp sketches from the simulation sequence, generating a texture-agnosticrepresentation, and employ a sketch-guided video diffusion model to synthesizehigh-quality animation frames. The resulting animations exhibit temporalconsistency and visual plausibility, demonstrating the effectiveness of ourmethod in creating dynamic anime-style animations. See our project page formore demos: https://xpandora.github.io/PhysAnimator/