T2DyVec: Leveraging Sparse Images and Controllable Text for Dynamic SVG

In this work, we introduce a controlled dynamic vector graphic generation method. While existing work mostly focuses on text-based generation of single-frame images, dynamic images, or single-frame vectors, there is a lack of research on generating dynamic vectors with complex elements and diverse styles. This is due to the unique challenges posed by dynamic vectors, which require coherent and seamless transitions of vector parameters between frames. To address these challenges, we propose T2DyVec, which leverages text prompts and sparse images as a control for vector generation. It incorporates Vector Consistency, Semantic Tracking, and VPSD to optimize the diffusion model for vector parameters, enabling the generation of multi-frame dynamic coherent vectors. This approach can significantly optimize creative workers' workflow, facilitating generation and further editing.