Vector Graphic

Vector graphics (VGs), using mathematical descriptions of shapes rather than pixels, offer scalable and editable image representations, driving research into efficient generation and manipulation methods. Current research focuses on developing models that convert raster images to VGs (often using superpixels or segmentation), generate VGs from text prompts (leveraging LLMs and diffusion models), and enable direct manipulation of VGs using LLMs. These advancements are significant for accessibility (creating tactile graphics), design (automating stylistic choices and generating complex scenes), and scientific visualization (producing publication-ready figures), improving efficiency and expanding creative possibilities.