OmniSVG: A model for generating scalable vector graphics (SVGs) using Vision-Language Models. It improves efficiency and quality over existing methods by tokenizing SVG commands and coordinates. The model is versatile, capable of producing various complexities from simple icons to intricate characters through multiple modalities (Text-to-SVG, Image-to-SVG). It includes the MMSVG-2M dataset, containing 2 million annotated SVGs, and outperforms other techniques, offering potential for professional use.
OmniSVG: a Unified Scalable Vector Graphics Generation Model
