DuetSVG: Unified Multimodal SVG Generation with Internal Visual Guidance

TL;DR

DuetSVG is a novel multimodal model that jointly generates images and SVGs, using visual guidance during decoding to produce more accurate, visually appealing, and semantically aligned SVG graphics.

Contribution

It introduces a unified end-to-end model for multimodal SVG generation with a novel test-time scaling strategy leveraging visual predictions.

Findings

Outperforms existing methods in SVG quality

Produces visually faithful and semantically aligned SVGs

Effective across diverse applications

Abstract

Recent vision-language model (VLM)-based approaches have achieved impressive results on SVG generation. However, because they generate only text and lack visual signals during decoding, they often struggle with complex semantics and fail to produce visually appealing or geometrically coherent SVGs. We introduce DuetSVG, a unified multimodal model that jointly generates image tokens and corresponding SVG tokens in an end-to-end manner. DuetSVG is trained on both image and SVG datasets. At inference, we apply a novel test-time scaling strategy that leverages the model's native visual predictions as guidance to improve SVG decoding quality. Extensive experiments show that our method outperforms existing methods, producing visually faithful, semantically aligned, and syntactically clean SVGs across a wide range of applications.