Problem
Research questions and friction points this paper is trying to address.
neural networks
human designers
Scalable Vector Graphics
collaboration
vector graphics generation
DesigNet: Learning to Draw Vector Graphics as Designers Do
🤖 AI Summary
Existing neural networks struggle to collaborate with human designers in vector graphics creation due to their inability to model essential design constraints such as geometric continuity and axis alignment. This work proposes DesigNet, a hierarchical Transformer-VAE architecture that directly processes native SVG sequences and incorporates differentiable self-refinement modules for continuity and alignment. The model supports prediction of C⁰, G¹, and C¹ continuity and refines Bézier control points to enable intelligent snapping of horizontal and vertical lines. DesigNet generates highly accurate, editable vector outlines that significantly outperform existing methods in both adherence to geometric constraints and compatibility with standard design workflows.
📝 Abstract
AI-driven content generation has made remarkable progress in recent years. However, neural networks and human designers operate in fundamentally different ways, making collaboration between them challenging. We address this gap for Scalable Vector Graphics (SVG) by equipping neural networks with tools commonly used by designers, such as axis alignment and explicit continuity control at command junctions. We introduce DesigNet, a hierarchical Transformer-VAE that operates directly on SVG sequences with a continuous command parameterization. Our main contributions are two differentiable modules: a continuity self-refinement module that predicts $C^0$, $G^1$, and $C^1$ continuity for each curve point and enforces it by modifying Bézier control points, and an alignment self-refinement module with snapping capabilities for horizontal or vertical lines. DesigNet produces editable outlines and achieves competitive results against state-of-the-art methods, with notably higher accuracy in continuity and alignment. These properties ensure the outputs are easier to refine and integrate into professional design workflows. Source Code: https://github.com/TomasGuija/DesigNet.
Innovation
Methods, ideas, or system contributions that make the work stand out.
DesigNet
vector graphics generation
continuity control
alignment snapping
differentiable refinement
🔎 Similar Papers
No similar papers found.
