NIV: Neural Axis Variations for Variable Font Generation

2026-06-03

AI Summary

Existing variable font generation relies heavily on manual expert adjustments, resulting in low efficiency. This work proposes NIV, the first end-to-end method capable of automatically generating standard OpenType variable fonts that support continuous interpolation across multiple design axes. Built upon vector glyph geometry, NIV employs a neural network to predict per-point displacements and introduces an attribute embedding mechanism to jointly model the intricate couplings among design axes such as weight and width. Trained on a million-scale variable font dataset, the model is fully compatible with existing rendering engines and demonstrates strong generalization to unseen characters, typefaces, and even handwritten inputs, achieving particularly robust performance on complex CJK glyphs.

Abstract

Variable fonts enable continuous variation of glyph geometry along semantic design axes such as weight, width, slant, and optical size. However, constructing a variable font from a static font remains a labor-intensive process requiring expert typographic design and manual specification of glyph variation data. We introduce NIV (Neural Axis Variations), a method that automatically converts a static font into a fully functional variable font. Given glyph outlines and a set of desired design axes, NIV predicts per-point displacements. The model operates directly on vector glyph geometry and employs a novel Property Embedding mechanism that captures interactions between multiple axes, enabling consistent multi-axis variation within a unified framework. We train NIV on a newly constructed dataset derived from variable Google Fonts, comprising over one million variation tuples. The resulting model generalizes across unseen code points, unseen font styles, high-complexity CJK glyphs, and even out-of-distribution handwriting inputs. The generated outputs are standard variable font files supporting continuous interpolation via existing rendering engines. To facilitate research, we release the dataset, the complete training and inference implementation, and trained models at https://github.com/ndvbd/NIV. Beyond typography, our approach demonstrates how structured geometric objects with continuous parametric variation can be synthesized using neural deformations.

Problem

• variable font
• glyph variation
• font generation
• design axes
• static-to-variable conversion

Innovation

• Neural Axis Variations
• Variable Font Generation
• Property Embedding
• Vector Glyph Deformation
• Multi-axis Interpolation