Search papers, labs, and topics across Lattice.
The paper introduces CASHG, a context-aware online handwriting generator that explicitly models inter-character connectivity for improved sentence-level trajectory synthesis. CASHG employs a Character Context Encoder and a bigram-aware sliding-window Transformer decoder to capture character identity, sentence context, and local transitions. A three-stage curriculum training strategy and new Connectivity and Spacing Metrics (CSM) are used to improve robustness and evaluation, demonstrating superior performance in connectivity and spacing compared to existing methods.
Achieve more natural-looking, sentence-level handwriting by explicitly modeling the connections *between* characters, rather than treating them as an afterthought.
Online handwriting represents strokes as time-ordered trajectories, which makes handwritten content easier to transform and reuse in a wide range of applications. However, generating natural sentence-level online handwriting that faithfully reflects a writer's style remains challenging, since sentence synthesis demands context-dependent characters with stroke continuity and spacing. Prior methods treat these boundary properties as implicit outcomes of sequence modeling, which becomes unreliable at the sentence scale and under limited compositional diversity. We propose CASHG, a context-aware stylized online handwriting generator that explicitly models inter-character connectivity for style-consistent sentence-level trajectory synthesis. CASHG uses a Character Context Encoder to obtain character identity and sentence-dependent context memory and fuses them in a bigram-aware sliding-window Transformer decoder that emphasizes local predecessor--current transitions, complemented by gated context fusion for sentence-level context.Training proceeds through a three-stage curriculum from isolated glyphs to full sentences, improving robustness under sparse transition coverage. We further introduce Connectivity and Spacing Metrics (CSM), a boundary-aware evaluation suite that quantifies cursive connectivity and spacing similarity. Under benchmark-matched evaluation protocols, CASHG consistently improves CSM over comparison methods while remaining competitive in DTW-based trajectory similarity, with gains corroborated by a human evaluation.
