Telligram: Text-Driven Calligram Generation via Diffusion-Guided Skeleton Optimization

Tianci Shi1, Pengfei Xu*,1
1CSSE, Shenzhen University, China
Computer Graphics Forum / Pacific Graphics 2026
*Corresponding Author
PDFPaper PDFSupplementary
Overview of Telligram calligram generation results

Overview of Telligram. Telligram generates semantic word-level calligrams directly from text prompts while preserving letter readability, without requiring a manually provided contour or input image.

Abstract

Compact calligram generation aims to form a semantic shape while keeping letters recognizable. Most existing methods are shape-conditioned and mainly solve downstream letter layout inside a given contour. We study text-only calligram generation without an input contour. This setting is difficult because semantic shape formation and letter readability strongly interfere with each other when optimized in a single stage. Pushing the word toward a clear figure can easily damage glyph structure, while preserving readable letters can weaken the target shape. To address this difficulty, we present Telligram, a training-free, low-tuning, two-stage framework composed of Semantic Occupancy Prior Formation and Readability-Constrained Glyph Realization. The first stage uses Variational Score Distillation (VSD) with structured skeleton optimization and hierarchical gradient projection to produce a semantic occupancy prior. The second stage converts this occupancy prior into per-letter regions and reconstructs readable glyph layouts through lightweight geometric processing. The framework generates coherent and creative word-level semantic calligrams directly from text prompts.

Method

Telligram method pipeline
Two-stage pipeline of Telligram. The first stage forms a semantic occupancy prior through diffusion-guided skeleton optimization, and the second stage converts this prior into readable per-letter regions with lightweight geometric processing.

Results

Qualitative Telligram results
Qualitative results. From only a target word and a semantic prompt, Telligram produces compact calligrams that balance the intended object shape with the recognizability of individual letters.
Diverse Telligram outputs
Output diversity. The framework supports diverse semantic concepts and word layouts while retaining a coherent whole-word structure.

Comparison

Comparison with prior calligram generation methods
Comparison with prior methods. Compared with image- or contour-conditioned calligram generation pipelines, Telligram directly infers semantic structure from text prompts and then realizes readable glyph layouts under the generated occupancy prior.
Comparison under matched semantic information
Comparison under matched semantic information. Telligram focuses on converting semantic structure into a writable whole-word calligram while maintaining recognizable letter identities.
Silhouette prior compared with skeleton prior
Silhouette prior and skeleton prior. Skeleton-based optimization provides a compact deformation space for semantic shape formation before the final glyph realization stage.

Limitations

Telligram limitations and failure cases
Representative failure cases. The current training-free pipeline can still produce ambiguous silhouettes, lose letter identity after severe intermediate deformation, or leave branch-like support in the final glyphs. Stronger identity-preserving guidance may further improve stability.

BibTeX

@article{shi2026telligram,
      title={Telligram: Text-Driven Calligram Generation via Diffusion-Guided Skeleton Optimization},
      author={Shi, Tianci and Xu, Pengfei},
      journal={Computer Graphics Forum},
      year={2026},
      doi={10.1111/cgf.70647},
      url={https://doi.org/10.1111/cgf.70647},
}