TexSketch: Bringing Texture-Aware Colorization to Sketches

2026-07-27

AI Summary

Existing reference-based sketch colorization methods suffer from limited generalization due to high annotation costs, restricted dataset scale, and fixed artistic styles. This work proposes TexSketch—a controllable procedural framework that, for the first time, leverages geometric analysis and shader-driven stylization to automatically generate a texture-aware colored sketch dataset. By integrating region extraction, semantic color prediction, and programmable shader rendering, TexSketch establishes a fully automated synthesis pipeline capable of producing highly diverse and scalable supervision data without manual annotation. Human evaluations demonstrate that the generated results exhibit strong visual plausibility and rich stylistic variation, offering high-quality synthetic training data for sketch colorization tasks.

Abstract

Reference-based sketch colorization methods rely on large paired datasets that preserve both the structural and stylistic characteristics of hand-drawn artwork. However, existing datasets are limited in scale, expensive to annotate, and bound to fixed, often inconsistent artistic style biases that propagate to downstream models and limit cross-domain generalization. We present TexSketch, a controllable procedural framework for generating colored-sketch datasets with programmable artistic styles via geometric analysis and shader-driven stylization. Our fully automatic pipeline integrates region extraction, semantic color prediction, and shader-based rendering. By defining artistic appearance procedurally rather than inheriting it from a static corpus, TexSketch enables scalable dataset generation without manual annotation or artist supervision. Human studies demonstrate that TexSketch generates perceptually plausible colored sketches with high stylistic diversity, providing a controllable, scalable source of synthetic supervision for sketch colorization.

Problem

• sketch colorization
• dataset limitation
• artistic style bias
• cross-domain generalization
• manual annotation

Innovation

• procedural generation
• sketch colorization
• shader-based stylization
• synthetic dataset
• texture-aware rendering