Project Summary: Developed a GAN-based pipeline system for generating comic line art with controllable thickness, utilizing Real-ESRGAN and APISR training to produce specified line thickness outputs.
This project creates a comprehensive pipeline for generating comic line art with precise thickness control based on vector comic line drawings. The system generates multiple line thickness variations from vector-based comic line art, creating a diverse dataset through systematic thickness manipulation and data augmentation techniques.
The training pipeline utilizes Real-ESRGAN and APISR models to learn the relationship between different line thicknesses, enabling the system to output comic line art with user-specified thickness parameters. This approach provides comic creators with precise control over line art characteristics while maintaining artistic style consistency across different thickness settings.