BLK-Assist: A Methodological Framework for Artist-Led Co-Creation with Generative AI Models

Researchers have developed BLK-Assist, a modular framework that enables artists to fine-tune AI diffusion models using their own artwork while maintaining privacy and stylistic control. The system includes three components for concept generation, transparency-preserving assets, and high-resolution outputs, demonstrating a consent-based approach to human-AI collaboration in creative work.

• →BLK-Assist provides a privacy-preserving framework for artists to train AI models on their proprietary artwork.
• →The system consists of three modular components: conceptual sketch generation, transparency-preserving asset creation, and high-resolution upscaling.
• →The framework uses parameter-efficient methods like LoRA adaptation to maintain stylistic fidelity to the original artist's work.
• →The approach emphasizes consent-based human-AI co-creation rather than unauthorized use of artistic content.
• →The methodology is designed to be reproducible and adaptable for other artists with similar privacy requirements.