RESEARCH ON A "CRAFT-AWARE" AI FRAMEWORK FOR LACQUER PAINTING BASED ON ARTIST-SPECIFIC MODELS AND STANDARDIZED COLLABORATIVE WORKFLOWS

Abstract

This paper examines how generative AI can participate in lacquer painting practice while respecting the internal logic of the craft. Lacquer painting is a highly material- and process-dependent art form in which aesthetic value emerges from layered procedures, embodied tacit knowledge, and sensitive material judgment. However, existing general-purpose AI models often fail to capture the deeper “lacquer quality” of the medium or to respond to craft-based decision-making in a meaningful way. To address this mismatch, the paper proposes a “Craft-Aware AI” framework composed of three core elements: an artist-specific LoRA model, a multimodal art knowledge base, and a five-stage human–machine workflow consisting of Conceptualization, Generation, Iteration, Pre-materialization Rehearsal, and Physical Realization.Tacit knowledge such as “three grindings and nine coatings” and lacquer drying rhythms is encoded via semantic tagging and trigger-word structuring, mapping material depth, gloss levels, and textural features into the model. This framework positions AI as a supportive collaborator while preserving artistic agency. A pilot study comparing general-purpose AI outputs and Craft-Aware outputs, supplemented with artist feedback, demonstrates practical effectiveness. The study provides a replicable methodological structure and a transferable pathway for other craft-oriented art forms undergoing digital transformation.

Keywords: Lacquer Painting, AIGC, LoRA, Human–machine Collaboration, Craft-Aware AI