Research on art education assisted painting training and image enhancement method based on generative AI technology

This study proposes an art education method that integrates generative AI technology to assist drawing training and enhance images, aiming to address the “learning disconnect” and “feedback deficiency” issues present in traditional AI drawing tools for educational applications. By constructing a “dynamic layered generation” architecture, the painting process is decomposed into three interactive stages: sketching, line art, and coloring. This mimics the human learning path of “observation-imitation-innovation,” enabling a collaborative training mechanism where students receive step-by-step intervention and real-time AI assistance. Additionally, a dual-dimensional “skill-style” evaluation matrix quantifies technical proficiency and creative expression, delivering interpretable personalized feedback. The experiment recruited 60 beginners to compare three training modes: our system, Stable Diffusion's one-shot generation, and traditional video instruction. Results show the system achieved a 26.2% improvement in technique scores for geometric sketching tasks, significantly outperforming the baseline (p < 0.01). In color composition tasks, it secured the highest creative scores and expert blind evaluations, with student feedback rating “feedback helpfulness” at 4.6/5. This research demonstrates the method's effectiveness in enhancing learning efficiency and artistic creativity, providing an expandable technical framework and empirical evidence for integrating generative AI into art education.