Given the challenges of simultaneously achieving style preservation, continuous rendering, and interactive feedback in the dynamic expression of traditional ink-wash styles, this study investigates an AI-driven interactive dynamic illustration generation system. It introduces a data construction and preprocessing pipeline designed for multi-source inputs, elaborates on the system’s overall architecture, user interaction controls, dynamic illustration generation, temporal rendering, and methods for the coordinated optimization of style consistency and generation efficiency, thereby establishing a complete operational closed-loop.Test results using landscape, flower-and-bird, and figure samples indicate that the constructed system achieves an FID of 18.74, a style consistency rate of 93.6%, a dynamic continuity score of 0.931, a temporal consistency rate of 94.8%, an average response latency of 68 ms, a control accuracy rate of 96.2%, and a peak VRAM usage of 19.5 GB.These results demonstrate that the system achieves a well-balanced technical implementation path between the expression of traditional ink-wash visual features, the stable output of consecutive frames , and real-time interactive control, making it valuable for the generation of digital ink-wash illustrations.
