Developers are facing challenges when using AI (like Gemini 3 Pro) to generate scenic tour route maps. Despite providing detailed data structure and prompt requirements (including size limitations, node styles, line curvature, etc.), the AI output remains highly random, leading to messy layouts and overlapping elements. The author shares hands-on experience, including multiple attempts to optimize prompts and generation results, but with poor outcomes. This case reveals the limitations of current AIGC technology in handling customized graphic design, raising questions about AI controllability and exploring whether manual drawing is the only alternative. The article provides valuable insights for developers focused on AI graphic generation, reflecting on the challenges and considerations of AIGC in complex scenarios.
