LLM-Empowered Image Generation in the Neko Painter App: A Preliminary Application for Producing Teaching Materials
Kaiyi Wu, Jiaoyang Ding, Jingsen Li, Yuke Yang, Chen Zhang, Jiaxin Cao · 2024
This paper introduces the Neko Painter app and its key features, demonstrates the Large Language Models (LLMs)-empowered image generation with diffusion models and ContorlNet to be smarter and more automatic to control and optimise the image generation process, and shares some cases of using it to produce teaching materials. A preliminary application for producing teaching materials on General Studies using the Neko Painter app was conducted with 36 pre-service teachers from Hong Kong. The results showed that using LLM-empowered features positively impacts pre-service teachers’ motivation in producing teaching materials by using image generation. Future work will further explore the potential of LLM-empowered image generation in more educational subjects and scenarios.