Pika, in collaboration with Peking University and Stanford, has open-sourced the RPG framework, which leverages LLM technology to enhance the diffusion model's understanding of complex prompts, surpassing the capabilities of Dall·E 3. This framework can generate images that more accurately meet the requirements of prompts and has sparked heated discussions online. The contributors include authors from Peking University, Stanford, and the co-founders of Pika. For detailed content, please visit the original article link.