As a practical application, image editing must meet diverse user demands while ensuring ease of use. In this paper, we introduce MagicQuill, an integrated image editing system that enables users to bring their creative ideas to life quickly. Our system features a streamlined yet powerful interface, allowing users to perform tasks such as inserting elements, erasing objects, and adjusting colors with minimal effort. A multimodal large language model (MLLM) continuously monitors interactions to anticipate user intent in real time, eliminating the need for manual prompts. To achieve precise control, we leverage a diffusion-based approach enhanced by a carefully designed two-branch plug-in module.
