Meta Launches Smart Body Image Model Muse Image

According to TrendForce Monitoring, Meta has launched the image generation model Muse Image developed by the Super Intelligence Lab. Unlike traditional models that directly map text to pixels, it exhibits behavior patterns similar to a design intelligence.

The model works in conjunction with the recently released Muse Spark language model. Before drawing, the model first creates a composition plan. If the prompt requires factual information, it will search the web to ensure accuracy; if precise typesetting is needed, it can directly render clear, readable text in the artwork and generate a real QR code based on custom prompts.

Users can also use the tagging feature to directly annotate on the artwork, instructing the model to understand context, fine-tune specific details, without the need to regenerate the entire image.

Muse Image is currently live in Meta's AI application, available for free for everyday creation, while more advanced creative needs require unlocking through a paid subscription. It will also gradually introduce commercial systems such as Advantage+ for advertising creativity.