Midjourney’s new feature for creating coherent characters
Unlock the power of narrative continuity with Midjourney's groundbreaking feature. Dive into the future of AI-generated characters that maintain their identity across your creative projects. Read on to see how
Midjourney has recently unveiled a feature that addresses a significant challenge in the realm of AI-driven image creation: the ability to generate a character with consistent appearance across multiple images. This development is particularly noteworthy as it overcomes a common limitation inherent in the technology used by most AI image generators.
These generators, including those based on the diffusion model like Stability AI’s Stable Diffusion, construct images pixel by pixel from textual descriptions provided by users. They draw upon a vast and sometimes contentious dataset of human-created images and associated text tags to learn how to create new visuals. However, ensuring that characters remain consistent in different scenes has been a complex task until now.
The quest for narrative continuity in AI-generated visuals
Generative AI, much like its text-based counterparts such as OpenAI’s ChatGPT or Cohere’s Command-R, has historically struggled with producing consistent outputs. This is particularly problematic when visual consistency is required, such as in storyboards for films, novels, graphic novels, or comic books, where characters must maintain their identity across various scenes and expressions.
Midjourney’s innovative solution to this problem is the introduction of a new tag, “–cref” (character reference), which allows users to guide the AI to maintain character traits such as facial features, body type, and attire. By appending this tag to text prompts within the Midjourney Discord, along with a URL for reference, users can achieve a level of consistency previously unattainable in generative AI imagery.
As this feature evolves and becomes more sophisticated, Midjourney is poised to transition from a novel concept to a valuable asset for professionals in creative fields. It’s important to note that while Midjourney’s Version 6 is currently in alpha, features may undergo changes. However, the official beta release is on the horizon, and the team at AI Media Cafe is eager to hear user feedback on the ideas-and-features channel. The early release of this feature is intended to enhance the user experience in crafting narratives and exploring new worlds.
