Midjourney’s ability to generate stunning visuals of people is well-established. However, ensuring consistent character appearances across multiple images presented a challenge. This hurdle has been addressed with the introduction of the --cref tag, allowing artists to maintain visual continuity for their characters.
By including --cref followed by a URL referencing a character image at the end of your prompt, subsequent prompts referencing that tag will generate images featuring the same character. This includes replicating facial features, body type, and even clothing from the reference image, notes NIX Solutions.
For optimal results, the reference image is best generated by Midjourney itself. This ensures compatibility and allows for a seamless workflow. The process involves first creating your desired character or obtaining a URL for an existing one.
The provided example showcases this process. After generating an image of a “muscular bald man with beads and an eye patch,” the --cref tag is used to maintain this baseline. Additional prompts refine the character’s appearance, such as adding a tuxedo (through text description) or removing the eye patch using the --cw tag with a numerical value (higher values signifying stronger emphasis).
Benefits and Future Potential
The developers envision this feature being particularly valuable for projects like graphic novels, comic books, and film storyboards. As Midjourney continues to evolve, this functionality has the potential to elevate it into a professional-grade character design tool.