Midjourney has finally released the long-awaited tool for creating consistent characters across generations. If you've ever tried to create multiple images with the same character, you know it was a significant challenge.
People who tried to use generative AI for this purpose employed various methods, such as flash and character sheets, seed references, and blending, to achieve a close representation of the same character. However, it required a lot of effort without always delivering satisfactory results… until now? Let’s put it to the test.
Character reference parameter, --cref
Similar to the image reference parameter we previously released, you can now use a URL of an existing image as your character reference image.
The great thing about this reference is that you can now identify the character in one image and then change your new prompt to reference that character in different contexts, and even with different styles. This means that this feature is probably the best solution we have for creating storyboards with a single character across different scenes and situations.
Here's an example of an image created in Midjourney (left) that was used as a character reference. On the right is a new generation image created in Midjourney with an adapted context. This feature transformed this woman from a florist into a restaurant owner:
You may notice a few things:
Keep reading with a 7-day free trial
Subscribe to Designing with AI to keep reading this post and get 7 days of free access to the full post archives.