Until now, image generators like Midjourney have not been very helpful if you wanted to use a specific, fictitious person in several images. However, such consistency is essential for many applications. Think of comics or marketing materials.
With Stable Diffusion, this can be solved with LoRAs. However, the technical effort is not trivial. Another option is to use celebrities from the AI training material. But this is not always useful or legally sound.
Midjourney is now experimenting with a feature that should make this possible with little effort: You reference in your prompt another image that already shows the person you want. You can then use it in your new work, adjusting the background, image composition, facial expression, posture, and more.
An article on VentureBeat explains it and shows examples.