Following the last Midjourney update for zoom and pan, here is my new method for generating more consistent characters. I prefer this method however you have to check you have a good start file as this video shows otherwise you’ll have to try redirect the AI to get what you want. I typically try to have the eyes, eyebrows and a bit of hair in the start image so Midjourney will work from there and not try to add its on ideas.
Also, be in Remix mode and go with the less variation option. Hopefully it’s useful for those also trying to get more consistent characters in Midjourney.
Song in the beginning is generated by Beatbot AI.