Creating Video Images with Artificial Intelligence (AI)

Tom, I've been playing with Midjourney quite a bit these past few months. I really love it, but it may not be the best tool (at least, as of this writing) for generating consistent characters. It seems like Stable Diffusion, which lets you train your own model with images of your character, may provide a stronger basis for generating consistent characters across multiple images, but I haven't tried Stable Diffusion yet.

I have used Midjourney in a couple of more limited ways, including:

1. To modify stock images. Feed Midjourney a stock photo of, say, someone who is NOT wearing a reflective safety vest, and add in the text prompt "Wearing a reflective safety vest" and you can usually get a picture that takes the general pose and other clothing from the stock photo but adds the safety vest. Very handy for adding things like safety glasses, hard hats, leather gloves, etc. to stock photos that don't have these things already.
2. To generate technical equipment for training scenarios. Need a photo of a hotplate-stirrer to place on a research-lab benchtop? Ask Midjourney to generate one for you. You sometimes have to do some cleanup in Photoshop, but it's still a nice way to get items like beakers, chemical bottles, voltmeters, and other items that you can use in Storyline to let learners interact with them.

But I haven't tried to use Midjourney to wholesale create characters unless I only need them in a single image. If/when you figure out a consistent way to do this in Midjourney, I hope you will post about it!