Experimenting with character consistency and permutations - Midjourney Tutorial

- [Instructor] In this video, we're going to do some experimentation with character consistency. So if you're following along with the Exercise Files, I have a reference for you. You simply want to click on the plus icon to upload a file to Midjourney, and inside that, you'll see a Consistent Characters V6 folder. In that folder, we have a couple references of the character that we'll be using. Let's pick the wider shots. We'll open that up and just press Return in order for us to have access to this and the link. Now before we copy the link, we're going to create an /imagine prompt with several permutations to try this character in different emotional poses at different scene backgrounds and from different shot perspectives. So just a warning, depending on if you're limited in your credits, you might want to dial back down on the permutations that you use here. But in this case, we're going to start with that /imagine prompt, and we're going to enter the following. And in squiggly brackets, establishing, medium, closeup, we'll follow with curly brackets again, shot of a bearded man, and then we'll do squiggly brackets again for another set of permutations, crying, laughing, yelling. Let's close those squiggly brackets, and we will now say on a, squiggly brackets, beach, mountain, so two different locations, and we'll type in wearing a, squiggly brackets, summer shirt, and followed by swim trunks, squiggly bracket again. So we have quite a few different things that we are running here. And now for the important part, let's do ,--cref for our character reference, right click that photo that we just uploaded, copy the link and paste it. And we're going to follow this with a character weight. And the reason for the character weight is that we want to have a bit more creativity with our prompt, with a lower character weight value. So that's --cw, this case, I wanted to set it to 10. And last but not least, our aspect ratio. I'm going to just stick with my desired 16.9. So again, this is quite a few permutations. You can cut back if you'd like. I'm going to press Return. It's going to warn us how many permutations we are creating, 36 prompts, and I believe the limit is 40. So let's see all the variations that we can come up with with our consistent character also using the dialed down character weight. You'll see it starts to populate with the image reference, but this is going to switch in a second to the grid of four variations that Midjourney is going to create from all of these prompts. Okay, so it's now completed the majority of the 36 jobs. Just wanted to go over a few of these. So here's the establishing shot of the bearded man laughing. You can see those four variations, and it just goes on. So we see the alternative here, wearing swim trunks. In many cases it didn't give as wide a shot as I'd like, but we can see that he is not wearing a shirt. Then we have him on a beach wearing his summer shirt, transfixed with different types of swimsuits in this case. And then on a mountain. And we have some jobs still processing here wearing the summer shirt, and how that's being continued there with his different expressions of the man yelling and crying. How great this could be for ideation, giving us all of these different perspectives as well as different facial reactions in multiple locations using these permutations. Another great way to experiment besides this, don't forget the repeat value where you could just have the same job repeat over and over again to give you multiple variations with a low character weight. Now last but not least, I'm going to take one of these jobs, a closeup of the bearded man yelling on a mountain and wearing his swim trunks. So I'm going to take this job and I'm going to remix it. And again, you could set up remix by enabling this in settings, which will allow you to re-roll the job, but then also tag some stuff onto it. And in this case, what we're going to do is add a style reference. So for this, I'm going to head to midjourney.com, and right now I'm looking just at random images that were generated and are shared by users publicly. I kind of like this animation style I'm seeing here, where I'll select that photo like we used before. And keep in mind, shortly in this course, I'll be giving more information about how you can use Midjourney on the web as it's being rolled out to more and more users each and every day. Heading here to the side of the options list, I'm going to copy, in this case, the image URL, head back here where we'll re-roll that job I was just chatting about. And at the end here, just append an SREF, so --sref followed by a link. We'll submit that and see what we end up with. Okay, so here's our end results, and as you can see, while trying to keep our character consistent, it did add a splash of overlay of some of the textural styles that we did reference with our style reference. And again, playing with this and character weight and some different scenarios and scenes, we can easily ideate with these two values. So there's a little bit about how we can ideate specifically with permutations and using our character reference, character weight, as well as style reference.