 | For almost two years now, I've been trying to create a series of 10 images that are going to be used in a narrative video. In the beginning, I couldn't get MidJourney to create the images with a consistent style. Mainly because there was no style ref parameter then. But now that there is... I still can't. It seems to interpret the style ref parameter differently each time. Most of the time it completely ignores it and simply creates a photo-realistic image of what I want. Then character refs... Yes, it works. Somewhat. It's actually now overdoing it. I have a picture of a face which I use as character ref. And ho-boy. Will it do its best to get that face in the picture. Here's the picture I use as character ref: And here's my prompt to create an image: 19th century victorian painting. An enigmatic victorian man walks through the dark, unlit victorian streets at night. He is seen from behind. He is dressed in victorian attire. Yes, I overuse the word "victorian". Because it's important. If I don't mention it everywhere, it will mix victorian stuff with modern stuff. But anyways, the results speak for themselves. Midjourney tries its damnest to get that face in the picture: Also, why does it insist on making the back of his attire look like the front? Long story short, I actually wanted a completely different style than this but I have found that if I just ask for a "victorian painting", at least it's somewhat consistent when it creates the images. With every other style I tried, it will vary wildly. Do I have a question too? Glad you asked! Yes! I realize all this is probably because I suck at prompting. So I was wondering if there's a really good tutorial out there for me to learn? I've watched several YouTube tutorials but they are either outdated or simply don't work for me. Also, prompting is of course done in english which is not my native language so sometimes I'm having a hard time conveying what I actually want... Thanks! [link] [comments] |
