I’m still learning all the time, which is why I think I’m so obsessed with it, I love the challenge and the surprises.
I tend to use the formula: subject, action, setting, style… as I tend to focus on portraits.
Sometimes I write in sentences, often I use :: to multiprompt and then refine that with weighting for each separate idea such as idea1::3 idea2::5 idea3::-1.5.
I sometimes have very brief prompts I write myself, sometimes I play with ChatGPT to give me a description of a character and use and/or edit that. I usually generate dozens of images with slight variations in the prompts each time until I get to what I’m imagining.
Sometimes I use the /shorten command to see the result and to see the keywords that MJ is picking up.
Sometimes I start off with an image and use the /describe command to see the keywords for style that MJ is seeing, this helps a lot as I avoid using names of real artists or people in my prompts.
Sometimes, if I’m having trouble getting an image right I will create one part (like the subject) and then another (like the setting), and then combine the images in one prompt. I often do this too with multiple versions of one subject to try to get just the right look or expression.
Variations are a blessing and I tend to use both types on any result that I like just to check if there might be a slight improvement.
Every image I make is also run through Lightroom where I do further editing and some images require more editing (or combining) so use Photoshop to composite before taking into Lightroom.
Cheers, and good luck (and have fun)!
Leave a Reply