Midjourney: An Image/Text-to-Image Primer
When I first learned that actress/singer Irene Cara had transitioned I immediately opened up my Discord server and private channel in Midjourney. Using version 4 (v4) I uploaded to photos of Cara and a string of words that included “sfumato” and “art deco.” Learning art history is helpful as far as knowing different terms but not copying or using the names of specific artists, which I don’t ever do. There are several online tutorials on how to get started using Midjourney, including this one:
The resulting image (see above left) wasn’t quite right. The shape of the face, nose and lips were off, so I used Adobe Photoshop to revise the image. Photoshop is a different tool or software and requires some time to learn. More specifically, I used the Liquify filter, the Stamp tool, and Layers to ‘mold’ the face features to look more like Irene Cara.
To try this yourself you will need to be in Discord and use the public (free) or your private channel, the latter of which is not free. In the text field enter “/settings” (without the quotation marks) and hit return. You’ll need to use MJ version 4 to upload and use images. You can use more than one but you will need to copy/paste the URL into a new text prompt.
After the pasted URL(s) you can add more text to describe how you want the image to be generated (ex. “art deco”). Don’t use filler words such as “she/he”, “they/them,” “her/him,” or even “the.” The more descriptive you are the better the results. I put my two semesters of color theory to use in many of my MJ images. For example, I used “analogous colors” to create this series:
An analogous color scheme involves three hues, all of which are positioned next to each other on the color wheel. The colors (text) comes after the URLs, if you are using an uploaded image or just use them with other text without the links. The images above were created using the same sketch and I modified the prompt to specify an age group.
In addition to using photo references in MJ prompts I also used my own sketches (see above). I upload them the same as the photos. You can mix it up, too: upload a photo and a sketch, then paste them in the same prompt. Here’s another example created from one of my pencil sketches:
In summary, tools such as Adobe Photoshop can be helpful if you need to modify output from Midjourney. Also, uploading photos, sketches, and other visual artwork (with text) can help generate images. Using specific color combinations are another way to create unique images. Additionally, you can experiment with “Make Variations” to discover more ways to generate an image based on the same text prompts.
The most important part: Give yourself time to play or experiment.