A Little Bird Told Me: Entering 2024 With Midjourney V6
She pauses to look over the faded newspaper. Looking back at us. Her brown eyes twinkling; prodding a question. She then asks me to help her draw a bird.
We are entering 2024 with an alpha test version of Midjourney V6 that comes with features such as
- More accurate prompt following as well as longer prompts
- Improved coherence, and model knowledge
- Improved image prompting and remix
- Minor text drawing ability
The developers are telling us that we will need to ‘relearn’ how to prompt. V6 is supposed to be better at understanding humans, so I put it to the test during winter break. I explored remixing and text drawing. I continued to reinterpret my favorite texts such as Song of Solomon by Toni Morrison.
My favorite scene from one of my favorite novels is the ending when Milkman (Dead) learns how to “ride” the air by surrendering to it, by letting it take control of him. His flight perpetuates the myth of Solomon’s Leap located at an outcropping of two rocks from which an ancestor (Solomon) began his flight home to Africa, to bury his son. Morrison creates a juxtaposition of the intense exaltation of an escape from oppression to a longed-for home in Africa. Also, the book explores the occupation of space between flying and living… surviving.
We can find similar stories such as “The People Could Fly” where enslaved African Americans who lost their wings under great oppression learn to fly freely once again. Because I had a keen interest in reading (and art) as a young person I find myself re-interpreting texts a lot through prompt engineering. In the 8th grade, I took on the persona of “Star Child” after hearing the song of the same name by Teena Marie.
I’m more familiar with the concept of remixing from my interest in hip-hop culture. Remixing is one of three creativity modes in my book, Techno-Vernacular Creativity & Innovation. This is one of the features of Midjourney that I expanded to include Deep Style (neural image transfer). First, I dug up an older prompt from 2023 to use in Midjourney V6: I subtly upscaled two thumbnails because I liked the hair in one, so I cut/pasted it into another. Next, I extended the left side using Photoshop’s generative fill.
I uploaded the composited image for use in Deep Dream Generator, with Deep Style. I applied a style reference that I’ve used for years. The images are part of my “Afro Surrealism” series. Next, I composited the prompt-generated source images from Midjourney V6 and the DDG output using Photoshop, along with generative fill — adding to and removing contents from images using Firefly — to get the following results:
Successful digital image remixes merge methods and styles. The images are generated using variations and sometimes text. My process is different for each image and this allows me to reinterpret texts, merge art styles, processes (ex. prompting, image style transfer, compositing), and use multiple tools (ex. Midjourney, Deep Style, Photoshop) to get myriad results. GenAI tools are enabling me to add more possibilities to my evolving remixing process.
For the image above, I applied a different DDG style reference to an image generated in Midjourney V6. If you look carefully you can see the deep style elements composited with the V6 source image (with text added in the prompt). My aim is to incorporate new features into this process, as desired, further developing my aesthetic.