For the past week I’ve been exploring Midjourney, an AI-based image generator tool that creates new images from simple text prompts. The text prompt for the image above is: black women in ethnic clothing dancing in a utopian fantasy botanical garden with insects. The tool then generates four images at a time such as this one:
Midjourney sits somewhere in the middle of the popular DALL-E 2 tool and Craiyon or DALL-E mini. The results in Midjourney are provocative, but not as photorealistic or closely text-matched as DALL-E 2. For my purposes, I don’t need DALL-E 2 because I have Deep Dream Generator, another AI image-generating tool that uses an entirely different process to create images.
I started by using text from my favorite non-fiction book, Song of Solomon by Toni Morrison. At the end of the book the main character Milkman learns how to “ride” the air by surrendering to it, by letting it take control. While flight can be an escape from constricting circumstances, it also scars those who are left behind. Midjourney did a great job with imagining the event.
In order to use Midjourney you need a Discord account. That’s where you will give text prompts to the Midjourney bot. You can sign up for a free account on Discord’s website. After that, you can use Discord in a web browser. The free trial gives you 25 images and after that you can pay $10 monthly to generate 100 images (per month). Note: DALL-E 2 is free (for now) but it has a waitlist and there’s no indication as to when you are taken off the list.
Tools such as Midjourney and DALL-E 2 use natural language processing, a branch of artificial intelligence or AI that is concerned with giving computers the ability to understand text and spoken words in much the same way human beings can. NLPs understand semantics (relationships between words). Deep Dream Generator uses neural (image) style transfer, which understands the relationships between images. Tools such as Dream by Wombo use a combo of NLPs, neural image styles, and generative adversarial networks (GANs) that generate images and other media based on an underlying distribution of data.
Today’s Midjourney text prompt was: Afrofuturistic Black woman botanist in trippy alien utopia. I imagined Black women scientists like in Wanuri Kahiu’s sci-fi film “Pumzi,” which features a character named Asha whose body becomes grounded to nature. Conceptualized in 1993 by Mark Dery, Afrofuturism has come to encapsulate works of Black creative and cultural production that treat futurist themes concerning Africa and its diaspora. In Afrofuturism 2.0: The Rise of Astro-Blackness, I wrote about the coming “semantic web” that
(I)ncludes software used to render digital images and objects in new media technologies
Or as William Gibson is quoted as saying, “The future is already here…” In a short time I wasn’t just writing about or predicting what was to come. I also dived head-first into the fray by creating my own AI artworks, algorithms, and possibly a new Afrofuturistic practice. The characters in my Midjourney botanist series live in a utopia that doesn’t need a specific geography. I used Deep Dream Generator and Adobe Photoshop to remix and layer or composite some of the Midjourney output.