My major in grad school was Art & Technology Studies but we were allowed to take courses in other departments, so I took an elective course titled Artist Techniques and Media. ATS focused on technological inquiry as both medium and subject. However, in the Artist Techniques course we were instructed to choose different artworks and attempt to recreate them using the same media/techniques the artists used. For example, I explored lithography (for the first time) and recreated Elizabeth Catlett’s “Madonna.” I had discovered Catlett’s prints in high school.
I had this assignment in mind when I decided to capture some of the features of Catlett’s prints through the use of generative AI. I had to create a prompt with words that described her technique and aesthetic: strong lines and bold shapes, textile patterns, layers, collage. Most, if not all of her subjects were Black men and women... the majority are of women.
Two of the images I upscaled in Midjourney were these:
I titled them “Usiku”, which means nighttime in Swahili. I was quite pleased with the results, that captured some of the essence of Elizabeth Catlett’s prints. However, the goal was not to recreate them. Nowhere in the prompt is the artist name, nor were there titles of her works in the prompts. What emerged from the generative imaging process was a unique series of images. I even added elements to the images using Photoshop (ex. the blue ribbon in the hair (second image above).
I cut and pasted parts of one image into another (see above), creating a digital collage of AI-generated images. For another series (commission) I wanted to capture the energy of jazz. This took me in an entirely new direction and I created a “constellation” of musicians using this specific style:
The secret lies in the prompt, more specifically in what words are used in the prompt. While AI writing tools are incredibly powerful, most tools have some drawbacks that can affect the quality of the content they produce. One such drawback is not being able to grasp the context and nuances of the artwork. Another is lacking the creativity and intuition of human creators. Thus, it is the role of artists to do these things. A lot of tech people count artists out but it is their/our knowledge and experience with materials and techniques that create amazing images.
And, yes, I added a finger to the violinist’s hand in the image above. Artists can do that, too.
The first time “artist’s hand” was mentioned in relation to AI art was back in 2019, when Wagner James Au interviewed me for his blog and it showed up on BoingBoing, a web aggregator. “Aaron_Hertzmann” worked for Adobe Research and we were debating Deep Dream Generator, the tool I was using at the time to generate images. Hertzmann wrote, “they show the artist’s hand in making them fully realized.” Anybody with access can use the tool but to get better output artists are needed. Always.