Project

Artists, designers, and tech enthusiasts were among the first to experiment with AI Image Generators, even before the associated artificially smooth aesthetic became overused and “AI slop” became ubiquitous on the internet.

When the technology first emerged, I was studying design. That gave me the time and freedom to explore this field early on. It allowed me to thoroughly test the new tools (some of which were still in the beta phase) and find ways to apply them to my design work. It also gave me the opportunity to gain early experience as a prompt engineer.

Using the text-to-image generator Midjourney I started creating images for my music uploads on SoundCloud, featuring motifs tailored to each track. Maintaining aesthetic consistency across the images posed a particular challenge at the time. This sometimes required numerous retakes and adjustments to the prompts. I also had to manually retouch unwanted artifacts and inconsistencies time and again. It became clear that some motifs were simply not suitable for AI-based image generation. For example, Midjourney was unable to accurately depict a piano octave. In this case, a CAD program had to be used to manually create the motif.

Not least because of the polarization surrounding the topic of AI and the shifting public perception of generative image AI, I ultimately decided to bring the cover series to a close.