MidJourney:

Midjourney is a software tool accessed through Discord that allows users to generate images using written prompts. At its most basic level, the process is simple: articulate an idea in words, submit it to the system, and an artificial intelligence model generates a set of visual interpretations within seconds.

Each prompt produces an initial grid of four images. From there, individual frames can be refined, upscaled, or reinterpreted through variations. The interaction is entirely text-based, relying on language, modifiers, and parameters to shape the results.

During my first weeks experimenting with the tool, what stood out wasn’t just the speed or novelty, but the way it translated language into imagery. Even loosely defined prompts often produced compelling—and sometimes unsettling—results. The process felt less like issuing commands and more like negotiating with the system: small changes in phrasing could lead to dramatically different outcomes. I later extended this exploration by focusing specifically on portraiture, which is documented in a separate journal entry.

Prompts themselves are built from parameters that influence scale, detail, composition, and style. MidJourney’s documentation outlines these options in depth, but much of what I learned came through trial and error—testing abstract ideas and seeing how the system interpreted them visually.

What interested me most was not realism, but suggestion. Some of my favorite results came from intentionally vague or abstract prompts, where the images hovered somewhere between the recognizable and the unfamiliar. The ambiguity is part of the appeal.

Tools like MidJourney open up new possibilities for visual artists, photographers, and filmmakers—particularly as a way to explore ideas, moods, or compositions before committing them to more traditional mediums. For writers, the process offers a different kind of feedback loop, where words quite literally shape images.

The images above and below represent my earliest experiments with AI-generated imagery. They aren’t meant as finished works, but as markers of curiosity—captured at a moment when generative image tools were still rough around the edges and full of unanswered questions.

MidJourney operates entirely through Discord, without a traditional user interface, and all images are generated by messaging the MidJourney bot directly. Additional experiments from this period were shared on my personal Instagram account, with bike-related imagery appearing on the BestRidesDC feed.

These images remain here as a record of that initial exploration.