Unprompted AI
What happens when MidJourney is asked to make an image without any description? I was able to get as close to an empty prompt by using a Zero Width Space character (U+200B) or a single punctuation.
This collection is a selection of my favorite MidJourney curations.
MidJourney v5
The results for V5 show a favor for food photography, skulls, reflective objects, old men, classic cars, birds, modern interiors, landscapes, bonsai trees, headphones, dress shoes, musical instruments, and jewelry. Compared to the previous model, results tend to be more photorealistic and less stylized and with a wider tonal range.
The results for V4 show a favor for women’s faces covered in paint, whales, Dalí-esque surrealism, birds, elephants, cats, things in spheres, faces split down the middle into two styles, buses, eyes, old men, water, strange buildings, masked images, and things sprouting tree growth, in addition to others.
MidJourney v4
Results for MidJourney v1 through v3 produce very similar results - very specific color tones, ghostly images of women and people standing in the distance, landscapes of trees and cities, often with stylized grainy chromatic offsets.
