Dive Into The Evolving AI Unconscious With Midjourney

Using Midjourney feels like plunging into a giant stew filled with countless images cooked up by humans over time. Like humans once used nature as inspiration, AI taps into a reservoir of psychic images handed down by its creators, i.e., us.

Before I discovered Midjourney, my artistic path was paved with unfinished projects. But things took a turn when I could finally bring my mental visions to life with just a few simple instructions. Suddenly, I was no longer hindered by the limitations of traditional mediums or my lack of artistic abilities. The AI processing platform was not only able to capture my blurry mental images; it could access the collective unconscious of humanity.

Here’s a quick rundown of how it works.

The images are generated from the latent space of a deep learning algorithm. To the eyes of a computer, images look like a bunch of numerical values representing the red, green, and blue components of each pixel. Unlike our brain, which perceive these variations in values as complete images, computers need to develop a strategy to efficiently process the same information. They start by making guesses and then learn through repetition, guided by our feedback. Metrics are collected to create a mathematical space that differentiates images based on attributes such as redness, shininess, roundness, etc.

As computers lack subjective judgment or biases, it is essential for these attributes to be varied enough to capture the unique characteristics of objects. But keep in mind that AI is not an artist, but a tool. Ultimately, the computer’s goal is to align with human preferences.

This preference is illustrated by two leading platforms: DALL-E and Midjourney. DALL-E tends to interpret prompts literally, resulting in vibrant and optimistic images. In contrast, Midjourney caters to creative individuals seeking atmospheric and fantasy-like aesthetics.

Its unique strength lies in its ability to understand ambiguous and abstract prompts, offering a fantastic playground breaking free from the constraints of physical reality.

Here are some examples of my experiments with the platform a few years ago

Dive into the Evolving AI Unconscious with Midjourney

When I started working on my “hyperrealistic underwater cathedral made of unicellular organisms”, Version 3 of the platform created a sense of space rather than focusing on the technical details, using organic pattern to hide any perspective issues.

In the second set, I added a “Cherenkov radiation” prompt to add an artificial, otherworldly glow for lighting. These prompt descriptors become brushes in the creator’s hand, and one can master their application.

In version 4, using the same prompt led to images that gained in realism, but lost the dreamy atmosphere and fleeting quality of the younger AI.

I have a sinking feeling that the innocence of childhood might fade away as the AI matures. Perhaps we’ve stumbled upon the golden era of AI at this turning point.

Style Fusion: AI’s ability to combine the distinct styles of diverse artists into a novel graphic style is one of its most fascinating capabilities. This works best with artists who have extensive libraries of work available for anyone to use.

Animal Fusion: When attempting to get “a wildebeest, a penguin, a cat, a coyote, and a vulture enjoying a Michelin-star 5-course meal at dusk, caught on wild camera”, I was expecting to see a gathering of animals not commonly seen together. To my surprise, the AI created entirely new animals instead.

Also interesting how the platform interpreted the prompt “Humans watching a garden grow” by imagining the garden as an advanced entity to make sense of the sentence.

Midjourney’s AI excels at producing lifelike photographs. It has the capacity to capture ambiguous scenes, like seen through a surveillance camera, or replicate the shallow depth of field found in vintage black and white photos, as seen in the depiction of “Ophelia summoning death”. Adding the descriptor “chaos 99” further disconnects the image from reality.

Above, a “neo-surrealist, neo-noir city in duotone green and purple” and “an old steam train riding the ocean, wide angle landscape photography” both resulted in enigmatic landscape, illustrating how the hyper-realistic depiction of surrealist elements can evoke a sense of unease and wonder.