Exploring Computer Generated Art

Harper was posting random things about Midjourney last night and it got me thinking about this entry.

What led me to Midjourney was an internal project at Brale intended to understand some different blockchains and functionality that could be supported in NFTs. The result of the project was a NFT series called Gone Missing and an attempt to re-imagine how headshots can function. The art created went through various forms of computer generated outcomes. The art turned out wonderfully and embedded into it is a bit of a goose chase. Here are a few of my favorites from the series and you can find some of the hints in the NFT properties.

As I’m sure you can imagine the project had nothing to do with art and everything to do with exploring various chain functions but the art was a good way to have fun with it.

This experiment led the team down a path to Midjourney and DALL·E 2. Once I got into Midjourney, it took me some time to understand how the different inputs could impact the outputs. After that, I decided to try to focus on exploring a topic, so I chose architecture.

I decided to go down the architectural path for a few additional reasons. The first is that architecture is difficult to imagine without divorcing yourself from the structural requirements required to support a structure. Assuming a computer is imaging hues and influences as opposed to physics is an interesting thought. When you say brutalist to a computer, it may understand greys, tones, and other attributes of that style of architecture. Still, it likely isn’t considering the weight per sq. ft. and the underlying soil’s ability to support that weight. That opens all kinds of fun and exciting doors.

Some were fanciful

Some were realistic and true to form

Some were more artist inspired

Some were beautiful nonsense

Others were curious

Something freeing about the Midjourney product is the permission to be wrong and have a terrible idea. I’ve written about this in the past and enjoyed the reminder.

The final output had me questioning and misremembering my inputs, wondering if there had been some influence of Asian culture in introducing the person in the scene. Or, if I’d referenced Dali’s Female Figure With a Head of Flowers and the algorithm interpreted that into the background. I couldn’t remember; by that point, it didn’t matter.

The actual answer is less curious than the possibility that it could be. This last scene is fascinating, and if the right person produced it with the right story, you never know what the outcomes would be.

Dali’s “Female Figure with Head of Flowers”

Using Midjourney felt reminiscent of design classes that require dozens of design concepts on paper as part of selecting a direction on any idea. I tested a few design ideas by throwing together thoughts that might be too complex for me to draw out independently. However, it took ol’ Midjourney a few minutes to kick out in variations I didn’t expect.

The outputs seem decoupled from the physics of the structures, but they are clearly influenced by the designs that adhere to them. Either way, Midjourney is amazing. I’m excited to try DALL·E 2 and explore how to implement some of this into new artistic projects.