Midjourney is a generative AI program which can convert simple natural language prompts into high-quality images. If you have an idea which you can pen (or rather type) down for the program, it will visualize it for you.
Right around the time the hype for this newly launched AI was building up I was finishing my exchange semester in Madrid, and like any other exchange student I made some stupid mistakes. My first mistake was to drop my phone from the fourth-floor balcony during New Year’s Eve. My second mistake was not making sure all my phone pictures are backed up on the cloud when I went to the repair store the next morning, still half dizzy. It was merely coincidental that during the two days my phone was kept at the store, I was bombarded with AI generated pictures on photography communities online. Upon further research, I found out that these were being created by inputting prompts into Midjourney. All you needed was a Discord account.
Thus, when I received my newly formatted phone back only to realize that all my pictures from the past six months of exchange have vanished, I decided to give Midjourney a try. Crestfallen that I had lost so many memories, I wanted these images to be as realistic as possible. The free version gives you 25 prompt tries, so I researched on the science behind these text prompts to make the most out of those tries. You enter “/imagine” into the text field and voila, you can describe your image.
Midjourney prompt text field
Using a bit of trial and error and building upon what I read on the Internet, here are some general ideas which helped me recreate the images of my choice:
- The more detailed the description, the better your image results usually are.
- Make use of commas, they act as soft breaks to your image description.
- Adding weights to your words, such as 0.5 or mentioning the axis ratio such as “–ar 16:9” can enhance the results.
Example of a typical Midjourney prompt
You can find the results of my journey with Midjourney below, which I believe are quite impressive. The only aspect where Midjourney struggled back when I made these pictures was recreating realistic humanistic features, this being continuously improved and functioning even better now. Whether AI generated images pose a threat to the professionals in the field is a matter of the consumer’s demand, and I have no opinions on that because the creative industry seems like an irrational vortex to me. However, I can definitely see photographers, film studios, and creatives making use of such programs for conceptualization, innovation and maximizing their creative potential.
What do you think?
