Midjourney is a research lab and the name of the lab's artificial intelligence program that creates images from textual descriptions, similar to OpenAI's DALL-E. The tool is currently in open beta thorugh their discord server, and I was given 25 prompts to test the software. Some results below:
These mask where the result of the following prompt: Sci-fi white skull mask based on bird of prey industrial design. I wanted to reproduce the masks of the giants from my webcomic Arcagen, and the results were pretty interesting.
After goofing around with the software (mixing stuff like the Simpsons, Beavis and Buttehead, Hugo Chavez and many other stupid prompts) I started testing the software to produce environments and characters similar to my comic. I started trying to produce some kind of movie poster/comic cover. Here’s some results of the prompt: girl in black bunny hoodie, walking through post-apocalyptic city, overgrowth tall buildings in background, epic movie poster, sunset, birds flying in background, artstation trending, 4k
One of the things my colleagues and I can agree on, is that the colors produced by the software are beautiful. I took the third image in there and made some variations:
Creating Paloma in Midjourney
After seeing the previous results, and inspired by Jake Parker’s experiments on Midjourney, I wanted to see how close could I reproduce Paloma using the software. Here’s the prompt: latina 11 year old girl, dark hair, with dark purple hoodie, bunny ears, big incisive teeth, big eyes, manga style, techwear fashion style, post apocalytic nature setting.
Pretty close uh? Not bad for a first attempt! I kept experimenting, and through some utterly horrible results, I started to get closer and closer. Next prompt: latina 11 year old girl, dark hair, tanned skin, dark purple hoodie with rabbit ears, big cute incisive teeth, big eyes, manga style, techwear fashion style, post apocalytic green nature setting.
Get ready for scary stuff! I tried to make variation on that first image, which was my favorite of the bunch, and these are the results:
Ok ok! That looks bad! Let’s go back. In the end, I decided to upscale the previous result, without any variations, and this was the result.
Ok I like this one. Of course, there are many results in between I skipped. Overall, it was a pretty fun experience. I tried to also reproduce Lechuza and Zorzal with Midjourney, but the results were so bad it’s not worth sharing haha.
Is A.I. taking our jobs as artists?
Maybe, maybe not. It much depends on the industry you’re in, your expertise and how these tools evolve in the future. Pay attention on how I emphasize the word tool. For me, these are no different from an electric hammer vs a traditional one, and iPad vs a sketchbook, or Pinterest vs an art gallery.
So far, Midjourney has been great for generating ideas, mixing different art styles, and emulating what’s already out there to create serendipitous results so we artists can use it as groundwork for future pieces. However, it still has a lot of limitations: First, it uses English, which by itself is a language that encompasses a certain view of the world, but cannot express emotonional concepts in the same manner as Spanish, German or Japanese. Second, it sucks creating perspective, as most characters or environments in creates are in “front view”. Third, it uses a limited pool of visual data, which will never compare to the personal human experience.
Of course I’m only speaking about Midjourney, and we don’t know how advanced A.I. can become in the future. Perhaps corporations, in order to cut their costs, will invest in these art generating tools more and more, at the point where popular art trends could be easily reproduced, in half the time, at half the cost. Imagine an A.I. that could reproduce what we see on Artstation front page everyday:
Maybe art jobs in the future will reward bigger research and idea generation than hours painting in Photoshop or sculpting Zbrush. Repeating my previous point, it depends on what kind of art job will you have.
Personally, I’m not afraid of A.I. (limited to art), as I see it as another way to bring beauty to the world. My prediction? It will never totally replace a high skilled artist which knows their fundamentals and is focused to create, just because it’s enjoyable. After all, guns didn’t make martial artists disappear, fast food didn’t replace gourmet chefs, and 3D printers didn’t destroy wood carving craftsmen. Will it change the current art economy and the rules of the game? Possibly. For better or worse? I believe that’s up to us to decide.
After one month of hiatus, Arcagen is back on Webtoon and Tapas. Remember that you can support the project joining my Patreon. You can also join my Newsletter where you can download a Free Art Pack with brushes, videos and PSD's.
Thank you for reading!