Artificial-intelligence models including Stable Diffusion, Midjourney and DALL·E 2 are cheering technology enthusiasts and causing tremors among artists — for good reason.
Those breakthroughs in software represent the best generative-art platforms, full of breathtaking images and videos created from users’ prompts.
Here are two examples of art generated by Midjourney, my favorite of the bunch:
To create this art, users write special keywords to guide the algorithm. These keywords, also known as prompts, describe the scene to the AI. (For example: “The Lady of the Lake rising from water.”). You then define the aspect ratio of the image or resolution (say, 16:9, 800 x 600), the preferred style (oil painting in the style of Salvador Dali, etc.) and other parameters that describe the scene in finer detail.
Here’s an example of one of my creations and the instructions.
The prompt: Path to salvation by Zdzislaw Beksinski, Craig Mullins, Yoji Shinkawa, complementary colors, intricate details, trending on ArtStation, symmetrical, silk red blue yellow orange black embroidery, Peter Mohrbacher, hyper-detailed, intricate, elite, art nouveau, ornate, liquid wax, marble, elegant, luxury, CGSociety, hyper-maximalist, golden ratio, environmental key art, octane render, Weta Digital, depth of field, ray trace, 8k.
Keywords have an intricate effect on the final creation, and each individual prompt needs to be fine-tuned to generate an image that closely corresponds to an artist’s vision. Although anyone can type, creating an individual artwork isn’t a straightforward process.
After you enter the prompt, the Midjourney AI will generate a 1×2 or 2×2 grid that consists of different interpretations of the provided input. It is up to the user to generate variations based on one of the suggestions, resubmit the entire prompt to the algorithm, or decide to pick one of the given images and upscale (resize) it. The upscaled images are finalized, and that is what I’ve been sharing with you here.
As a portrait artist who spent most of my time working with a stylus and a tablet, I’ve found Midjourney to be one of the best tools I could possibly imagine. While most users stop after generating a piece, my creative journey is just starting. I take Midjourney’s interpretation and refine it, or use the composition and color scheme and start laying in details of my own. This way I’m able to accelerate my process from something that took month(s) to mere days.
Not all artists share my sentiment. Many complain that generative art is nothing more than an amalgam of art that’s already floating around internet, which is then processed, remixed and spat out by an algorithm without rhyme or reason. They say the result is soulless because it wasn’t entirely generated by humans.
However, is that really the case? Coming up with a prompt, a set of variations and a final image may not take months, but it still takes time. Hours, sometimes a few days, if you really want to get it right. The AI does the painting, but the way it does it is heavily affected by your prompt.
Saying generative art isn’t art is like saying movie directors aren’t artists. They have an entire crew putting their vision to life. Such “prompts” are what make the difference between a good and a bad movie. The same is true with Midjourney and other generative-art platforms.
Now you can buy these prompts online and use them to replicate their creator’s result. While purchasing prompts and getting the same result isn’t exactly art — it’s more like purchasing a piece of computer code or a royalty-free image — it shows just how valuable they are in the generative-art process.
Generative art goes far beyond enthralling viewers. One can apparently win awards using it. And make money with it.
If we agree on the premise that generative art is truly art, then you can make money by selling it or using it to create beautiful NFT mints for projects. (That’s what I’m doing in my own beta version of the NFT Business Builder token-gating project.) Finally, you can use generative art to enhance your creative process, giving more quality and value to your creative output.
While the present already looks bright and promising, what about the future? Where is generative art going from here? Think animation — 3D integration and fully fledged, explorable worlds, generated purely by AI, with no traditional 3D artwork required.
This last step will bring us AI-powered games, interactive worlds filled with characters who respond to your inputs and actions the same way well-trained AI chatbots react to your queries.
We are already seeing glimpses of this new future of entertainment (see this video) and visualization, so it’s only a matter of time before we enjoy our first, fully AI-based virtual world. As one of my favorite YouTubers would say, “what a time to be alive!”
What are your thoughts on generative art? Is it art? Let me know in the comment section below.
A long-time Tesla bull cuts his price target on shares of the electric-vehicle maker.
Jurica Dujmovic is a columnist for MarketWatch. He is a business publisher, consultant, designer and gamer. Follow him on Twitter @JuricaDujmovic.
Visit a quote page and your recently viewed tickers will be displayed here.