Describe something, anything, even things that don't exist! Have fun and play! If you keep your expectations low, you will be amazed, enjoy yourself, and may even find that creating cool stuff is addictive.
Don't worry, all the images are synthesized, not “copied.”
One of many, many arthropods designed for a variety of tasks, in dry and wet environments
Every day, there are more free tools online for generating text and images, logos, layouts, wireframes, animation, and videos.
If you are a designer, you can use AI to generate enough ideas, quickly enough, to speed up your concept development or find an entirely new direction for a project. That alone makes it worth learning these new tools. But plan to edit, re-write, or re-paint, because the AI will make hilarious mistakes. It will certainly surprise you! Take advantage of those surprises, and use AI as a way to try out new concepts. If you already have strong writing, editing, digital painting, drawing, photo editing or graphic design skills, you’ll be intrigued instead of frustrated. Read on, for my Tips on Training Your Djinn, below.
But creating exactly what you or your client needs for a designed communication, or even getting reproducible results, requires learning how to train AI. To actually master AI, you'll need to learn a great deal more. See the list of AI Generators and Tutorials.
About the Ethics
Although an AI can synthesize new images from billions worldwide, it leans toward imagery and styles from popular culture, unless trained to use your model. That's why your towns may look like Tokyo or New York. If you prompt requests "a female warrior in armor" the characters
may resemble Disney's Mulan, even if your prompt is very specific. If you request a “mermaid,” the characters may resemble Disney’s Little Mermaid. (You really, really want to avoid getting a "cease and desist" letter from Disney's lawyers.) To avoid having your AI-generated character look like it might be copied, you must provide your own work as a model. You could also start with a composite of images that are copyright-free from an image bank, such as Unsplash or Pexels.
This is true of text as well. If you think that your writing style has unique and recognizable qualities, such as your descriptions or dialogue, provide AI with sample—the more, the better!
One of 50 iterations of an iridescent arthropod checking the eggs.
TIP #1. Watch Your Language
On writing prompts: Asking an AI to do anything is like negotiating with a Djinn. It will do its best to grant your wish (prompt), but you have to be very careful. If you ask a Djinn to make you live forever, you had better be more specific, “as a young, healthy, wealthy woman of 27 in New York” or you’ll be stuck in a creaky, 68-year old body in the desert for all eternity. Even if you provide very precise instructions, AI can't know what you really want.
For example, some AI image-generators may not understand similes. If your prompt states “skin like a fish,” and you get unexpected results—such as fishes everywhere—try prompting, “skin with small silver scales.” To create original mermaids, instead of Disney's Little Mermaid, I described each major body part, chose the best of dozens of results, and with each iteration I adjusted the description further as the AI and I both learned what results I would accept or reject, and what prompt phrases worked best.
For example:
“A female with a human face, with orange hair, red eyes, and a human torso with two arms and slim hips, merging seamlessly into a fish tail with orange scales and orange fins. Full figure, swimming, side view. In the background is a kelp forest in blue water, with rays of light. In the foreground are pink shells and red coral. Full image, detailed, concept art style.” And then you need to go through another hundred iterations (see the mer-people, below).
TIP #2. Curb AI's Adolescent Impulses
Unless trained out of it, AI adds full lips, sultry eyes, and very large curves to every female character, and huge muscles with lots of armor and big weapons to every male character. It seems as if AI is being raised by programmers with the fantasies of 12 year-old boys, and it is heavily influenced by popular culture. With non-human characters, the AI can become very confused. I have seen eye shadow and lipstick on monsters, bugs and reptiles. As well as spikes, bat wings, dripping saliva, and fangs, of course, on any character not actually defined as "friendly" or "beautiful".
Therefore, when prompting in Perchance and other free AI services, I may specify these negatives: “legs, teeth, fangs, claws, tongue, armor, breasts, big butts, clothes, weapons, scary, malformed" and add positives, such as, "beautiful, friendly, serene, joyful."
Adobe apps and ChatGPT both have a censor. You can't request an image of a public figure, or offensive imagery. Photoshop's Generative AI is careful not to sexualize images:
ChatGPT had a similar problem when I requested that the "chest be flattened and the arms be slightly more muscular." It suggested that I rewrite my prompt to say "streamlined body consistent with a non-mammalian aquatic swimmer."
The claws and legs were not in my prompt. Oddly, they were in my negative prompt, along with breasts
Rejected. The results of my first prompt, mermaid with purple hair swimming in blue water with coral, were too similar to Disney's artwork in tone and style, and she also looks a little too much like a SeaWorld employee wearing a stretchy bikini mermaid costume.
TIP #3. Be Prepared to Layer, Remove, In-paint and Re-paint
Perchance's free Stable Diffusion AI and Adobe's AIs can get confused about the number of limbs, even when creating a human. See the dogs, the astronaut, and the green witch, below. You may get an extra leg. Or three. You may also get extra eyes or fingers, or one ear. In particular, Photoshop's AI has a lot of trouble with hands and faces. If you want your character's hands to clasp the hilt of a sword, be prepared to re-paint the fingers, if you have that skill, or ask AI to in-paint hands. This may take many iterations. If your character's face is lop-sided, re-paint it. Or. . . keep refining and training your AI.
TIP #4. Get Help. Join a Club. Take a Class!
I used Perchance's Stable Diffusion and Adobe Photoshop's Generative AI to create the mermaid images above, while trying to train the AI. But it took forever. And most of the results needed fixing. Then I was taught that to get predictable and reproducible results, you have to download an instance of an AI to your own computer, and then train it on a model, and that takes a whole different level of knowledge. So you have to train yourself first. And that takes time. A lot of time.
Your education in using AI will go more smoothly if you find support, join a club, or take a class. I asked the advice of a friend who has been using AI to build his own characters and environments for two years, and he offered to tutor me. (Thank you, Delonzo!) If I get any of this wrong, it is entirely due to my own errors.
This image of Bogey required combining two images, repainting, and adding filters.
Our dog, a boxador, had a very hard life before we found him. So he is a tough guy with an attitude and we named him Humphrey Bogart. I didn't upload photos and instead pushed a free AI to generate an image of a black dog with smooth fur, folded ears, and a gray muzzle, wearing a tuxedo on a rainy night. There are several errors, including malformed ears, eyebrows like scars, and no sleeves. Edited image is at right.
TIP #5. Get Your Own AI and Build a Model
As you might imagine, I am most interested in creating my own images. So created a compressed file of original paintings, sketches, and photos of a clay bust of my character, Ixtania. I uploaded the Zipped file to Replicate, which built a LoRA (Low-Rank Adaptation) composite model for me. This will guide my AI in building characters and environments. Then I downloaded Diffusion Bee and Draw Things (two different instances of AI) to live locally on my computer, and started building character models I can rely on going forward.
Draw Things AI based my arthropod character at right on my LoRA, which was in turn built from 10 incomplete photos. (Note the extra arm.) My prompt specified "feathery" antennae, but I did not provide a sketch of the antennae to create the LoRA. So you get what you ask for. Some day, I'll be able to rotate, apply textures, add lighting, and more. (Thank you, CutScene Artist!)
See more arthropods, architecture, and examples of repainting below.
Full, green elytra (wing covers) were in my sketch. The petal-like wings and green blade were not
My prompt for Photoshop's Generative AI Fill: A man with dark hair and light brown skin, approximately 35 years old, is wearing a spacesuit and kneeling on one knee in the dock of a space station, side view, full figure, full image, detailed. He is not looking at anything specific so I decided to add a different environment.
I prompted the free Perchance AI through iteration after iteration until I got something close to my goal. This is interesting, but not a biped.
The prompt: "male and female college graduates in purple robes and purple mortarboards with gold tassels, in the background is the Manhattan skyline." I didn't upload a reference photo, and AI's reconstruction of faces looks unrealistic and those fixed robotic grins are disturbing. They might be acceptable to the client but are malformed. Don't try to to edit errors like this in Photoshop; requesting photos from the client is faster. Note that unless gender was specified, AI showed women.
Yes, you can ask an AI to synthesize a scene that does not exist in the real world. But when asked to build a spiral building, the AI found the concept surprisingly difficult to produce. It also assumes that a spiral building must be architecture of the future, not the Tower of Babel, or Pisa, or The Guggenheim. Note the airship hovering in the sky. A spaceship appears in another variation.
I wrote a prompt requesting a town with water and gardens. But a common AI interpretation of "town" includes skyscrapers. I needed to specify a small number of buildings with fewer floors, and add "skyscrapers" and "skyline" to the negative prompt. I rejected these and 40 other variations.
No, really, these are just for fun. Try adding "iridescent," "friendly," "detailed," and "beautiful concept art style" to your prompt. For all of these, I requested a "lightly armored exoskeleton." Remember to describe the environment—in this variation, a flowering meadow.
I'm editing these and will combine the best of AI-generated images with my own sketches into two different LoRAs for Administrative and Nurse castes of arthropods.
6/23/25. These pages are my sketchbook and a notebook for students and colleagues who are teaching AI-assisted design tools for concept development, image and text generation, wire-framing, and analysis. See also AI Document Layout and AI Vector Illustration.
Copyright © 2018 Li Gardiner - All Rights Reserved.