For those of you who know me, I’m a geek at heart who loves to write code. At this stage of my career, I mostly focus on IntelliTect’s core business needs, but coding still calls to me and I love spending an evening creating something beautiful from code.
As you (likely) know, the AI revolution continues to accelerate, and I would say that I am cautiously optimistic. This opportunity has proven too irresistible for several of us here at IntelliTect, and in our free time we’ve jumped into AI development, both with AI agents building code and baking complex AI into the core of the app. I’m happy to announce our passion has resulted in a product we are excited about, and I’m looking forward to getting your feedback.
Read about IntelliTect’s innovative work with Data & AI.
Over the last few months, this skunkworks project came together with the help of a crack crew of AI developers here at IntelliTect. Everyone has been so eager to help and dedicated to the product, we’ve all shared the passion to let it bleed into our personal lives. As the pieces fell into place, we realized that we had practically stumbled on something really unique and exciting. The momentum rapidly built, and AIS Images took on a life of its own. Read more about how the project came together here: The Accidental Image Library: A Story of Curiosity, Creativity, and 100,000 AI-Generated Pictures
If you want to skip all the details about what’s under the hood, you can explore for yourself HERE.
Origin Story
This project originally started as an interactive venue experience in collaboration with James Lowe over at the Spokane Symphony for an awesome Beethoven vs AI concert. We realized that we needed loads of visual content and that generating it with AI on the fly didn’t work well. Read more about our collaboration HERE.
So we dreamed up some new and novel ways to generate compelling images. We were bad at this at first, but as we practiced and learned the nuances of AI image generation, the results got better and better. As we became more creative with our artistic ideas, what started as a library of 9,000 images grew to over 100,000. I found myself just scrolling through what I found were shockingly beautiful images, and I wanted to use in my presentations and other communications.
Why Browsing Beats Prompting
The biggest question we had to wrestle with was, “Why not just generate the image you want on the fly?” Image generation is pretty straightforward. Provide a prompt and get an image a minute or so later. Well, it turns out that simple prompts generate simple images and good prompts are hard to build. It also takes time. There is a delay with every iteration and sometimes iteration is painful as AI wanders around after what you want.
Browsing was the key. People love to browse whether in the produce aisle, on social media, or at a buffet. We love to see the options. For most of us we need stimulus to help us generate ideas. How many times have you thought, “I’ll know it when I see it.” This is exactly what AIS Images provides, a massive library of images you can browse to get ideas and use freely.
We wanted a place where someone could enter an idea and see a variety of images related to that idea. Then you can just keep scrolling and getting new ideas. Maybe you find the perfect image or maybe you get ideas to help refine your search. Maybe you use the image as inspiration for your own image generation.
We’re having fun around the office playing “human or AI generated”. Can you tell which is which?
Curated for Creativity
So, what can I find in AIS Images? People, things, background, nature, art, emotion, and the list goes on. We are adding new images as we find gaps in our library.
Deciding what the software isn’t and what we’re not going to do is just as important as deciding what it is. We are a collection of high-quality, free-to-use AI images that are generally enjoyable and useful to a broad range of people. But what aren’t we?
In one of the first demos I did, I asked for input from the audience about an image search term. Someone shouted out “A hotdog on the moon.” I didn’t think we had any of those, but I searched anyway. It was abysmal on that search, but reflecting afterward I realized that we actually don’t want hotdogs on the moon. That is a niche need and something that people can prompt AI for individually. While fun, it has limited usefulness for us in a more artistic sense.
Two heads and 7 fingers? It turns out that AI image generation has some struggles with getting everything right. Initially, we left everything in the library, but quickly realized the awkwardness caused by seeing weird images. We determined that we would remove weird images, but also ones that were just boring and lame. This is really subjective, but we generated a set of criteria for curating our images.
- Is it useful?
- Is it interesting and engaging?
- Is it too weird?
- Is it appropriate for public use?
- Is it immediately apparent that it is AI?
This included things like subjects facing away from the camera, boring composition, central text that isn’t known words, missing body parts, merged bodies, too many fingers, etc. Ironically, we found that AI likes adding butterflies to everything, so we had to turn that setting down. I guess we all like our cheese.
Explore, Get Inspired, Create
This curation process drove us to intentionally push the artistic boundaries of AI image generation. Is it art? While I’m not qualified to answer that question, I know how I feel when I look at the images. I feel inspired and interested in seeing more. I get new ideas to expand the library. And I want to share a great resource with friends.
Read about AI & The Arts.
So, if you want some inspiration, need some free images for a project, or just want to browse around, check out AIS Images and let us know what you think.