Welcome to the 8th in a series of posts where I will test AI image generators and see how they handle making maze art. I will be asking 10 prompts and seeing what gets generated. My goal is to evaluate different AI image sites against each other to see how they perform. If you need to catch-up here is the series so far:
An exploration into Al Image Maze Generation
Making mazes with AI: DallE2 (Updated to compare 2023 to 2026)
Making mazes with AI: Stable Diffusion (Updated to compare 2023 and 2024 to 2026)
Making mazes with AI: Craiyon (Updated to compare 2023 to 2026)
Making mazes with AI: Dream by Wombo (Updated to compare 2023 to 2026)
Making mazes with AI: Nightcafe (Updated to compare 2023 to 2026)
Making mazes with AI: StarryAI (Updated to compare 2023 to 2026)
Update from 2026:
When I did this series in Spring of 2023 I found that none of the text to image generators were able to generate actual mazes, including this one. They were able to make images that looked maze-like, but none were solvable puzzles. The 10 prompts below are a record of what was possible then, but each generation of AI pushes the limit further on what is possible. I remade each maze in 2026 for comparison. All of these were made using text to image generation and not coding.
The first change we see after 3 years is a new logo:
Making Maze Art with DeepAI
2023: Today I test out DeepAI. You can access the DeepAI website here. You can start generating images with prompts without signing in. Each prompt, or generation will generate 1 images. There are 12 styles to choose from (for free - 26 in total). All prompts I use are the same for each site !
2026: I used the website without signing in and it generated 1 image per prompt. There are 3 models available and I used the free default option (the other 2 are for subscribers). The styles have expanded from 12 in 2023 to over 100 today ! I used the default and let my prompt guide the generator. You can also choose 1 of 5 possible output shapes.
Prompt 1 - Make a medium difficulty maze of the Eiffel Tower in black and white with arrows at the start and finish
2023: I was confused at first. Did I get 4 images or 1. I guess this is one.
2026: Much improved. I see the tower and I like the reverse coloring and the mazelike lines in the sky.
Prompt 2 - Draw a medium difficulty large maze of the Empire State Building with the start and goal embedded in the structure
2023: Not a maze. It looks like the bottom and middle section of the ESB. Why no complete building ? Not sure.
2026: If I asked for a line drawing of NYC featuring the the Empire State Building from a drone view this would be perfect ! I see a minor attempt at a maze.
Prompt 3 - Draw a difficult maze of the White House pixel art style
2023: No.
2026: Yes (ish). I like that very Americana looking solid orange flag ! Not in pixel style. Maze on the lawn.
Prompt 4 - Draw a difficult maze that looks like a drawing of a famous building in sketch style
2023: What famous building is this supposed to be ?
2026: I am going to guess this is Big Ben surrounded by a fortress like maze, in a sketch style. It got a lot right this time.
Prompt 5 - Draw a maze in the style of doyoumaze.com of a skyscraper in NYC
2023: Almost cool, or kind of cool.
2026: Really good interpretation of what was asked for. I like that there are mazes both in the sky and on the ground.
Prompt 6 - Draw a maze in the style of Sean C Jackson of a scene from a large outdoor market
2023: This looks very interesting but misses the mark of what was asked for.
2026: If I blindly showed you these 2 images with no information would you guess they were the same prompt ? The market must be closed. Cool illustration.
Prompt 7 - Make a maze of a slice of an orange in color
2023: Hmmm. No.
2026: Deepai has the image part down very well - that is a slice of an orange. The maze part ? Not so much.
Prompt 8 - Make a maze integrated on top of a photograph of a king sitting on his throne looking cantankerous beside his beautiful queen
2023: Cantankerous looking guy in a charcoal jacket in an almost passport photo!! No queen. Fun circular labyrinth.
2026: So much improvement ! This is much closer to what the prompt requested.
Prompt 9 - Make a solvable maze that is very large and very difficult to solve because it is so complex
2023: Weird border on the right side. I wonder what made the AI put this image off center ?
2026: I like the creativity, the layers and the shadows. Not a maze but gives the feeling of one well.
Prompt 10 - Make a 3d render of a red and blue glossy cube maze
2023: This is a big miss for me. Where is the cube ?
2026: Lots of cubes.
Summary and Conclusion:
2023: How did DeepAI do ? A few hits with the black and white versions. Not so good with mazes. I was generally completely unimpressed. Needs work.
2026: A lot of improvement. Missed some things that were asked for but never spit out something totally odd like many generations from 2023. Seems to struggle with not only with generating a real solvable maze (all struggle with this), but anything that looks like a maze is often forgetten.
Coming Next: Canva
