Name: Towards AI Legal Name: Towards AI, Inc. Description: Towards AI is the world's leading artificial intelligence (AI) and technology publication. Read by thought-leaders and decision-makers around the world. Phone Number: +1-650-246-9381 Email: [email protected]
228 Park Avenue South New York, NY 10003 United States
Website: Publisher: https://towardsai.net/#publisher Diversity Policy: https://towardsai.net/about Ethics Policy: https://towardsai.net/about Masthead: https://towardsai.net/about
Name: Towards AI Legal Name: Towards AI, Inc. Description: Towards AI is the world's leading artificial intelligence (AI) and technology publication. Founders: Roberto Iriondo, , Job Title: Co-founder and Advisor Works for: Towards AI, Inc. Follow Roberto: X, LinkedIn, GitHub, Google Scholar, Towards AI Profile, Medium, ML@CMU, FreeCodeCamp, Crunchbase, Bloomberg, Roberto Iriondo, Generative AI Lab, Generative AI Lab Denis Piffaretti, Job Title: Co-founder Works for: Towards AI, Inc. Louie Peters, Job Title: Co-founder Works for: Towards AI, Inc. Louis-François Bouchard, Job Title: Co-founder Works for: Towards AI, Inc. Cover:
Towards AI Cover
Logo:
Towards AI Logo
Areas Served: Worldwide Alternate Name: Towards AI, Inc. Alternate Name: Towards AI Co. Alternate Name: towards ai Alternate Name: towardsai Alternate Name: towards.ai Alternate Name: tai Alternate Name: toward ai Alternate Name: toward.ai Alternate Name: Towards AI, Inc. Alternate Name: towardsai.net Alternate Name: pub.towardsai.net
5 stars – based on 497 reviews

Frequently Used, Contextual References

TODO: Remember to copy unique IDs whenever it needs used. i.e., URL: 304b2e42315e

Resources

Take our 85+ lesson From Beginner to Advanced LLM Developer Certification: From choosing a project to deploying a working product this is the most comprehensive and practical LLM course out there!

Publication

Random Generative AI Art for Endless Creativity
Artificial Intelligence   Latest   Machine Learning

Random Generative AI Art for Endless Creativity

Last Updated on November 3, 2024 by Editorial Team

Author(s): Nicolas MARTIN

Originally published on Towards AI.

One of the greatest features of many Artificial Intelligence algorithms is its randomness. This randomness gives a more natural approach to answering different needs and can be applied to a new layer: creating random prompts for limitless creativity. I will present different ways to benefit from this randomness, from a specific field or set of words to full random generations.

Foucault’s Pendulum

I was always fascinated by Foucault’s Pendulum, a book by Umberto Ecco. This book has extraordinary intellectual and inspiring stories that make readers think about many ideas. One of the most interesting ones I remember is when the main protagonists try to create a story from scratch and lack inspiration initially. The solution was simple but brilliant: write random words on small pieces of paper, mix them in a bag, and select a few of them randomly to make a sentence. Here is where all the magic happens! The brain tries to create meaning between words, explore new possibilities, and hence create innovative ideas. They repeat the same process several times, creating different innovative scenarios thanks to the magic of randomness. Eventually, they pick their best one and create the scenario for a great story.

We will use this process to generate random art and explore different solutions to cover all possibilities of the randomness advantage.

Do we need to know programming or any technical skills to make random generative art? Not at all! Some basic knowledge of prompting is necessary. Here are five different solutions…

1. Experience-based

Internet is full of valid and complete prompts to create stunning images, why not use them with AI? Therefore, we need to use Microsoft’s Co-Pilot or Google’s Perplexity because they get their knowledge directly from the Internet.

For instance: β€œGet 5 great prompts for image generation from different sources on the Internet.”

Here is the result I get:

  1. A futuristic cityscape at sunset: Imagine a sprawling metropolis with towering skyscrapers, flying cars, and neon lights reflecting off glass buildings. The sky is painted in hues of orange and pink as the sun sets behind the horizon.
  2. A serene forest with mythical creatures: Picture a dense forest with tall, ancient trees. Among the foliage, you can see unicorns grazing, fairies fluttering around, and a gentle stream flowing through the scene, illuminated by soft, magical light.
  3. A steampunk-inspired airship: Visualize a massive airship floating above a Victorian-era city. The ship is adorned with brass gears, steam engines, and intricate metalwork. Below, the city is bustling with people in period attire, and the sky is filled with other airships and hot air balloons.
  4. A cozy, rustic cabin in the mountains: Envision a small wooden cabin nestled in a snowy mountain landscape. Smoke rises from the chimney, and warm light glows from the windows. Surrounding the cabin are tall pine trees, and in the distance, you can see the majestic peaks of the mountains.
  5. A bustling alien marketplace: Imagine a vibrant market on a distant planet, filled with various alien species. Stalls are overflowing with exotic goods, strange fruits, and futuristic gadgets. The atmosphere is lively, with colorful banners and lights strung overhead, and the sky is a deep shade of purple with multiple moons visible.

If I take any of those prompts, the result is already great. We can generate them with Co-Pilot, Meta.ai, Leonardo, or X’s Grok.

For example, I try the first prompt using Grok (based on Flux):

Imagine a sprawling metropolis with towering skyscrapers, flying cars, and neon lights reflecting off glass buildings. The sky is painted in hues of orange and pink as the sun sets behind the horizon.

But we want something more creative using randomness.

To do that, we ask the chatbot: β€œCreate 3 random prompts based on the previous ones.” Once we have the result, we copy/paste it in the next query to see the result.

I get the following renderings:

Visualize a tranquil underwater forest with tall, ancient trees made of coral and seaweed. Among the foliage, you can see unicorn-like sea creatures grazing, mermaids swimming around, and a gentle stream of bioluminescent light flowing through the scene. The entire forest is bathed in a soft, magical glow.
Picture a grand medieval castle with a library that combines steampunk and gothic elements. The castle is perched on a cliff, with dark storm clouds gathering overhead. Inside, the library is filled with brass gears, intricate metalwork, and bookshelves carved from dark wood. Lightning strikes illuminate the castle’s towers, creating a dramatic and magical atmosphere.
Imagine a vibrant marketplace on a floating island in a distant alien world. The market is filled with neon signs, holographic advertisements, and stalls selling exotic goods and futuristic gadgets. The sky is a deep shade of purple with multiple moons visible, and the atmosphere is lively with various alien species interacting.

Those examples might not be as beautiful as some non-random ones but they are much more creative:

  • An underwater world with trees and licorn creatures doesn’t exist.
  • A castle with an external library is also something new.
  • A floating alien marketplace in a purple world is not something that everyone could imagine.

Those are interesting examples as creative and successful prompts inspire them.

We can go a step further using a similar approach for a specific theme…

2. Theme-based

Some projects or challenges require a high creativity with a specific theme like trains or blue or robots.

Getting creativity with randomness is possible by applying the previous solution, just adding the desired theme.

First, ask: β€œGenerate 5 very different prompts about trains to generate images.”

Here is the result:

Here are five distinct prompts about trains that can be used to generate different kinds of images:

1. A futuristic high-speed bullet train traveling through a neon-lit city at night. The train’s sleek silver body reflects the vibrant lights of the towering skyscrapers, with glowing billboards and flying cars visible in the background. The scene is fast-paced, emphasizing advanced technology and a utopian atmosphere.

2. An old steam locomotive crossing a rickety wooden bridge over a misty, deep canyon at dawn. The train’s thick black smoke swirls into the early morning sky as the sun’s rays begin to pierce through the fog. The setting is a lush, green mountainous region, giving the scene a sense of adventure and historical nostalgia.

3. A luxurious vintage train interior from the early 20th century, with polished wood paneling, velvet upholstery, and golden chandeliers. The atmosphere is opulent, with passengers dressed in elegant period clothing, sipping tea, and enjoying the grandeur of travel in a bygone era.

4. A post-apocalyptic train station, where a rusty, decaying train sits on the tracks. The sky is overcast, and the environment is filled with abandoned buildings, overgrown vegetation, and makeshift shelters. Survivors are seen scavenging supplies around the station, hinting at the collapse of society, with a somber, dystopian tone.

5. A magical train flying through the sky, above the clouds, surrounded by stars and a glowing moon. The train has an ethereal, steampunk design with gears and glowing runes. Wisps of light trail behind the train as it glides through the sky, creating a whimsical, dreamlike atmosphere full of wonder and fantasy.

Each previous prompt creates beautiful pictures, but we want something more challenging to imagine.

So, we ask: β€œCreate 3 prompts with random words from the previous prompts.”

And here are the results with each prompt below:

A silver, rusty steam train flying above skyscrapers in a dystopian neon-lit city. The train’s decaying body contrasts with the futuristic lights and digital billboards. Thick black smoke rises into the night sky, blending with vibrant colors, while abandoned stations are scattered among the towering buildings.
A luxurious train interior set in a post-apocalyptic world. The velvet-upholstered seats and golden chandeliers are tarnished and overgrown with vines. Survivors dressed in tattered clothing sip tea amid the wreckage, while sunlight filters through the broken windows, casting a surreal glow on the scene.
An ethereal steam locomotive crossing a misty wooden bridge high above a canyon at night, surrounded by glowing runes and stars. The train’s thick black smoke swirls into a sky filled with constellations and a shimmering moon, creating a magical, dreamlike atmosphere in a forgotten, overgrown landscape.

As we can see, the results are creative, but even if they have a random factor, they are based on classical prompts.

What if we bring this randomness to the next level?

3. Parameter-based

In general, image prompting follows more or less the same structure:

  • The main focus. It could be a person, an animal, an object, a landscape, a concept. It should reflect what is on the center of the picture.
  • The details. Those are extra, but information about the main focus. Does it has an action? Is there some specific things on it? What is its color?
  • The background or the environment. Every picture can be created with only one main focus, but the AI will suggest any environment. That’s why it is important to specify what background or environment we want. I could be a forest, a city, a couch shop, etc.
  • The style. It is the form or the tone of our picture. Do I want a traditional, an Asian, or a cyberpunk style?
  • The emotion. Many options are possible: Dramatic, happy, sad, chill, etc.
  • The camera features like the focus mode, or the type of camera (any known model).
  • The camera angle. Low angle, high angle, wide angle, etc.
  • The camera shot. Full shot, close-up, half-body shot.

If I ask ChatGPT or Grok a list for each parameter β€œcreate a list of 6 different examples for each feature about image prompting”, here is the result:

Main Focus:

  1. A wizard casting a spell β€” Emphasizes a character in action.
  2. A vintage car β€” Focuses on an object with historical significance.
  3. A coral reef β€” Highlights a natural underwater landscape.
  4. A futuristic drone β€” Centers on a piece of technology.
  5. A philosophical debate β€” Represents an abstract concept or activity.
  6. A phoenix rising from ashes β€” Symbolizes rebirth or a mythical creature.

Details:

  1. The wizard has a glowing blue staff β€” Adds color and magical elements.
  2. The vintage car is rusted but with shiny chrome bumpers β€” Specific textures and conditions.
  3. The coral reef is teeming with neon-colored fish β€” Adds vibrancy and life.
  4. The drone has four rotors and a camera with a red blinking light β€” Details on the drone’s design.
  5. The debate takes place in an ancient Greek forum β€” Setting the scene for the concept.
  6. The phoenix is red and gold, with flames that don’t burn the surrounding flora β€” Specific colors and magical properties.

Background or Environment:

  1. In a misty forest at dawn β€” Sets a natural, serene environment.
  2. On a bustling city street in the 1950s β€” Places the car in a specific time period.
  3. In the deep ocean with faint light from above β€” Creates depth and a sense of isolation.
  4. Hovering above a smart city skyline at night β€” Suggests a futuristic setting.
  5. Under the shade of an old oak tree in a sunlit field β€” Peaceful, rural environment.
  6. Against a backdrop of a volcanic eruption β€” Adds drama and danger.

Style:

  1. Oil painting β€” Traditional art style.
  2. Manga β€” Japanese comic style for dramatic expressions.
  3. Steampunk β€” Combines Victorian era and mechanical elements.
  4. Watercolor β€” Soft, flowing, and translucent effects.
  5. Pixel art β€” Retro video game aesthetic.
  6. Art Deco β€” Geometric shapes, bold outlines, 1920s flair.

Emotion:

  1. Mysterious β€” The wizard’s spell casting feels like uncovering ancient secrets.
  2. Nostalgic β€” The car evokes a sense of past eras.
  3. Serene β€” The coral reef suggests tranquility and beauty.
  4. Innovative β€” The drone imparts a feeling of technological advancement.
  5. Intense β€” The debate conveys intellectual rigor and passion.
  6. Awe-inspiring β€” The phoenix’s rebirth is majestic and inspiring.

Camera Features:

  1. Macro lens for intricate details of the wizard’s staff.
  2. Vintage film camera for the car to enhance the old-time feel.
  3. Wide-angle lens to capture the expanse of the coral reef.
  4. High-speed camera to catch the drone in motion without blur.
  5. Tilt-shift photography for the debate to focus on the philosophers.
  6. Long exposure to capture the movement of the flames and the phoenix.

Camera Angle:

  1. Eye-level to engage directly with the wizard.
  2. Low angle looking up at the vintage car to give it grandeur.
  3. Top-down view to show the layout of the coral reef ecosystem.
  4. Bird’s eye view for the drone, emphasizing its surveillance capability.
  5. Side angle to capture the dynamic of the debate.
  6. From below to emphasize the phoenix’s ascension.

Camera Shot:

  1. Close-up on the wizard’s face showing concentration.
  2. Full shot of the car to see it in entirety with the environment.
  3. Extreme wide shot to show the vastness of the ocean around the reef.
  4. Medium shot of the drone to see it in relation to its surroundings.
  5. Two-shot of the philosophers engaged in discussion.
  6. Over-the-shoulder shot capturing the phoenix from behind as it rises, showing its perspective.

Although we can extend the list of main focus objects, we will use this list to create random prompts thanks to the query: β€œNow, pick single random examples of each feature to generate 3 random prompts.”

Here are the results we get:

In a bustling 1950s city street drawn in vibrant Manga style, a futuristic drone with four rotors and a red blinking camera light hovers, capturing the scene with a high-speed camera. The drone is viewed from a low angle in a full shot, embodying the spirit of innovation amidst an anachronistic setting.
Capture an awe-inspiring scene with a long exposure watercolor of a red and gold phoenix rising from ashes under an ancient oak tree. The flames flicker gently without harming the flora, all viewed from below in an extreme wide shot that includes the sunlit field around.
A close-up, steampunk-style image of a wizard at eye-level, casting a spell in a misty forest at dawn. The wizard’s staff glows blue, captured with a macro lens to highlight the intricate details of his mysterious, mechanical attire.

4. Semi-Random

After several trials, I wanted to get random suggestions about things thanks to ChatGPT, Claude, or Grok. Unfortunately, some words came back, such as β€œserenade”, don’t ask me why. Consequently, current chatbots cannot suggest total random words yet, but we can try alternative solutions to reach this objective.

One of the best one, is to ask to an available service such as the Random Word Generator, get a dozen random words, and ask a chatbot this:

β€œGenerate a prompt with 5 random words in this list: gainful puzzled morning kindhearted pin influence zonked ground time real wretched identify”

Here are the results, repeating the process 3 times:

Create an image showing a kindhearted character in the morning light, sitting on the ground while looking puzzled at an old, mysterious pin they’ve just found.
Create an image where an ordinary day turns psychedelic with flowery patterns emerging from a sophisticated device, while a furry creature watches in wonder.
Design an image of a gorgeous yet lethal device that emits a nebulous light, set in a utopian landscape, where a tramp is entertaining onlookers with tricks.

I find the results interesting, and the results are very different if we ask the same prompt again. This is probably due to the fuzzy view that those prompts suggest.

This is not completely random, as the chatbot selects words to extract some meaning, but this is a good approach for creativity. You can reduce the suggested words or increase them, but the fewer choices the AI has, the more effort it will make to create meaning.

5. Crazy Prompts

Now, let’s explore the most creative side of AI using crazy prompts.

To achieve that, we create 5 random words randomly with the Random Word Generator, and ask the chatbot: β€œGenerate a crazy image prompt with those key words: historical fire illegal petite steer”, then apply the same logic twice. Here are the results:

In a vast historical battlefield frozen in time, a towering bonfire crackles with bright orange flames, casting eerie shadows on ancient stone ruins. Among the ruins, a petite figure dressed in medieval armor stands defiantly, holding the reins of an enormous steer with glowing red eyes, seemingly out of place in this setting. Nearby, a scroll marked with illegal symbols floats in the air, flickering with sparks as if threatening to catch fire. The air is thick with tension, as if a powerful, forbidden event is unfolding under the night sky.
At the early break of dawn, a suspect stands on the edge of a vast desert, but instead of sand, the ground is covered in reflective mirrors that shimmer with the first light. A massive, floating silver bucket above the desert slowly tips over, releasing a surreal rinse of glowing, iridescent water that cascades down in slow motion. Despite the bizarre scene, the air feels strangely hospitable, as if welcoming the suspect to a secret world. In the distance, giant floating scales of justice hover over a mountain, their movements conveying something profoundly truthful about the universe.
Envision a rainy countryside where the air is thick with rolling fog, barely revealing the outlines of a rusted car stranded on a narrow dirt road. In front of the car, a massive, surreal block of golden butter sits inexplicably in the middle of the road, glistening under the rain. The country landscape is both eerie and peaceful, with distant mountains fading into the fog. Raindrops splash on the butter, creating a glossy sheen, while strange, ghostly shapes seem to emerge from the mist, watching the scene unfold in the silence of the rain-soaked morning.

As we can see, the results are even more creative than before. This is due to the word β€œcrazy” in the prompt to let the chatbot create anything without common sense and to limit the request to 5 random words. Of course, we can use more random words or mix Crazy Prompts with Semi-Random or Parameter-Based solutions (e.g., a crazy prompt with a random camera angle). Last tip: This applies to music, stories, videos, etc. The sky is the limit!

Here are random AI-generated pictures brought to life thanks to Hailuo AI:

Conclusion

Randomness is one of the main functions used in Artificial Intelligence that allows more creativity within acceptable limits, including in image generation. That explains why AI-generated images are well-balanced and always different with the same prompt. However, we can go a step further in randomness by creating random prompts thanks to chatbots. Those new prompts can be adjusted to be more or less random, allowing for highly creative outputs.

Do you want other random generative AI art examples? Check my account on X.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming aΒ sponsor.

Published via Towards AI

Feedback ↓