Generate AI Art

Get started with DALL-E, Midjourney, Stable Diffusion, or other image tools. Learn basic prompting, then develop your own style.

How AI Image Generation Works

You type a description of what you want to see, and the AI creates an image. It sounds simple, but the results can be stunning, strange, or surprisingly specific depending on how you phrase your request.

These tools have been trained on millions of images and their descriptions. They've learned relationships between words and visual concepts: what "sunset" looks like, what "art nouveau" means, how "melancholy" might appear in a portrait.

Choosing a Tool

Free Options

Microsoft Copilot (DALL-E 3)

Leonardo.ai

Ideogram

Paid Options

Midjourney

DALL-E 3 via ChatGPT Plus

Adobe Firefly

Your First Image

Let's start with Microsoft Copilot since it's free and easy.

  1. Go to copilot.microsoft.com
  2. Type: "Create an image of a cozy reading nook with a cat sleeping on a velvet armchair, afternoon sunlight through the window, warm colors"

NOTE You may be asked a few questions about your identity. Yes it's annoying. Unfortunately bad actors have pushed websites like this, and others, to extreme security. Don't think of it as a nuisance, but as a way we and our work are being protected.

  1. Wait about 30 seconds
  2. You'll get several variations to choose from

That's it. You've just created AI art.

The Anatomy of a Good Prompt

A detailed prompt typically includes:

Subject: What's in the image

"a lighthouse on a rocky cliff"

Style: How it should look

"watercolor painting" or "photorealistic" or "Studio Ghibli style"

Mood/Atmosphere: The feeling it evokes

"dramatic and stormy" or "peaceful and serene"

Lighting: How the scene is lit

"golden hour sunlight" or "neon lights" or "soft diffused light"

Composition: How elements are arranged

"wide shot" or "close-up portrait" or "bird's eye view"

Details: Specific elements you want included

"seagulls in the distance, waves crashing, old wooden door"

Example Prompts to Try

Cozy and warm:

"A small bakery in Paris at dawn, warm light glowing from the windows, fresh bread visible inside, cobblestone street wet from rain, watercolor style with soft edges"

Dramatic landscape:

"Ancient temple ruins in a jungle, shafts of golden sunlight breaking through the canopy, vines and moss covering the stone, cinematic wide shot, photorealistic"

Whimsical character:

"A friendly robot gardener tending to glowing mushrooms, steampunk aesthetic, warm copper tones, illustrated children's book style"

Abstract concept:

"The feeling of nostalgia visualized as a landscape, faded photographs floating in a golden sky, childhood toys scattered across rolling hills, dreamy and surreal"

Style Keywords That Work Well

Artistic Styles

Photography Styles

Reference Styles

Mood and Lighting

Common Mistakes and How to Fix Them

Problem: Image is too generic Solution: Add specific details. Instead of "a forest," try "an ancient redwood forest with shafts of morning light, ferns covering the forest floor, a winding dirt path"

Problem: Wrong style Solution: Be more explicit about medium. Add "oil painting" or "3D render" or "pencil sketch"

Problem: Elements are missing Solution: Put important elements early in your prompt. AI pays more attention to the beginning.

Problem: Weird anatomy on people/animals Solution: This is a known limitation. Try different angles, simpler poses, or crop to avoid problem areas.

Problem: Text in the image looks garbled Solution: Most AI struggles with text. Either avoid it, use Ideogram (which handles text better), or add text afterward in an image editor.

Iterating and Refining

Your first result is rarely your final result. The process:

  1. Generate an initial image
  2. Note what you like and don't like
  3. Adjust your prompt
  4. Generate again
  5. Repeat until satisfied

You might also:

Building Your Style

As you practice, you'll develop preferences. Keep notes on:

Save your best prompts. They become your personal toolkit.

Ethical Considerations

AI image generation raises real questions:

Different tools handle this differently. Adobe Firefly, for example, was trained only on licensed content. Consider these issues as you explore.

Next Steps

Ready to apply these skills? Try Design Your Own Tarot Deck for a structured creative project with 22 themed images, or Create Album Covers for Imaginary Bands for pure creative play.

Or browse all Projects.