Stable Diffusion Prompt Dictionary

Stable Diffusion is an AI system that generates images from text descriptions. The text descriptions are called prompts – they allow you to guide the AI to create the kind of image you want. Writing effective prompts is key to getting good results from Stable Diffusion.

Prompts can range from simple (“a red flower”) to complex scenes with multiple elements, styles, and attributes specified. As you become more experienced with Stable Diffusion, you can write prompts that give you more control over the output.

What makes a good prompt?

A good prompt clearly describes the content, style, and attributes of the desired image. Using lots of adjectives and descriptive phrases allows you to specify details. Being too vague often results in random or nonsensical images.

Some key elements of effective prompts:

  • Describe the main subject first
  • Use adjectives and descriptive phrases
  • Specify the medium or style
  • Add relevant context or scene details
  • Use formatting for clarity (commas, dashes etc.)

Prompt structure

Stable Diffusion prompts generally follow this basic structure:

[Main subject], [attributes], [style], [details]

Let’s break this down with an example prompt:

A siamese cat with blue eyes, digital art, fantasy style, intricate details, trending on artstation
  • Main subject: A siamese cat
  • Attributes: with blue eyes
  • Style: digital art, fantasy style
  • Details: intricate details, trending on artstation

This structure helps the AI understand the key elements to include in the generated image.

Styles

Specifying an artistic style is important for guiding Stable Diffusion’s output. Here are some common styles you can reference:

  • Digital art
  • Oil painting
  • Watercolor
  • Pen and ink
  • Vector art
  • Pixel art
  • Low poly
  • Vintage
  • Film noir
  • Impressionist
  • Baroque

You can also specify a medium like “acrylic painting” or “colored pencil drawing”. Referencing specific artists can help define a style too.

Subject matter

The key is choosing descriptive terms and details that the AI will comprehend. Here are some prompt ideas for different subject matter:

Portraits

  • A stylized portrait of a gnome with a red hat
  • An oil painting of a girl with freckles and curly red hair
  • A vector illustration of a body builder flexing his muscles

Landscapes

  • Japanese countryside, oil painting by Claude Monet
  • Aerial view of Manhattan with neon lights, vaporwave aesthetic
  • The Eiffel Tower with a vibrant purple sky, digital art

Still life

  • An arrangement of fruits in a bowl, 17th century Dutch still life painting
  • Glass bottles and flowers on a table, photorealistic rendering

Advanced prompt techniques

As you get comfortable with basic prompts, you can start incorporating more advanced techniques for greater control.

Stylize an existing image

You can make Stable Diffusion stylize an existing image by providing the image and describing the desired style.

For example:

The Mona Lisa in impressionist style with thick visible brushstrokes

Then upload the image of the Mona Lisa.

Mix styles and elements

You can combine multiple styles in one prompt:

Film noir scene with a vibrantly colored pop art woman holding a gun

Or blend elements that don’t normally go together:

A businessman in a suit sitting on a floating inner tube in a swimming pool

Tell a story

For generating compelling scenes, tell a story in your prompt:

A giant robot firing lasers at alien spaceships over a futuristic city, digital art

Use brackets for options

Give the AI options to choose from using brackets:

A treehouse in a [jungle/giant oak tree/magical forest]

Useful Stable Diffusion resources

Here are some great sites to explore for more Stable Diffusion examples and guidance:

The key is experimenting with different prompts and fine-tuning based on the results. With practice, you’ll be creating incredible AI-generated art in no time!