Stable Diffusion is a powerful AI image generation model that can create amazingly realistic and creative images from text prompts. However, writing good prompts can be challenging. In this article, I will provide various prompt examples and tips to help you generate better Stable Diffusion images.
Basic Prompt Structure
A basic Stable Diffusion prompt has the following structure:
(seed)[parameters]: detailed text description of the image
(seed)
: An optional random number seed for reproducibility.[parameters]
: Optional parameters enclosed in square brackets to control image properties like size, sampling method etc.: detailed text description
: The actual textual description of the image you want to generate. This is the most important part of the prompt.
Here’s an example prompt:
(seed 1234)[512x512, oil painting]: a scenic view of a lake surrounded by tall pine trees, in Bob Ross's style
Prompt Tags
You can add special tags in your prompt enclosed in parenthesis to control different aspects:
(masterpiece)
: Generates higher quality, more coherent images(best quality)
: Maximizes image quality(ultra-detailed)
: Adds more fine details(cinematic lighting)
: Adds dramatic lighting
For example:
(masterpiece, best quality): girl with long golden hair wearing a red gown, intricate details, ultra-realistic
Describing Subjects
When describing the main subject, be as specific as possible about visual attributes like hair color, clothing, facial expressions etc. For example:
A woman with long blonde hair, bright green eyes, wearing a white sweater and jeans, laughing
Vague descriptions like “a happy woman” will not work well.
Describing Styles
You can make the AI generate images in different art styles using the following formats:
X style: [Prompt]
in the style of Y: [Prompt]
[Prompt], reimagined as Z
For example:
oil painting style: an astronaut walking on Mars
in the style of Van Gogh: a busy city street
a bowl of fruit, reimagined as a sculpture
Adding Modifiers
Use modifiers enclosed in parenthesis to tweak attributes like lighting, detail level etc. For example:
(cinematic lighting), (intricate details), (8k resolution), (depth of field)
Prompt Weighting
Wrap words/phrases in parenthesis followed by a colon and weight (0-1) to make the AI focus more on those elements. For example:
(woman:1.2)(smiling:0.8): a portrait painting
Providing Examples
You can provide 1-2 example images in the prompt for the AI to use as references. Use markdown image syntax:
![Example1](https://image-url1)
![Example2](https://image-url2)
A cat wearing steampunk goggles, in the style of the examples
Useful Websites
I hope these Stable Diffusion prompt examples and tips will help you create better AI images. Let me know if you have any other questions!