Stable Diffusion is a powerful AI image generation model that can create amazingly realistic and creative images from text prompts. However, writing effective prompts requires some skill and practice. This article provides examples of prompts across different categories to help you get started with Stable Diffusion prompt engineering.
The key to writing good prompts is being descriptive, specific, and providing relevant details and context to guide the AI model. Prompts should communicate not just the subject matter but also the desired style, composition, lighting, angle, and other aesthetic qualities of the image.
Most Stable Diffusion prompts follow a simple structure:
[Subject], [Style], [Details]
A still life painting of fruit in a bowl, oil on canvas, by Paul Cézanne
Let’s explore prompt examples for different categories and use cases.
Portrait prompts should describe facial features, expressions, angle, framing, and lighting:
A high resolution close up portrait of a beautiful young woman with long blonde hair, bright green eyes, rosy cheeks, and a sweet smile, soft lighting
Landscape prompts can specify geographic location, time of day, weather conditions and other details:
A misty autumn morning in Kyoto, Japan, golden sunrise lighting up red maple trees and a tranquil lake, depth of field
Conceptual and Surreal
For imaginary or metaphorical images, set the theme or concept first:
A symbolic image depicting the chaos of modern life, surreal abstract mixed media art
Logos and Graphics
Include the company name, industry, and visual style for logos:
A modern and minimalist logo for a machine learning startup, using a stylized hexagon shape
Advanced Prompting Techniques
You can use special keywords and syntax to control other aspects of the generated image:
Image Size and Aspect Ratio
A cute baby sloth hanging from a tree branch, National Geographic photography, 4k
Style and Medium
Obama portrait in the style of a Renaissance oil painting
Viewpoint and Perspective
A city skyline viewed from above, aerial perspective
There are many more advanced prompt engineering methods like insertion, masking, image embeddings, aspect guidance and more.
Here are some great sites to find and share Stable Diffusion prompts:
- PromptBase – Marketplace to buy and sell prompts
- PromptHero – Search engine for top prompts
- Stability AI – Official Stable Diffusion blog
- Lexica – Collaborative prompt wiki
- Reddit – Stable Diffusion subreddit
Prompt Engineering Challenges
While Stable Diffusion can produce amazing images, it has some limitations to be aware of:
- Repetitive outputs – The AI can get stuck generating very similar images. Varying prompts helps.
- Bias – Like any AI, Stable Diffusion reflects biases in training data. Mitigations are being developed.
- Coherence – Images can sometimes have distorted faces or objects. Using more descriptive prompts reduces this.
With practice, you will get better results by crafting prompts tailored to the strengths and weaknesses of Stable Diffusion.
I hope these AI prompt examples spark your creativity with Stable Diffusion! Start simple, be descriptive, and don’t be afraid to experiment. Bookmark prompts that work well so you can reuse and build on them. Prompt engineering is an iterative process, have fun with it!