Stable Diffusion prompt basics
Diffusion workflows vary by model, checkpoint, and interface. A descriptive positive prompt tells the model what to depict; a negative prompt can discourage unwanted properties in workflows that support it.
Positive prompt example
A handmade leather notebook on a wooden desk, overhead product photograph, warm morning sunlight, realistic material texture, clean composition, subtle shadows, high-detail editorial aesthetic
Negative prompt example
blurry, low resolution, duplicate objects, distracting watermark, distorted geometry, unreadable text
Settings are starting points, not rules
Sampling steps, guidance scale, resolution, and negative-prompt support differ across models and interfaces. The settings suggested by this website are illustrative defaults, not universal recommendations. Check your model documentation before using them.
A practical iteration workflow
- Start with a concise subject and composition.
- Generate a baseline image.
- Adjust one visual attribute or setting at a time.
- Compare results and keep the details that improve the outcome.
Draft a prompt in the Stable Diffusion mode of our free tool.