Stable Diffusion Prompts and Negative Prompts Explained
How tag-style prompts work, what negative prompts are for, and how to keep both short and effective.
5 min read · Updated Nov 6, 2026
Tag-style prompts
Many Stable Diffusion workflows use comma-separated tags rather than sentences. Order matters: tags near the start tend to carry more weight, so put the subject and style first.
portrait of an elderly fisherman, weathered skin, knitted wool hat, overcast harbor, soft diffused light, muted blue-gray palette, 85mm, shallow depth of field
Negative prompts
A negative prompt lists what you don't want. It's most useful for recurring defects and unwanted elements rather than as a long checklist.
blurry, low quality, watermark, text, extra fingers, distorted face
Weighting
Interfaces such as AUTOMATIC1111 and ComfyUI support emphasis syntax like (word:1.2) to strengthen a term or (word:0.8) to soften it. Use it sparingly — heavy weights often cause artifacts.
Keep it focused
- Remove tags that contradict each other (“night” and “bright midday sun”).
- Prefer a few precise tags over many vague ones.
- Change one thing per test so you can see its effect.
Put it into practice — free, no sign-up.
Get SD tags and a negative prompt from an image