Negative prompt
Also called negative prompting
What is a negative prompt?
A negative prompt is a list of things you want an AI image or video model to avoid in its output. Where the main prompt describes what to include, the negative prompt names what to exclude, such as 'text, watermark, blurry, extra fingers, low quality'. The model steers each generation step away from those concepts.
Negative prompts are common in tools built on open diffusion models, where they appear as a separate field. Some newer hosted image tools do not have a negative prompt field and expect you to describe what you want positively instead.
Use them sparingly. A long list of generic negatives can flatten the image or have little effect. Targeted negatives that fix a problem you actually saw tend to work best.
Why a negative prompt matters
For marketing visuals, the most common problems are unwanted text, logos, watermarks and distorted people. A short negative prompt removes many of these and saves regenerations.
It also helps keep a set of images on-brand, for example by excluding colors that clash with your palette.
How to use a negative prompt
- Generate an image with only the main prompt first.
- Note what is wrong: stray text, clutter, wrong colors, distortions.
- Add only those items to the negative prompt.
- Regenerate and compare, keeping the seed the same if you can.
Negative prompt example
Prompt: flat lay of yoga props on a light oak floor, soft daylight, sage and cream tones
Negative prompt: text, letters, logo, watermark, people, clutter, dark shadowsNegative prompt FAQ
Do negative prompts work in every AI image generator?
No. Many open-model tools support them, but some hosted generators ignore negatives or have no field for them. In those, describe what you want instead.
What should I put in a negative prompt?
Only the problems you actually see in your results, such as text, watermark, blur or extra limbs. Short and targeted works better than a long generic list.
Related terms
A prompt is the instruction or input you give an AI model, such as a question, a task or a description of the image you want it to create.
Prompt engineering is the practice of designing and testing prompts so an AI model gives accurate, useful and consistent output.
Text to image is AI that turns a written description into a picture. You type a prompt and the model generates matching images.
A diffusion model is a type of generative AI that creates images, video or audio by starting from random noise and removing it step by step.
An AI image generator is a tool that creates images from a text description, and sometimes from a reference image, using a trained AI model.