Negative prompts are optional, but are helpful for describing what you don’t want in an image.
As of early 2026, many state-of-the-art image models, including Google's Nano Banana pro, do not support (or require) negative prompting.
For these models, when you need to specify what not to generate, you can simply include these any unwanted elements in your base prompt, followed by the word "no". For example: "Business man on a pristine beach. A wide screen horizontal composition for a high quality website hero banner. No text or typography."
That said, quite a few leading models continue to support negative prompting, as a powerful way to achieve better creative control over AI-generated imagery.
Adding a negative prompt can help you get rid of a recurring problem in image generations. For example, in the image below, we’ve generated an image of a pizza, but we don’t want to show the ingredients beside it.
Prompt: Generate a top-down photo of a large pepperoni pizza with green basil leaves on a stoneware plate in the centre of the shot. Pizza on a dark wooden table. Brightly lit, natural daylight. Medium-wide shot. Lens 85mm f/5.6.
In the negative prompt, add the elements that you want to remove, e.g. “Extra tomatoes on the table. Extra basil on the table. Additional ingredients.”
Please note that there's no need to write “do not include” or “remove” in your negative prompts, as these are implicit in negative prompting. Including words like "no", "none" or "exclude" is discouraged, as it effectively creates a double negative, (it's like telling it 'do not include do not include tomatoes!') which can confuse the image model.
Simply write the things you don’t want to see, and tweak your prompt until the AI understands what you’re looking for.
