What Is the Gemini Nano Banana Watermark? Everything You Need to Know

If you've generated images with Google Gemini, you may have noticed a small, subtle mark in the bottom-right corner of your images. The internet has affectionately nicknamed this the "nano banana" watermark due to its shape. Here's everything you need to know about it.

What is the nano banana watermark?

The "nano banana" is a small, semi-transparent watermark that Google Gemini adds to every AI-generated image. It typically includes:

• A small stylized logo or symbol
• The text "Gemini" or a similar identifier
• A semi-transparent overlay in the bottom-right corner

The community calls it a "nano banana" because certain versions of the mark have a vaguely banana-shaped curve to them.

Why does Google add it?

Google adds visible watermarks to Gemini-generated images as part of their responsible AI initiative. The goals are:

1. Transparency: Viewers can tell the image was AI-generated
2. Provenance: Combined with C2PA metadata, the origin of the image is traceable
3. Misinformation prevention: Makes it harder to pass off AI images as real photographs

Is the nano banana watermark always in the same place?

Yes. The watermark is deterministically positioned at:

• Horizontal: Approximately 22% from the right edge, with a 1.5% margin
• Vertical: Approximately 8.5% from the bottom edge, with a 1.5% margin

This consistency is what makes automated removal possible — we know exactly where to look.

How to remove the nano banana watermark

Since the position is fixed, specialized tools can automatically detect and cleanly remove the watermark using AI inpainting. Gemini Watermark Remover does exactly this:

1. Detects the watermark region automatically
2. Generates a mask around the watermark
3. Uses LaMa AI or OpenCV to fill in the area naturally
4. Returns the clean image at original resolution

The process runs entirely in your browser — your images never leave your device.

Does removing the visible watermark remove all traces?

No. Google also embeds C2PA metadata inside the image file. This is invisible data that persists even if you remove the visible watermark. Tools that check C2PA provenance (like Content Credentials) can still identify the image as AI-generated.

Removing the visible watermark only removes the visual overlay. If full provenance removal is needed, the metadata would need to be stripped separately — though this raises ethical considerations around AI content transparency.