What Is the Gemini Watermark?

The visible Gemini watermark is a semi-transparent sparkle-shaped logo placed near an image corner on some Google AI image outputs. It is separate from SynthID, Google DeepMind’s invisible watermarking technology.

What does the visible watermark look like?

The mark is a small white four-point sparkle with a semi-transparent alpha edge. It is composited over the image pixels rather than stored as a separate editable layer.

Observed size and position

These are bounded observations from one Gemini Web account on July 15, 2026, not a universal specification. This site offers reconstruction only for those exact decoded sizes.

Which plans and surfaces add it?

Google’s November 2025 Nano Banana Pro announcement says images generated by free and Google AI Pro users retain the visible sparkle, while Google AI Ultra subscribers and the Google AI Studio developer tool receive a clean visual canvas without that visible mark. Google still applies provenance technology such as SynthID, and newer model announcements also discuss C2PA Content Credentials. Plan behavior can change, so verify the current Google product terms for the surface you use.

Sources: Google — Introducing Nano Banana Pro and Google — Nano Banana 2.

Visible logo versus SynthID

The visible logo is an overlay a person can see. SynthID is designed as an invisible signal in generated content. Removing a corner logo does not establish that SynthID is gone, and this tool does not attempt to detect, remove, or alter it.

Can the visible mark be removed?

If a known semi-transparent overlay is at an exact known position, reverse alpha blending can reconstruct the affected region without generative repainting. It cannot recover information already lost to resizing or lossy compression, which is why this tool shows the exact output before download and always keeps crop available.

Is removing it legal?

There is no blanket worldwide answer. Only edit images you own or are authorized to edit. Google’s Generative AI Prohibited Use Policy bars deceptive provenance claims, including claiming generated content was created solely by a human in order to deceive. In the United States, 17 U.S.C. §1202 regulates some removal or alteration of copyright management information; whether a particular mark qualifies and whether the statutory elements are met depends on the facts. Other jurisdictions can impose AI-content transparency duties.

Primary references: Google Generative AI Prohibited Use Policy, 17 U.S.C. §1202, and EU Regulation 2024/1689. This is general information, not legal advice.

Can Google provide an image without the visible mark?

Google’s published Nano Banana Pro plan description identifies Google AI Ultra and Google AI Studio as official routes without the visible watermark. That can be preferable to editing a marked export when the current plan and tool fit your workflow. We compare every option — plans, developer surfaces, local reconstruction, and crop — in how to get a Gemini image without the watermark.