Google has added an image editing module to its AI movie tool Flow today, deeply integrating the Gemini 2.5 Flash image model (codename Nano Banana), which supports natural language one-click background removal, subject separation, and scene replacement. It also allows direct dragging into the timeline to generate 8-second dynamic shots. The feature is available for users with the free version of Gemini and above, priced at $0.039 per image, and the enterprise-level Vertex AI is also launched simultaneously.
After users upload photos and input prompts such as "remove the background and add soft light" or "put it on the lunar surface," the model outputs PNG transparent images or composite images, retaining edge details down to individual strands of hair. In practical tests, street photos processed by Nano Banana were imported into Flow, and a "person + fantasy forest" dynamic shot was completed in 30 seconds, maintaining consistent facial and clothing textures.
Google stated that all generated images default to embedding a SynthID invisible watermark and plans to open an API batch interface next month, targeting high-output scenarios such as short videos and e-commerce posters.
