AI filmmaking tool 'Flow' gets Google's brand new video generation model

It's been a few months since Google launched its AI-powered filmmaking tool, Flow, and the upgraded Veo 3 video generation model. Filmmakers and enthusiasts have used the tool to generate over 275 million videos so far.

Google has now upgraded Flow with the freshly-baked Veo 3.1 model, which the company describes as its "most advanced video generation model." The AI model offers more narrative control, richer audio, enhanced realism in videos, stronger prompt adherence, and improved audio-visual quality when converting images into videos.

Veo 3.1 has a deeper understanding of storytelling, cinematic styles, and character interactions, according to Google. It's available as Veo 3.1 and Veo 3.1 Fast, with text-to-video and image-to-video support in horizontal (16:9) and vertical (9:16) formats.

The search giant is bringing audio support across existing features in Flow's Scenebuilder, alongside more granular control over the final output. You can use multiple reference images in "Ingredients to Video" to have more control over the characters, objects, and style of the final output, complete with audio.

You can use "Frames to Video" to supply a starting and ending image in Flow, and Veo 3.1 will create a video with smoother creative transitions. The feature uses Veo 3.1 to build on the existing audio capabilities offered by Veo 3.

The new model also powers the "Extend" feature that lets you create longer clips based on the final second of the original clip. Google said that "audio continuity is limited to that final second, making it most useful for scenes where you want to continue the background audio, such as the sound of ocean waves."

In addition, Google has baked new editing features into the AI filmmaking tool. One of them is the "Insert" option to add new elements to any scene. Flow is now capable of handling complex details like shadows and scene lighting to make those elements look more natural.

Another feature set to arrive soon is the ability to remove objects and characters from a scene. Flow will make up for the removed elements by reconstructing the background and surroundings.

Not just technical upgrades, Google is taking other routes to improve the AI filmmaking tool. It hired Belgian filmmaker and creative director Henry Daubrez, who has been associated with Flow since its early stages.

Other than Google Flow, the Veo 3.1 video generation model is also available through the Gemini API, Vertex AI, and the Gemini app. Google said it's also bringing "Ingredients to Video," "Frames to Video," and "Extend" features to the Gemini API.

Note that Google Flow is an experimental tool currently available via Google Labs and will require a Google AI Pro or Ultra subscription to unlock higher usage limits.