Google introduces Veo 3.1 and advanced editing in Flow

Google has introduced Veo 3.1, an update to its AI video generation model, along with new editing and audio tools for Flow, its audiovisual creation platform. The company says Flow has generated more than 275 million videos in its first five months.

The update responds to a specific user request: more control over what happens in each scene, without having to rely on a single attempt generated from scratch.

Audio and more control over each scene

Veo 3.1 improves generated audio, instruction following and image realism. It also aims to represent textures more accurately and turn still images into videos with more consistent audiovisual quality.

Audio is now coming to features that already existed in Flow, although Google warns that these capabilities are experimental and still in development. They include:

• Ingredients to Video: lets you use several reference images to define characters, objects and style. For example, you can provide an image of a character, one of a vehicle and another showing the visual look you want to create a single scene.
• Frames to Video: uses a starting image and an ending image to generate the transition between them. This can be used to create scene changes or more controlled visual movement.
• Extend: expands a video and continues the action from its final second. Google says it can create sequences of one minute or more, although the feature is particularly designed to extend establishing shots.

In practice, these features bring Flow closer to an editing process like the one used in audiovisual production: first you define the scene, then you adjust the camera movement and finally you work on continuity.

Inserting and removing elements

Flow also adds new tools for correcting a scene without having to generate the entire thing again.

With Insert, you can add objects, details or even fantasy creatures to an existing video. The tool attempts to adapt shadows and lighting so the added element fits with the rest of the image.

Google is also preparing a feature to remove objects or characters. Once available, Flow will reconstruct the background and surroundings to make it look as though the element was never there.

This difference matters. If a scene has the right framing but includes an unwanted object, you will no longer have to start over. You will be able to modify only the problematic part.

Where Veo 3.1 will be available

Veo 3.1 will be available through the Gemini API for developers, Vertex AI for businesses and the Gemini app. The new capabilities will also come to the Gemini API and Vertex AI, according to Google.

The most important change is not simply that AI can generate longer or more realistic videos. It is beginning to offer more detailed control over the result: what appears, how the scene develops, what sound accompanies the action and which elements can be corrected afterward.

The audio and editing tools may still change as Google gathers feedback. What to watch now is whether this greater control produces videos that are genuinely coherent from beginning to end, rather than just individually striking scenes.