Veo 3.1 Brings Surgical Video Editing to Flow: Insert, Remove, and Reshoot Without Re-rendering
Google’s Veo 3.1 adds object-level in-scene editing to Flow: insert or remove elements, doodle-prompt frames, and reshoot camera angles without re-rendering the full clip.
Google just made the most frustrating part of AI video creation significantly less painful. With Veo 3.1 and a matching update to its Flow filmmaking tool, you can now insert or remove objects from a generated clip, doodle-prompt a frame to adjust characters, and reshoot a scene with a different camera angle, all without scrapping the whole thing and starting over. For anyone who has spent 20 minutes generating a beautiful clip only to notice a misplaced lamppost, this is genuinely good news.
Google responded directly to creator feedback with the Veo 3.1 update. The company heard users wanted “more artistic control within Flow, with increased support for audio across all features.” The result is a cluster of precision editing tools that, combined with the model’s improved audio and realism, push Veo closer to an actual production workflow and further from a novelty prompt box.
What’s Actually New in Veo 3.1
The headline capability is object-level in-scene editing. Veo 3.1 lets users add an object to a video and have it blend into the clip’s style. The Remove Object feature, which lets you surgically delete an unwanted element, is rolling out soon. These tools work directly inside Flow’s interface without touching anything else in the scene. The feature is still experimental and works best when the target object isn’t moving much, which is worth keeping in mind before you try to vanish a running actor.
The other new editing tools round out the package. A doodle-prompting feature lets you draw directly on a frame and Flow interprets your sketch, letting you shift a pose, adjust a hairstyle, or drop in accessories without typing a single word. On the camera side, Flow now supports more camera motion presets so you can reshoot a scene with a new perspective instead of regenerating from scratch. First and Last Frame control, which lets you define the opening and closing image of a clip and have Veo 3.1 generate the motion in between, is also available via the Gemini API, Vertex AI, and Flow.
Veo 3.1 Is the Full-Stack Upgrade, Not Just an Edit Layer
The editing tools arrive alongside a broader model upgrade. Veo 3.1 generates richer native audio, from natural conversations to synchronized sound effects, and offers greater narrative control with improved understanding of cinematic styles. Critically, Veo 3.1 now adds audio to previously audio-free features like Scene Extension, Ingredients to Video, and Frames to Video, which were already available in Veo 3 but previously generated silent clips.
The model family has also expanded. Google offers three tiers: Veo 3.1 for top-tier visual fidelity, Veo 3.1 Fast for standard production workflows, and Veo 3.1 Lite for high-volume, cost-effective applications. All three tiers include native audio generation. A Veo upscaling capability is also in private preview, letting you push existing videos up to 1080p or 4K regardless of whether they came from Veo, another AI model, or a real camera.
Here are a few prompts to try once you get into Flow or Google AI Studio with Veo 3.1:
A busy coffee shop interior, morning light through large windows, a barista preparing espresso behind the counter. Cinematic 16:9, warm tones, ambient café sounds.Insert a vintage red bicycle leaning against the brick wall on the left side of the frame, matching the existing shadow and lighting of the scene.First frame: a lone hiker standing at the edge of a forest at dawn. Last frame: the same hiker emerging at a mountain summit in golden hour light. Generate the journey between them, cinematic slow motion, natural wind sounds.Extend this clip: continue the motion of the ocean waves breaking on the shore, preserving the ambient sound and maintaining the same camera angle for 7 additional seconds.Where You Can Use It
Veo 3.1 is rolling out to Flow, the Gemini app, and Vertex and Gemini APIs. The object editing features, however, are not fully exposed in the public API yet. Full editing features like insert and remove object may not be available via the public API at the time of writing, so Flow is currently the best place to actually use them. You’ll need a Google AI Pro subscription to access the new models in Flow.
Flow itself has grown faster than most expected. Users have created more than 500 million videos in Flow since its May launch. That’s a platform that clearly needed better editing tools, and now it’s starting to get them.
Why It Matters for Creators
The bigger shift here isn’t any single feature. It’s that Veo 3.1 stops treating video generation as a one-shot gamble where a flawed result means starting over. This kind of surgical editing is rare in AI video generation, and it signals a shift toward more creator-level control. The features are still experimental, the object removal is still rolling out, and edge cases will inevitably trip up the model. But the direction is clear: Google is building AI video that you can iterate on, not just regenerate and hope for the best.
