Upload multiple reference images to guide characters, objects, and style. Generate vertical outputs for mobile-first storytelling.
8s
Max clip duration
3
Reference images
48kHz
Audio sample rate
24fps
Frame rate
Reference images
Character and style direction
Use multi-image references to direct scene consistency across characters, props, and tone.
Veo 3.1 does not bolt audio on after generation. Dialogue, music, ambient sound effects, and environmental layers are all synthesised in a single unified pass — timed at the model level with 10ms audio-video synchronisation precision. The result is sound that feels physically present in the scene, not layered on top of it.
Dialogue & Voice
Realistic speech with proper lip sync and intonation.
Music & Score
Generative background music matched to scene mood.
Ambient SFX
Environmental soundscapes layered with precise timing.
10ms Sync
Industry-leading audio-to-video synchronisation precision.
Audio timeline — Scene 01
48kHz · Stereo · AAC 192kbps
Choose the creation path that fits your workflow — start from pure text, anchor with a reference image, or interpolate between two keyframes.
Describe your scene in natural language. Veo 3.1 interprets camera direction, motion style, character behaviour, and atmosphere from a single prompt — producing a complete 8-second clip with native audio in one pass.
Lock character identity, visual style, and environment simultaneously. The more references you provide, the tighter the consistency across every frame.
Single reference
Lock a character or object. The model anchors identity and appearance across the full clip.
Prompt-driven examples inspired by the same storytelling style shown on the Gemini Veo experience.
Moonlit forest follow shot
A follow shot of a wise old owl above moonlit clouds, circling a forest clearing, then diving beside a nervous badger on a quiet path. Include layered ambient audio.
Result: coherent camera motion, atmospheric woodland tone, and dialogue-friendly scene continuity.
Opera cat with full orchestra
A cat singing opera with full orchestra, expressive close-up performance, dramatic lighting, and playful emotional beats.
Result: stylized, meme-ready musical motion with character focus and synchronized energy.
Elderly sailor spaghetti sequence
Eye-level medium shot of an elderly sailor by the pier, lifting spaghetti from a white ceramic plate in warm natural light with subtle ocean ambience.
Result: cinematic realism, consistent framing, and strong continuity across multi-step action.
For Exploring
Play with styles, animate characters, and combine ideas you normally could not test quickly.
Common specs
Veo 3.1 Fast
Google AI ProCreate videos with sound using a model optimized for speed while keeping high visual quality.
SynthID watermarking — AI generation transparency
Every clip produced by Veo 3.1 carries an invisible SynthID watermark embedded by Google DeepMind. The watermark is imperceptible to viewers but machine-readable, enabling provenance verification at any point in the content lifecycle.
Native audio. Up to 3 reference images. First + last frame mode. 1080p output. Every generation ready to publish.