Google Veo｜Photos That Talk
One sentence or one photo becomes a realistic ~8s Veo 3.1 clip with dialogue and sound. Single photos work; vertical goes straight to TikTok.
Veo AI video tool: turn one photo or sentence into a realistic ~8s clip with native audio via Google Veo 3.1 — dialogue and SFX in one pass. Try it free now.
Features
- Native dialogue audio——Dialogue, ambience and music in one pass — the duck interrogation and the rain need no dubbing.
- One photo performs——Upload a phone photo and the person looks into the lens, blinks and talks — still recognizably them.
- First-to-last frames——Upload opening and ending frames; Veo fills a smooth 8 seconds between them, rain and street intact.
- Landscape + vertical——16:9 cinematic or 9:16 phone vertical — post to TikTok with no re-cropping.
- Stitched 8s segments——Each clip runs about 8 seconds; longer stories generate in consistent segments and stitch together.
- Quality or Fast——Veo 3.1 Quality for detail, Fast for quick drafts — inspiration never waits in line.
How it works
- Upload or describe Got a photo? Upload it as the first frame (optionally add a last frame). Got nothing? Pure text works too.
- Pick a preset + format Photo speaks, viral vlog, candy ASMR… then 16:9 or 9:16, about 4 or 8 seconds.
- Generate with sound Hit generate — Veo 3.1 outputs picture, dialogue and SFX at once. Longer stories stitch from segments.
FAQ
- Can one photo move and talk?
- Yes. Upload the photo as the first frame and put the line in quotes, e.g. Dialogue: “Hi, it's really me!” Veo 3.1 returns about 8 seconds of the person facing the lens, blinking and talking. Front-facing, well-lit photos lip-sync best.
- How do I write Veo prompts?
- Write like a director's shot list: subject plus one small action, camera, light and texture detail, dialogue in quotes, sound on its own line. English keywords work best. One small action per clip is the number-one anti-glitch rule.
- Lips out of sync, stiff motion?
- Three fixes: cut lines short — one or two sentences at a time; use medium close-up front-facing shots, no walking while talking; write small moves (smile, blink, turn), never big ones (run, jump). For long Chinese dialogue, generate silent and dub instead.
- Is there a free Veo version?
- Yes, all capped (as of Oct 2026): the Gemini app gives a few free Veo 3 Fast clips a month, Google Flow grants starter credits, and AI Studio has trial credits. Free tiers usually watermark and queue. In Loomi you generate with Veo 3.1 directly — no extra accounts or network tricks, pay per use.
- Veo vs Sora vs Kling — which wins?
- Veo 3.1 leads on photorealism plus native audio — pick it when you want footage that looks real. Sora wins on long narrative and imagination, Kling and Runway on control and editing. For one sentence into a clip with sound, Veo 3.1 is currently the laziest option.
- Can one clip run 1–2 minutes?
- No. Veo 3.1 caps a single clip at about 8 seconds (about 4 seconds optional) — a model limit, not ours. Longer stories generate in segments inside Loomi and stitch together with matched cast, light and sound.
- How to spot fake Veo videos?
- Check four things: stable fingers and on-screen text; light and reflections obeying physics; lips matching the voice exactly; Veo or AI watermarks at the source. When in doubt, assume a too-real viral clip is Veo until proven otherwise — you will be right most of the time.
Try it free