Is this the native Veo 2 API?

Not exactly — full disclosure: Veo 2 is the previous generation (December 2024), and Loomi has no native Veo 2 endpoint. This tool is fulfilled by the newer same-family Veo 3.1, covering all Veo 2 uses — text-to-video and image-to-video — with stronger realism. Every case on this page is a real render from that pipeline.

Can one photo become a video?

Yes — that is image-to-video's signature move. Your photo becomes frame one: faces, clothing and composition locked, only expressions, hair, water and clouds set in motion. Old prints, portraits and product shots all work — 'bring old photo to life' is exactly this.

Do the videos have sound?

No. The Veo 2 generation has no native audio — that is the biggest difference from Veo 3. You get a clean 5–8s silent clip, which is a feature for editing: add music or voiceover later in about a minute.

Veo 2 vs Veo 3.1 — difference?

One line: Veo 2 is the older model (famous for cinematic physics, no native audio); Veo 3.1 is the newer same-family model with better quality and stability. If you specifically want the old model's free quota, the official doors are Gemini, AI Studio and Flow; if you want maximum quality right now, generate here.

Hands deform — how to fix?

Three tricks. First, slow the hands down — write 'slowly lifts the cup', never big fast gestures. Second, reduce hand acting — hands in pockets, holding a fixed object, or wider framing. Third, lock the hand pose first with image-to-video, then move. Render several variants and keep the steadiest; avoid crowds and heavy background text.

How long and how much?

A 5–8s clip takes roughly half a minute to two minutes, including multi-variant best-pick. After login each render costs about 30 credits; complex shots auto-render parallel variants. Start with a 5s test for framing, then run the 8s final.

Spot an AI-made fake?

Slow the footage down and check fingers, teeth and background text for warping; watch shadow directions and objects vanishing mid-shot. Official Google renders carry an invisible SynthID watermark verifiable with detection tools. Explosive news, disaster or celebrity clip? Do not forward — verify with authoritative media first.

Veo 2 Video Generator ｜ Upgraded Veo

One sentence or one photo becomes a realistic short. Veo 2 has been upgraded — fulfilled by the new-gen Veo 3.1, vertical-ready in 5–8s.

Veo 2 is Google's cinematic video AI. This tool runs on the newer same-family Veo 3.1, covering all Veo 2 text-to-video and image-to-video uses. Try it free.

Features

One sentence, one clip——No editing, no setup. Write one sentence in plain words and get a 5–8s cinematic short.
Photos start moving——Upload a first frame to set the tone: blinking eyes, drifting clouds, flowing water. Old prints and portraits come alive.
Cinematic camera moves——Slow push-ins, tracking shots, aerials — realistic light and physics, B-roll you can cut straight in.
Vertical, ready to share——Pick 16:9 or 9:16 before you render. The vertical cut is made for TikTok, Reels and group chats.

How it works

Upload an old photo Dig out the family portrait or vintage print. It becomes frame one: faces, clothes and composition locked, nothing drifts.
Apply the preset Hit “bring old photo to life” — duration and style auto-set: blinks, nods, swaying hair, with slow-hand anti-morph baked in.
Export the living cut Render several takes, keep the most natural: check the blinks, the smile, the steady background — then score it and send to the family chat.

FAQ

Is this the native Veo 2 API?: Not exactly — full disclosure: Veo 2 is the previous generation (December 2024), and Loomi has no native Veo 2 endpoint. This tool is fulfilled by the newer same-family Veo 3.1, covering all Veo 2 uses — text-to-video and image-to-video — with stronger realism. Every case on this page is a real render from that pipeline.
Can one photo become a video?: Yes — that is image-to-video's signature move. Your photo becomes frame one: faces, clothing and composition locked, only expressions, hair, water and clouds set in motion. Old prints, portraits and product shots all work — 'bring old photo to life' is exactly this.
Do the videos have sound?: No. The Veo 2 generation has no native audio — that is the biggest difference from Veo 3. You get a clean 5–8s silent clip, which is a feature for editing: add music or voiceover later in about a minute.
Veo 2 vs Veo 3.1 — difference?: One line: Veo 2 is the older model (famous for cinematic physics, no native audio); Veo 3.1 is the newer same-family model with better quality and stability. If you specifically want the old model's free quota, the official doors are Gemini, AI Studio and Flow; if you want maximum quality right now, generate here.
Hands deform — how to fix?: Three tricks. First, slow the hands down — write 'slowly lifts the cup', never big fast gestures. Second, reduce hand acting — hands in pockets, holding a fixed object, or wider framing. Third, lock the hand pose first with image-to-video, then move. Render several variants and keep the steadiest; avoid crowds and heavy background text.
How long and how much?: A 5–8s clip takes roughly half a minute to two minutes, including multi-variant best-pick. After login each render costs about 30 credits; complex shots auto-render parallel variants. Start with a 5s test for framing, then run the 8s final.
Spot an AI-made fake?: Slow the footage down and check fingers, teeth and background text for warping; watch shadow directions and objects vanishing mid-shot. Official Google renders carry an invisible SynthID watermark verifiable with detection tools. Explosive news, disaster or celebrity clip? Do not forward — verify with authoritative media first.

Try it free

2026-10-27