Veo 3.1 TikTok workflows work best when you generate vertical 9:16 clips first—not landscape videos you crop later. Google’s Veo 3.1 can turn text or images into short video with native audio (sound effects, ambience, and dialogue where it holds up). On Movby.ai’s Veo page, you can run that model as Text to Video or Image to Video, set 9:16, pick 720p or 1080p, and often get a reviewable clip in about two minutes. This guide is for creators, marketers, and freelancers who need Reels-ready footage without a crew—plus 10 copy-paste vertical prompts.
Why 9:16 matters (with data)
Short-form video is where attention already is. In Pew Research Center’s 2025 surveys, 37% of U.S. adults said they use TikTok (up from 21% in 2021), and 63% of adults under 30 report using the app (Pew Research Center, March 2026 summary of 2025 data). In a separate Feb–March 2025 Pew survey, 24% of U.S. adults said they use TikTok daily (Pew Research Center, Americans’ Social Media Use 2025).
Engagement is still competitive. Socialinsider’s analysis of about 2 million TikTok videos (Jan 2024–Dec 2025) found average engagement rate by views rose from 3.85% in 2024 to 4.20% in 2025, while brand posting frequency rose about 40% year over year—more supply fighting for the same scroll (Socialinsider, 2026 TikTok Benchmarks).
Completion is brutally hard. Dash Social’s H1 2025 TikTok benchmark sample reported that viewers typically watch about 30% of a video and that only about 10% finish a full TikTok (Dash Social, Social Media Benchmark Report — TikTok, 2025 H2). Note: the public PDF path above may redirect; the report’s TikTok section is the cited source for the 10% / 30% figures.
Format reality: Instagram’s help docs allow Reels between 1.91:1 and 9:16, with a minimum of 720p and 30 FPS (Meta / Instagram Help: Reel size & aspect ratios). For full-screen Reels placements, Meta’s ads guidance recommends 9:16 and keeping critical elements out of UI-heavy edges (commonly cited safe margins: leave clear space at top, bottom, and sides so captions/profile controls don’t cover faces or products) (Meta Ads Guide — Instagram Reels video).
Why native 9:16 beats cropping 16:9: Landscape-first AI clips put the subject in a wide frame. Force-cropping to vertical often chops heads, hides the product, or leaves dead space. A Veo 3.1 9:16 prompt that frames for a phone from the start keeps the subject centered, preserves headroom for TikTok/Reels UI, and matches how people actually hold their phones.
What you need
Short checklist:
- Account on Movby.ai (new accounts get 30 credits on signup; no card required to start—confirm on pricing)
- One clear idea: hook + single visual payoff (not a five-scene story)
- Optional still for Image to Video (product packshot, selfie, room photo)
- Caption plan: first line = the scroll-stop; CTA secondary
- Optional: CapCut / Instagram editor for burned-in captions after export
Step-by-step: Movby.ai + Veo 3.1 for TikTok / Reels
You can do this in under 15 minutes once you’ve written the prompt.
-
Open Veo 3.1 at movby.ai/ai-video-generator/veo. Movby.ai is an independent all-in-one AI image & video platform that provides access to Google’s Veo 3.1.
-
Pick a mode
- Text to Video — concept from scratch (UGC ad vibes, POV, mood)
- Image to Video — animate a still you already trust (product, outfit, location)
- Use Text to Video / Image to Video hubs if you’re comparing models; for this workflow stay on the Veo page.
-
Set output for short-form
- Aspect ratio: 9:16
- Resolution: 720p (faster tests) or 1080p (sharper posts)
- Confirm duration and any other toggles on the current Veo settings UI—options can change; don’t assume 4K or a specific second length unless you see it selected.
-
Write a vertical-first prompt (use the formula below). Explicitly say vertical 9:16 framing, subject position, and “leave headroom for UI overlays.”
-
Generate and wait. Most Veo 3.1 runs on Movby.ai finish in about two minutes, depending on prompt, settings, and queue (Movby.ai Veo FAQ).
-
Review on a phone-sized preview. Check: face/product in the middle third, no baked-in text, motion readable in the first second.
-
Iterate one or two variables only (camera move or lighting or subject action—not all three). Re-roll until the hook frame works.
-
Download → light edit → post. Add captions, a cover frame, and a CTA in CapCut or the native editor. Post natively when you can; cross-post second.
For a wider model menu later, browse the main AI video generator.
Vertical prompt formula (reusable)
Use this skeleton for every Veo 3.1 9:16 prompt:
[shot type], [subject + action], [setting], [camera move], [lighting], [style],
[duration feel], vertical 9:16 framing with subject centered and headroom for UI,
[audio/mood], [negatives: no on-screen text, no watermarks, no deformed hands, no letterboxing]
| Part | What to write | Why |
|---|---|---|
| Shot type | Close-up, medium, POV, slow orbit | Locks framing before the model invents a wide establishing shot |
| Subject + action | Who/what + one verb (“applies serum,” “pours glaze”) | One action reads in 1–2 seconds |
| Setting | Bathroom mirror, street at dusk, kitchen counter | Grounds physics and reflections |
| Camera move | Slow push-in, gimbal follow, locked tripod | Prevents random whip pans |
| Lighting | Soft window light, neon, harsh noon | Strongest quality lever after framing |
| Style | Clean UGC, editorial, documentary | Keeps look consistent across a series |
| Duration feel | “single continuous 8-second beat” | Discourages mid-clip scene jumps |
| Vertical framing | Centered, headroom, no letterboxing | Phone-native composition |
| Audio/mood | Soft ASMR sizzle, city hush, upbeat bed | Veo can generate native audio; describe it |
| Negatives | No text, logos, watermarks, extra fingers | Reduces common AI clutter |
Google model note: Per Google DeepMind and the Google Developers Blog (Oct 15, 2025), Veo 3.1 generates video with richer native audio—conversations, synchronized SFX, ambient sound—and improved image-to-video adherence (Google Developers Blog; DeepMind Veo). DeepMind also states that natural, consistent spoken dialogue—especially short speech—remains an area of active development. Treat dialogue prompts as experimental; prefer SFX + ambience for client-safe posts.
10 copy-paste 9:16 prompt templates
1) Beauty / skincare — product texture tease
Prompt: Extreme close-up, vertical 9:16, glass dropper releasing a single clear serum droplet onto the back of a hand, slow macro push-in as the droplet spreads into a glossy sheen, soft bathroom window light, shallow depth of field, clean UGC beauty aesthetic, continuous 8-second beat, subject centered with headroom for UI, gentle ASMR liquid drip and soft room tone, no on-screen text, no watermarks, no logos, no deformed fingers.
Hook caption: “The texture check before I commit to a serum…”
Use: Product tease / ASMR loop opener
2) Food — skillet sizzle
Prompt: Medium close-up, vertical 9:16, chef’s hands tossing glazed vegetables in a hot wok, steam and oil shimmer rising toward camera, handheld but stabilized gimbal feel, warm kitchen practical lights, documentary food-commercial style, single continuous motion, leave safe margins at top and bottom, loud sizzle and metallic wok clank, no captions burned in, no brand logos, no letterboxing.
Hook caption: “If you hear this sizzle, dinner’s already winning.”
Use: Food POV / retention loop
3) Fashion — haul vibe (mirror try-on)
Prompt: Full-body mirror shot, vertical 9:16, young adult turning slowly in a fitted trench coat over jeans, natural phone-mirror framing with phone edge subtly out of frame, soft overcast daylight through a window, candid fashion-haul UGC style, slow 180° turn then pause on silhouette, centered subject with headroom, quiet fabric rustle and city ambience outside, no on-screen text, no watermarks, realistic hands and proportions.
Hook caption: “Outfit math: one coat that makes everything look expensive.”
Use: Fashion haul vibe / GRWM B-roll
4) Fitness — form check
Prompt: Medium side-angle shot, vertical 9:16, athlete performing controlled dumbbell Romanian deadlifts in a bright home gym, smooth stabilized camera locked on hip hinge form, hard but flattering overhead gym lighting, crisp athletic instructional look, continuous single set pacing, full body visible with headroom, gym ambience and measured exhales (no long dialogue), no text overlays, no deformed limbs, no shaky handheld chaos.
Hook caption: “Fix your hinge before you add weight.”
Use: Educational fitness clip / form demo
5) Travel — street POV
Prompt: First-person POV, vertical 9:16, walking through a narrow European alley at golden hour, warm stone walls and hanging laundry, gentle gimbal forward move with natural bounce suppressed, cinematic travel documentary style, continuous stroll, horizon level, leave margins for UI, footsteps on cobblestone and distant chatter, no subtitles, no watermarks, no fisheye distortion.
Hook caption: “Golden hour in an alley you’ll never find on a tour bus.”
Use: Travel POV / wanderlust hook
6) Product unboxing — desk ASMR
Prompt: Top-down to 45° hybrid angle, vertical 9:16, hands carefully opening a matte black retail box on a clean oak desk, foam insert lift revealing a minimalist gadget, slow deliberate movements, soft north-window light, premium unboxing commercial style, continuous reveal beat, product stays in center safe zone, cardboard and foam ASMR textures, no logos invented on packaging text (blank or abstract markings), no burned-in captions, no extra fingers.
Hook caption: “The satisfying part is never the gadget—it’s the foam.”
Use: Unboxing / DTC ad opener
7) SaaS / app — metaphor demo
Prompt: Clean desk scene, vertical 9:16, smartphone on a stand showing a glowing abstract dashboard UI (no readable words), a hand taps once and soft neon charts animate upward as floating light particles, slow push-in, cool blue accent lighting in a dark office, modern tech-commercial style, single continuous gesture-to-payoff, UI remains illegible by design, subtle UI chime and soft synth bed, no real brand names, no readable text, no watermarks.
Hook caption: “What ‘saving 2 hours a week’ looks like in one tap.”
Use: SaaS metaphor / paid social tease
8) Real estate — mood walkthrough
Prompt: Slow dolly forward, vertical 9:16, entering a sunlit empty living room with wide oak floors and linen curtains drifting, afternoon light stripes across the floor, architectural mood-film style, continuous walk-in, furniture sparse and centered in frame, soft curtain flutter and quiet room tone, no text, no agent lower-thirds, no fisheye, no warped walls.
Hook caption: “When the light does the selling for you.”
Use: Listing mood / realtor Reels
9) Pet — joyful reaction
Prompt: Medium shot at dog-eye level, vertical 9:16, golden retriever tilting its head then excitedly spinning as a treat appears from frame bottom, sunny backyard bokeh, handheld-stabilized pet-UGC style, continuous playful beat, dog face centered with headroom, happy panting and soft collar jingle, no on-screen text, no watermarks, natural fur detail, no distorted snout.
Hook caption: “The head tilt before the chaos.”
Use: Pet entertainment / brand-adjacent cuteness
10) Educational explainer — cinematic micro-lesson
Prompt: Macro to medium transition, vertical 9:16, a clear glass of water with a single drop of colored ink blooming into swirling patterns, locked tripod then subtle push-in, bright soft studio light on white seamless backdrop, clean educational-cinematic style, continuous diffusion process as visual metaphor, subject centered, soft liquid ambience only (no lecture dialogue), no diagrams, no on-screen labels, no watermarks.
Hook caption: “Diffusion in 8 seconds—watch the ink take over.”
Use: Educational explainer / science metaphor
Prompt iteration cheat sheet
| Problem | Fix in the next prompt |
|---|---|
| Face or product cut off | “full subject centered, headroom for UI, vertical 9:16, no tight crop on forehead” |
| Looks like cropped landscape | “native vertical composition, no letterboxing, no widescreen framing” |
| Shaky / nauseating motion | “smooth stabilized gimbal, slow push-in, no whip pan” |
| Too many scene changes | “single continuous shot, one action only” |
| Unreadable AI text appears | “no on-screen text, no captions, no logos, no subtitles” |
| Hands look wrong | “anatomically correct hands, five fingers, negative: deformed hands” |
| Dark / muddy | “soft window light, exposure for face/product, gentle fill” |
| Subject too small | “medium close-up, fill 60% of frame vertically” |
| Dialogue sounds off | Drop spoken lines; use “ambient + SFX only” (spoken audio still variable per Google’s notes) |
| Watermark-looking artifacts | “no watermarks, clean edges” + regenerate; don’t fight it in CapCut first |
Posting for TikTok vs Reels
Shared rules (evidence + practice): Export 9:16. Keep faces/products inside the center safe area so UI chrome doesn’t cover the payoff. Because completion rates are low on TikTok (Dash Social’s ~10% full-watch figure above), design the first second as the entire thesis—then let the rest reward staying.
TikTok (practical):
- Captions often work punchier and more conversational.
- Hashtags: commonly recommended by creators to use a small set (topic + niche), not a block of 20—treat as best practice, not a platform law.
- Native upload usually preserves quality and avoids watermark issues from other apps.
- Cover: pick the sharpest mid-action frame before posting.
Instagram Reels (practical + docs):
- Stay within Instagram’s allowed aspect range; prefer full 9:16 (Instagram Help).
- Cover photo guidance from Meta lists 420×654 as a recommended cover size (same Help article)—design a still that reads as a tiny poster.
- Captions can skew slightly cleaner/brand-safe; keywords in the first line still help discovery.
- Cross-posting TikTok→Reels is common, but creators often report better distribution when posted natively (best practice; not a published Meta/TikTok guarantee).
Length heuristic (creator best practice, not a universal study): Use AI clips as hook modules (often one short generation), then stitch 2–3 in CapCut if you need a 15–30s story. Don’t ask one generation to be a full tutorial.
Common mistakes
- Landscape-first prompts (“cinematic wide shot of…”) then cropping—subjects die in the crop.
- Multi-scene scripts in one generate—Veo shines on one continuous beat.
- Asking for readable on-screen text—add captions in post instead.
- Ignoring safe zones—hooks hidden under username / CTA bars.
- Over-editing before testing the hook—if second 0–1 fails, color grade won’t save it.
- Changing five prompt variables at once—you won’t know what fixed the shot.
- Assuming free commercial rights without checking—Movby’s Veo FAQ states paid plans include commercial use; confirm current terms on pricing.
FAQ
Is Veo 3.1 good for TikTok?
Yes for short vertical hooks, product motion, POV, and mood clips—especially when you generate in 9:16. It’s weaker if you need long multi-scene stories or perfect spoken VO in one pass.
How do I set 9:16 on Movby.ai?
On the Veo 3.1 page, choose ratio 9:16 before generating (alongside 16:9). Also say “vertical 9:16 framing” in the prompt.
Text to Video vs Image to Video?
Text to Video invents the scene. Image to Video locks identity/product from your still—better for consistent packaging, faces, or rooms.
Can Veo 3.1 generate audio?
Google documents native audio (SFX, ambience, dialogue). On Movby, describe the sound in the prompt. Treat precise speech as hit-or-miss; prefer sound design for ads.
Free vs paid on Movby.ai (watermark / commercial)?
Signup includes 30 free credits with no card required to start. Movby’s Veo FAQ states paid plans include commercial use for generated videos. Always verify watermark and license details on movby.ai/pricing before client work.
How long do clips take?
About two minutes per generation on Movby.ai is typical, varying with settings and queue.
What resolution should I pick for Reels/TikTok?
Start 720p for prompt tests; export 1080p when the take is final. Instagram requires at least 720p for Reels (Instagram Help).
Try a 9:16 Veo 3.1 clip on Movby.ai with your free credits and post the best hook the same day.