AI Video Prompts for Instagram: 20 Copy-Paste Prompts for Reels, Stories and Shoppable Video
AI video prompts for Instagram work best when they encode the platform’s constraints instead of describing a pretty picture: 9:16 framing, a 1–3 second hook, motion written as subject action + camera move, and a safe-zone budget that keeps text out of the UI gutters. Below you’ll find 20 ready-to-paste prompts across Reels aesthetics, Stories, shopping-tag video and IG ad specs — plus 10 fill-in-the-blank hook templates and the four metrics that tell you a prompt family is worth repeating.
Instagram punishes generic output. The fix is specificity in the prompt, not more adjectives.
Why This Matters in 2026
Short-form video returns 49% ROI — the highest of any format marketers run (HubSpot marketing statistics), and 73% of consumers say they prefer learning about a product through short video (Sci-Tech Today). Instagram is where that preference meets a shopping surface: a Reel can carry a product tag, a Story can carry a link sticker, and a paid Reel can carry a CTA button. Prompts that ignore those surfaces waste the only formats that convert on-platform.
How to Write Instagram-Specific AI Video Prompts
Every prompt below uses the same four-slot skeleton. Learn it once and you can rewrite any idea in this list.
Slot 1 — Subject motion. What physically changes: “the serum drops from the dropper and blooms into the water” beats “serum product shot.”
Slot 2 — Camera path. Shot size, lens, move. “Slow 20° arc right, 50mm, eye level” is actionable. “Cinematic” is not.
Slot 3 — Environment reaction. Light shifts, steam rises, fabric settles. This stops a clip reading as a still with a wobble.
Slot 4 — Constraints. Duration, aspect ratio, text-free zones, and what must stay identical (label, logo, cap colour).
A short syntax cheat sheet:
[shot size/lens] + [camera move] + [subject action] + [environment reaction]
+ [light + palette] + [duration + aspect] + [what must NOT change]
Keep prompts to four to eight seconds per shot. Longer clips drift — the label warps, the logo swims. A stable 6-second cut beats a 20-second one you can’t post.
Instagram Reel / Story spec constraints worth writing into prompts
| Constraint | Value to bake into the prompt | Why |
|---|---|---|
| Aspect ratio | 9:16, 1080×1920 | Reels and Stories crop square/landscape |
| Hook window | First 1–3 seconds | Scroll decides before the caption reads |
| Safe zone | ~250px top, ~420px bottom | Profile UI, caption, CTA sticker, audio badge |
| Sweet-spot length | 30–60s for demos | 51% of companies say 30–60s performs best (Sci-Tech Today) |
| Text | Large, ≤7 words per card | Sound-off is the default |
| Product tag | Keep lower third clear | Tag bubbles land bottom-left/centre |
Now the prompts.
Reels Aesthetic Prompts: Lighting, Palette and Pace
These set the look of a Reel. Each one is a single-shot prompt you can feed to an image-to-video model with a product still as the source frame, or to a text-to-video model from scratch.
Prompt 1 — Warm window-light hero
50mm, eye level, slow 15° arc right around [product] on a pale oak table.
Late-afternoon window light from camera left, soft falloff, warm 4200K palette.
The [product] cap catches a moving highlight as the camera arcs; dust motes
drift through the light shaft. 6 seconds, 9:16. Do not alter the label text,
logo or cap colour.
Use case: Skincare, ceramics, coffee, homeware — anything tactile. Tool fit: Image-to-video (feed a clean product still). Text-to-video works but costs you label fidelity.
Prompt 2 — Cold clinical palette for tech
Macro, 85mm, shallow depth of field. Camera dollies forward 30cm toward the
[product] ports while a thin blue rim light sweeps left to right across the
matte black surface. Cool 6500K palette, seamless dark grey background, faint
atmospheric haze. 5 seconds, 9:16. Keep all port shapes and printed text
identical to the source image.
Use case: Audio gear, chargers, smart devices. Tool fit: Image-to-video, first/last-frame models.
Prompt 3 — Fast-cut pace prompt for fashion
Three beats in 6 seconds, 9:16, hard cuts on the beat:
Beat 1 (2s): close-up of the [garment] fabric being pulled taut, 35mm.
Beat 2 (2s): wide, model turns away from camera, camera whip-pans with her.
Beat 3 (2s): extreme close-up of the [detail: zip/stitch/hardware], static.
Consistent daylight, slightly desaturated palette. No text overlays.
Use case: Drops, lookbooks, “3 ways to wear” Reels. Tool fit: Multi-shot / multi-clip generators; script generators for beat timing.
Prompt 4 — Slow luxury pour
Top-down 45°, 50mm. A slow tilt down as [liquid/cream/powder] pours onto raw
linen in slow motion at 120fps. Warm directional key from the right, long soft
shadows, sand-and-cream palette. The poured material keeps settling for 3
seconds after the pour ends. 7 seconds, 9:16. Product label faces camera and
stays legible throughout.
Use case: Fragrance, food, supplements, foundations. Tool fit: Image-to-video with slow-motion; text-to-video for the pour physics.
Prompt 5 — Pattern-interrupt opening frame
First frame (0.5s): the [product] frozen mid-air, dropped from just out of
frame, motionless against a solid [brand colour] background — a frozen
pattern-interrupt. From 0.5s: gravity resumes, the product lands on a soft
surface and settles with a small bounce. 35mm, static camera, hard studio key
light. 4 seconds, 9:16.
Use case: Any product that reads instantly as a silhouette. Tool fit: Text-to-video (you need control of the first frame); image-to-video with a “frozen” source image.
Prompt 6 — Lifestyle hand-entry Reel
POV, 24mm, handheld micro-shake. A hand enters frame from the bottom right,
picks up [product] from a cluttered real desk (mug, keys, notebook), and turns
it toward the light. Natural room light, no colour grading, slight sensor
grain. 6 seconds, 9:16. Do not clean up or stylise the background.
Use case: The “real life” Reel that out-performs studio cuts for cold traffic. Tool fit: UGC ad generators, image-to-video with a lifestyle source frame.
Stories Prompt Sets: Vertical, Fast, Tappable
Stories are consumed one-handed and fast. Prompts should produce 2–4 second beats that stack.
Prompt 7 — Sticker-ready demo beat
Static camera, 35mm, 9:16. A hand squeezes [product] onto fingertips and
presses it into the back of the other hand. Two seconds of contact, then the
hand turns to show the finish to camera. Bright even daylight, true-to-life
colour. Keep the bottom 30% of frame free of detail — it is reserved for a
link sticker.
Use case: Skincare, cosmetics, consumables. Tool fit: UGC ad generators; image-to-video.
Prompt 8 — Before/after swipe beat
Split-screen 9:16. Left half: [problem state] shot flat and grey. Right half:
[result state] shot warm and bright. At 2s a vertical wipe travels left to
right, fully replacing the grey side with the warm side by 4s. Same framing,
same angle, same subject in both halves. No text. 5 seconds.
Use case: Cleaning, skincare, organisation, furniture. Tool fit: Text-to-video with first/last frames; image-to-video with two source frames.
Prompt 9 — Countdown / poll scaffold
Three 2-second clips, 9:16, static camera, high-key white studio.
Clip 1: [product] rotates 90° on a turntable.
Clip 2: the product is picked up and held at chest height.
Clip 3: the product is set back down, hand exits frame.
Clean negative space top-left in all three clips for a countdown or poll sticker.
Use case: Launch countdowns, “which colour?” polls, engagement bait that feeds the algorithm. Tool fit: Text-to-video; batch UGC generators.
Prompt 10 — Behind-the-scenes packing Story
Overhead 24mm, natural workshop light. Two hands fold [product] into
packaging, add a tissue layer, then drop in a handwritten card. Camera nudges
forward 10cm over the 6 seconds. Ambient sound bed: paper, tape, room tone.
9:16. Keep hands and packaging un-manicured and real.
Use case: Small brands, indie sellers — the single highest-trust Story format. Tool fit: Image-to-video; real-shoot + AI cleanup hybrid.
Shopping-Tag Video Prompts: Keep the Product Clickable
Product-tagged Reels need the object identifiable in under a second and the lower third clear for the tag bubble.
Prompt 11 — Clean tag-anchor product pass
50mm, product centred and large in frame (occupying ≥55% of frame height).
Camera makes a slow push-in of 20cm over 5 seconds while [product] rotates 45°
on a matte white surface. Flat even key light, no dramatic shadows, palette
matched to the brand sheet. 9:16. Label faces camera at all times. Leave the
bottom 35% of the frame uncluttered for the product tag.
Use case: Any SKU you intend to tag in-feed. Tool fit: Image-to-video from a PDP hero shot.
Prompt 12 — Three-SKU carousel-style pass
Static 35mm, 9:16, seamless [brand colour] backdrop. A hand places [SKU A],
then [SKU B], then [SKU C] left to right on a line, each landing with a soft
tap, 1.5 seconds apart. Final 2 seconds: all three in frame, evenly spaced,
label-forward. Even softbox light. Do not change label typography or cap
colours.
Use case: Bundles, “pick your shade,” collection launches. Tool fit: Multi-subject text-to-video; UGC ad generators with asset slots.
Prompt 13 — Use-in-context tag clip
Medium shot, 35mm, natural kitchen daylight. [Product] is used once, clearly:
[one action — pour, press, wipe, charge]. Camera holds still; the action is
the movement. Real surfaces, real wear, no styling. 5 seconds, 9:16. Product
label visible from second 1 to second 5.
Use case: The tagged Reel that converts, because the tag has context. Tool fit: Image-to-video; UGC generators.
Prompt 14 — Scale-and-detail reference
9:16. Cut 1 (2s): [product] held between two fingers against a wide open
background to establish scale, 35mm. Cut 2 (2s): macro 100mm on [texture /
stitch / grain] with a slow rack focus from front to back. Cut 3 (2s): full
product on a plain surface, static. Consistent warm-neutral light across all
cuts.
Use case: Apparel accessories, jewellery, small electronics — anything people can’t size from a photo. Tool fit: Multi-shot generators; image-to-video per cut.
IG Ad-Spec Prompts: Bake the Format In
Paid placements have hard constraints. Put them in the prompt so you don’t burn a render.
Prompt 15 — Feed-safe safe-zone ad
9:16 vertical, 30 seconds, built in four 7-second beats. All key text and the
product sit inside the central 70% of frame height. First beat is a
pattern-interrupt: [unexpected action] within the first 1.5 seconds. Last beat
holds [product] plus empty space bottom-centre for the CTA button. 35mm,
natural light, high contrast captions burned in ≤7 words per card.
Use case: Meta/IG Reels ads where UI eats the edges. Tool fit: UGC ad generators; script generators for the beat map.
Prompt 16 — Sound-off caption-first ad
Text-overlay-first vertical ad, 9:16, 20 seconds. Opening frame is a bold
text card reading "[HOOK, ≤8 words]" over a blurred moving background of
[product in use]. Text burns in word-by-word over 2 seconds, then cuts to the
product demo. Every subsequent caption card ≤7 words, white on 60% black scrim,
placed in the upper-middle safe zone. No reliance on audio.
Use case: Cold traffic, sound-off feeds. Tool fit: Text-to-video + caption tools; script generators.
Prompt 17 — Spec-compliant demo with claim guardrails
Vertical 9:16 demo, 30 seconds. Show only these approved claims:
[claim 1], [claim 2]. For each claim, show the physical evidence on camera —
do not animate or dramatise the result. Neutral studio light, true colour,
no beauty filters on skin or surfaces. Do not invent benefits, awards,
statistics or testimonials. 51% of companies find 30–60s the best-performing
length, so hold to that window.
Use case: Regulated categories — supplements, cosmetics, skincare claims. Tool fit: Script generators with a claims allowlist; UGC generators.
Prompt 18 — Testimonial-frame ad shell
UGC-style vertical 9:16, 25 seconds. Front-facing handheld framing, a real
room behind [creator description], soft window light from camera left.
Camera is a phone held at arm's length, slight natural shake. The creator
looks at the lens, not the screen. Do not smooth skin, do not grade the
footage, do not add music in the render. [Product] enters frame at second 8.
Use case: The ad shell you fill with real or generated spokespeople. Tool fit: UGC ad generators with avatar libraries.
Prompt 19 — A/B hook-swap template
Generate 4 versions of this 8-second vertical ad, identical from second 3 onward. Only the first 3 seconds change:
onward. Only the first 3 seconds change:
A: [verbal hook variant 1]
B: [verbal hook variant 2]
C: [text-overlay hook variant]
D: [visual pattern-interrupt variant]
Same product, same lighting, same camera move, same CTA end card across all
four. 9:16.
Use case: Structured hook testing where the body is a constant. Tool fit: Batch UGC ad generators; any tool with seed/variant controls.
10 Fill-in-the-Blank Instagram Hook & Script Templates
Prompts build the picture. Hooks buy the first three seconds. Use these as text overlays, verbal openers, or as the first line inside the prompt itself.
Pattern-interrupt opener
Stop scrolling — [unexpected claim about the product category].POV confession
POV: you finally found [product type] that [specific outcome].Three-reason stack
3 reasons I switched from [old solution] to [product] — reason 2 is the one nobody mentions.Objection-first
I didn't believe [product] could [claim] either. Here's the 7-day test.Comment reply (IG’s highest-trust Reel format)
"@[handle] asked if [product] works for [use case] — watch this."Before/after tease
This is [problem] on Monday. Swipe for Friday.Price-per-use
Everyone says [product] is expensive. It's [price] ÷ [uses] = [cost] a day.Founder-to-camera
I make [product]. Here's the one thing I'd never let us cut corners on.Silent demo overlay
Watch what happens when I [action] — no talking, just [product].Story-series serialiser
Part [n] of testing every [product type] under [price]: today, [SKU].
Script skeleton for a 30-second Reel (fills in brackets):
[0–2s] HOOK: "[template line]" — product visible in frame
[2–8s] PROBLEM: show [pain] happening, don't describe it
[8–18s] DEMO: one action, one benefit, one proof point
[18–24s] OBJECTION: "[the #1 doubt]" answered in one sentence
[24–30s] CTA: "[verb] [product] — link in bio / tap the tag"
Production Notes and How to Measure Prompt Families
Ship in prompt families, not one-offs. Change exactly one variable — camera move, palette or hook — and render four. Keep a log: prompt hash, hook, model, seed, 3-second retention, saves, tag taps. After 20 posts you’ll see which slot moves the needle. In practice the camera path and the first 1.5 seconds dominate; palette changes rarely move anything.
The four numbers that judge a prompt family:
| Metric | What it tells you | Healthy signal |
|---|---|---|
| 3-second view-through | Did the hook frame land? | Rising across variants |
| Saves per 1,000 views | Is it useful, not just pretty? | Above your feed median |
| Product-tag taps | Did the object read as shoppable? | Tag tap rate ≥ 1.5% |
| Profile visits → link clicks | Did the Reel carry intent downstream? | Improving week over week |
Do the human pass. AI video still fails on hands, text on packaging, and reflections in glass. Zoom to 200% and check the label lettering before anything goes into Ads Manager. If the label is wrong, no amount of good lighting saves the ad.
Repurpose, don’t re-render. A 30-second Reel becomes four Story frames, one feed teaser and one retargeting bumper. Render wide and crop, or re-prompt only the frames that need it.
If you’d rather start from what’s already working, VideoIdeas reads live trends across Instagram, TikTok, YouTube and Facebook and hands back video concepts plus AI-ready prompts — see how it works and browse example outputs, or generate ideas free.
Frequently Asked Questions
What are the best AI video prompts for Instagram Reels? The best ones specify four things: subject motion, camera move, environmental reaction, and constraints (duration, aspect ratio, what must not change). A prompt like “50mm, slow 15° arc right, serum drops and blooms, warm 4200K window light, 6s, 9:16, keep label identical” will out-perform “cinematic skincare product video” almost every time.
Can I use AI video prompts for Instagram Stories as well as Reels? Yes, but shorten the beats. Stories are consumed in 2–4 second taps, so prompt for short stacked clips with negative space reserved for link stickers and polls. Reels prompts can hold 5–8 second shots with a single continuous camera move.
How do I keep product tags working in AI-generated Instagram video? Keep the product large, label-forward, and identifiable within the first second. Prompt for a clean lower third so the tag bubble doesn’t cover detail, and write “do not alter label text or logo” into the constraints slot — models drift on typography long before they drift on colour.
What aspect ratio and length should AI Instagram video prompts target? 9:16 at 1080×1920 for Reels and Stories. For demos, prompt 30–60 seconds: 51% of companies report that window performs best. For cold hook tests, 6–15 seconds is cheaper to iterate and usually tells you what you need.
How do I stop AI video looking obviously AI on Instagram? Add imperfect detail to the prompt: handheld micro-shake, sensor grain, real clutter, un-manicured hands, no beauty filters. Polished perfection is the tell. 76% of marketers say authentic content beats polished production, and feed behaviour agrees.
Written by the VideoIdeas Editorial Team. We build tooling that turns live social trends into product video concepts and AI-ready prompts.
Sources:
- HubSpot — Marketing Statistics
- Sci-Tech Today — Short-Form Video ROI Statistics
- Webtonic — E-commerce Video Marketing Statistics
- Instagram Business Blog
Share Article
