AI Product Video Ad Prompts
A motion-only prompt library for product video ads built around the Martini canvas. Each recipe assumes you have already wired a brand-approved product still into a reference slot, then describes camera, light, and pacing — never the bottle, label, or color the still already covers. Pick the model that matches the shot intent (Seedance 2 for label-locked hero spins, Runway Gen-4 for editorial lifestyle, Kling 3 for cinematic reveals, Hailuo for fast variant sweeps), fan multiple recipes off the same anchor, and assemble in the sequence builder.
When to use this prompt
- Shipping creative for a new SKU drop on Meta, TikTok, and YouTube without booking a studio shoot.
- Producing the Amazon-required 16:9 PDP video plus a 9:16 social cutdown from one product still.
- A/B-testing 6–10 motion treatments off the same brand-approved hero photo before paid spend.
- Refreshing legacy product photography into 6–10 second motion clips for retargeting flights.
- Replacing per-SKU video shoots with a templated canvas the next launch reuses end-to-end.
Required inputs
- A brand-approved product still at the highest resolution you have (artifacts amplify across 24fps).
- Brand color script and any logo / typography constants as a separate reference node.
- Aspect ratio target per platform (1:1, 9:16, 16:9) decided before generating.
- Shot intent in plain language ("hero spin", "rim-light reveal", "lifestyle pour") so the prompt stays motion-only.
- Optional: a scene reference image for lifestyle treatments (wood counter, marble vanity, gym floor).
Prompt recipes
Slow orbit hero spin (label-locked)
The signature ecommerce hero spin. Label and packaging stay legible because the prompt only directs camera and light — the reference still does the rest.
Slow 360-degree orbit around the referenced product, 8 seconds, even rotation, soft front-key plus rim light, shallow depth of field, label remains crisp and centered, subtle floor reflection, locked focal length, no zoom, no cuts.
Variations
- Tighten the arc to 180 degrees for a faster 4-second cut.
- Switch to backlit reveal for category-leader brand pages.
Aspect ratio 1:1 for PDP, 9:16 for Reels.
Lifestyle pour-shot insert
Editorial lifestyle insert that mixes well with on-set plates. Use Runway Gen-4 when the cut needs to feel director-led rather than studio-clean.
Hand pours referenced product into frame from upper right, slow-motion droplets, soft daylight from camera left, kitchen counter context, 6 seconds, ambient steam rising, hand exits frame on final beat, no dialogue.
Variations
- Swap pour for drizzle for sauces and condiments.
- Replace daylight with warm tungsten for evening / cozy positioning.
Rim-light cinematic reveal
Brand reveal cut for top-of-funnel hero placements. Kling 3 holds the cinematic camera language without losing the product.
Camera dollies in slowly toward referenced product over 5 seconds, dramatic rim light from behind, deep shadows, atmosphere haze, product transitions from silhouette to fully lit, locked subject, 24fps cinematic, no rotation, no cut.
Variations
- Add a subtle lens flare on the final beat for premium fragrance / spirits.
- Switch to side rim light for tall bottles and elongated SKUs.
Macro material detail loop
Texture-forward detail cut for skincare, beverages, packaging materials. Pair with the hero spin to build a 3-cut sequence.
Macro lens hovers across textured surface of referenced product, 4-second seamless loop, soft top light, depth of field shifts focus left to right, no rotation, no zoom out, ends matching first frame for clean loop.
Best at 1:1 for paid social carousels.
Editorial vertical cutdown
Photoreal vertical hero for IG Reels and TikTok. Veo holds natural light better than synthetic studio fill.
Static wide shot of referenced product on minimal plinth, natural side light shifting subtly across 6 seconds, faint dust motes in beam, no camera movement, no cuts, 9:16 vertical framing, premium editorial mood.
Variations
- Swap dust motes for smoke / steam for hot foods or fragrances.
Fast A/B variant sweep
Light, repeatable prompt you fan across 5–10 generations to A/B-test motion treatments before committing render budget.
Slow push-in toward referenced product, 4 seconds, soft studio key plus subtle bounce fill, neutral seamless backdrop, locked subject, no rotation, no cut, end on full-frame product.
Run 6 variants, pick winner, scale up on Seedance 2.
Unboxing reveal
Unboxing-style narrative cut for retail / DTC launches. Kling 3 handles the multi-action reveal without identity drift.
Camera tracks down toward referenced packaging on table surface, lid lifts off slowly revealing referenced product inside, 7 seconds, soft overhead daylight, hands enter and exit naturally, ends on hero product framed center.
Variations
- Substitute hands-out-of-frame version for trademark-sensitive brands.
Brand-color backdrop punch
High-saturation paid-social hero. Pair with on-image headline downstream in the NLE for full ad creative.
Static referenced product centered on brand-color backdrop, 5 seconds, light pulse syncs at 1.5 second mark, subtle shadow shift, no camera movement, locked subject, ends on full-frame product with backdrop saturation peak.
Wire your brand-color script as a separate reference node.
Talent-with-product lifestyle
Lifestyle variant for skincare, beauty, and food categories. Runway Gen-4 keeps the model and product anchored together.
Mid-shot of model hands holding referenced product at chest height, soft window light from camera left, slight handheld breathe, 6 seconds, model turns product gently to camera, no dialogue, no cuts, kitchen context.
Variations
- Swap kitchen for vanity, gym, or desk per category.
Martini canvas workflow
Drop the brand-approved product still into an image node and label it "hero ref". Add the brand-color script and any scene reference as separate nodes — multi-anchor wiring is the differentiator vs single-prompt tools, and prevents palette drift.
Wire the product still into one Seedance 2 video node for the label-locked hero spin, a Runway Gen-4 video node for editorial cutaways, a Kling 3 node for the cinematic reveal, and a Hailuo node for fast variant sweeps. One canvas, parallel fan-out — no per-tool tab juggling.
Write motion-only prompts in each video node (camera, light, pacing). Never re-describe the bottle, label, or color — the reference does that, and re-describing causes label drift. Keep clips between 6 and 10 seconds for the paid-social sweet spot.
Run audio in parallel: ElevenLabs voiceover node for the explainer cut, brand SFX bed or Hunyuan Foley for product sounds. The audio nodes generate while the video nodes render.
Sequence the winning cuts in the canvas sequence builder (hero → lifestyle → detail loop), duplicate the sequence for 1:1, 9:16, and 16:9 aspect ratios, and export through ai-video-nle-export to Premiere or DaVinci. Save the canvas as a SKU template — the next launch reuses the chain.
Variations
9:16 paid social (Reels / TikTok / Shorts)
Vertical framing, 6–8 seconds, headline-safe top third, hook beat in the first 1.5 seconds.
1:1 ecommerce PDP loop
Square framing, 4-second seamless loop, label centered, locked focal length for clean autoplay.
16:9 YouTube pre-roll
Widescreen, 6–10 seconds, room for product plus environment, voiceover-friendly pacing.
Studio (white seamless)
Catalog-clean treatment for marketplace and PDP — soft front-key, neutral backdrop, no environmental cues.
Lifestyle (in-context)
Real-world scene with talent or environment cues (kitchen, bathroom, gym) for paid social and CRM.
Dramatic / editorial
Rim light, deep shadow, premium fragrance / spirits / fashion mood for top-of-funnel awareness.
Related features
Related how-tos
Related models
Related blog posts
Related docs
Open-source prompt library
Frequently asked questions
- Why are these prompts motion-only?
- Because the brand-approved product still is wired into a reference slot, you do not need to re-describe the bottle, label, or color in the prompt. Re-describing is the #1 cause of label drift and palette shift across generations. Direct only camera, light, and pacing — let the reference carry product identity.
- How long should each clip be?
- 6–10 seconds is the paid-social sweet spot. Most product video models hold coherence cleanly through that window; longer single takes drift. For 30-second spots, generate 4–6 short cuts and assemble in the sequence builder rather than asking the model for one long take.
- How do I match one product to multiple aspect ratios?
- Generate the hero motion once, then duplicate the canvas sequence for 1:1, 9:16, and 16:9 outputs. Each duplicate gets a framing-specific recut. Export all three through ai-video-nle-export so Premiere or DaVinci finishing happens once across all platforms.
- Which model should I pick first?
- Default to Seedance 2 for label fidelity on the hero shot, then add a Runway Gen-4 cut for editorial flavor and a Kling 3 cut for the cinematic reveal. Use Hailuo only for variant sweeps. Mixing models per shot intent is a feature, not noise — same canvas, different model per cut.
- Can I trademark-protect this motion treatment?
- No. Generated motion is not eligible for trademark protection. Use the canvas template approach to get visual consistency across SKUs and campaigns, but treat each cut as a paid-media asset rather than a brand-IP asset.
- Where does this fit in the Martini workflow?
- These prompts plug into the ai-product-video workflow as the per-shot motion library. Upstream is product photography (or your studio still); downstream is sequence builder + ai-video-nle-export. The prompts are reusable; the canvas around them is the lasting win.
Try this prompt on Martini
Copy a recipe above, drop it into a node, and run it inside a full canvas workflow.