Create a Video with Seedance 2.5
One product photo. One 15-second Seedance 2.5 prompt. That was enough to turn a plain moisturizer jar into a TikTok-style skincare ad with five fast cuts, natural handheld movement, and a final product-plus-face frame.
The result is below. I used a fictional jar and an AI-generated creator, so this is a prompt case study, not a real brand endorsement.
Below is the exact prompt, the settings I used, and a shot-by-shot explanation of why each block earns its place.
What 15 seconds forces you to do differently
A 30-second skincare UGC ad has room for a problem, a routine, an emotional change, and a selling frame. A 15-second vertical video does not. It has to get to the product immediately and still prove texture, application, and finish before the payoff.
This prompt is written for that shorter shape: the first two seconds open on the cream, the middle four shots prove texture and application, and the final three seconds lock the product beside the face.
The difference from a longer Seedance 2.5 UGC prompt tutorial is timing density. Every shot carries a product reveal, a texture close-up, or a face finish. There is no warm-up, mirror setup, or bathroom walk-in.
The setup: one reference image, one prompt
The video was generated from one reference image and one prompt. The reference image is a white-background studio-style moisturizer photo with fictional packaging, made first so the video would not need to invent a label or logo.
| 設定 | 価値 |
|---|---|
| モデル | シードダンス 2.5 |
| モード | Image to video |
| Length | 15 seconds |
| アスペクト比 | 9:16 vertical |
| 解決 | 720p |
| オーディオ | の上 |
| Watermark | Off |
We generated several takes and kept the one that held the packaging most clearly across all five shots.
The full Seedance 2.5 skincare UGC prompt
Copy the prompt below and replace the product, scene, dialogue, and timing lines to match your reference image.
Morning Moisturizer UGC — 15s
FORMAT: 15s · 9:16 vertical · Native iPhone camera
STYLE: Fast-paced American beauty TikTok UGC, authentic bedroom morning routine, engaging hook, realistic product demo.
CAST & ENVIRONMENT
A gorgeous young adult white woman with blonde hair, fresh natural skin, wearing a white tank top. Cozy bedroom, slightly messy white bedding, soft natural daylight. Real handheld iPhone footage, candid and relatable, no beauty filter.
PRODUCT REFERENCE LOCK
Use the uploaded product image as the exact visual reference. Accurately preserve the original packaging, shape, colors, label, logo, and proportions. Include clear, recognizable packaging shots. Never invent or redesign the product.
SHOT 1 — SCROLL-STOPPING HOOK (0–2s)
Instant close-up: she quickly brings the moisturizer into frame beside her face, then jump-cuts to a fingertip holding a generous dollop of cream close to the lens. Immediate movement, tight framing, no intro.
Dialogue: “Wait—look at this moisturizer!”
SHOT 2 — PACKAGING + TEXTURE (2–5s)
Two rapid cuts: first, a sharp close-up of the actual product packaging in her hand; second, an extreme close-up of the cream being scooped and slowly pulled apart between her fingertips, revealing a thick, silky texture.
Dialogue: “First of all, this texture? So good.”
SHOT 3 — SATISFYING APPLICATION (5–9s)
Quick jump cuts: she dots cream onto both cheeks, forehead, and chin, then massages it in with both hands. Switch from a close-up to a handheld selfie angle. Natural facial movement, realistic skin texture.
Dialogue: “It melts right into my skin.”
SHOT 4 — THE SKIN FINISH (9–12s)
Close-up of her cheek as she gently presses her fingertips against it. She turns her face toward the window, showing a fresh, naturally moisturized finish. Quick cut to her smiling at the front camera.
Dialogue: “And look at that finish!”
SHOT 5 — PRODUCT + FACE PAYOFF (12–15s)
A quick, natural selfie shot with her holding the actual product beside her face. She gestures toward her cheek, smiles, and brings the product slightly closer to the camera for the final frame.
Dialogue: “Okay, this is going in my morning routine.”
EDITING & AUDIO
Fast jump cuts every 1–2 seconds, immediate opening hook, alternating product close-ups and selfie footage, dynamic reframing, natural handheld movement. Keep spoken English energetic, conversational, and slightly spontaneous, like a real beauty creator sharing a discovery. Layer subtle cap opening, cream scooping, fingertip rubbing, and application sounds under the voice. No dead air, no slow-motion sequences, no background music.
NEGATIVE PROMPT
No invented packaging, no altered labels or logos, no generic substitute product, no beauty filters, no plastic skin, no artificial glow, no fake cream physics, no excessive cinematic lighting, no studio-ad aesthetic, no robotic delivery, no long static shots, no unnecessary transitions, no subtitles, no watermarks.The prompt is deliberately long. Each block gives the model a different job: lock identity and product, create the hook, show texture, show application, show the finish, then end on the product beside the face.
Five shots, five jobs
| Time | Shot | What the prompt asks for | Dialogue |
|---|---|---|---|
| 0–2s | Hook | Bring the jar beside the face, then cut to a dollop of cream close to the lens. | “Wait—look at this moisturizer!” |
| 2–5s | Packaging and texture | Show the real packaging, then stretch the cream between fingertips so the texture reads as thick. | “First of all, this texture? So good.” |
| 5–9s | 応用 | Dot cream on the cheeks, forehead and chin, then massage it in with both hands. | “It melts right into my skin.” |
| 9–12s | Finish | Press fingertips against the cheek and turn toward the window to show the moisturized finish. | “And look at that finish!” |
| 12–15s | Payoff | Hold the actual product beside the face, gesture to the cheek, and move the jar closer for the final frame. | “Okay, this is going in my morning routine.” |
The paragraph that stops the jar from changing
The prompt contains one paragraph that does more work than the rest: “Use the uploaded product image as the exact visual reference. Accurately preserve the original packaging, shape, colors, label, logo, and proportions.”
That is there because image-to-video models often change a product when the camera gets close to it. The jar can gain a new label, lose its cap, or change color between shots. By naming the packaging details, the prompt gives the model a tighter target to hold.
It also helps to use a reference image with a clean background and no real brand. We generated a fictional jar first, then used that image as the video input. The prompt can then say “never invent or redesign the product” without asking the model to reproduce a brand it has never seen.
The negative prompt is what keeps it from looking fake
The most important negative lines are not the ones blocking text or watermarks. They are the ones blocking the production look:
No invented packaging, no altered labels or logos, no generic substitute product, no beauty filters, no plastic skin, no artificial glow, no fake cream physics, no excessive cinematic lighting, no studio-ad aesthetic.
A skincare ad falls apart when the skin looks plastic or the cream behaves like CGI. Those negative lines keep the result closer to a real iPhone video: visible skin texture, normal hand movement, and a cream that sits and spreads instead of floating.
For a slower, more cinematic product reveal, see our AI bullet time product video prompt. This skincare prompt is built for the opposite register: quick cuts, fast delivery, and a creator-style payoff.
Make it yours
- Replace every moisturizer reference with the exact product name and form: serum, cleanser, mask, lip balm, or sunscreen.
- Change the texture line to what that product should show, such as a gel spread, an oil drop, a foam lather, or a tinted cream finish.
- Keep the timing blocks, but shorten the application section if the product has fewer visible steps.
- Generate several takes and keep the one where the product stays the same from hook to final frame.
If you change the length from 15 seconds, rescale the timestamps in the prompt. Seedance 2.5 can generate videos from 4 to 30 seconds.
よくある質問
Can I use a real branded product image?
You can, but this prompt is written for a fictional or owned product image. If you use a real brand, the model can still alter its label or logo, and you should check the result before publishing anything that implies a real endorsement.
Why does the prompt include spoken lines?
The five spoken lines give the video a creator-ad rhythm and a reason to cut from the cream to the face. The generated voice can vary from take to take, so treat the dialogue as direction and keep the version that sounds most natural and matches the action.
Why is the reference image made first?
Making a clean fictional product image first gives the video a stable jar, label, and proportions to hold. If the reference image already has a busy background or real brand, the video has more room to invent details.
Make your own 15-second skincare UGC video
Start from the prompt, swap in your product reference, and generate a few takes until the packaging stays locked through the final frame.
Create a Video with Seedance 2.5
著者について
