A streetwear drop teaser works by showing less than the product, not more: a texture, a silhouette, a place, a date. That makes it one of the best fits for AI video in fashion, because the hardest thing for a model to get right (the exact print, the exact graphic, the exact label) is the thing a teaser holds back until release day. The approach that holds up: plan a short sequence of clips, generate the atmosphere and the hints, and keep the reveal for real photos or for shots made from them. On 8frame that is the clip maker's Fashion preset for each teaser, a few fixed-price single shots on the board, and the Video Editor for the date and the drop time as text.
TL;DR
- Tease with mood, reveal with truth. Generated clips for the hints; real product photos, or clips made strictly from them, for the reveal.
- Plan four clips: the hint (a week out), the detail (a few days out), the silhouette (the day before), the reveal (drop day).
- On 8frame: the Fashion preset ("The look in motion, editorial light") opens at 4 scenes, 15 seconds, 9:16. Slots: Person (3 photos recommended), Outfit (2), Place (1), Style (1).
- Graphics and prints drift. Models redraw screen prints, embroidery and woven labels. Never let a generated version of your graphic stand in for the real one.
- Fixed prices on the board: Veo 3.1 Lite 54 credits for 8 s with sound, Gemini Omni Flash 85 for 5 s, Kling v3 Standard 86 for 5 s with sound, Veo 3.1 Fast 168 for 8 s with sound.
The four-clip teaser sequence
Drops differ, but most teasers follow the same arc: create a question, narrow it, then answer it. Here is a sequence you can adapt, with what each clip shows and hides.
| Clip | When | What it shows | What it hides | Made how |
|---|---|---|---|---|
| 1. The hint | About a week out | A place and a mood: an empty underpass at dawn, a rooftop, a laundromat at night. The brand mark at the end | The product entirely | Fully generated |
| 2. The detail | A few days out | Extreme close-ups: a zip pull, a seam, fabric moving in wind | The full garment and the graphic | Generated from your fabric and detail photos |
| 3. The silhouette | The day before | A person in the piece, backlit or in shadow, walking through the place from clip 1 | The front graphic and colour detail | Fashion preset, with the real pieces in the Outfit slot |
| 4. The reveal | Drop day | The pieces clearly, the price, the time and where to buy | Nothing | Real photos and video, or shots made strictly from them and checked frame by frame |
Two rules make the sequence work. First, keep one visual thread across all four (the same place, the same light, the same colour grade), so the audience recognises clip 4 as the answer to clip 1. The Style slot is how you hold that grade: use the same style frame for every clip. Second, put dates and times on screen as text, not in the generated picture, where models misspell lettering.
Why prints and labels stay real
A streetwear piece is often sold on its graphic: the chest print, the back hit, the woven label, the embroidered logo. Video models do not copy these; they reconstruct them on every frame, and lettering and fine artwork are among the things they reconstruct least reliably. A two-word chest print can come back with a letter missing; an intricate back graphic can come back as something that looks similar from a distance and wrong up close. Your audience will screenshot the teaser, zoom in and compare it with what arrives. Our explainer on why AI changes product labels covers the mechanism.
So the teaser plan above keeps the graphic hidden until the reveal, and the reveal uses real photos or clips you have checked against the real piece. If clip 3 shows the front of a tee, frame it in shadow or from behind.
Making each clip on 8frame
Clips 1 and 2: atmosphere and detail
For the hint, you do not need a preset at all. Put a single video node on the board, describe the place and the light, and run it. For the detail clip, upload close photos of the real fabric, zip, trim or label and use an image-to-video model so the texture comes from your photo.
Clip 3: the silhouette, with the Fashion preset
Open the Fashion preset in the clip maker. Its question is "Which look are we filming?".
- Person (three recommended). Photos of the model wearing nothing that conflicts with the drop, with their permission.
- Outfit (two recommended). Photos of the real pieces from the drop: flat lays or on a hanger, front and back.
- Place (one). The location from clip 1, so the sequence stays connected.
- Style (one). The same style frame you used for clips 1 and 2.
Write the idea in a sentence ("the model walks through the underpass at dawn wearing the hoodie, backlit, the graphic in shadow") and let the clip maker draft a script by scene. Edit any line that describes the piece wrongly: fabric weight, colour name, fit. The render button shows its credit price before you press it; the clip maker renders all scenes as one video. A scene that drifted can be re-shot; a failed scene can be retried on its own, and failed generations are refunded automatically.
Clip 4: the reveal
Use real photos of every piece, and real video if you have it. If you generate any reveal shot, start from a photo of the real piece, keep the move small, and check the graphic, the colour and the label at full size before posting. Then add the drop time, the price and where to buy as text layers in the Video Editor. Captions come from a script you paste; the editor does not transcribe.
Prompts for the teaser clips
Prompt:
An empty concrete underpass at dawn, cold blue light, a single sodium lamp flickering on, slow push in toward the far end where a shadow crosses quickly, no people visible, no text in frame
Prompt:
Extreme close-up of the heavyweight cotton from the reference photo moving slightly in the wind, then a slow slide along the zip pull from the reference, hard side light, shallow focus, colours exactly as in the photos
Prompt:
The model from the reference photos walks slowly toward camera through the underpass at dawn wearing the hoodie from the outfit photos, strong backlight so the front of the hoodie stays in shadow, fabric moving with each step, cold blue grade
Prompt:
A rooftop at night in the rain, neon from the street below reflecting on wet concrete, a hand places a folded garment on a bench and leaves frame, camera holds on the bench, the garment's colour as in the reference photo
Which model for which clip
Fixed prices per clip, in credits, for single shots on the board:
| Model on the board | Credits | Good for | The catch |
|---|---|---|---|
| Veo 3.1 Lite | 54 for 8 s with sound | The hint: a place and a mood from text | 720p; 16:9 or 9:16 only |
| Gemini Omni Flash | 85 for 5 s | Short atmospheric beats | Opens at 5 s and 16:9 on the board; change the ratio for vertical |
| Kling v3 Standard | 86 for 5 s with sound, 57 silent | The detail clip from your fabric or trim photo | Needs a start image; 720p |
| Veo 3.1 Fast | 168 for 8 s with sound | A richer hint or silhouette shot with room sound | From an image it is locked to 8 s |
| Seedance 2.5 | 157 for 5 s at 720p | An extra silhouette scene on the model family the clip maker calls | Tops out at 720p |
No video model on 8frame outputs 4K and there is no video upscaler. If you join several clips with sound, use the Video Editor timeline: the Montage node (10 credits) joins the picture only, and its file has no audio track.
Platforms and labels
- Instagram Reels ads. Meta's ads guide lists 9:16 at 1440 x 2560 and asks you to keep 14% at the top, 35% at the bottom and 6% at each side clear of text and logos. Put the drop date in the middle band. More in Instagram Reels ad specs.
- TikTok. Its ads policy allows AI-generated or significantly edited content with its AIGC label or "a clear disclaimer, caption, watermark, or sticker" of your own, and says undisclosed AI content will be rejected or restricted. A fully generated hint clip needs the label.
- Your store. If the drop sells on Shopify, product videos can be up to 10 minutes and 1 GB as MP4, MOV or WebM, so the reveal clip can go straight onto the product page.
Honesty in a drop
This is general information, not legal advice. The FTC's small-business advertising guide says it judges an ad in context, "words, phrases, and pictures". For a drop, that means two things beyond the product itself. If the teaser says "limited", the quantity should really be limited. And if a generated silhouette suggests a fit, length or colour, the piece should match it. A teaser that builds expectation the drop cannot meet costs more in returns and comments than it earned in hype. For the wider disclosure question, see do you have to disclose AI-generated ads.
Rather have the clips made for you?
If you run a streetwear label and want the teaser clips without learning a new tool, the 8frame team can make them for you: 10 clips in 5 business days for $490, with one round of changes, made in the same Fashion preset this guide uses. Send photos of the pieces and a short brief on the done-for-you fashion clips page. To make them yourself instead, open the Fashion preset in the clip maker and follow the sequence above.
FAQ
How long should a drop teaser video be?
Short. Each teaser in the sequence above works at 5 to 15 seconds; the Fashion preset opens at 15 seconds with 4 scenes, and you can set 5 or 10 instead. The reveal can run longer if it shows several pieces.
Can I make the whole drop campaign with AI?
The hints and the atmosphere, yes. The reveal should use real photos of the real pieces, or clips made from them and checked at full size, because models redraw prints, graphics and labels and your audience will compare.
Why does my graphic look different in the AI video?
Video models reconstruct artwork and lettering on each frame instead of copying it. Keep the graphic in shadow or out of frame in generated teasers, and show it in real photos on release day.
Do I have to label an AI teaser?
On TikTok ads, significantly AI-edited or fully generated content needs the AIGC label or your own clear disclaimer. Other platforms and markets differ; see our guide on disclosing AI-generated ads.
What does a teaser cost on 8frame?
The clip maker shows its price on the render button before you render. Single shots on the board have fixed prices, such as Veo 3.1 Lite at 54 credits for 8 seconds with sound and Kling v3 Standard at 86 for 5 seconds with sound. Plans start at $19 a month for 1,000 credits; Creator is $49 for 3,000.
Sources
- TikTok ads policy, misleading and false content (AIGC label or own disclaimer; undisclosed AI content rejected or restricted): ads.tiktok.com, checked 2026-11-04
- Meta ads guide, Instagram Reels video (9:16, 1440 x 2560, safe zones 14% top, 35% bottom, 6% sides): facebook.com/business/ads-guide, checked 2026-11-04
- Shopify Help Center, product media types (videos up to 10 minutes and 1 GB; MP4, MOV, WebM), checked 2026-11-04
- FTC, Advertising FAQ's: A Guide for Small Business ("words, phrases, and pictures"), checked 2026-11-04
- 8frame prices, presets, defaults and limits: the 8frame app, checked 2026-11-04
Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.