AI Product Video for Etsy: My Full 15-Second Workflow

TL;DR

• An AI product video for Etsy has to fit a hard box: 5 to 15 seconds, under 100 MB, 1080p or better, one video per listing.
• Etsy strips the audio automatically, so anything you build around sound is wasted work.
• My workflow: one real product photo, one directed camera move, render high, trim to 12 seconds, mute, disclose.
• Picking the right model per shot matters more than writing a longer prompt.
• Etsy permits seller-prompted AI work, but you have to be transparent about it. Read their policy before you list.

I make AI video for brands, not for Etsy shops. But the question keeps landing in my DMs from people who sell candles, ceramics, and printed tees: can I use AI video for my listings without a camera, a studio, or a videographer?

Yes. The catch is that almost every AI video tutorial online is built for a 30-second social clip with sound design. Etsy’s format is the opposite of that, and if you ignore the constraint you will produce something beautiful that the platform refuses to publish.

Here is the workflow I would actually run.

What Etsy Allows for Listing Video

Start with the box you are working inside. These are the current listing-video specs, cross-checked against soona’s Etsy specs guide and Aiarty’s Etsy video guide:

Read that audio line again. You get no voiceover, no music, no sound design. Every AI video tool sells you on generated audio and lip sync, and Etsy throws all of it away. That single rule changes what you should generate in the first place.

The 15-second ceiling is not a limitation to fight. It is the brief. One product, one move, one look.

Step 1: Shoot the Product Once, Badly

Do not start from a text prompt. Text-to-video invents a product that looks like yours and is not yours, and on a marketplace where the buyer receives a physical object, that gap is a refund waiting to happen.

Start from a photo of the real thing. Phone camera, window light, plain surface, product filling most of the frame. It does not need to be good. It needs to be accurate, in focus, and honest about color.

That photo becomes the reference. Higgsfield’s product video generator accepts a product photo, a store link, or a prompt, and the photo path is the one that holds your actual label, shape, and finish frame to frame. Upload one to three references and the model locks the product across the shot.

If you sell 40 SKUs, this is the step that scales. One photo per SKU, batched.

Step 2: Pick One Camera Move, Not a “Video”

Twelve seconds fits exactly one camera move plus a settle at the end. That is it. Sellers who try to fit a spin, a lifestyle shot, and a detail macro into a listing video end up with three seconds of each and a clip that reads as chaos on a phone screen.

Choose one:

1. 360 spin for anything with dimension a photo cannot show. Ceramics, bags, sculpture, furniture.
2. Slow push-in for anything where the selling point is a surface detail. Engraving, stitching, glaze, grain.
3. Macro detail pass for texture-led products. Fabric weave, hand-lettering, resin inclusions.
4. Lifestyle placement for anything a buyer needs to scale mentally. Mugs, prints, jewelry, home goods.

Match the move to the objection. If people message you asking “how big is it?”, you need a lifestyle placement, not a spin. If they ask “is the finish glossy or matte?”, you need a slow push-in under raking light. Your DMs are the shot list.

Step 3: Pick the AI Video Model Per Shot

This is where most people lose time. They pick one model, fight it for an hour, and blame the prompt. Different models are built for different motion, and switching is faster than rewriting.

Those are the model strengths Higgsfield documents on its own product video page, and they track with what I see in my own renders. Running them under one subscription is the reason I stopped stacking tool bills.

I went deeper on the two I use most in my 30-day Kling 3.0 versus Seedance 2.0 test, including where each one falls apart.

Runway is the other one I keep on the bench for clean, controlled product motion, and code BYMICHYDEV25 takes something off the first month if you go that route. For a marketplace seller running a handful of listings a month, one subscription is enough. Pick the one whose model lineup covers your product category and stop there.

Step 4: Trim to 12 Seconds and Mute It

Render high, then cut down. I export the generation at the highest resolution the tool gives me, then downscale to 1080p on the way out. Etsy compresses on upload, and starting from a bigger master survives that compression better than starting at exactly 1080p.

Then:

• Trim to 12 seconds. The cap is 15. Leave yourself a buffer so a rounding difference in the export does not bounce the upload.
• Mute the track. Etsy removes it anyway. Muting before export also shrinks the file.
• Check the file size. Under 100 MB. A 12-second 1080p MP4 lands far under that unless you exported something exotic.
• Cut the first frame ruthlessly. Frame one is doing the job a thumbnail does. If the product is not already in frame and readable, recut.

CapCut handles the trim, the mute, and the export in about ninety seconds on the free tier. There is no reason to open a real editor for this.

Step 5: Handle the AI Disclosure on Etsy

Etsy does not ban AI. It regulates how you describe what you made.

Per Etsy’s Creativity Standards and their seller handbook piece on Etsy’s stance on AI creations, seller-prompted AI work is allowed, it falls under the “Designed by a seller” attribution, and sellers are expected to be transparent about using AI. Etsy also prohibits selling AI prompt bundles.

Where it gets murky for the workflow in this article: those standards are written about the item the buyer receives, not about your marketing assets. A listing video for a real ceramic mug you threw by hand is a different situation than a digital download generated from a prompt.

My position, and I am a video producer and not your lawyer: read the policy pages yourself before you list, and if AI touched anything, say so plainly in the description. A line of copy costs you nothing. A removed listing costs you the shop.

One hard rule regardless of policy: never show a finish, color, or texture the buyer will not receive. AI models flatter surfaces by default. They add gloss, they deepen color, they smooth flaws. If the render looks better than the object in your hand, you did not make a listing video. You made a returns problem.

Show Your Work

# SHOW YOUR WORK — Etsy listing video, 360 spin
Reference: 1 clean product photo, plain background, window light

PROMPT:
Slow 360 rotation of the [product] on a seamless [color] surface.
Soft diffused studio light from the upper left, gentle falloff.
Camera locked, product rotates. No cuts. Product stays centered
and fully in frame the entire time. Realistic shadow contact
with the surface. Neutral color grade, true to reference.

Settings: 1080p or higher, no audio needed
Trim: 12 seconds, mute, export MP4
Tool: Higgsfield (WAN 2.7 for spins) → higgsfield.ai/?ref=michydev

Swap “slow 360 rotation” for “slow push-in toward the [detail]” and you have the second template. Swap it for “static frame, hand enters and picks up the [product]” and you have the lifestyle one.

The prompt discipline here is the same one I use for multi-shot work: describe the shot, not the mood. I broke that down properly in my post on the 20 to 30 word rule for multi-shot prompts.

FAQ

Do I need a subscription to make one Etsy listing video?
No. Most of these platforms run a free tier with daily credits. A single 12-second product clip is well within that. Subscriptions start making sense somewhere around 10 listings a month, when you are batching.

Can I use AI video for a print-on-demand shop?
Technically yes, and this is exactly where the disclosure question gets sharpest, because the design itself may also be AI-generated. Read Etsy’s Creativity Standards for your specific case rather than copying what another seller does.

Should the video be vertical or horizontal?
Etsy does not mandate an aspect ratio. I would build 16:9 or square for the listing, then re-export the same generation vertically for Instagram and TikTok. One render, three placements.

Why not just film it on my phone?
If you can, do. A real 12-second phone clip of a real product in real light beats a generated one on trust, and trust is what sells handmade. AI earns its place when you have 40 SKUs, no studio, and no time, or when the shot needs a lifestyle setting you cannot build.

The Honest Summary

AI video does not make your Etsy shop work. Your product, your photos, and your listing copy do that. What this workflow removes is the excuse that video is too expensive or too technical to bother with, when Etsy is asking you for twelve silent seconds.

Start with one listing. Your best seller, the one where people keep asking the same question in the messages. Answer that question in twelve seconds, mute it, upload it, and watch what happens before you batch the other 39.

If you want the full system I use to plan, prompt, and ship AI video without burning credits on renders you throw away, that is what the AI Video Blueprint ($27) covers. It is the same process, just applied to reels instead of listings.

Written by Michele De Vivo, AI video producer. I make cinematic AI video for brands and break down the workflow on Instagram and YouTube.
Last updated: July 31, 2026.