← Back to blog

How to Make App Store Preview Videos with AI

A hybrid workflow for App Store and Google Play preview videos: where Apple's device-capture rule blocks AI, where AI b-roll and localization fit, verified specs, and cost math.

App Store preview videos have a rule that surprises most people the first time they read it: Apple requires the footage to be real on-device screen capture. You cannot fill an App Store preview with AI-generated hero shots of your app, because Apple bans people, hands, fake scenes, and recreated interfaces from the preview itself. That sounds like it rules AI out entirely. It does not. It tells you exactly where AI fits and where it doesn't, and once you know the boundary you can build a hybrid workflow that captures the screen honestly and uses AI for the lifestyle b-roll, the intro and outro, and the localized variants that would otherwise cost a fortune to reshoot per market. Google Play is more permissive and changes the equation. This guide covers both platforms' verified specs, where AI is allowed, and the cost math.

What you'll end up with

A compliant Apple App Store preview built from real device capture with AI handling the parts Apple permits (and the parts Google Play permits, which are broader), plus localized variants for every market at near-zero incremental cost. Knowing the boundary up front saves you from generating footage Apple will reject in review, which is the expensive way to learn this. The AI portions cost a few dollars in compute; the alternative is a motion-graphics studio charging four figures per app preview and more per localized cut.

TL;DR

Verified specs: Apple vs Google Play

The two stores have genuinely different rules, and the difference is the whole strategy.

Apple App Store app preview:

Google Play promo video:

The key difference: Apple's preview must be on-device capture, so AI cannot be the body of it. Google Play's is a YouTube marketing video, which gives you far more room for AI-generated storytelling as long as you never fake the UI or invent features.

Where AI fits (and where it doesn't)

Be honest about the boundary, because it's what keeps you out of App Store rejection.

Think of it as: the screen is real, the story around the screen is AI. On Apple you have a narrow story lane; on Google Play it's wide.

The hybrid workflow

Step 1: Capture the real screens

Record clean, high-resolution screen capture of the actual app flows on-device. This is your compliant core for Apple and your product-truth footage for Google Play. It has to be real, so make it good: smooth navigation, the strongest features first, no dead time.

Step 2: Generate lifestyle b-roll and brand beats

For the human and context shots the stores allow (lifestyle around the app for Google Play, intro/outro and permitted context for Apple), generate with Veo 3.1 for cinematic native-audio beats, or Nano Banana Pro stills animated with Seedance 2.0 for cheaper context.

Tested prompt, a lifestyle beat:

A person smiling while checking their phone at a sunny cafe table, relaxed and satisfied, natural candid motion, shallow depth of field, warm daylight, phone screen not visible. Cinematic, 9:16 vertical, 5 seconds.

Note the phone screen is deliberately not visible, so this shot never implies a fake UI. That keeps it clean for both stores.

Step 3: Assemble to each spec

Cut the Apple preview at 15 to 30 seconds, device resolution, with the real screen capture as the body and AI beats only as permitted framing. Cut the Google Play video up to 120 seconds landscape, where AI b-roll can carry more of the narrative around the real screen footage. Burn in captions for both, since both can play muted.

Step 4: Localize at scale

This is where AI earns its place even under Apple's tight rules. Every market needs its own preview with localized captions and often localized lifestyle b-roll. Regenerating the AI lifestyle beats with region-appropriate context and swapping caption text costs cents per market, versus reshooting a lifestyle scene per country. Apple allows up to 3 previews per localization, so you have room to test per market.

Cost math vs a studio

Here is a hybrid preview with real screen capture plus AI framing, then localized to five markets:

Asset Model Cost
Screen capture On-device recording $0
Intro/outro brand beats Veo 3.1 ~$2.00
Lifestyle b-roll (3 clips) Seedance 2.0 ~$1.60
Context stills Nano Banana Pro ~$0.30
5 localized b-roll variants Seedance 2.0 ~$2.70
AI portion total ~$6.60
Equivalent studio, base + 5 locales $5,000-$15,000+

Roughly $7 in compute plus your own screen recording replaces a motion-graphics studio for the base preview and every localized cut. A studio typically charges four figures for one app preview and more per localized version, which is why most apps ship one English preview and skip localization entirely. AI is what makes per-market previews affordable. Budget human time for the capture and edit.

FAQ

Can I use AI-generated video in an App Store preview?

Only for permitted framing, not for the app interface. Apple requires the App Store preview to be real on-device screen capture and bans people, hands, fake scenes, and recreated interfaces in the preview, so AI cannot be the body of it. AI is fine for permitted intro/outro and context beats and, on the more permissive Google Play, for lifestyle b-roll and narrative around real screen footage, as long as you never fake the UI.

What are the specs for App Store and Google Play preview videos?

Apple app previews are 15 to 30 seconds, device-specific resolution, delivered as .mov, .mp4, or .m4v with H.264 and AAC under 500 MB at 30 fps max, up to 3 per localization, and must be real screen capture. Google Play promo videos are 30 to 120 seconds, delivered as a public or unlisted YouTube URL shown landscape, with no fake UI or unsupported feature claims and captions recommended for muted autoplay.

How much does an AI-assisted app preview video cost?

The AI-generated portions (intro/outro, lifestyle b-roll, context, and five localized variants) run roughly $7 in model credits on 8frame, plus your own free screen capture. A motion-graphics studio typically charges four figures for a single app preview and more per localized version, which is why AI-assisted localization is where the savings are largest.


Capture the screen honestly and generate everything around it. Open a canvas on 8frame with Veo 3.1 and Seedance 2.0, cut a compliant Apple preview and a longer Google Play video from one set of assets, and localize every market for cents. Reusable workflows live at /workflows.

Related articles

use caseHow to Make an ASMR Product Video with AIuse caseHow to Make a Kickstarter Video with AIuse caseHow to Make a Tech Product Video with AI

Make it
move.

Stay in the loop

Be the first to hear about our launch and get product updates