AI Zombie Video: Timing, Length, and What Breaks the Clip
How long an AI zombie video should run, how the three beats split the time, and the failures that make a clip look broken instead of sad.
Last updated: 2026-10-09
An AI zombie video is about twenty seconds of finished footage, and almost all of it is decided before you press generate. The model handles the faces; you handle three things — how the two photos are framed, how the runs are divided, and whether the cut between them is hard or soft.
This page is the timing half. If your clip already looks wrong and you want to know why, jump to the failure list.
How Long It Should Run
Twenty seconds, vertical, three beats. Not two, not forty.
- The standoff — 6 to 8 seconds. The two of you facing each other, one already turned. Long enough for the viewer to read what they're looking at, short enough that they don't start looking for the settings. Most clips that feel slow are slow here.
- The memory — 3 to 4 seconds. Sunlight, both faces unchanged. This is the shortest shot and the most important one, and plenty of people cut it to a flash. Don't.
- The lowered gun — 6 to 8 seconds. Back to the standoff, the gun coming down. Give it the room the first shot got.
The two halves should be close to equal. If the standoff runs twelve seconds and the memory runs two, the clip reads as a zombie video with an editing mistake at the end, which is not the same thing as a sad clip.
Where the Cut Goes
Hard cut, on the beat. No fade, no dissolve, no transition. The whole move of this format is putting the same person on screen twice in two opposite states, one frame apart — and a soft transition is the sound of someone sanding that edge off. Every version that works does it with a straight cut.
Cut on a recognizable frame. Cut away from the standoff while the turned face is still readable, not mid-turn and not on a blurred frame. If the memory shot opens on mush, the hard cut reads as a glitch.
Don't cut back inside the memory. No flash of the standoff inserted mid-memory, no drifting back and forth. The three beats run in order, once.
What Breaks a Clip
Six failures cover almost everything, in rough order of how often they happen:
- The zombie look leaks into the memory shot. Grey skin, clouded eyes or a torn collar showing up in the sunlight half. This is the most common failure of the format and it turns a sad clip into a broken one. Re-generate — don't try to fix it in an editor.
- The two photos are framed differently. A tight headshot paired with a waist-up photo gives you a clip that reads as a slideshow. Headshot with headshot, waist-up with waist-up.
- The memory shot is too short. A one-second sunlight flash reads as a scene change, not a memory. Give the two halves comparable time.
- A soft transition where the hard cut belongs. A dissolve kills the emotional snap faster than anything else on this list.
- Two people in one photo. Group shots and screenshots-of-screenshots fail on the first generation nearly every time. One person per photo, eyes visible, straight to camera.
- Heavy filters. The model has to invent the daylight version of a face it can barely read. Take the filters off the input photos.
Four of the six are photo problems, which is why the photo guide is worth ten minutes before you spend anything.
Audio
Keep it to one track with no dialogue. Anything with words competes with the two beats that carry the clip, and anything with a hard drop turns the lowered gun into a punchline.
If you're adding audio yourself: instrumental, low, and rising into the memory shot, then out. Don't reuse a track from someone else's clip — the terms cover it, and on short-video platforms the audio match is what gets a clip muted.
Reels, Shorts, TikTok
Vertical is the default: 9:16, framed so both faces sit in the upper half — that's where the app's interface covers the picture. Nothing else changes between platforms. The template page has presets for length and framing.
What It Costs
No free tier, no subscription. You pay for the generations you run; one payment covers one video; length is 10 or 15 seconds at 720p, and packs are on the pricing page.
Output quality depends on the model and on your two photos. The photos you control — the guide above covers which ones work. Anything outside that list is a coin flip, which is why we ask you to read it before you pay rather than after.
FAQ
Is it free? No — there's no free tier, and no subscription to cancel. You pay per video; the pricing page has the current numbers.
What if I don't like the result? The output depends on the model and on the photos you upload. We don't offer refunds based on how much you like the video, and we don't promise a specific likeness or a specific look. If a file is broken, the payment goes through without a video, or the export fails, that's a different case: write to support@zombieai.video and we'll sort it out.
Why does my clip look like a slideshow? The two photos are framed too differently. Pair headshot with headshot, or waist-up with waist-up, and re-generate.
How long does generation take? Thirty seconds to two minutes. The queue moves at video-generation speed.
Can I fix a bad clip in an editor? The leak and the framing are generation problems — re-generate instead. Timing and audio you can adjust afterwards.
Can I make one on my phone? Yes. Upload and export both happen in the browser.
Ready to time one properly? Start a generation — or see the format breakdown for the beats themselves, shot by shot.