How to Make a
Zombie Couple Video
Two photos in, a six-second zombie love story out. Here's exactly how the trend works — and how to make yours in minutes.
What is the zombie couple trend?
It's an AI video trend on TikTok and Instagram: two people — one turned into a zombie, one still human — meet again in the fog. The zombie walks out slowly, the survivor freezes, recognizes them, and they hug. Six seconds, one emotional beat, endlessly re-makeable with your own couple.
The key thing to understand: it's image-to-video. You don't need a reference video, a template, or any editing. Two portrait photos are the entire input — the AI generates the whole scene from scratch around them.
The 4 steps
Pick your two photos
Clear, front-facing, one person per photo, even lighting. Faces large enough in frame to hold identity through the fog.
Decide who turns
The first photo becomes the zombie — same face, same hair, but pale skin, cloudy eyes and torn clothes. The second photo stays human. Order is the whole decision.
Generate the clip
Use the generator above (no prompt writing), or any image-to-video AI tool with a prompt like the one below. Expect a few minutes of rendering.
Post with sound
AI clips render silent — that's normal. Add a trending audio in TikTok or Reels when you post. The trend lives on the audio + the hug.
The prompt behind the trend
If you're doing it yourself in a general AI video tool, here's the prompt structure that produces the trend's beats (this is the shape our generator uses internally):
"Create a cinematic vertical zombie love story video from the two reference photos. The FIRST photo becomes a zombie: same face, hairstyle and proportions, but pale grey-green skin, cloudy eyes, torn clothes. The SECOND photo stays human. In 6 seconds: the zombie walks out of rolling fog, the survivor freezes, recognizes them, and they embrace. Moody blue-grey lighting, handheld camera, film grain."
The two load-bearing phrases: "FIRST photo becomes… SECOND photo stays…" (role assignment by upload order) and the explicit beat list (fog → recognition → embrace). Without the beat list, models drift into generic zombie chaos.
Do you need a reference video?
No. This is the most common confusion. Some AI video editors have a mode literally called "Reference to Video" — there, "reference" means your reference photos, not a video. The zombie couple workflow never needs an existing clip.
That's also why making one got cheap: two images in, one short clip out. If a tool asks you to find a template video first, you're in a video-editing workflow — a different (harder) trend format.
Free vs. one-click
Two honest routes to the same clip.
DIY with any AI video tool
Free-ish: use an image-to-video model that accepts two reference images, paste the prompt above, mind each tool's daily limits. Expect some trial and error on identity preservation.
One-click here
$4.99 a clip: the story is pre-prompted, identity preservation is tuned, output is 720P vertical with no watermark, and failed generations never charge. You upload two photos and press one button.
What both share
Consent. Both faces come from your photos and stay recognizable — only use photos of people who agreed to appear.