The zombie AI trend, and how to make your own

Two people, one short film: one has turned into a zombie, the other has the gun and cannot pull the trigger. It took off on TikTok and Instagram at the end of September 2026. Its signature is a match cut: the zombie leaps at the survivor, and mid-air the film cuts to a happy memory where the same leap lands as a hug.

The four shots

The videos run 12 to 15 seconds. Everything builds to the cut: dark and grey until the leap, then warm light and the two of them alive.

Make it from two photos

1. Add a photo of the survivor (the one holding the gun) and a photo of the one who turns.
2. Choose where the memory happens and vertical or wide.
3. Pay $9.90. The video is ready in about five minutes, and we email you the link.

Make my zombie video

The prompt behind it

If you want to try it in a video model yourself, this is the shape of the prompt we use: one generation with time codes for every shot, so motion stays at real speed and both faces carry across the cuts. Reference image 1 is the survivor, reference image 2 the one who turns.

Photorealistic cinematic short film, 35mm film grain, shallow depth of field, natural motion at real-time speed, consistent faces across all shots. No text, no subtitles, no watermark.
Person A is the person in reference image 1. Person B is the person in reference image 2. Keep both faces, hairstyles and builds exactly as in the references.
Shot 1 (0–4s): inside a dim, abandoned wooden cabin, dust in the light. Tearful close-up of Person A behind a raised rifle, hands shaking. Past the barrel, Person B, already turned (ashen grey skin, dark veins, clouded white eyes), shuffles closer across the room.
Shot 2 (4–6s): extreme close-up of Person B as a zombie: grey skin, white eyes, bared teeth.
Shot 3 (6–10s): Person A lowers the rifle, lets it drop to the floor and opens both arms. Person B sprints across the cabin and leaps at Person A.
Shot 4 (10–15s): match cut in mid-air. The same leap continues into a snowy hillside under soft winter sun, both in warm coats and scarves, in warm golden light. Person B, alive and healthy again, lands in Person A’s arms and they hug, laughing. Hold on the embrace.
Sound: cabin ambience and breathing, then a soft, warm score in the memory. No dialogue.

Photos that work

• One person per photo, face filling a good part of the frame.
• Facing the camera or slightly turned; no sunglasses or masks.
• Even light. Night shots with flash work less well.
• No heavy beauty filters: the model copies what it sees.

Questions