Zombie love story AI prompts you can copy
These are the prompts we wrote for the trend. Paste them into any video model that accepts two reference images. Image 1 is the person who turns; image 2 is the survivor.
The full story in one prompt
Use a 12 to 15 second, vertical 9:16 setting. If your tool numbers references differently, keep the same order: the one who turns first.
Cinematic vertical short film, 12 seconds, a zombie-apocalypse love story. 0-3s: In a grey, abandoned city street at dusk with drifting fog, the person from image 2 steps back in fear as a figure walks out of the fog: the person from image 1, now a zombie. 3-6s: The person from image 2 recognises them, tears falling, stops and opens their arms. 6-8s: The person from image 1 rushes forward; at the moment of contact the scene cuts. 8-12s: Hard match cut to the same embrace on a beach at sunset in warm orange light, both healthy and laughing. Keep both faces, hairstyles and body shapes recognisable in every shot. The zombie look is only pale grey skin, dark veins and clouded eyes. No blood, no wounds, no biting, no text on screen. Handheld camera, shallow depth of field, film grain.
Pet version
Cinematic vertical short film, 10 seconds. In an empty small town at night under falling snow, the person from image 2 sees the pet from image 1 walking out of the dark as a zombie: pale fur, clouded eyes. The person kneels, tears falling, and opens their arms. The pet runs to them; at the moment of contact, hard match cut to the same hug on green grass in warm summer light, the pet healthy and happy. Keep the pet's markings and the person's face recognisable. No blood, no wounds, no biting, no text.
Two-part version for editing
Generate the two halves separately and join them in an editor at the hug. This gives you more control over the cut.
Vertical 9:16, 6 seconds. Abandoned highway at dawn in cold blue light. The person from image 2 faces the person from image 1, who is now a zombie with pale grey skin and clouded eyes. Recognition, tears, arms open; the zombie rushes forward and they begin to embrace. Keep both faces recognisable. No blood, no biting.
Second half: the memory
Vertical 9:16, 5 seconds. The person from image 1 and the person from image 2, both healthy and smiling, finishing the same embrace at a sunny summer picnic in golden light, slow push-in, laughing together. Match the pose and camera angle of a hug that is already in motion.
Tips that make the prompts work
- Say which image turns. Models swap roles when it is left open.
- Describe the zombie look narrowly. Asking for gore often triggers safety filters or ruins the faces.
- Name the cut: “hard match cut to the same embrace” keeps the pose continuous.
- Results vary between models and runs. Expect to try a few times, and check faces before you post.
Questions
Which model works best?
Models that accept two or more reference images and generate 10+ seconds keep both faces best. Our own generator uses Seedance 2.0 Mini.
Can I add a gun to the prompt?
Many viral versions show a pistol that is lowered, never fired. Some tools refuse weapons; the version without one tells the same story.
Updated 2026-10-15