The zombie love story AI prompt (full text and shot list)

The full prompt (copy and paste)

This is the production prompt behind the Zombie love story scene in our generator. "Image 1" is the survivor and "Image 2" is the person who turns. It is written for an image-to-video model that accepts two reference photos and generates audio.

Create a 12-second photorealistic cinematic zombie love story using the two uploaded character reference photos. Character A (Image 1) is the human survivor. Character B (Image 2) is the infected lover. Maintain both characters' facial identities throughout.
0–3s: Inside a dark, abandoned kitchen, Character A tearfully aims a handgun at infected Character B, who has pale skin, milky white eyes, and a frightening expression. Cold blue-green cinematic lighting, emotional close-ups.
3–5s: Character A breaks down crying, lowers the gun, and opens both arms wide. Character B suddenly rushes aggressively toward Character A.
5–7s: Just before impact, a dramatic match cut transforms the scene into a beautiful golden-hour alpine meadow. Character B is now healthy and smiling, running into Character A's arms. Character A catches and embraces them.
7–10s: The happy couple spins together, tumbles gently onto the grass, laughing and holding each other lovingly. Warm golden sunlight, distant mountains, romantic cinematic atmosphere.
10–12s: Cut to a wide sunset shot of the couple sitting together on a grassy mountain ridge, leaning against each other. The camera slowly pulls back to reveal their silhouettes against the sunset.
Photorealistic cinematic quality, natural movements, consistent faces, emotional storytelling, dramatic contrast between cold horror and warm romantic memories.
No text, subtitles, watermarks, logos, extra characters, distorted faces, gunfire, or graphic violence.

Audio Direction:

Generate an original, emotionally powerful cinematic instrumental soundtrack synchronized with the visual storytelling.

0–5s: Soft, sorrowful piano and low, suspenseful strings accompany the tragic zombie confrontation. Subtle breathing, trembling sobs, and distant atmospheric wind.

5–6s: The music builds into an emotional orchestral swell, perfectly synchronized with the zombie's rush and the dramatic match cut into the romantic memory.

6–12s: Transition into warm, beautiful piano and gentle soaring strings, creating a nostalgic, bittersweet feeling of love and loss. The soundtrack softly resolves during the final mountain sunset.

Include subtle natural sound effects, but keep the music as the emotional centerpiece.

No dialogue, no singing, no lyrics, and no imitation of recognizable copyrighted melodies or commercial songs.

Skip the prompting: our generator runs this prompt for you, picks the beats for the length you choose (5–15 seconds), and refunds the credits if a render fails.

Shot list with timestamps

The trend works because it always moves through the same four beats. Keep the timings tight: if the confrontation runs long, the memory has no room to land.

Shorter cuts drop beats rather than squeezing them. Our 5- and 8-second versions end on the embrace and skip the sunset; the 15-second version lingers on eye contact before the cut. The showcase page breaks the sample down second by second.

Why each line is there

• "Maintain both characters' facial identities throughout." Image-to-video models drift toward a generic face, especially after a cut. Saying it once at the top and again at the end keeps both people recognizable.
• Explicit seconds per beat. Models follow timestamps far better than words like "then" or "later". Without them the confrontation usually eats half the clip.
• "A dramatic match cut." The cut from horror to memory is the emotional turn. Naming it as a cut stops the model from morphing the kitchen into a field over two seconds.
• A negative list. "No gunfire, no graphic violence, no text, no extra characters" removes the most common failures: a muzzle flash, burned-in subtitles and a stranger in the background.
• Separate audio direction. Seedance generates sound with the picture. Timing the music to the beats (swell at 5s, resolve at 12s) is what makes the clip feel edited.

Running it in other models

We run this prompt on Seedance through our generator. The notes for other models below describe what to change; they are not test results, because we have only run it on Seedance.

Why the gun beat sometimes gets filtered

Video models run safety filters on both the prompt and the output, and a weapon pointed at a person can trip them. The prompt is written to make the intent unambiguous: the gun is lowered, nobody is hurt, and the negative list says "no gunfire, no blood, no graphic violence".

• Keep the gun pointed and lowered, never fired. Do not describe a trigger pull, a shot or a wound.
• Avoid words like "kill", "shoot" and "headshot" in your own direction.
• Use photos of adults only. Our terms forbid photos of minors.
• If you would rather avoid the gun entirely, choose Your own story in the generator and describe the standoff without a weapon, for example a raised flashlight or a locked door.

On our site every photo and every finished video passes an automated moderation check. If the provider or the check rejects a render, the credits go back to your balance automatically.

Adding your own direction

The generator has an optional direction box (up to 600 characters) that is appended to the preset prompt. Short, concrete additions work best:

• "Pouring rain in the kitchen, wet hair" (weather)
• "She is wearing a wedding ring; close-up on it as she lowers the gun" (a detail)
• "The memory is a beach at sunset instead of mountains" (a different memory)
• "Both wear the same clothes as in the photos" (wardrobe)

Every option costs the same as the plain preset: $6.99 for one 12-second video, or $2.83 per video in the Crew Pack (35 videos for $99.00). See pricing for the other models.