Zombie scene AI is the part of this pipeline people underestimate. The effect supplies the story, but the scene, the mood, and whether the shot reads as cinematic at all come almost entirely from the photo you feed it. This page is about that half of the job: what actually makes a short clip feel like a scene rather than a filter, how to compose the source photo for it, and a set of scene ideas you can copy.
What makes a clip read as a scene
A five-second clip becomes a scene when it has three things: one clear subject, one setting the viewer can name, and one change. The subject is the face or the animal, and it has to be legible. The setting is where the shot is supposed to be, and it should be readable without explanation. The change is the turn the clip is built around, whether that is a transformation, a hug, or a cut into daylight.
Strip any of the three and the clip flattens. A clear subject with no setting is a portrait. A setting with no change is wallpaper. The change with no readable subject is an effect demo. The effects on this site are designed to hit all three, but only if the photo gives them something to work with.
Composing the source photo for a scene
The model reads the geometry of your photo, so the way you frame it decides the geometry of the scene.
Fill the frame with the subject
A face that occupies a good share of the frame gives the model detail to animate. A person standing small in a wide landscape gives it almost nothing, and the clip will bury the likeness. Crop in before you upload.
Keep the background simple and nameable
A porch, a car at night, a kitchen, a wheat field: these are settings a viewer reads in a glance. A cluttered room full of half-recognisable objects muddies the scene, because the model has to guess where the shot is. One clean backdrop beats a busy one every time.
Use light that already tells a story
Golden-hour light says warm memory. A single window or a car interior says intimacy. Flat, even daylight says neutrality, which is a fine default when the scene is the point rather than the mood. Whatever the light is doing in the source photo, it will carry into the clip.
Keep it to one subject
Two faces make the model choose, and the result gets vaguer. If the scene calls for two characters, that is the effect's job, not the photo's; feed it one clear subject and let it build the pair.
Scene ideas you can copy
These are briefs, not prompts, and they are grouped by the effect that fits. Each one describes a single moment, which is the right scale for five seconds.
Love story scenes
- A survivor aims a weapon at a turned partner on a front porch, then lowers it and hugs them before the shot cuts to a summer memory.
- A couple in the same room before and after the outbreak, the second half shot through the first, curtains open onto daylight.
- Two friends in a car at night, one turned in the passenger seat, a hand on the shoulder, then a cut to a bright road trip.
Pet scenes
- A dog on a porch at dusk, turned and staring at the camera, mist rolling across the boards.
- A cat on a kitchen windowsill, eyes milky, rain on the glass behind it.
- A large dog standing in a wheat field at golden hour, one eye catching the light, turned.
Costume scenes
- A cracked-porcelain look, pale base with fine cracks across the cheeks and a run of red at the mouth.
- An undead bride, chalk skin, dark veins climbing the neck, a veil casting a shadow over the eyes.
Scale scenes
- A person walking a normal street, scaled up until their shoulders reach the shop signs.
- A thumb-sized copy of someone standing on a kitchen counter while the full-size them looks on.
Which effect fits which scene
The scene idea should decide the effect, not the other way around. If the brief needs a second character who turns and is embraced, that is the zombie love story. If the subject is an animal, it is the zombie pet effect. A scene built around makeup is the AI Halloween makeup effect, a scale joke is make-things-giant, and a shrunken double is the AI mini me effect.
Doing it in that order stops the most common frustration: uploading a photo and then trying to force a scene onto it, when the scene you actually wanted asked for a different picture to begin with.
Why the photos still matter more than the words
However good a scene idea sounds, the clip is only as cinematic as the photo behind it. A brief like "a survivor lowers a weapon and hugs a turned partner" is a plan, not a magic spell; the model still needs one clear face to build it from. That is why the practical workflow is scene first, photo second.
It also helps to change one variable at a time. If a scene does not land, change either the photo or the scene idea, not both, or you will not know which one helped.
What a scene costs
One clip is 300 credits, and a clip is five seconds. The entry tier is $4.99 for a single clip, $9.99 for five, and $19.99 for eleven, and the same prices apply to the monthly plans. The studio shows the credit cost before you generate, video is paid from the first clip, and the file you download has no watermark on it. The pricing page lists the tiers.
Build your first scene
Pick one idea from the list, match it to the effect it needs, and upload the photo that idea calls for. The studio shows the cost before you generate, so you can see which of your photos produces the scene you pictured. Start with the zombie video generator if you want the love story, or flip through the zombie AI video prompts for more copyable briefs before you commit.