A recognizable pair and a small, readable action matter more than complicated effects. This guide walks through the free preview, explains which photos to choose, and shows how to avoid spending an attempt on a preventable mistake.
The on-site tool requests a 5-second, 480P preview. It creates one short reunion scene. The longer story workflows and five prompts are collected on this page; a multi-shot film is not included in the free preview.
1. Choose two clear adult portraits
Use a separate photo for each person, with permission from both adults. A face that fills a useful part of the image, even lighting and a simple background give the model less to guess. A front-facing or gentle three-quarter angle is easier to interpret than a tiny face in a group shot.
Photo 1: the human survivor. This character stays human. Keep their eyes, mouth and outline clearly visible.Photo 2: the zombie partner. This character receives subtle zombie makeup. These examples use fictional AI-created adults.
Prefer
Avoid
Why it helps
One visible face
A crowded group picture
Reduces ambiguity about which person to use.
Even light and sharp features
Backlighting, blur or sunglasses
Keeps facial details available to the model.
A relaxed expression
Heavy filters or extreme poses
Leaves room for a small expression change.
Similar framing for both portraits
A close-up paired with a distant full-body photo
Gives both people a comparable level of detail.
Images may be up to 10 MB each. At least 512 × 512 pixels is recommended. The preview shows the entire selected photo; before submission it is fitted onto a square reference with padding, without cropping. That preserves the input, but does not guarantee the model will reproduce every detail.
2. Assign the roles, scene and video format
Upload to the correct role. Choose the human survivor first or second—it does not matter. Each role has its own photo picker, replacement and remove control.
Choose the setting. Classic reunion uses a candlelit country house at dusk. Rainy city uses a deserted wet street at night. Golden memory uses a ruined garden in warm light. Golden memory is a setting, not a promise of a separate flashback shot.
Choose the frame. Portrait (9:16) fits a phone screen. Landscape (16:9) leaves more room beside the pair. Output dimensions can vary slightly with the model; 480P is a resolution tier, not an upscale.
Complete human verification. Generation unlocks only when both photos and verification are ready. Read the current Usage Limits before starting.
Generate, then inspect. Watch both faces and hands. Listen to any generated audio before sharing. AI output varies; do not assume that every attempt will preserve likeness or motion perfectly.
Download promptly. Keep the MP4 on your device because provider links are temporary. Story HD is a separate $5.99 USD one-time purchase for an approximately 15-second, 768P story, in portrait or landscape. It creates a new video rather than extending this free preview. Compare the two options.
Some iPhone photos use HEIC or HEIF. The picker accepts these files when the browser can decode them, then prepares a JPEG locally before upload. Support depends on the browser and device; a filename extension alone does not make the image readable.
If the photo cannot open, export a JPEG or PNG copy and select that copy. On a Mac, Preview can open the photo and export it as JPEG. On an iPhone, use a trusted photo export or conversion workflow you already have. Renaming .heic to .jpg does not convert the file.
A conversion failure does not send a generation request. Your other selected photo remains in place. Check orientation and the full preview before proceeding.
Fix a problem without accidentally starting over
The generate button is disabled
Check that both role cards contain a photo and the human-verification message says it is complete. If you have a saved task with a connection error, use Check existing task first. This checks progress and does not start another generation.
I refreshed or closed the page
Reopen this site in the same browser. It can recover the latest submitted task for up to 24 hours while its original session cookie and browser storage remain available. You do not need to upload the photos again to check that task. Photo selections themselves are not stored and cannot be restored.
Recovery does not transfer to another device or browser. Cleared site data, private-browsing cleanup and expired provider files can prevent recovery. Make another or Forget this task removes the local recovery record.
The connection failed after I pressed Generate
Do not repeatedly submit. The provider might already be working on the original request. The site saves the attempt identifier before submission and checks for the existing task. If its outcome remains uncertain, wait or contact us with the approximate time and error message.
The result is not what I expected
Check the role assignments and photo quality first. For an unclear face, use a sharper, more evenly lit reference. For confused identities, avoid group photos. The short preview supports a restrained reunion; complex action is better tested in a separate longer workflow.
Download does not save a file
Use the offered link to open the video and your browser’s save controls. On a phone, check where your browser stores downloads. The site can observe that a download was initiated, but cannot verify that the operating system saved the file.
Plan a longer zombie love story
Build a short storyboard before buying or generating multiple clips. A useful sequence is recognition, hesitation and an embrace; a memory can follow if the timing allows it. Keep the same role assignments, clothing, screen direction and lighting cues across shots.
Test the hardest transition first. If a cut changes someone’s identity, simplify the makeup and movement, then test again. Check the audio transition too: an abrupt music change or unexpected voice can distract from an otherwise convincing image.
The two longer-story workflows and five editable shot prompts are provided below. These are planning resources, not extra features promised by the free preview. Use only music and photos you have permission to use.
01 / The story behind the scene
Why This Trend Exploded
The zombie couple video puts a familiar question into an impossible scene: would you still recognize the person you love if everything about them changed? Its appeal comes from four emotional beats that make sense even without dialogue.
One person turns. A partner becomes a zombie, but their face is still recognizable. Fear arrives before the other person can process the loss.
The other raises a gun. Survival seems to demand a simple choice. The camera makes the distance between them feel painfully small.
They cannot shoot. The weapon trembles, then lowers. Recognition interrupts the expected horror ending.
They embrace, and memories return. A hug cuts to warmer moments together, changing the scene from a threat into a zombie love story.
That reversal gives the format its emotional pull. Two recognizable faces make the question personal; a brief flashback gives the ending room to linger.
02 / Choose your approach
How to Make It
Start with the people, then build the atmosphere around them. Decide who turns, who hesitates, and which ordinary memory will close the story. Keep those roles consistent from the first frame to the last. You can use a prepared story template or create each shot yourself; the right route depends on how much control you want over the edit.
Route A / The simplest setup
Use an AI video generator
If your chosen AI video generator offers a two-photo story template, you can submit the scene in one generation step. Check that the template supports both people and the sequence you want before uploading.
On-site free preview.Create a 5-second, 480P preview above. The guide below describes longer storytelling workflows; those are separate from the free preview.
Prepare two clear, front-facing photos: one of each adult participant. Keep the eyes, nose, mouth, and jaw visible. Avoid sunglasses, heavy beauty filters, motion blur, and hands covering the face.
Use soft, even light and simple backgrounds. A window-lit portrait against an uncluttered wall is a useful starting point. Avoid deep shadows across one face or bright backlighting that hides facial detail. Similar lighting between the photos makes the references easier to compare.
Assign the roles and choose the mood. Make clear which reference becomes the zombie. Aim for an approximately 8–10-second finished video; available output lengths depend on the generator, so trim a longer result if needed.
Review before sharing. Look for identity changes, changing clothes, extra fingers, and abrupt transitions. A short, readable story matters more than a crowded scene with elaborate effects.
Route B / More creative control
Build your own prompt sequence
For a more personal AI zombie video, separate the process into still images, motion, and editing. One prompt per shot gives you a clearer way to adjust a weak moment without rebuilding the entire story.
Create the still frames first. Use the same two photo references throughout. Choose consistent outfits, a ruined street, and cool evening light. Establish each composition before asking for movement. The flashback can switch to warm daylight while keeping the same faces.
Animate with image-to-video. Feed each approved still into an image-to-video workflow. Describe a single main action and a restrained camera move. A trembling hand or a slow approach is easier to read than several competing actions in one clip.
Edit the clips and add music. Arrange the transformation, standoff, hesitation, and embrace in order. Cut to the memory at the moment their arms meet, then hold the closing frame. Use music you have permission to use.
Set the final pacing. Treat 8–10 seconds as a compact editing target, rather than the required length of every generated shot. Leave enough time for the hesitation to register and shorten the surrounding footage.
03 / Your shot-by-shot starting point
5 Copy-and-Paste Prompts
Use these prompts as a connected sequence. Reference photo A is the partner who remains human; reference photo B is the partner who transforms. Replace the wardrobe or location details with your own choices, then repeat those choices across every shot. The prompts describe the intended result, so check which reference-image and motion controls your chosen tool actually supports.
For a still-image workflow, use the composition and appearance directions to establish the frame. For image-to-video, start from that frame and emphasize the action and camera movement. A zombie couple video prompt works best when the emotional beat is specific: recognition, hesitation, or relief. Broad instructions such as “make it epic” leave the most important part of the scene undecided.
Shot 01 / The transformation
The moment they turn
A slow, subtle transformation, ending with the two partners making eye contact.
Read full promptHide full prompt for shot 1
Create a cinematic, non-graphic zombie transformation shot using two adult photo references. Photo A is the human partner in a charcoal jacket; photo B is the partner who transforms, wearing a faded brown coat. Preserve their recognizable facial features and keep their identities separate. Set the scene on an abandoned city street at blue hour, with light fog, scattered leaves, and distant amber streetlights. Frame photo B in a medium close-up while photo A watches from the soft-focus background. Photo B pauses, exhales, and slowly develops pale skin, subtle dark veins, and clouded eyes while retaining a trace of recognition. Use a gentle camera push toward the face, shallow depth of field, and restrained movement. End on eye contact with photo A. No blood, exposed wounds, text, logos, or exaggerated facial distortion. Widescreen 16:9.
Clipboard access is unavailable. The prompt is selected; use your device’s copy command, or select the text manually.
Shot 02 / The confrontation
A familiar face across the gun
An over-the-shoulder confrontation: fear gives way to a first sign of recognition.
Read full promptHide full prompt for shot 2
Create the next shot of a cinematic zombie love story using the same two adult photo references. Photo A remains human in a charcoal jacket; photo B is a pale, clouded-eyed zombie in a faded brown coat. Keep both faces recognizable and their outfits consistent. On the same abandoned street at blue hour, place the partners a few steps apart in light fog. Use an over-the-shoulder composition from behind photo B, keeping photo A's face and upper body in focus. Photo A slowly raises a fictional handgun toward the approaching partner, then holds still, visibly frightened and grieving. Photo B stops and looks back with a faint expression of recognition. Move the camera slightly closer without shaking or changing the viewpoint. Keep the scene non-graphic, with no firing, impact, blood, text, or logos. End before photo A makes a decision. Widescreen 16:9.
Clipboard access is unavailable. The prompt is selected; use your device’s copy command, or select the text manually.
Shot 03 / The choice
They cannot pull the trigger
A close-up of the human partner choosing recognition over fear and lowering the weapon.
Read full promptHide full prompt for shot 3
Create an intimate hesitation shot using the same two adult photo references and established street setting. Photo A is human, wearing a charcoal jacket; photo B is the recognizable zombie partner in a faded brown coat. Begin with a close-up of photo A's face, with the raised fictional handgun visible at the bottom of the frame and photo B softly blurred beyond it. Photo A's eyes fill with tears, their breathing slows, and their expression changes from fear to recognition. Their hands tremble slightly before they lower the weapon out of frame without firing. Photo B remains still rather than lunging. Use cool blue evening light, a soft amber rim light, subtle film grain, and an almost stationary camera. Preserve facial proportions and natural hands. End on photo A looking directly at their partner. No blood, violence, text, logos, or sudden camera movement. Widescreen 16:9.
Clipboard access is unavailable. The prompt is selected; use your device’s copy command, or select the text manually.
Shot 04 / The embrace
A hug becomes a memory
A gentle embrace that cuts to a warm memory of the same couple.
Read full promptHide full prompt for shot 4
Create the emotional embrace in a cinematic zombie couple video using the same two adult photo references. Photo A remains human in a charcoal jacket; photo B is the pale zombie partner in a faded brown coat. On the abandoned street, photo A opens their arms and takes one slow step forward. Photo B approaches gently, and they hold each other with relief rather than aggression. Keep both faces identifiable, hands natural, and body positions physically coherent. Frame the hug in a steady medium two-shot. As their arms meet, match cut to a warm memory of the same embrace in a sunlit meadow: both partners are healthy, wearing the same outfits, smiling softly in golden light. Maintain matching framing across the cut. No biting, blood, weapons in the embrace, text, or logos. If the transition is unsupported, generate the two matching scenes separately for editing. Widescreen 16:9.
Clipboard access is unavailable. The prompt is selected; use your device’s copy command, or select the text manually.
Shot 05 / The last frame
Let the memory linger
A quiet closing moment in the meadow, with a steady frame to hold in the edit.
Read full promptHide full prompt for shot 5
Create a closing memory shot using the same two adult photo references from the previous scenes. Both partners are healthy and recognizable, with photo A in a charcoal jacket and photo B in a faded brown coat. Continue the sunlit meadow setting immediately after their embrace. Show them resting their foreheads together, eyes gently closed, with relaxed expressions and a light breeze moving their clothes. Use a close medium two-shot, warm backlighting, softly blurred grass, and subtle film grain. Keep the camera steady and movement minimal so the final moment can be held in the edit. Preserve consistent faces, natural anatomy, and the position of their hands. Finish on a calm, balanced composition with room around their silhouettes. No zombie effects in this memory, no text, logos, blood, or dramatic camera moves. Hold the last frame during editing for a quiet ending. Widescreen 16:9.
Clipboard access is unavailable. The prompt is selected; use your device’s copy command, or select the text manually.
Before generating all five shots, test the transformation and the hesitation together. Compare the faces, wardrobe, and direction of the light. If the result loses a person’s identity, simplify the zombie makeup and reduce motion before adding more detail. For vertical delivery, change every prompt to the same 9:16 composition and reframe the stills; changing the ratio only at the end can crop away the expressions that carry the story.
04 / Before you make your story
Frequently Asked Questions
How did the zombie couple video trend become popular?
The format has an immediately understandable conflict: someone must choose between survival and the person they love. The raised gun sets up a horror ending, then the hesitation and hug overturn that expectation. A warm flashback adds a second interpretation, inviting viewers to think about what the couple had before the world changed.
That is an explanation of the story’s appeal, rather than a claim about a single original upload. Putting a couple’s own faces into the same sequence makes it easy to reinterpret the idea as a personal message. You do not need to reproduce another creator’s exact edit or soundtrack to tell your version.
How many photos should I prepare?
Prepare two photos: one clear, front-facing portrait of each person. Choose adults who agree to have their images used, and label the references so the human and zombie roles stay consistent. You can keep an extra photo of each person nearby if your chosen workflow accepts additional references, but two portraits are the starting point for this guide.
A recent, evenly lit image is more useful than a heavily edited selfie. Avoid crowded group pictures that leave the model guessing who the subject is. If your workflow accepts only one starting image, compose both people into an approved still before animating it.
Which AI model works best for this video?
There is no universal best model for every pair of faces or every shot. Compare the qualities that matter here: recognizable identities, two people remaining distinct, natural hands during the hug, controlled camera movement, and a believable change in expression. A dramatic background is less valuable if the partners stop looking like themselves.
Test the same short scene with your own references before committing to the full sequence. Check whether the workflow supports image-to-video, reference images, and your preferred aspect ratio. Choose based on the result and the controls you can actually use, rather than assuming a model name guarantees a convincing zombie love story.
Will the generated video blur or censor the faces?
Face blurring is not something this guide can promise or rule out. Output behavior depends on the service, its settings, and the material you submit. A service may reject a scene, modify it, or produce a soft face as a generation artifact; these are different outcomes, so inspect the preview rather than treating all of them as censorship.
Keep the intended effect non-graphic and ask for recognizable features, as the prompts do. If a face becomes muddy, try clearer references and gentler motion. If the service blocks the scene, review its rules and adjust the concept within them.
How long does it take to make one?
The free on-site tool creates a 5-second, 480P preview. Five seconds is the requested clip length. Generation usually takes a few minutes, depending on the service’s queue. The tutorials on this page also describe longer stories that you can assemble separately with your own editing workflow.
Allow time to review each face and transition before sharing. The quickest useful test is one short shot, followed by a second shot that checks continuity. Once those work, build the rest around them. Save your accepted stills and prompt text so you can revise a single scene without starting the whole project again. For on-site free preview allowances, see the Usage Limits.
Can I use the finished video commercially?
Check the current terms of the generator you use and the rights attached to every input. Permission to use an AI output does not automatically give you permission to use someone else’s photo, likeness, music, or footage. Get agreement from both participants and choose a soundtrack licensed for your intended use, especially for advertising.
Commercial permission and copyright ownership are separate questions, and rules vary by jurisdiction. For a U.S. reference on human authorship and AI material, see the U.S. Copyright Office’s AI guidance. Keep records of source permissions, relevant tool terms, and music licenses with the final edit so you can review them before publishing a campaign.
Before uploading, read our Privacy notice. Prepared photos are sent through our server to fal.ai for generation; free-preview photos are not kept in site photo storage. Story HD references are saved privately before checkout and removed after submission, or within about 24 hours for incomplete orders. Paid video access lasts 7 days from order creation; save your private recovery link. See the free and paid storage details.
Make the full Story HD video
Open Create Story HD, choose two consenting adult portraits, and select portrait or landscape.
Save the photos and copy your private recovery link before paying $5.99 USD through Stripe.
Return with that link to watch and download the approximately 15-second, 768P video. Generation starts after payment confirmation and may take several minutes.
Review the refund policy before purchasing. A confirmed generation failure starts a full refund; an uncertain result needs support review. Appearance, motion and exact timing can vary. No completion email is sent.
