Preparing Two Photos for a Zombie Couple Video
Match lighting, crop and face angles before combining two adult portraits.
Zombie Couple Video uses two separate portrait uploads. Photo 1 is the human survivor and photo 2 is the zombie partner. The photos identify the two people; the selected template supplies the reunion scene. You can use portraits taken separately, and you do not need a source video or zombie makeup.
Prepare one adult portrait for each role
Choose a clear front-facing or slight three-quarter portrait with visible eyes, nose, mouth and chin. A simple expression and even lighting give the model more useful facial information than a dark screenshot or a heavily filtered selfie. Use JPEG, PNG or WebP in the generator.
Each upload should show one intended adult. If you only have a couple or group photo, crop it into two separate images before uploading. Keep each face large enough to inspect, with the head and chin inside the crop. The current workflow does not automatically select two people from a group picture. Do not mirror a face to make it fit a slot.
| Input | Intended role | What to check |
|---|---|---|
| Photo 1 | Human survivor | The adult who should retain a human appearance |
| Photo 2 | Zombie partner | The adult who should receive the fictional zombie styling |
Use the photo-order control to exchange the roles, then reread the labels. A role assignment guides the model; it does not guarantee a fixed left/right position in the finished frame. See photo roles and placement.
Compare the two portraits before uploading
- Keep face size and lighting reasonably similar. A sharp daylight portrait paired with a blurred night image gives uneven references.
- Remove sunglasses, masks and obstructions when choosing a source. Avoid extreme side profiles and heavy beauty filters.
- Use a separate portrait instead of enlarging a tiny face until its features become unclear.
- Choose photos that represent the people as they want to appear. The generator can reinterpret clothing, makeup, age cues and backgrounds.
There is no need to create a special portrait crop ratio that matches the final video. The output request is 10 seconds in a 9:16 vertical frame. A portrait is an identity reference, not a locked first frame or an exact clothing replacement.
Confirm uploads and permission
Get both adults' permission to upload their photos for AI generation, and agree separately on whether the result may be shared. Avoid sensitive images and keep source-upload URLs out of public posts. References pass through configured storage and the AI provider; read photo processing and privacy.
Wait until both uploads finish. An image visible in a local preview is not enough if its upload reports an error. Retry that upload or choose a fresh file before submitting a generation.
Choose a scene, then change one input at a time
Open the six-template collection, watch the supplied scene previews and choose a setting. They show the intended reunion direction, rather than a personalized free test of your portraits. The current presets pair one human with one zombie; see what transformation means here.
Keep the scene and quality fixed while replacing one portrait. That makes it easier to judge whether the reference helped. Every new generation uses the displayed credits: 20 at 480p, 45 at 720p or 80 at 1080p. A higher tier starts another generation and cannot guarantee the same performance or a better likeness. Review the credit examples first.
Compare both faces throughout the result, including head turns and the embrace. If identity changes during contact, use the face-drift checklist. These are preparation guidelines, not a measured promise of face accuracy or a report of a new paid test.