AI Zombie VideoThis guide explains preparation for the current studio. The guide covers input selection, framing and role assignment.
Start with two separate portraits
Prepare one photo of each person. The studio has two input slots, so a single couple photograph is not the same as providing the two required portraits. Choose files that let you identify each face independently, without another person competing for attention in the frame.
You do not need zombie makeup, tears or a ruined background in the reference. Use ordinary portraits. The configured workflow supplies the fictional styling and story instructions. A natural expression and a readable face give you a useful reference for evaluating the delivered video; they do not guarantee that the model will reproduce every detail.
Quick preparation list: two files, one adult in each, a clearly visible face, permission from both people, and a final check of the role labels. You do not need to prepare a text prompt.
If your only source is a group photograph, use an image editor you trust to prepare a separate, usable portrait of each adult. Check that the crop still contains enough detail; enlarging a tiny face does not recover information that was never present. Do not include a cropped-out stranger, a child or another person who has not agreed to this use.
Check faces, eyes and lighting
Look at the image itself before looking at its file size. Can you see the eyes, facial outline and expression? Is the face in focus? Are strong shadows or objects hiding important features?
The preparation examples below illustrate visible versus obscured faces, even versus poor lighting, and unobstructed versus obstructed inputs.
Portrait preparation examples.
| Check | Prefer | Reconsider |
|---|---|---|
| Face visibility | Eyes and facial outline are visible | Sunglasses, masks, an object crossing the face |
| Focus | Facial details are readable at normal viewing size | Motion blur, a very small face or a compressed screenshot |
| Light | Soft, even light across the face | Strong backlight, deep shadow or an overexposed face |
| Editing | A photograph with restrained processing | Heavy beauty filters, face reshaping or pasted features |
| People | One authorized adult | A crowd, an unidentified second face or a minor |
These are practical input checks, not a scoring system. The studio does not promise to repair a blurred face or preserve a likeness perfectly because a photograph passes these checks.
Leave room around the head and shoulders
A tight crop can remove hair, the jaw or other details you will later use to compare the character with the reference. Prefer a photo that includes the complete face and enough head-and-shoulder detail to recognize the person.
Portrait framing guide.
Avoid spending time trying to match the reference photo's background to the apocalypse. The encounter and memory settings are chosen separately in the studio. Those choices guide the scene; a real location in a portrait is not a promise that the same location will appear in the result.
Do not stretch a portrait simply to make it fit the requested video frame. The reference photograph and the delivered video's aspect ratio serve different purposes. You will choose portrait or landscape video framing later.
Match each person to the correct role
The first input is assigned to the Zombie character, and the second to the Human character. The labels beneath the upload slots are the authority for the current draft. Read them after adding or replacing a photo.
The studio's Swap roles control reverses the assignments. You do not need to upload both files again just to change who plays which role. After swapping, check the labels once more before continuing.
| Input or action | What to check |
|---|---|
| First photo | The person intended for the current first-slot role |
| Second photo | The other person, with no accidental duplicate |
| Swap roles | Labels now match the story you want |
| Replace a photo | The new image belongs to the role shown beneath it |
| Final review | Both people agreed to this fictional use of their likeness |
Our current direction is an encounter followed by a separate memory, with no weapons or physical contact in the instructions. Read the story and its limits.
Meet the file requirements
| Requirement | Current studio input |
|---|---|
| Number of files | Two separate portraits |
| People | One authorized adult per file |
| Formats | Still JPG / JPEG, PNG or WebP |
| File size | Up to 2 MiB per file |
| Image dimensions | Up to 20 megapixels per file |
| Animation | Animated files are not supported |
| Custom text prompt | Not required and not accepted by this studio |
A file can be small in bytes while still having excessive pixel dimensions. Conversely, a very compressed file can meet the size limit but contain little usable facial detail. Check both properties rather than treating the upload limit as a quality recommendation.
If a file is rejected, export a still copy in a supported format or reduce its dimensions with your usual image editor. Preserve a separate original. Renaming an unsupported file to end in .jpg does not convert its contents.
Understand photo access before submitting
Choosing a local file creates a browser preview. It does not by itself mean that you have submitted a paid generation. Read the confirmation dialog when you are ready to proceed.
Originals have a separate private archive, but processed photos and delivered videos may use public links. Anyone holding such a link may view the media. This is different from a promise of end-to-end private processing, and a consent checkbox does not make a public link private.
Use photos of yourself or adults who have agreed to this specific fictional use. Permission to possess a photograph is not automatically permission to publish or transform someone's likeness. Do not upload minors or prohibited material. Read the Content Policy and Privacy Policy for the service's conditions.
Stored photos and creations can be managed through Settings → Media and privacy. Deleting a file we control does not necessarily erase a downloaded copy, a provider-held copy or every cache. The privacy policy explains those limits.
A final check before opening the studio
- You have two different, supported still files.
- Each file shows one adult who agreed to this use.
- Both faces are readable, with visible eyes and no heavy obstruction.
- The crops retain enough detail to identify the people.
- You know who should be the zombie and who should remain human.
- You understand that processed media may use public links.
- You will review the actual settings and charge before submitting.
Ready? Open the creation studio. For the next steps, read How to Create and Find Your AI Zombie Video.
Questions about preparing your files
Should I add zombie styling myself?
No makeup or editing is required. Start with an ordinary adult portrait. Existing makeup or a heavy filter can make the original facial details harder to compare with the result.
Can I use one photo of the two of us?
The current studio expects two separate portrait files. Prepare an appropriate individual portrait for each input instead of putting the same couple photograph in both slots.
Does a clear photo guarantee a recognizable result?
No. Clear photos are useful references, but AI can change faces, anatomy and expressions. Inspect the complete delivered video rather than assuming an input choice guarantees quality.
Is the upload preview the final output?
No. It is a local view of your selected reference. The generated video is a separate result after you confirm a task and it completes.