Current form: This site requires two photo files, one consenting adult per role. The discussion of one-photo workflows below concerns other tools or input limitations; it is not a single-photo option offered by this checkout.
The format takes two portraits: one per person. If you only have one usable photo, you have a decision to make, and it's worth knowing what each option actually produces before you spend a render.
What the two photos do
Each portrait anchors one identity in the frame. The model uses them as references: this is the person on the left, this is the person on the right, keep them recognisable while something happens.
With one photo, one of those anchors is missing. There's no path where the second person appears correctly, because there's nothing describing them.
So "can I make a couple video with one photo" has a precise answer: you can make a video, but not of two specific people.
The three situations people are actually in
1. You have a photo of yourself and none of your partner.
Almost always solvable. Their phone has hundreds. Their family has more. If it's a surprise, ask a friend or a relative. Ten minutes of asking beats an hour of trying to make one photo do the work of two.
2. Your partner doesn't want to appear.
Then don't. A clip of one person is a legitimate thing but it isn't this, and using someone's face against their preference is the one prohibition that matters more than the output quality.
3. The second person isn't available to be photographed.
This is the situation where the "one photo" question is really a different question, and it deserves a straight answer rather than a workaround.
If someone has died, or is otherwise not available, a photo of them can still be used — a photo is a photo, and the format doesn't require consent from someone who can't give it. But it's worth being honest with yourself about what you're making and who you're showing it to. A ten-second joke and a memorial are different objects, and only one of them is what this format is built for.
What one photo actually produces
If you upload one portrait and leave the second slot empty or filled with something else, expect one of two outcomes:
- A single figure. The model renders one person, sometimes with a faint or blurred second shape.
- An invented second person. The model fills the gap with something generic, which is worse — you get a stranger standing next to you.
Neither is what you want, and neither is a bug. The input described one person.
The option that does work
If you genuinely only have one person to include, the honest reframe is: make a single-portrait clip and use it differently.
A ten-second clip of one person turning toward the camera is a perfectly good thing. It works as:
- A profile image with motion
- A reaction clip in a group chat
- One half of a pair — you each make your own, then send them to each other
That last one is better than it sounds. Two separate clips, made from two separate photos, sent to each other, is a nicer object than one clip with a forced second figure in it.
The composite trap
The workaround people try: cut the second person out of a group photo, or stitch two faces into one image, then upload that.
It fails for a specific reason. A cut-out face has different lighting, a different angle, and a visible edge — and the model treats inconsistent input as ambiguity, so the result drifts more than either photo would have alone. You've spent more effort to get a worse clip.
If you have a second photo somewhere, use it. If you don't, change the plan rather than the input.
The checklist
Before you try to make one photo do two jobs:
- Is there a second photo anywhere — their phone, a family member, a shared album?
- Would asking for one be reasonable? (Almost always yes.)
- If not, is a single-person clip actually fine for what you want?
Two of those three questions usually end the problem.
Try it at the generator; price at pricing.
Frequently asked questions
Can I make a couple video with only one photo?
You can make a video, but not one with two recognisable people. The second identity has no reference to work from.
What happens if I only upload one?
Either a single figure, or a generic invented second person — the second is usually worse.
Can I cut a face out of a group photo?
It processes, but the mismatched lighting and angle make the result drift more, not less.
Where do I find a second photo?
Their phone, a family member, a shared album from an event. Ten minutes of asking usually solves it.
Is a single-person clip worth making?
Yes, as its own thing. Two people each making one and sending them to each other works better than forcing a second figure in.