Face Swap
Puts one or two faces into a scene it generates from your prompt, or into a photo you upload. In a two-person photo it can replace just the left or the right person.
Overview
Face Swap is Wiro's face replacement tool, built on FaceFusion. Give it one or two face photos and it either generates a scene around those people with FLUX.1 Dev, or puts the faces into a photo you already have. Every result then goes through GFPGAN v1.4 face restoration.
Before generating, it reads each face photo for a man or woman estimate, skin tone, eye color, hair and face shape, and adds those notes to your prompt so the generated body and hair fit the swapped face.
What you can build
- A "put yourself in the picture" feature: users upload a selfie, your app picks the prompt and style.
- Couple portraits from two separate selfies.
- Themed photo booths built on presets such as retro80s, 90sTV or vintageFilm.
- Two-person photos where only the left or the right person is replaced.
Inputs
Face Photo 1 is required: a clear, front-facing photo of one person. If several faces are visible, the largest is used. In two-person results this face goes on the left.
Face Photo 2 is optional and becomes the person on the right.
Target Photo is optional. With it, nothing is generated and the faces go into this photo, so it must already show the people being replaced. With Face Photo 2 or Swap Position it needs exactly two people side by side; with Face Photo 1 alone, one person is enough.
Swap Position (left or right) replaces one person in a two-person Target Photo. It cannot be combined with Face Photo 2.
Prompt, Output Style, Width and Height only matter for generated scenes. Write about the setting and the outfit; the faces come from your photos. Output Style defaults to hyperRealistic and has more than a dozen other looks. Width and height default to 768, up to 2048.
Photos can be JPG, PNG or WebP.
Outputs
Generated scenes return two PNG images, and Target Photo runs return one. A two-person image where both faces were not found is dropped.
Two-person images are cut between the faces, swapped half by half and joined again. With Swap Position the person you keep is restored too, so their face can look lightly retouched.
Generated scenes use a fixed seed: the same inputs give the same scene. Change the wording or style for a new one.
Recommended settings
- Face photos with even light and no sunglasses. Steep head turns lower the quality.
- For two-person scenes, keep crowds out of the prompt. A third clear face in a generated image stops the run.
Limitations
- Two-person generated scenes always get a red carpet and chandeliers added to the prompt, and both show up whatever setting you describe. For other two-person looks, use a Target Photo.
- In generated scenes, a wrong man or woman estimate gives a body that does not match the face.
- Only the face area changes. Hair, head shape and body come from the target photo or generated scene.
- With one face and a group Target Photo, you cannot choose which face is replaced. Crop first.
Safety & compliance
Only swap the faces of people who have agreed to it, and never use the results to impersonate or deceive. Label swapped images as edited where they could pass for real photos.
Example prompts
Great starting points for Face-Swap.
API quick start
Run Face-Swap with a single API call.
{
"prompt": "in a futuristic city plaza at night, surr…",
"inputImage": "https://your-cdn.com/input.png",
"inputImage2": "https://your-cdn.com/input.png",
"inputImage3": "https://your-cdn.com/input.png"
}curl -X POST "https://api.wiro.ai/v1/Run/wiro/face-swap" \
-H "Content-Type: application/json" \
-H "x-api-key: YOUR_WIRO_API_KEY" \
--data-binary @- <<'JSON'
{
"prompt": "in a futuristic city plaza at night, surr…",
"inputImage": "https://your-cdn.com/input.png",
"inputImage2": "https://your-cdn.com/input.png",
"inputImage3": "https://your-cdn.com/input.png"
}
JSONOther models you might like
Explore similar image generation models.



