How to Swap Faces in Images with an AI API

Face swapping has moved from novelty Snapchat filters to serious production use cases — entertainment apps, e-commerce virtual try-on, and video post-production all rely on it. Building a face swap pipeline from scratch means dealing with face detection, landmark alignment, blending, and artifact removal. The Face Swap API handles all of that in a single request: you send two images and get back a JPEG with the faces swapped in seconds.

What the API Can Do

The API offers five endpoints that cover the full face manipulation workflow:

• Swap Face — Replace the face in a target image with a face from a source image
• Swap Face by Index — Target a specific face in a group photo by its index
• Detect Faces — Find all faces in an image and get their bounding box positions
• Enhance Face — AI-powered face restoration and 2x upscaling without swapping
• Enhance Face by Index — Enhance a specific face in a group photo

Quick Start: Swap Two Faces

The simplest use case: take the face from one photo and place it onto another. The API accepts file uploads or public URLs and returns the result directly as a JPEG image.

cURL

# File upload
curl -X POST "https://deepfake-face-swap-ai.p.rapidapi.com/swap-face" \
  -H "x-rapidapi-key: YOUR_API_KEY" \
  -H "x-rapidapi-host: deepfake-face-swap-ai.p.rapidapi.com" \
  -F "source_image=@source.jpg" \
  -F "target_image=@target.jpg" \
  --output result.jpg

# JSON with URLs
curl -X POST "https://deepfake-face-swap-ai.p.rapidapi.com/swap-face" \
  -H "x-rapidapi-key: YOUR_API_KEY" \
  -H "x-rapidapi-host: deepfake-face-swap-ai.p.rapidapi.com" \
  -H "Content-Type: application/json" \
  -d '{"source_url": "https://example.com/source.jpg", "target_url": "https://example.com/target.jpg"}' \
  --output result.jpg

Python

import requests

HOST = "deepfake-face-swap-ai.p.rapidapi.com"
HEADERS = {
    "x-rapidapi-host": HOST,
    "x-rapidapi-key": "YOUR_API_KEY",
}

# Option 1: File upload
with open("source.jpg", "rb") as s, open("target.jpg", "rb") as t:
    resp = requests.post(
        f"https://{HOST}/swap-face",
        headers=HEADERS,
        files={"source_image": s, "target_image": t},
    )

with open("result.jpg", "wb") as f:
    f.write(resp.content)

# Option 2: JSON with URLs
resp = requests.post(
    f"https://{HOST}/swap-face",
    headers={**HEADERS, "Content-Type": "application/json"},
    json={
        "source_url": "https://example.com/source.jpg",
        "target_url": "https://example.com/target.jpg",
    },
)

JavaScript (Node.js)

const HOST = "deepfake-face-swap-ai.p.rapidapi.com";
const HEADERS = {
  "x-rapidapi-host": HOST,
  "x-rapidapi-key": "YOUR_API_KEY",
};

// File upload
const form = new FormData();
form.append("source_image", sourceFile);
form.append("target_image", targetFile);

const resp = await fetch(`https://${HOST}/swap-face`, {
  method: "POST",
  headers: HEADERS,
  body: form,
});
const resultBlob = await resp.blob(); // JPEG image

Targeting a Specific Face in a Group Photo

When the target image contains multiple people, you probably don't want to swap every face. The workflow is: first detect all faces to get their indices, then swap only the one you want.

Step 1: Detect Faces

# Detect all faces in a group photo
resp = requests.post(
    f"https://{HOST}/detect-faces",
    headers=HEADERS,
    files={"image": open("group_photo.jpg", "rb")},
)
faces = resp.json()
print(faces)
# {"total_faces": 3, "faces": [
#   {"index": 0, "bbox": {"x": 120, "y": 50, "width": 160, "height": 160}},
#   {"index": 1, "bbox": {"x": 400, "y": 80, "width": 120, "height": 120}},
#   {"index": 2, "bbox": {"x": 650, "y": 90, "width": 100, "height": 100}}
# ]}

Step 2: Swap by Index

# Swap only face at index 1
resp = requests.post(
    f"https://{HOST}/target-face",
    headers=HEADERS,
    files={"source_image": open("source.jpg", "rb"), "target_image": open("group_photo.jpg", "rb")},
    data={"face_index": 1},
)

with open("result.jpg", "wb") as f:
    f.write(resp.content)

Face Enhancement Without Swapping

The /enhance-face endpoint is a standalone face restoration tool — no swapping involved. It uses AI upscaling to fix artifacts, sharpen details, and output a 2x resolution image. This is useful for restoring old or low-quality photos. Pair this with Face Restoration for even more enhancement options.

# Enhance all faces in a photo
resp = requests.post(
    f"https://{HOST}/enhance-face",
    headers=HEADERS,
    files={"image": open("old_photo.jpg", "rb")},
)

with open("enhanced.jpg", "wb") as f:
    f.write(resp.content)  # 2x upscaled with enhanced faces

Real-World Use Cases

1. Entertainment and Social Apps

Let users swap faces with celebrities, friends, or fictional characters. Face swap features are among the most shared features in social apps — each share drives viral growth. Combine with anime style transfer for even more creative options.

2. E-Commerce Virtual Try-On

Fashion and beauty brands can let customers see how they would look wearing different hairstyles, glasses, or makeup by swapping their face onto model photos. This reduces return rates and increases purchase confidence.

3. Video and Film Post-Production

Replace stunt doubles, de-age actors, or fix continuity issues in post-production. Process individual frames through the API to achieve consistent face replacement across video sequences.

4. Photo Restoration

Restore damaged or low-quality faces in old family photos using the enhancement endpoint. The 2x upscaling and artifact removal can bring new life to faded prints and scanned photos.

Tips and Best Practices

Use Similar Angles

Face swaps look most realistic when the source and target faces are at similar angles. A front-facing source photo works best with a front-facing target. Extreme angle differences (profile vs. front) will produce less natural results.

Multi-Face Swapping

Set many_faces: true to swap all faces in a group photo with the source face. For selective swapping, use the detect → target workflow to pick specific faces by index.

Handle Binary Responses

The swap and enhance endpoints return raw JPEG bytes, not JSON. Make sure to write the response as binary (resp.content in Python, resp.blob() in JavaScript). The detect endpoint is the only one that returns JSON.

Respect Privacy and Consent

Face swap technology is powerful and comes with responsibility. Always ensure you have consent from individuals whose faces are being used. Clearly communicate your usage policies to users and comply with local regulations around synthetic media.

Optimize Input Quality

Higher resolution source images produce better swaps. The face should be clearly visible, well-lit, and unobstructed. Maximum input size is 10 MB per image, and the output preserves the original resolution of the target image.

With five endpoints covering swap, detect, and enhance workflows, the Face Swap API lets you build a complete face manipulation feature in an afternoon. Grab your free API key and start swapping.