AI Photo Editing API

Turn a difficult portrait into a cleaner ID-style photo in one request. The API evaluates the whole image, corrects only the problems it finds, and leaves already suitable regions unchanged.

This endpoint is for portrait normalization, not beautification. Its highest priority is keeping the person recognizable as the same real individual.

The problem it solves

Real users rarely submit a perfectly controlled studio photo. A single image may combine a turned head, a tilted posture, visible teeth, glasses, hard shadows, compression artifacts, or a shoulder that was cut off during an earlier background-removal step. Fixing these issues manually is slow, and chaining separate edits can introduce inconsistent results.

POST /v2/aiEdit applies the necessary corrections together while protecting details that should not change: identity, natural skin texture, age, hair, clothing, background, crop, and overall photographic character.

Use it before specification-based processing

This endpoint improves the portrait itself. It does not select a country-specific document size or guarantee acceptance by an issuing authority. For sizing, background processing, validation, and issue codes, pass the normalized result to the Passport Photo API.

What the API corrects

Input problem	Correction	What remains protected
Head turned or tilted; uneven posture	Makes the face naturally front-facing, keeps the head upright, and levels the shoulders when needed	Recognizable facial structure, natural anatomy, framing, and scale
Broad smile, open mouth, or visible teeth	Produces a calm, natural, closed-mouth expression	Characteristic lip shape, eye shape, and the person's likeness
Eyeglasses, sunglasses, or unsuitable novelty headwear	Removes the obstruction and conservatively reconstructs only hidden facial regions	Hairline, ears, head shape, ordinary jewelry, and religious or cultural head coverings that do not obscure the face
Harsh shadows, color casts, overexposure, or underexposure	Balances facial light and color while preserving real skin tone	Pores, age, complexion, and natural texture—without whitening or makeup
Blur, noise, compression, or weak local contrast	Restores only the clarity needed for a usable ID-style portrait	Real detail—without invented features, aggressive sharpening, or plastic skin
A shoulder or garment ends at an unnatural internal edge	Reconstructs only the missing continuation of the existing clothing and silhouette	Body proportions, clothing, lighting, background, and the existing crop

If a region already meets these requirements, the API is instructed to leave it unchanged.

Before and after

These synthetic examples illustrate the kinds of corrections the endpoint is designed to make. They are not real customers or identity documents, and individual results vary with the input image.

Before · turned head and visible teeth

After · front-facing and neutral

Pose and expression. The face is brought toward the camera and the expression is neutralized while the subject's age, hair, skin texture, clothing, and background remain consistent.

Before · glasses and hard side light

After · clear face and even light

Face visibility and lighting. The glasses are removed, only the covered facial areas are reconstructed, and the strong light imbalance is corrected without changing the background or adding glamour retouching.

Before · eyes hidden by dark glasses

After · obstruction removed

Dark-glasses removal. The frames and lenses are removed and only the hidden eye and eyebrow region is reconstructed. The pose, expression, face shape, hair, clothing, and background remain consistent.

Before · blur, noise, and compression

After · conservative quality recovery

Recoverable image quality. Noise, compression artifacts, softness, and weak local contrast are reduced without aggressive sharpening, invented facial detail, or artificial skin smoothing.

Before · shoulder cut off inside frame

After · silhouette reconstructed

Incomplete silhouette. Only the missing shoulder and sweater continuation are repaired. A shoulder that already ends naturally—or continues through the image boundary—is not changed.

Identity-preserving boundaries

The endpoint is designed to preserve facial proportions, face shape, eyes, eyebrows, nose, lips, jawline, cheeks, ears, skin tone and texture, hairstyle, hairline, age, and distinctive characteristics.

It does not intentionally:

• slim, reshape, rejuvenate, beautify, or cosmetically retouch the person;
• change ethnicity or gender presentation;
• whiten skin, add makeup, or create glamour lighting;
• replace the background or clothing;
• add objects or alter regions that are already suitable;
• invent fine facial detail when the source does not contain enough information.

API endpoint

Send the request to the region where your account and data processing are configured:

• https://api-us.idphotoapp.com — US region
• https://api-eu.idphotoapp.com — EU region
• Other regional deployments are available on request

POST /v2/aiEdit

Request

Field	Type	Required	Description
apiKey	string	Yes	Your API key. Send requests from your server; do not expose credentials in browser code.
apiSecret	string	Yes	Your API secret.
imageBase64	string	Yes	Base64-encoded JPEG or PNG, with or without a data URI prefix. Use a single-person portrait.
outputFormat	string	No	JPEG or PNG. When omitted, the input image format is used.
outputType	string	No	Currently only IMAGE_BASE64 is supported.

{
  "apiKey": "your-api-key",
  "apiSecret": "your-api-secret",
  "imageBase64": "data:image/jpeg;base64,...",
  "outputFormat": "JPEG",
  "outputType": "IMAGE_BASE64"
}

Response

The edited image is returned as Base64 without a data URI prefix. width and height report the output pixel dimensions.

{
  "output": "/9j/4AAQSkZJRgABAQ...",
  "width": 600,
  "height": 800
}

To display the result as a JPEG in an application, prepend data:image/jpeg;base64, to output. For storage or further API processing, decode the value as binary image data instead.

Integration guidance

• Access to this endpoint may need to be enabled for your account.
• Use a clear, single-person portrait and keep the complete image payload on your server.
• Set an appropriate timeout for image processing and do not log credentials or full Base64 values.
• Treat authentication failures, invalid image data, and processing failures separately from a successful edit.
• Review the edited portrait before passing it to the Passport Photo API. A successful edit does not by itself select or validate a document specification.