Outfit Try-On API

Overview

The standard POST /v1/virtual-tryon endpoint accepts one garment at a time. If you want to dress a model in a coordinated outfit — a shirt and trousers — you previously had to run two separate calls and thread the intermediate result manually.

The /outfit endpoint removes that complexity. Send the person photo, the top garment, and the bottom garment in a single request. The API runs two sequential VTON passes internally and returns one composite result.

Endpoint

POST https://api.pixelapi.dev/v1/virtual-tryon/outfit
Authorization: Bearer <your_api_key>
Content-Type: application/json

Request body

Response (202 Accepted)

{
  "job_id": "1bf21d2d-7432-4930-a6d3-e941d2157d80",
  "status": "processing",
  "eta_seconds": 90,
  "poll_url": "/v1/virtual-tryon/jobs/1bf21d2d-7432-4930-a6d3-e941d2157d80",
  "credits_used": 22,
  "processing_stages": 2,
  "stage_order": "lowerbody first, then upperbody layered over result"
}

Polling for results

Use the standard job endpoint — no changes needed:

GET https://api.pixelapi.dev/v1/virtual-tryon/jobs/{job_id}
Authorization: Bearer <your_api_key>

When status === "completed", the output_url field contains the final composite image URL (or result_image_b64 for inline base64). On failure, credits are auto-refunded.

Pricing and timing

Preflight gates applied

All the same quality and safety gates from the standard endpoint are applied to both garments and to the person photo:

• NSFW detection — both garments and person photo
• Person framing — must show upper body (not headshot-only)
• Garment validity — both garments must be actual wearable clothing
• Placeholder / texture-only garment rejection
• Category mismatch detection (upperbody_garment must not be pants; lowerbody_garment must not be a top)
• EXIF orientation correction on all three images

No credits are charged if any gate rejects the input.

Known limitation

Full example (cURL)

PERSON=$(base64 -w0 person.jpg)
TOP=$(base64 -w0 shirt.jpg)
BOTTOM=$(base64 -w0 trousers.jpg)

curl -s -X POST https://api.pixelapi.dev/v1/virtual-tryon/outfit \
  -H "Authorization: Bearer $PIXELAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "person_image":       "'"$PERSON"'",
    "upperbody_garment":  "'"$TOP"'",
    "lowerbody_garment":  "'"$BOTTOM"'"
  }' | jq .job_id

Full example (Python)

import requests, base64, time

def b64(path):
    return base64.b64encode(open(path,'rb').read()).decode()

resp = requests.post(
    "https://api.pixelapi.dev/v1/virtual-tryon/outfit",
    headers={"Authorization": "Bearer " + API_KEY},
    json={
        "person_image":      b64("person.jpg"),
        "upperbody_garment": b64("shirt.jpg"),
        "lowerbody_garment": b64("trousers.jpg"),
    },
)
job_id = resp.json()["job_id"]

# Poll until done
while True:
    time.sleep(5)
    r = requests.get(
        f"https://api.pixelapi.dev/v1/virtual-tryon/jobs/{job_id}",
        headers={"Authorization": "Bearer " + API_KEY},
    ).json()
    if r["status"] == "completed":
        print("Result:", r["output_url"])
        break
    if r["status"] == "failed":
        print("Failed:", r["error_message"])
        break
    print(f"  {r['status']} — ETA {r.get('eta_seconds','?')}s")

Lensora / mobile integration

The /outfit endpoint is a plain JSON REST call, identical in authentication and polling to the existing single-garment endpoint. From the Lensora Android app:

1. Accept three image inputs from the user in the virtual try-on screen: person photo, top garment, bottom garment.
2. Base64-encode each image before sending (same as the existing single-garment flow).
3. POST to /v1/virtual-tryon/outfit with Authorization: Bearer <user_token>.
4. Poll /v1/virtual-tryon/jobs/{job_id} every 5 seconds. ETA is ~90s.
5. Display the result when status === "completed". Show the user a cost notice of 22 credits before submitting (2× the single-garment cost).

Last updated: 2026-07-13 · [email protected]