FLUX Virtual Try-On (VTO)

Documentation Index

Fetch the complete documentation index at: https://docs.bfl.ml/llms.txt

Use this file to discover all available pages before exploring further.

​

Examples

​

Prompting tips

• Core prompt formula: The person of image 1, maintaining exactly their face and pose, wearing the {YOUR GARMENT DESCRIPTIONS} of image 2.
• A good default is even a static prompt like The person of image 1, maintaining exactly their face and pose, wearing the garments of image 2.. But to get higher quality results, we recommend to add a concise description of the garments. Especially for complex garments, a detailed description can significantly improve the outcome.
  You can for instance specify the fit and category of the garment like for example:
  • the oversized tee
  • the cap with Keep FLUXing text
  • the 7/8 length pants.
• When using another model shot as garment reference, only describe what should be tried on from that image and don’t specify in the prompt what should be kept from the model image.
  For instance, if the garment reference image also shows a model wearing pants, but you only want to try on the top, your prompt could be: The person of image 1, maintaining exactly their face and pose, wearing the green jacket of image 2.

​

Reference images

​

General guidance

• Input images (both model and garment) up to 2 Megapixels are used as-is. Larger inputs are automatically downscaled to a ~1 Megapixels target while preserving aspect ratio.
• For best results and latency, keep the model image and the garment image at around 1 Megapixels.
• Some artifacts from existing clothing may be retained in the output. If you are consistently seeing this, you can try preprocessing the model image to get a more neutral base for the try-on. A neutral base could mean just wearing tight-fitting, plain clothing.
• For the garment reference, clean, well-lit images with a clear view of the garment details work best. Both packshot and on-model references are supported, but packshot references with a plain background tend to produce the best results.

​

Designing the garment image

​

Endpoint

curl -X POST https://api.bfl.ai/v1/flux-tools/vto-v1 \
  -H "x-key: $BFL_API_KEY"

curl https://api.bfl.ai/v1/get_result?id=<TASK_ID> \
  -H "x-key: $BFL_API_KEY"

​

Regional endpoints

• https://api.eu.bfl.ai/v1/flux-tools/vto-v1 — Europe
• https://api.us.bfl.ai/v1/flux-tools/vto-v1 — US

​

Quick start

#!/usr/bin/env python3
import base64
import os
import time
import requests

API_KEY = os.environ["BFL_API_KEY"]
BASE = "https://api.bfl.ai"
HEADERS = {"accept": "application/json", "x-key": API_KEY, "Content-Type": "application/json"}

PROMPT = "The person of image 1, maintaining exactly their face and pose, wearing the white t-shirt of image 2."
PERSON_PATH = "/path/to/person.png"
GARMENT_PATH = "/path/to/garment_top.png"

with open(PERSON_PATH, "rb") as f:
    person_b64 = base64.b64encode(f.read()).decode()

with open(GARMENT_PATH, "rb") as f:
    garment_b64 = base64.b64encode(f.read()).decode()

payload = {
    "prompt": PROMPT,
    "person": person_b64,
    "garment": garment_b64,
    "output_format": "webp",
}

submit = requests.post(f"{BASE}/v1/flux-tools/vto-v1", headers=HEADERS, json=payload)
submit.raise_for_status()
meta = submit.json()

task_id = meta["id"]
poll_url = meta.get("polling_url", f"{BASE}/v1/get_result?id={task_id}")

while True:
    r = requests.get(poll_url, headers={"accept": "application/json", "x-key": API_KEY})
    r.raise_for_status()
    result = r.json()

    status = result.get("status")
    if status == "Ready":
        print("Result URL:", result["result"]["sample"])
        break
    if status in {"Error", "Request Moderated", "Content Moderated", "Task not found"}:
        raise RuntimeError(f"VTO failed with status: {status} | payload: {result}")

    time.sleep(1)

​

Request parameters

​

Response format

​

Initial response

{
  "id": "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
  "polling_url": "https://api.bfl.ai/v1/get_result?id=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
}

​

Polling response (success)

{
  "id": "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
  "status": "Ready",
  "result": {
    "sample": "https://delivery.bfl.ai/..."
  }
}