Prompt Try-On

The Prompt Try-On endpoint allows you to describe a new outfit using text to seamlessly drape it onto an AI fashion model.

Model Name: prompt-tryon

Lifecycle: stable

Processing Time: ~15s–40s (increases when using style or model references)

Output Format: Auto (JPEG or PNG)

Delivery Methods: URL

Credits: 1 to 5 per generation (depends on quality)

Request

Submit your try-on configuration to the universal /v1/run endpoint:

POSThttps://fititon.app/api/v1/run

Request Examples

Authentication You can pass your API key either as a Bearer token (Authorization: Bearer YOUR_API_KEY) or via the x-api-key header (x-api-key: YOUR_API_KEY).

curl -X POST https://fititon.app/api/v1/run \
     -H "Content-Type: application/json" \
     -H "Authorization: Bearer YOUR_API_KEY" \
     -d '{
           "model_name": "prompt-tryon",
           "inputs": {
             "personImage": "https://example.com/person.jpg",
             "prompt": "A stylish red summer dress with floral patterns.",
             "quality": "2k",
             "sampleCount": 1,
             "ratio": "16:9",
             "modelImageUrl": "https://example.com/model-reference.jpg",
             "styleId": "movie"
           }
         }'

Response

The Try-On generation happens asynchronously in the background. Your request will immediately return a 200 OK status with a tracking object.

{
  "id": "123a87r9-4129-4bb3-be18-9c9fb5bd7fc1",
  "status": "starting",
  "created_at": "2026-07-17T11:00:00Z",
  "error": null,
  "output": null
}

Polling for Results You must use the returned id to poll the status endpoint (GET https://fititon.app/api/v1/status/{id}) to retrieve your generated image URLs. See the API Setup guide for detailed instructions.

When the generation finishes, the polling endpoint will return status: "success" with the following output array containing the final image URLs:

{
  "id": "123a87r9-4129-4bb3-be18-9c9fb5bd7fc1",
  "status": "success",
  "output": [
    "https://pub-r2.com/.../result-0.png"
  ],
  "error": null,
  "created_at": "2026-07-17T11:00:00Z",
  "updated_at": "2026-07-17T11:00:45Z"
}

Request Parameters

Required Parameters

personImageRequiredstring

The base image of the model you want to dress. Can be a publicly accessible URL or a Base64-encoded image string. Supported formats: JPEG, PNG, WEBP. Max size: 25MB.

Pro Tip: Front-facing poses with arms visible produce the most realistic draping.

Formatting Requirement When submitting a Base64 string, it must include the standard data URI prefix (e.g., data:image/jpeg;base64,...). The /v1/run endpoint only accepts JSON payloads.

promptRequiredstring

Text description of the outfit you want the model to wear.

Pro Tip: Include fabric type (silk, cotton, denim) and fit style (slim, oversized) for the best results. Specify colors precisely ("navy blue" works better than "blue") and explicitly reference what the model is already wearing so the AI knows exactly what to change.

Optional Parameters

quality'1k' | '2k' | '4k'

Output resolution tier for Prompt Try-On (Gemini models). Higher resolutions consume more credits. Default: 1k

sampleCountinteger

Number of image variations to generate per request. Must be between 1 and 4. Additional images consume more credits and increase processing time linearly. Default: 1

ratiostring

Defines the width-to-height ratio of the generated image. If empty, the system defaults to the aspect ratio of the personImage.

Supported values: '1:1', '3:4', '4:3', '9:16', '16:9', '2:3', '3:2', '4:5', '5:4', '21:9'.

modelImageUrlstring

Optional. URL of a human reference model to use instead of preserving the original person's features.

Preset Models Available:

styleIdstring

Optional. Apply a specific photographic style to the output image.

Available Styles:

Style Name	Style ID (`styleId`)

Valid values: y2k, studio, iphone, professional, lifestyle, analog, streetwear, flash, movie, minimalist, ugc, editorial.

Note: Complex styles may increase the overall generation time.

Runtime Errors

Runtime errors for this feature use the shared set documented in Error Handling. If an error occurs during processing, the status will update to "error" and the error field will contain the specific failure reason.

Credit Cost

The cost depends on the selected quality tier and is multiplied by your sampleCount.

Formula: Cost = QualityCredits × sampleCount

Parameter	Credits per Image
`quality="1k"`	1
`quality="2k"`	3
`quality="4k"`	5