Model Create
Model Create enables you to generate realistic fashion models with simple, intuitive prompts or reference images.
- Model Name:
model-create - Lifecycle: experimental
- Processing Time: 12 seconds
- Output Formats: PNG, JPEG
- Delivery Methods: URL or Base64 encoding
- Credits: 1 per image (4 per image with
face_reference)
Request
Generate fashion models by submitting your prompt and optional reference assets to the universal /v1/run endpoint:
Request Parameters
Required Parameters
promptRequiredstring
Prompt for the model image generation. Describes the desired fashion model, clothing, pose, and scene.
Optional Parameters
image_referenceimage URL | base64
Optional image to guide composition and pose. The AI won't copy the exact details from the image, but will use it as inspiration.
You can control whether to copy just the pose or the exact silhouette using the prompt parameter with natural language.
Base64 images must include the proper prefix (e.g., data:image/jpg;base64,<YOUR_BASE64>).
aspect_ratiostring
Defines the width-to-height ratio of the generated image.
When image_reference is provided, the output inherits the reference image's aspect ratio. Supply an explicit aspect_ratio to override that default.
Supported ratios: "21:9", "16:9", "9:16", "4:3", "3:4", "3:2", "2:3", "4:5", "5:4", "1:1".
Default: 1:1
face_referenceimage URL | base64
Optional portrait image that locks in a specific identity across generations.
- Only available when resolution is set to
1k. - Consumes
4credits per image. - Adds roughly
20-30seconds of processing time.
Use this to achieve model consistency across generations without training a custom checkpoint.
face_reference_mode'match_base' | 'match_reference'
Controls how the provided face_reference shapes pose and expression.
-match_base prioritizes the prompt and base generation, keeping the original pose while adapting the reference face to those instructions.
-match_reference aligns the generated model closely with the reference face’s pose, gaze, and expression for maximum resemblance.
Default: match_reference
resolution'1k' | '4k'
Chooses the generation profile. '1k' produces precise, instruction-following results suited for consistent catalog imagery. '4k' unlocks creative, ultra high-definition renders with richer detail but slightly less control over pose and styling.
Supported values: '1k', '4k'
Default: '1k'
seedinteger
Sets random operations to a fixed state. Use the same seed to reproduce results with the same inputs, or different seed to force different results.
Default: 42
Range: 0 to 2^32 - 1
num_imagesinteger
Number of images to generate per request. Must be between 1 and 4. Each image consumes one credit (or four credits when face_reference is present).
Default: 1
output_format'png' | 'jpeg'
Specifies the desired output image format.
-png: Delivers the highest quality image, ideal for use cases such as content creation where quality is paramount.
-jpeg: Provides a faster response with a slightly compressed image, more suitable for real-time applications.
Default: png
return_base64boolean
When set to true, the API returns the generated image as a base64 string instead of a CDN URL. The base64 output is prefixed according to output_format (e.g., data:image/png;base64,... or data:image/jpeg;base64,...).
Enables stricter privacy because images are never uploaded to our CDN and are only available for 60 minutes after completion.
Default: false
Response Polling
After submitting your request, poll the status endpoint using the returned prediction ID. See API Fundamentals for complete polling details.
Successful Response
When your model creation completes successfully, the status endpoint will return:
The output array contains URLs to your generated fashion model images based on your prompt and optional reference parameters.
Runtime Errors
Runtime errors for Model Create use the shared set in Error Handling.
Related Guides
For detailed implementation guidance and best practices:
- Image Preprocessing Best Practices - Optimize your reference images