API Documentation (1.0)

Lists the available avatars. These include stock avatars as well as avatars created by the account associated with the supplied API key.

Lists the available voices. These include stock voices as well as voices created by the account associated with the supplied API key.

Lists the videos that have completed generation.

Fetches metadata for a single completed video.

Downloads a single completed video.

Checks the status of a video generation job. The status will be one of:

• "inference_started": Job has been queued
• "inference_working": Job is in progress
• "inference_complete": Video generation is complete
• "inference_failed": Job failed
• "inference_error": Unknown error occurred

Note: The ID parameter should be the jobId that was returned from the video creation endpoint.

Starts a new video creation job. The specified avatar will be used, and either a speech can be created using a specified voice and script, or an uploaded audio track can be used.

Starts a new text-to-image generation job. Provide a text prompt describing the image you want to generate.

This endpoint generates images from text descriptions without requiring any reference images.

Workflow:

1. Call this endpoint to create a job. You'll receive a jobId in the response.
2. Use the jobId with GET /api/image/{jobId} to check the job status.
3. Poll the status endpoint until the job is complete (status: "inference_complete").
4. When complete, the response will include a downloadUrl field containing the image URL.

Starts a new image-to-image editing job. Upload one or more reference images and provide a prompt describing how you want to edit them. The AI will transform the images based on your prompt.

You can either upload files directly (multipart/form-data) or provide media IDs of previously uploaded images (application/json). You can provide a single image or multiple images.

Workflow:

1. Call this endpoint to create a job. You'll receive a jobId in the response.
2. Use the jobId with GET /api/image/{jobId} to check the job status.
3. Poll the status endpoint until the job is complete (status: "inference_complete").
4. When complete, the response will include a downloadUrl field containing the image URL.

Starts a new image inpainting job. Upload an original image and a mask image (black and white), then specify what you want to replace in the masked areas. The mask should have white areas indicating what to replace and black areas indicating what to keep.

You can either upload files directly (multipart/form-data) or provide media IDs of previously uploaded images (application/json).

To create a mask image, you can use the frontend interface at /create-ai-image:

1. Upload your original image
2. Switch to Inpaint mode
3. Draw mask areas with the blue brush
4. Click "Download Mask" to download the black/white mask image
5. Use the downloaded mask image in your API request

Workflow:

1. Call this endpoint to create a job. You'll receive a jobId in the response.
2. Use the jobId with GET /api/image/{jobId} to check the job status.
3. Poll the status endpoint until the job is complete (status: "inference_complete").
4. When complete, the response will include a downloadUrl field containing the image URL.

Gets the status or result of any job. This is the general entry point for job status management.

When the job is in progress or failed, returns status information:

• "inference_started": Job has been queued
• "inference_working": Job is in progress (may include progress percentage)
• "inference_failed": Job failed

When the job is completed:

• For image jobs (GEMINI_IMAGE): Returns image metadata and download URL
• For other job types: Returns completion status

Note: The jobId parameter should be the jobId that was returned from the job creation endpoint.