Skip to main content
POST
Create Render Job
Kick off a video render. Send the request as multipart/form-data.
Two workflows available:
  • On-Demand: Upload ref_image and driving_video directly — no preprocessing, no stored assets. Costs 1 credit/second of video.
  • Reusable Assets: Pre-create characters and scenes for reuse across renders. Each render costs 1 credit/second.
Video limits: driving_video uploads are limited to 10 minutes duration and 100 MB file size.

Parameters

Send as multipart/form-data. Supply the character as either ref_image/ref_image_url or a pre-created character_id, and the motion as either driving_video/driving_video_url or a pre-created scene_id.

Response

Error Handling

If the request fails, the response includes a structured error with an error_code:
Common error codes for this endpoint: See Error Codes for the full list. When contacting [email protected], include the error_code and your job ID.

Examples

On-Demand Render (upload files directly)

Reusable Assets Render (pre-created character + scene)

Next Steps

After creating a render job, poll Get Job Status every 3-5 seconds (no auth required). When status is complete, use the cdn_url field to download the final video.

Authorizations

Authorization
string
header
required

API key passed as Bearer token

Body

multipart/form-data
ref_image
file

Character reference image file (PNG, JPG). Use this or ref_image_url.

ref_image_url
string<uri>

URL to a character image. The server fetches it. Use instead of ref_image file upload.

Example:

"https://example.com/character.png"

driving_video
file

Driving video file (MP4). Use this or driving_video_url. Maximum 10 minutes duration and 100 MB file size.

driving_video_url
string<uri>

URL to a driving video. The server fetches it. Use instead of driving_video file upload.

Example:

"https://example.com/dance.mp4"

character_id
string

ID of a ready character (for preprocessed renders)

Example:

"char_550e8400e29b41d4"

scene_id
string

ID of a ready scene (for preprocessed renders)

Example:

"scene_660e8400e29b41d4"

model
enum<string>
default:V3_Preview

Avatar model. Defaults to V3_Preview. Scenes preprocessed on one model cannot be rendered on the other.

Available options:
V4_Preview,
V3_Preview
background_mode
enum<string>
default:original

Background handling: original (default, person removed via AI inpainting), solid (single color), transparent (alpha mask for compositing)

Available options:
original,
solid,
transparent
bg_color
string

RGB string for solid background color, e.g., "0,255,0". Only used when background_mode is solid.

Example:

"0,255,0"

Response

Render job created

job_id
string
Example:

"job_abc123xyz"

status
string
Example:

"queued"

mode
string

The render mode (e.g. full_pipeline, preprocessed)

Example:

"full_pipeline"

enqueued_at
string<date-time>
poll_url
string

URL to poll for job status

Example:

"/api/render/job_abc123xyz"

status_url
string

URL to get job status

Example:

"/api/render/job_abc123xyz"