Input Formats
Media input specification for the API
Input Methods
The API accepts media through image_uri, video_uri, and audio_uri parameters using three methods:
Cloud Storage Upload
Upload via v1/upload endpoint. Returns a storage_uri for use in generation requests.
200 MB is the upload endpoint cap. Each feature applies its own size limit when the file is used in a generation request (for example, images are capped at 15 MB), so a large upload may be rejected at generation time depending on the endpoint. See Per-endpoint video limits.
HTTPS URL
Direct link to publicly accessible files. API fetches during request.
Requirements:
- HTTPS only
- Domain names (no IPs)
- Publicly accessible
- No redirects
Data URI (Base64)
Inline base64-encoded data. Format: data:{mime-type};base64,{encoded-data}
Base64 encoding increases file size by ~33%.
Per-endpoint video limits
Video input is capped at 32 MB on most endpoints. These accept more:
To send a file this large, use Cloud Storage or an HTTPS URL. Data URI input keeps its own lower limit on every endpoint.
Supported Formats
Images
Videos
Audio
AAC audio must use AAC-LC profile (1024 samples/frame). HE-AAC and HE-AACv2 (2048 samples/frame) are not supported. PCM audio is not supported.