Skip to content

API_CONTRACT.md

Latest commit

 

History

History
56 lines (45 loc) · 1.61 KB

API_CONTRACT.md

File metadata and controls

56 lines (45 loc) · 1.61 KB
Raw

TARDIS Studio proxy contract

The browser talks only to the local Express server. The proxy keeps the TARDIS inference credential private and forwards the documented asynchronous video API.

Create

POST /api/generations

{
  "prompt": "A cinematic ...",
  "imageData": "data:image/png;base64,...",
  "settings": {
    "quality": "speed",
    "size": "1280x720",
    "fps": 30,
    "duration": 5,
    "withAudio": false
  }
}

prompt is required and is limited to 512 characters. imageData is optional, but when present it must be a PNG/JPEG Data URL whose decoded bytes are at most 5 MB. The proxy maps this to the TARDIS service's reference-image field. The response is a small client-facing envelope containing id, requestId, taskStatus, and model.

The upstream request is sent to the configured TARDIS inference base URL at /tardis/v1/videos with model: tardis. The bearer credential is read only from TARDIS_INFERENCE_API_KEY (or an equivalent secret-manager entry) in the server environment; it is never accepted from the browser.

Poll

GET /api/generations/:id

The proxy calls GET /tardis/v1/videos/{id} on the configured TARDIS inference service. The normalized response is:

{
  "id": "...",
  "status": "PROCESSING|SUCCESS|FAIL",
  "videoUrl": "https://.../video.mp4",
  "coverUrl": "https://.../cover.jpg",
  "error": null,
  "model": "tardis"
}

videoUrl and coverUrl are populated from the first item of upstream video_result after SUCCESS. These are temporary provider URLs, so a client that needs durable history should download/copy the asset to local storage.