Image to Video - HyperGate AI

Overview

Image-to-video models animate a reference image based on a text prompt. The output video preserves the subject and composition of the input image while adding motion.

Which models support this?

Check the id field from GET /v1/models — image-to-video models contain image-to-video in their ID. The required fields differ by model family:

Model family	How to pass the image
Kling v1–v3, Seedance, Omni	`image` (start frame). Also accepted: `images[0]` as shorthand
Kling Video O3	`images[]` (reference array — `image` is not used)

Requirements

Kling/Seedance/Omni: either image or images[0] is required
Kling Video O3: images[] array is required; image is ignored
The URL must be publicly accessible from international servers — Shopee (susercontent.com), TikTok, Google Drive, and similar CDNs that geo-restrict or require cookies/auth will fail with invalid_parameters. Use Upload Media to get a reliable URL first.
Supported formats: JPEG, PNG, WebP

Example — Seedance (start frame)

curl -X POST https://hypergate.store/v1/videos \
  -H "Authorization: Bearer <your-token>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance/2.5/image-to-video",
    "prompt": "The character slowly turns their head and smiles",
    "image": "https://your-cdn.com/keyframe.jpg",
    "duration": 5,
    "aspect_ratio": "16:9"
  }'

Example — Kling Video O3 (reference-to-video)

curl -X POST https://hypergate.store/v1/videos \
  -H "Authorization: Bearer <your-token>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling/video-o3-pro/image-to-video",
    "prompt": "The product rotates slowly on a white background",
    "images": ["https://your-cdn.com/product-ref1.jpg", "https://your-cdn.com/product-ref2.jpg"],
    "duration": 3
  }'

For Kling/Seedance/Omni: omitting both image and images[] returns HTTP 400. For Kling Video O3: images[] is required.