Gemini Omni first and last frame: image_url + end_image_url

On Sume, Gemini Omni's first-and-last-frame control is the image_to_video shape with an extra field: send image_url as the first frame and end_image_url as the last frame to gemini-omni-flash-1.1 on POST /v1/video-router/generate.

Google's changelog for the August 27, 2026 general availability lists "Interpolation (first + last frame)". Sume's side is from the Video Router docs and catalog code, read 2026-09-30.

How does Google describe interpolation?

The Omni page says to provide two images in the input list and describe the desired transition in your prompt. Google's page allows up to 2 images for interpolation. The names differ, but the idea matches: two frames in, a clip that moves between them out.

What do I send on Sume?

The catalog sets end_frame: true for this model, and the docs list end_image_url as optional on image_to_video. Use the same envelope as the other modes.

curl -X POST https://api.sume.com/v1/video-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: omni-first-last-001" \
  -d '{
    "model": "gemini-omni-flash-1.1",
    "prompt": "The door slowly opens and warm light fills the room",
    "image_url": "https://example.com/first.png",
    "end_image_url": "https://example.com/last.png",
    "resolution": "720p",
    "duration": 6,
    "aspect_ratio": "16:9",
    "mode": "async"
  }'

What limits apply?

Do other models take an end frame too?

Capabilities differ per model, and the docs say to read capabilities from GET /v1/video-router/models instead of assuming one envelope. For other models see Seedance 2.5 first and last frame and Veo 3.1 first and last frame.