Wedding save-the-date clip: first frame or reference photos

A first frame animates the exact photo; reference images keep the couple's look but let the model compose a new scene. On Sume's POST /v1/videos, frame_images with first_frame is image-to-video, and input_references is reference-to-video. Gemini Omni Flash 1.1 takes up to 10 reference images, and an 8-second 720p clip costs $1.00.

Pick the first frame when the photo itself is the keepsake, and the references when you want the couple in a setting you describe. Sume's catalog also states that a single reference image with no frame counts as reference-to-video, not as a pinned first frame. Source: Video generation and Video Router, read 2026-10-06.

What does each route cost?

Eight seconds at 720p on three models that accept both inputs.

What does the request look like?

The reference version sends the photos in input_references and no frame_images.

curl -sS -X POST https://api.sume.com/v1/videos \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-omni-flash-1.1",
    "prompt": "The couple walk through a lantern-lit garden at dusk, gentle camera drift",
    "duration": 8,
    "resolution": "720p",
    "aspect_ratio": "16:9",
    "input_references": [
      {"type": "image_url", "image_url": {"url": "https://example.com/her.jpg"}},
      {"type": "image_url", "image_url": {"url": "https://example.com/him.jpg"}}
    ]
  }'

Is it fine to use real people?

Use photos of people who have agreed to appear, and do not use someone else's likeness. Sume does not rank faces or promise a match, so test two takes and keep the one that looks like them.