Wedding save-the-date video with Omni: 8 seconds, vertical
A 9:16 save-the-date from an engagement photo with Gemini Omni Flash: $1.00 at 720p on Sume, with names and date added as captions, not generated text.

A vertical save-the-date is one Gemini Omni Flash clip: an engagement photo as the start frame, 8 seconds, 9:16. On Sume that is $1.00 at 720p ($0.125 per second) and $1.50 at 1080p. Add names and the date as captions afterward. Google's Omni page, read 2026-10-08, tells prompt writers to give exact text, placement and animation, but generated lettering is the part most likely to need a retry, and a misspelled name is not a retry you want to pay for.
The photo sets the faces and the clothes; the prompt sets the motion and the sound.
Prompt and request
Use the photo as image_url. Describe a small motion rather than a new scene, and tell the model to keep everything else. Google's tips use the phrase Keep everything else the same for that purpose.
curl 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":"Gentle slow push-in on the couple, petals drift down, soft piano. Keep everything else the same. No dialogue.","image_url":"https://example.com/engagement.jpg","aspect_ratio":"9:16","duration":8,"resolution":"720p"}'Faces and regions
A photo of two adults is the normal case. Google's Omni page says editing images that contain minors is not supported in the EEA, Switzerland or the UK, and that certain recognizable people are unsupported. If the couple appears with children or flower girls in the photo, expect a refusal in those regions, and crop the photo before you upload it.
Cost of a draft round
Check the framing at 360p before you pay for the 720p version.
| Step | Resolution | Per second | Cost |
|---|---|---|---|
| Draft | 360p | $0.0375 | $0.30 |
| Final | 720p | $0.125 | $1.00 |
| Final | 1080p | $0.1875 | $1.50 |
| Three drafts and a 1080p final | mixed | n/a | $2.40 |
Captions and the watermark
Put the names, date and a link on the video in your editor. Every Omni output carries SynthID, which Google says is invisible and built to survive cropping and compression, so adding captions does not remove it. Guests will not see a mark. If you send the video to a printer or a venue screen, ask for 1080p, noting that Google labels its 1080p output as upscaled.
Delivering it
Download the finished file and keep it with your project. Share it as an MP4 link or send it through the messaging app your guests use. Most phones play a 9:16 video full screen with no editing.
Submit with an Idempotency-Key header so a retry after a dropped connection returns the original job instead of a second charge; the same key with a different payload returns 409. Then poll GET /v1/jobs/:id/status and fetch GET /v1/jobs/:id/result when it finishes.
Sources
Related posts
More in Use cases
- Weekly TikTok trend sweep: 3 regions x 4 keywords is $1.20
Loop four keywords over US, GB and KR with Sume trending search: 12 calls, $1.20 a week. A short Python script and a month of cost.
- Which caption style wins? Test slam, punch, tiktok-green
Burn three caption styles onto one clip for $0.60 and compare them. Latin text works on all three; Korean text is rejected on these styles. Runnable loop.
- Vertical 9:16 Seedance ad at 720p: $1.512 on Mini, $4.6224 on 2.5
An 8-second 720p vertical ad costs from the Mini price to the Seedance 2.5 price on Sume. All four tiers at 6, 8, 12 and 15 s, and the 720x1280 frame size.
- Wine tasting invite video: Wan 3.0, 9:16, 12 s, audio reference
A 12-second vertical invite on Wan 3.0 costs $0.75 at 480p and $1.50 at 720p on Sume, and takes up to 5 audio references. Age gates and limits.
Written by Sume