Food truck weekly special clip: Gemini Omni Flash 1.1, vertical, sound
Make a 3 to 10 second vertical clip of this week's special with Gemini Omni Flash 1.1 on Sume: native synced audio, 9:16, from a dish photo as a reference.

A food truck can make a weekly special clip by sending a dish photo and a short prompt to POST /v1/videos with model: "gemini-omni-flash-1.1". Sume lists this model at 3 to 10 seconds, in 16:9 or 9:16, at 360p, 720p, 1080p or 4K, with native synced audio, so a sizzle comes with the picture.
That sound is the point for food. A silent clip of a burger is a photo that moves; a clip with a sizzle and a crunch is closer to a story-ready ad.
What the model is good for
Sume's docs give a 3 to 10 second window, and say this model accepts image and video references but not audio references.
Ten seconds is long enough for one dish: the plating, a steam shot, the box closing. If you need a longer menu reel, make several clips and cut them together.
The request
Send the dish photo as an input_references entry and describe sound in the prompt. Use 9:16 for stories and reels. Keep resolution at 720p while you iterate, and move up only for the take you will post.
Do not promise the dish looks exactly like the clip. The photo is a guide, and the generated food may differ in garnish or portion, so check it against what you actually sell.
curl -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": "Close shot of a loaded street-taco plate on a truck counter, steam rising, sizzle of the grill and street ambience",
"input_references": [
{ "type": "image_url", "image_url": { "url": "https://example.com/taco-special.jpg" } }
],
"aspect_ratio": "9:16",
"resolution": "720p",
"duration": 8
}'Poll, download, and add the words
The job is asynchronous. Poll the polling_url until completed and download unsigned_urls[0]. A completed video has its own audio track, so check the volume and any generated speech before you post.
For the special's name and price, add text yourself or burn captions onto the clip with Sume's video-captions route. Captions work from audible speech, and a silent clip fails with caption_no_speech; for authored overlay text without speech, send cues with text, start and end.
| Item | Value | Note |
|---|---|---|
| Duration | 3 to 10 seconds | Whole seconds |
| Resolutions | 360p, 720p, 1080p, 4K | Choose per take |
| Aspect ratios | 16:9 or 9:16 | 9:16 for stories |
| Audio | Native synced audio | Check it before posting |
| References | Image and video, no audio | Dish photo as reference |
Keep it truthful
Use the clip for atmosphere and the real photo for the menu. If the clip shows an ingredient you do not serve, an allergen claim that is untrue, or a portion bigger than yours, it is a bad ad even if it looks good. Re-run with a tighter prompt rather than posting a close miss.
Sources
Related posts
More in Use cases
- Founder story video with an AI avatar: who is on screen?
An About-us video built from an avatar raises one question: whose face. Three Sume avatar inputs for a founder story, a 40-second script, cost, and a label.
- Four 360p Omni drafts then one 1080p final: $1.69, not $4.69
Four 5-second 360p Omni drafts cost $0.75 on Sume and the 1080p final $0.94, so $1.69 against $4.69 for five finals. The catch: the final is a new generation.
- Four characters, two locations: a Wan 3.0 series bible in ten images
Wan 3.0 on Sume takes 10 reference images per job. Spend them as four characters at two views each plus two sets, and reuse the same set for every episode.
- Four seasons from one venue photo: four Ideogram 4.5 edits for $0.30
Make spring, summer, autumn and winter versions of one venue photo with four ideogram/ideogram-v4.5 edits at $0.075 each on Sume, $7.50 for 25 venues.
Written by Sume