AI time-lapse video prompt for Gemini Omni Flash on Sume
Write a time-lapse prompt for Gemini Omni Flash 1.1: Google's guide on lighting and camera, a locked-off frame, 10-second limit, and a request to test at 360p.

For a time-lapse in Gemini Omni Flash 1.1, name a locked-off camera, one subject that changes over time, and the light that shifts, then ask for a single continuous shot. Google's Omni guide lists these building blocks but does not describe a time-lapse mode, so test the wording at 360p and treat the speed of change as something the model chooses.
Sources: Google DeepMind's Omni prompt guide, the Gemini API Omni docs and Google's Omni 1.1 Flash build post, all read on 2026-10-03. Output is 3 to 10 seconds, so a time-lapse here is short by design.
Ingredients from Google's guide
Pair a fixed camera ("static", "locked off", "fixed") with a continuous-shot phrase to avoid cuts, then describe the light source and its quality, since light is what carries the passage of time: dawn to full sun, or dusk to street lamps.
| Part | Wording to use | Why |
|---|---|---|
| Camera | static, locked off | Time-lapse reads from a fixed frame |
| Continuity | single continuous shot, no scene cuts | Stops automatic cutting |
| Subject | one thing that changes | Clouds, a crowd, a plant, a building |
| Light | name source and quality | Carries the time passing |
| Length | 3 to 10 seconds | Model limit |
A prompt that avoids the traps
Say what changes and in which direction: "clouds move quickly left to right while the light shifts from soft dawn to bright midday". Avoid asking for people to move naturally in a time-lapse; fast crowd motion is a hard case and Google lists no guarantee for it. Do not include on-screen clocks, because rendered text may drift.
Draft at 360p first
Google's build post calls 360p a draft mode: up to 60% faster and one-third the cost of 720p. Send the draft, watch it, then rerun the winning prompt at 720p.
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-timelapse-001" \
-d '{
"model": "gemini-omni-flash-1.1",
"prompt": "In a single continuous shot, locked off, a city skyline from a rooftop. Clouds race quickly left to right while the light shifts from soft dawn to bright midday. Natural ambient city sound.",
"resolution": "360p",
"duration": 6,
"aspect_ratio": "16:9",
"mode": "async"
}'Speeding up real footage instead
If you have real footage and want it faster, that is a different job. Sume has no speed control in its trim surface, which a related post covers. Use Omni when you want a generated time-lapse, and see the Video Router docs for the request fields.
Sources
Related posts
More in Use cases
- AI tote bag mockup: design file plus blank bag as 2 references
Make a tote bag mockup with the Sume image API: send a blank-bag photo and your design as two references, keep the text yours, and compare the price per mockup.
- AI trading card art: generate at 2:3, then crop to a 5:7 card
Sume has no 5:7 ratio. Ask for 2:3, keep art out of the margins, and crop 6.7 percent off the height. Models, request and crop math.
- AI try-on for sunglasses, bags and scarves from a photo
ChatGPT Try On covers accessories as well as clothes. The same edit on Sume: several photos of the item as references, and what to check on scale and shape.
- AI ultrawide wallpaper and banner API: which Sume models list 21:9
Seven Sume image rows list 21:9, from Flux 2 Pro at $0.0375 to Qwen Image Max at $0.09375. Nano Banana 2 adds 4:1 and 8:1; Ideogram adds 3:1. Request example.
Written by Sume