Change the weather in a video with Gemini Omni Flash edit
Try turning a sunny clip into rain or snow with a Gemini Omni Flash 1.1 edit on Sume: a prompt that keeps the rest unchanged, request and a still check.

To change the weather in a clip, send it to Gemini Omni Flash 1.1 as video_url and prompt a single change: "Make it rain heavily. Keep everything else the same." Google's guide recommends simple edit prompts and names this closing phrase; it does not list weather as a tested edit, so run it on your own footage and inspect the result.
Facts here are from Google DeepMind's Omni prompt guide, the Gemini API Omni docs and DeepMind's Omni page (read 2026-10-03), plus Sume's Video Router docs.
What the sources support
Google's pages list editing existing video, swapping characters or objects, and lighting control among Omni capabilities. Weather is not named, so frame it as a prompt to test. The model works on clips of 10 seconds or less, and the output carries a SynthID watermark.
| Item | What the page says |
|---|---|
| Edit source length | 10 seconds or less |
| Prompt style | Simple; add "Keep everything else the same" |
| Lighting | Describe source and quality |
| Region | Uploaded-video editing unavailable in EEA, Switzerland, UK |
| Watermark | SynthID on all videos |
Prompt wording to try
Change one condition at a time: rain, fog, snow or overcast. Add the light that goes with it, since a rainy scene with hard sunlight reads as wrong: "Overcast, flat grey light, wet pavement." Avoid asking for weather and a time-of-day change together in one pass.
Request
Sume's edit mode is chosen by video_url: it cannot be combined with image_url, end_image_url or reference_*_urls, takes no aspect_ratio or duration, and resolution defaults to 720p. Native audio is always on.
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: weather-rain-001" \
-d '{
"model": "gemini-omni-flash-1.1",
"prompt": "Make it rain heavily with flat grey overcast light and wet pavement. Keep everything else the same.",
"video_url": "https://media.sume.com/artifacts/artf_demo/source.mp4",
"resolution": "720p",
"mode": "async"
}'Inspect the result
Pull stills from the source and result at matching timestamps with Video inspect. Check that people, signs and the camera path survived, that shadows match the new light, and that the audio still fits. If it fails, rerun with a more specific description of the weather, not a longer one.
Sources
Related posts
More in Use cases
- ChatGPT Try On won't promise your size: what sellers publish
ChatGPT Try On is reported to carry no size or fit guarantee. What apparel sellers should still publish, and where Sume's fitting preview helps and stops.
- ChatGPT Try On saves a reference photo: what your own flow should keep
OpenAI's Try On keeps a reference photo you can delete. If you build your own try-on on Sume, decide what you store, what you send, and what you link back.
- Check a video for SynthID in Gemini: 100 MB and 90 s, trim first
Gemini scans uploaded video for SynthID, up to 100 MB and 90 seconds. Cut longer clips with Sume video trim first, then read the segments it reports.
- Chrome heavy ad intervention for VAST video: 4 MB and 60 s limits
Microsoft's VAST guidance warns Chrome can block video ad iframes over 4 MB, 60 s of main thread or 15 s in 30 s. What a finished Sume MP4 changes.
Written by Sume