Add steam to a coffee or food clip with Omni edit, check the sound
Make a plain food or drink clip look hot by editing in steam with Gemini Omni Flash. Prompt wording, 8-second cost at 720p, and why the audio needs a listen.

To add steam to an existing coffee or food clip, send the clip as video_url to Gemini Omni Flash 1.1 with a prompt such as "Add soft steam rising from the cup. Keep everything else the same." An 8-second edit at 720p costs $1.00 on Sume. Because the edit regenerates the clip with native audio, listen to the result before you publish.
Prompt wording
Steam is a motion effect, so describe where it comes from, how it moves and how strong it is. "Thin steam rising from the surface of the coffee, drifting left, fading before it reaches the top of the frame" is more useful than "add steam". Say what stays: the cup, the hands, the table and the camera angle.
- Source of the effect: the surface of the liquid, the sauce, the bun.
- Direction and speed: slow, drifting, curling.
- Strength: light, not a fog.
- Light: backlit steam reads best in a window-lit scene.
Cost
Edit output follows the length of the source. The edit mode takes video_url and a prompt, with resolution optional and 720p by default.
| Resolution | Per second | 8 seconds |
|---|---|---|
| 360p | $0.0375 | $0.30 |
| 720p | $0.125 | $1.00 |
| 1080p | $0.1875 | $1.50 |
Sound is part of the output
Native audio is always on for this model, and Sume rejects generate_audio: false. A food clip that arrived with a quiet kitchen may come back with different ambience, or with a sound the new steam seems to call for. Probe the result with video inspect and check probe.has_audio, then listen. If the audio is unwanted, remove it afterwards: video trim takes audio: "drop", at $0.02 per job. See why audio cannot be turned off for the same fix described on its own.
Keep the original, add a second pass
Keep the unedited clip. Steam is easy to overdo, and a second pass with a lighter prompt starts from the original, not from a clip that already has steam in it. Editing an edit stacks regeneration on regeneration, and small errors in hands and edges compound. Name each version with the prompt that made it, so the one you approve can be rebuilt.
For a menu or an ad with many dishes, run a 360p test on the dish with the most complicated surface first. If steam looks right on a bowl of ramen at $0.30, a flat white will be easy.
Where it fails
Steam on a hot surface is easy. Steam on a drink that was filmed with ice in it is not, and the model may add a mismatch that reads as a mistake: condensation next to fog. Choose clips where heat is plausible. Close the loop by pulling frames at the start, middle and end with video frames and checking that the effect is steady and does not appear from nowhere on frame one.
If the effect is wrong, change one word in the prompt and rerun at 360p for $0.30 before returning to the higher tier. The product swap guide has the same loop for objects.
Sources
Related posts
More in Use cases
- Add sunglasses or a hat to a person in an existing video (Omni edit)
Use Gemini Omni Flash 1.1 video_url edit on Sume to add an accessory to a person already in a clip. Prompt wording, a frame check list, and 4-second prices.
- Aerial establishing shot with Gemini Omni Flash: 1080p or 4K cost
A drone-style opening shot for a location video: prompt order, the two aspect ratios, and the 5, 8 and 10 second cost at 1080p against 4K on Sume.
- AI play-by-play voiceover for a sports highlight reel: how to time it
Write one line per play, ask TTS for sentence timings, and place each clip on its line: a 60-second reel costs about 14 cents on Sume. Not live commentary.
- AI spokesperson release notes, 90 seconds: two jobs, cost by tier
A 90-second spokesperson video is two Sume avatar jobs because one job tops out at 60 seconds. Cost at standard, plus and max, and how to split the script.
Written by Sume