One ad, five cities: swap the backdrop per market with Omni edit
Localize a street-scene ad by editing the backdrop per market with Gemini Omni Flash. Prompt pattern, five 8-second edits at 720p for $5.00, and what to verify.
To localize one street-scene ad for five cities with Gemini Omni Flash 1.1, send the same finished clip as video_url five times, each with a prompt that swaps the backdrop and keeps everything else. Five 8-second edits at 720p cost $5.00 on Sume, or $7.50 at 1080p.
The prompt pattern
Use one template and change only the city line: "Replace the background with a street in Lisbon: tiled facades, yellow tram, narrow cobbled road. Keep the person, the product, the camera movement and the lighting the same." The Sume docs example for edit mode, "Replace the bottle with an apple. Keep everything else the same.", has the same shape: replace this, keep the rest.
What it costs
Edit output follows the source length, and the rate is per second of output. For an 8-second source:
| Resolution | Per second | One edit | Five edits |
|---|---|---|---|
| 360p | $0.0375 | $0.30 | $1.50 |
| 720p | $0.125 | $1.00 | $5.00 |
| 1080p | $0.1875 | $1.50 | $7.50 |
Run it as a batch
Give each market its own Idempotency-Key, such as ad-city-lisbon, so you can retry a single market without paying for the other four. Send the five requests async and collect the results from the jobs. Keep a table of market, key and job id, so a late result is matched to the right city.
Keep the master clean
Start from the clip without captions, logos burned in, or end cards. Those elements pass through the edit and may be redrawn. Add them back after the edit on a timeline, per market, with the local language text. Keep the master at the highest resolution you have and let the 720p edit be a review copy; only the approved cities need a 1080p pass.
The output follows the source length, so cut a longer master to the 8 seconds that contain the backdrop before you send it; video trim does that for $0.02 per job.
Order of work
Run Lisbon first as a 360p test, since a bad template costs $0.30 rather than $1.00. Once the template holds, run the other four at the tier you will review. Approve one city before you spend on the rest of the set at 1080p; this ordering turns one bad prompt into a $0.30 mistake.
Keep the template text in version control next to the Idempotency-Keys. A localized set tends to be re-requested a month later when someone changes the offer.
What a backdrop swap cannot fix
The prompt can change what is behind the subject. It cannot make the clip culturally right. Signage in the wrong language, drive-on-the-left traffic, or a plug type on a countertop will need a person from that market to look at it. The model may also render storefront text as nonsense. Check the stills at three moments with video frames and send the set to the local reviewer.
If the person in the ad also has to change by market, that is a different tool: one ad, two recast jobs describes it. For the background-only case on one clip, see changing the background behind a person.
Sources
Related posts
More in Use cases
- One approved backdrop for 200 product photos: reference edit cost
Use one approved scene as the first reference and each product shot as the second, in Ideogram 4.5 on Sume. 200 photos cost $7.50 at low and $15 at medium.
- One voiceover, two ratios: render 9:16 and 1:1 from one audio file
Reuse one voiceover for a 1080x1920 and a 1080x1080 cut in Sume Timeline. Two renders at $0.10 per output minute each; only output width and height change.
- How do I make one intro bumper and reuse it in every course lesson?
Make a 5-second bumper once ($0.83 with a sting), then add it to 24 eight-minute lessons at $0.82 each: $20.51 for the course on Sume.
- Outage status update video with an AI avatar: script and cost
A 20 to 30 second avatar clip can front a status-page update. What to script, what it costs by tier, and why the written update must go out first.
Written by Sume