38-second ad: stills, clips, voice, music, render, cost per second
One 38-second ad from 6 stills, five 8-second Wan clips, a 640-character voiceover, a music bed and a render costs $5.8554, or $0.1541 per finished second.

A 38-second ad built from six Nano Banana 2.1 stills, five 8-second Wan 3.0 clips at 720p, a 640-character voiceover, one Music 1.0 track and one timeline render costs $5.8554 on Sume list prices. That is $0.1541 per finished second. The video line is $5.00 of it, so 85 percent of the price is footage you mostly cut down.
Line items
Every number comes from the catalog and the docs. The footage is 40 seconds because five 8-second clips are shot for a 38-second cut; the render bills the 38-second output as one minute (ceil, minimum 1 minute).
| Line | Basis | Cost |
|---|---|---|
| Stills | 6 x nano-banana-2.1 1K | $0.6000 |
| Video | 5 x 8 s wan-3.0 720p = 40 s x $0.125 | $5.0000 |
| Voiceover | 640 characters x $0.0475 / 1,000 | $0.0304 |
| Music | 1 x Music 1.0 | $0.1250 |
| Render | 38 s output = 1 billed minute x $0.10 | $0.1000 |
| Total | $5.8554 |
Cost per finished second is the better metric
A per-clip price hides the shot ratio. Here 40 seconds of footage ends as 38 finished seconds, a ratio of 1.05, and the cost per finished second is $5.8554 / 38 = $0.1541. If you shoot 2 clips for every one you keep, the footage line doubles to $10.00 and the total becomes $10.8554, or $0.2857 per finished second.
The non-video lines (stills, voice, music, render) add up to $0.8554. Together they are 14.6 percent of the total. Optimizing them is worth less than choosing the right resolution for the clips.
Running it as a series
Four such ads a month cost 4 x $5.8554 = $23.4216. Move the footage from 720p to 1080p on Wan 3.0 and the video line becomes 40 x $0.25 = $10.00, so one ad is $10.8554 and four are $43.4216.
The other lines barely move with scale. A longer voiceover adds $0.00475 per 100 characters. A second music track adds $0.125. A second render minute, if the ad goes over 60 seconds, adds $0.10. The footage line is the one that matters, so use the table to decide where to spend: shots a viewer sees for two seconds do not need the same resolution as the hero shot.
Keep the shot ratio (footage seconds divided by finished seconds) in your own log. It was 1.05 here and 2.1 when half the takes were rejected.
Swap one line and recompute
- Clips at 480p Wan instead of 720p: 40 x 0.0625 = $2.50, total $3.3554.
- Clips at 720p Seedance 2.5: 40 x 0.5778 = $23.112, total $24.4674.
- Add a second music track: +$0.125, total $5.9804.
What to log
For each ad, record these five numbers in your own sheet so the next quote is a lookup instead of a guess.
- Footage seconds generated and finished seconds delivered, to get the shot ratio.
- Stills generated and stills used. Six stills at $0.10 each is only $0.60, so do not over-optimize this line.
- Voiceover characters. At 640 characters the voice costs $0.0304, which is a rounding error.
- Render minutes billed. A 38-second ad bills one; a 61-second ad would bill two.
Sources
Related posts
More in Pricing
- 39 vertical Omni Flash clips of 7 s: $34.13 at 720p, $51.19 at 1080p
39 clips of 7 seconds on Gemini Omni Flash 1.1 is 273 seconds: $10.24 at 360p, $34.13 at 720p, $51.19 at 1080p and $102.38 at 4K on Sume, with native audio.
- A $4.44 ceiling per clip at 1080p or 4K: which Sume models still run
Under a $4.44 cap per clip, Wan 3.0 1080p runs 17 s and H3 Max 1080p its full 15 s, but Seedance 2.5 1080p cannot start. Table of the longest clip per row.
- 4-second 1080p clip: Veo 3.1 Fast $0.48 at Google vs Sume
Google lists Veo 3.1 Fast at $0.12/s for 1080p, so 4 s is $0.48. Sume's cheapest 4 s 1080p-class row is Gemini Omni Flash 1.1 at $0.75.
- 4-second 1080p clip: Veo 3.1 $1.60 at Google vs Sume
Google lists Veo 3.1 at $0.40/s for 1080p, so 4 s is $1.60. Sume's cheapest 4 s 1080p-class row is Gemini Omni Flash 1.1 at $0.75.
Written by Sume