Which entries say when the output disappears
Two entries publish a clock on the finished file. One states that generated videos are stored for two days and then removed; the other returns a video URL valid for twenty-four hours. On the rest, collection has to be planned on an assumption. As of 2026-09-22.
| Item | What is published | Detail |
|---|---|---|
| Two days, then removed | One entry | Re-generation is subject to that page's tier conditions |
| A URL valid for 24 hours | One entry | Re-generation is unconditioned on that page |
| Nothing stated | Thirteen entries | A gap, not a guarantee of permanence |
Inclusion rule. Statements each page makes about its own output remaining available. Input ceilings and storage a buyer provides are different subjects and are not recorded here. Order. Longer window first, then the silent rows.
1A published clock makes collection a scheduled step
Both windows are generous for a single piece and thin for a batch. A queue that runs overnight and is reviewed the next afternoon fits inside either; one that runs on a Friday and is reviewed on a Monday fits inside neither.
The failure is quiet. Nothing errors at generation time and the pieces were produced correctly, so the problem appears when somebody opens a link, which is exactly the kind of specification that belongs beside the length and the resolution.
2The shorter window has the cheaper recovery
On the two day row a repeat request is subject to the same tier and length conditions that page attaches to its longest run. On the one day row nothing is coupled, so a repeat is a plain request at any tier.
Reading the windows alone would rank them the other way round. It is the reason these sheets keep the conditions in the cell rather than collecting them somewhere else.
3Silence is not permanence
Thirteen entries say nothing, and an unstated retention policy is a gap. A pipeline built on the assumption that a file will still be there next week fails late and expensively, and no page in this register supports that assumption.
The practical rule is the same everywhere: copy the output into storage the buyer controls as soon as the job finishes. On two rows the deadline is published and on the rest it has to be guessed at.
- How long the output stays fetchableGenerated videos are stored on the server for 2 days, after which they are removed
- How long the output stays fetchableA video URL valid for 24 hours is returned
4A generator's ad pipeline
For ad work around “How long output stays”, SceneMixer's pages explain how its ad projects are produced:
You can recomposite with different end-card text, and in step-by-step mode you can reshoot individual segments. The finished file can also be downloaded and shared with a link.
SceneMixer, How to Make an Ad from One Photo — “Can I change the ad after it is finished?” (read 2026-10-13)
Measured across every job that finished successfully on scenemixer.com in the thirty days to 20 September 2026: analysis of a whole book runs about 2 minutes 49 seconds at the median, one episode's storyboard 1 minute 54 seconds, one video segment 3 minutes 20 seconds, and compositing an episode 3 minutes 52 seconds.
SceneMixer, Turn a Novel into a Video Series with AI — “How long does it take to produce one episode?” (read 2026-10-13)
5Sources
Format requirements come from Google Ads Help, read 2026-09-12, and tool limits from each vendor page listed on published limits, read 2026-09-22. The rule against mixing the two is on method. Nearby: Floors, not just ceilings, Length in frames, Single values.