Handing a season to someone else
A season in progress is not a folder of videos. It is a cast, a set of locations, a prompt idiom, a set of tool settings and a record of which of those produced which episode. Hand over the folder alone and the series restarts visually at whatever episode the new hands begin. As of 2026-09-12.
| Item | Lost if it does not travel |
|---|---|
| Cast reference images | Faces drift from the handover onward |
| Episode record | Nobody can explain why an episode looks different |
| House prompt phrasing | The look changes while the new operator relearns it |
| Location references | Rooms change between episodes |
| Tool settings and versions | Later shots cannot be matched to earlier ones |
Inclusion rule. Artefacts that cannot be reconstructed from delivered video. Anything recoverable from the files themselves is not listed. Order. Alphabetical by item.
1The cast travels as files, not as descriptions
Every recurring character needs its reference images handed over, not a paragraph describing them. A new operator writing their own description of the same character produces a different person, and the difference shows at the first cut between old and new material.
Wardrobe variants count as part of the cast. A character who appears in two outfits has two reference sets, and knowing which episodes use which is part of what is being handed over.
2Locations are the half that gets forgotten
Recurring rooms and exteriors need the same treatment and rarely get it, because they feel like background. A room re-described comes back with different furniture, and a season where the same apartment changes between episodes reads as carelessness.
Include the time of day each location was established at. The same space at a different hour is a different picture and an easy thing to lose in a handover.
3The prompt idiom is an asset
Productions develop phrasing that works: how shots are described, which words the model responds to, which constructions it ignores. That vocabulary is worth more than most of the settings and is almost never written down.
A page of house phrasing, with two or three examples of prompts that produced good shots, saves the new operator a week of rediscovery and keeps the look continuous while they learn.
4Tool settings, versions and accounts
Model and version, resolution, aspect ratio, clip length, and whatever quality tier was in use. Also which account produced the work, because permissions and history attach to accounts rather than to files.
Where the platform pins model versions, hand over the pin. Where it does not, hand over which version was in force and when, so a later mismatch can at least be explained.
5The episode record
For each delivered episode: which reference set it used, which settings, what was regenerated and when. This is the document that answers why episode six looks different, and it is the one that cannot be reconstructed afterwards.
It is also short. A row per episode with five columns covers nearly every question a new operator asks in their first week.
6Hand over the failures too
Which shots took many attempts, which ideas never worked, which prompts reliably produced artefacts. Negative knowledge is expensive to acquire and free to transfer, and without it the new operator spends budget rediscovering the same dead ends.
A short list of what not to try is often the most useful page in a handover pack.
7What makes a handover possible at all
Some of this only transfers if the tool documents the mechanism it uses. Where a cast is held by a named, exportable reference sheet, it can be handed over; where consistency comes from something the product does not describe, there is nothing to pass on but the output.
In the register that distinction is visible across the three season-scale steps: holding a cast across episodes, turning one script into episodes, and assembling an episode. SceneMixer publishes a sentence on all three — a reference sheet injected into every segment, episodes produced in one pass, and assembly priced per second — which is what makes a mid-season handover describable rather than improvised. Kling AI, Vidu and Runway each document part of it, and their rows record which part. For a production that expects to change hands, a tool documenting the mechanism is worth more than one that merely performs well on a single episode.
8A handover pack that fits in an afternoon
Reference images by character and by location. A page of house phrasing. A settings sheet. A row per episode. A list of dead ends. Account and version notes. Names of who to ask.
Seven items, none of which need writing from scratch if they were kept during production. That is the actual argument for keeping them: the handover is when the cost of not having them arrives.
9Where the register holds these tools
The techniques above are general craft. Which tools document support for them, and in what words, is held in the register, each cell carrying the page it was read from and the date.
- The register — which tools document a season-scale mechanism, with sources
Craft notes, not a register entry. No line here is attributed to a tool, and none should be read as a claim about one. The sourced material is on the register. Related: Checking an episode, Planning a season.