Order before footage
In a conventional edit, the sequence is the last thing you decide because the footage has to exist first. Here the sequence is the first thing, because each card is rendered as a separate take and the join reads the timeline. That turns a question that used to be expensive to answer, what if the reveal came earlier, into a drag and a re-join for two credits.
Cards show their shot size and seconds on the face, so a sequence can be read without opening anything: W 8, M 6, OS 6, C 5, W 10. If that reads like a shot list, it is because it is one.
Rendering independent takes
Each card renders as its own job. That matters for sequencing because a change to one card never forces a re-render of the others. Swap the fourth shot for a different line, render only that card, and join again. The other takes are reused as they are.
Cast, outfits, props and the set are passed as reference images on every take, which is what keeps shot two and shot eleven looking like the same people in the same room. A voice style is a description in the prompt, so expect a voice in that style rather than a particular voice.
Joining and re-joining
The join conforms every take to one 9:16 frame at the sharpest resolution on the timeline, adds a silent track where a take has none, and concatenates them losslessly in card order. It runs in under a minute for a full twelve-card episode and costs two credits whether you join once or ten times.
The previous cut is retired when a new one is made, so the library does not fill with stale versions. The takes are never deleted by a join; they are separate files the episode references.
Sequencing for the small screen
Vertical episodes are watched on phones, where a cut lands harder than it does on a wall. Short shots tolerate rapid sequencing; a fifteen-second card should be a shot with somewhere to go. The runtime meter helps here too: a sixty-second target with twelve cards means five-second shots, which is a brisk rhythm, and the meter makes that arithmetic visible.
The default aspect ratio is 9:16 and every join outputs it. Cards do not have an aspect ratio of their own; the episode does.