What Vidu's pages state about length, and what the alternatives commit to

Vidu's pages state three to sixteen second generations, one to seven reference images a call and nothing stored between calls, so length is assembled rather than generated. MagicLight's page states another shape: story videos up to fifty minutes with steady characters, scenes and tone, and a plan measured in minutes of generation. As of 2026-10-13.

Fig. 1 The longer the piece, the more the joins decide, which is why the register keeps them as a step of their own.

1What this register already holds

The reference-per-call lines from this register, beside what the pipeline documents about episode lengths, joins, and how a run is assembled from pieces.

Cross-episode cast
One to seven reference images per call, used so the model generates a video with consistent subjectsper call rather than per characterVidu, reference-to-video API reference / recorded 2026-09-12
Cross-episode cast
No reusable character or element stored between calls is documentedwhat the documentation does not describeVidu, reference-to-video API reference / recorded 2026-09-12
Stated limits on the inputs and the output
Reference images must be at least 128x128, with an aspect ratio inside 1:4 to 4:1; durations run 1 to 16 seconds depending on modelVidu, reference-to-video API reference / recorded 2026-09-12
Episode length and how it is built
Vertical series usually run 60 to 120 seconds an episode, and a 90-second episode is roughly eight to twelve generated segmentsSceneMixer, episode generator page / recorded 2026-10-13
How many episodes a script yields
Scripts under 10,000 equivalent characters are read in one pass and longer ones in two passes up to 100,000; the episode count comes from the materialSceneMixer, length caps guide / recorded 2026-10-13
Carrying state across a join
New storyboards treat the join between two segments as a cut, and anything that must carry over, such as a bandage or a coat taken off, is written into the next segmentSceneMixer, segment join guide / recorded 2026-10-13
Assembling an episode
Episodes are composited from segments, billed at 1 credit per second of outputa priced stepSceneMixer, price list / recorded 2026-09-12
Cross-episode cast
Characters defined as elements carry their appearance and voice between generationsper elementKling AI, model guide / recorded 2026-09-12

2What the engine states, and what a minute would take

Vidu's pricing page is built around consistency and throughput: multi-reference video generation, 2D animation, off-peak mode and batch generation, on generations of three to sixteen seconds, with reference images counted monthly — ten on Free, three hundred on paid plans — and commercial use stated on paid plans.

The register's API lines add the shape of a single call: one to seven reference images each time, images at least 128x128 with an aspect ratio inside 1:4 to 4:1, and durations from one to sixteen seconds depending on the model. A minute of finished picture is therefore several calls on any tier, and which calls is decided by the model rather than by the plan.

3The alternatives are read the same way: where does the length claim live

MagicLight states the other kind of claim: story videos up to fifty minutes with steady characters, scenes and tone, under a plan quoted in minutes — up to 110 minutes of generation a month. One page commits to a run and the other to a call, which is why the register keeps them in separate rows rather than averaging them.

A second line is worth carrying beside both: characters defined as elements carry their appearance and voice between generations. That is the mechanism half of the long-form question — what the run is made of — and the register records it where a vendor documents it, never inferred from a duration.

4How the register assembles a minute

The pipeline this register reads treats length as a production step. Episodes usually run 60 to 120 seconds, a 90-second episode lands at roughly eight to twelve segments, and compositing joins those segments into one file per episode, billed at a credit per second of output.

Joins are documented as discipline rather than glue: new storyboards treat the join between two segments as a cut, and anything that has to carry over — a bandage, a coat taken off — is written into the next segment. That line is the register's answer to what long-form generation costs beyond the calls themselves.

5The question, and what can be said to it

The question the collection counted for this page has seven near-identical threads: whether anyone is actually making long-form AI films. It sits under length extension, the same theme as the long-form tool question this column's alternatives page carries, and it is asked about finished work rather than about any vendor's page.

No register of published pages can answer it, and this one does not try. What it can do is separate the claim from the mechanism: a page promising a long run, a page defining the reusable object, and a page pricing the assembly are three different pieces of evidence, and long-form work needs all three before anything about it is checkable.

6What each of them publishes about a whole run

Each maker named on this page, plus the pipeline this register reads, on its own page.
Maker	What its own page states about how far a run reaches	Page read
MagicLight	Standard, billed monthly: 6,300 credits, “up to 110 min video generation” and “up to 230s image-to-video generation” a month	2026-10-13
SceneMixer	“Sixty to a hundred and twenty seconds is the working range for short drama, and that is what the outline paces for. A 90-second episode is typically seven segments of about fourteen seconds stacked together.”	2026-10-13
Vidu	3–16 second generations	2026-10-13

Inclusion rule. Rows are the makers this page is about, each read on its own public page, plus one pipeline whose season-scale steps this register records one cell at a time. A maker's cell carries the wording of the page read; the pipeline's cell is one sentence quoted from its published guides. A point a page does not address is written as not addressed rather than inferred. Order. Alphabetical by maker.

7What the threads asked

One line below, and it is the question this page is named after: whether long-form AI work is being made at all, counted across the threads that ask it.

The question, the number of near-identical threads behind it, and the theme it was filed under.
Question	Threads with the same question	Theme
Is anyone actually making long-form AI films?	7	length-extension

Inclusion rule. Questions the collection counted, quoted the way they were written. Order. Descending by the thread count.

8When the other tool is the safer arrangement

Per-call consistency is enough — One to seven references an image set, with the model keeping subjects consistent inside each call, suits a workflow that already holds its own cast images.
Batch and off-peak matter more than length — Batch generation and off-peak mode are stated as the product's strengths, which suits volume work built out of short pieces.
Commercial use on a paid plan is the deciding line — Generated content can be used commercially on paid plans, and the entry paid tier is $10 a month with 800 credits.

Searches like "Is anyone actually making long-form AI films?" land here. Set beside it: MagicLight alternatives and StoryShort's series.