Give every storyline its own people
An ensemble falls apart when the audience cannot tell who belongs to which plot. Series Maker fixes that at the structural level: a thread owns a cast subset, the drafts only write those people into that thread’s scenes, and every take is rendered from the same reference portraits.
Casting by thread, not by episode
When you create a thread you choose up to three characters for it. The harbour pilot and the radio host might live in thread A; the auctioneer and the lawyer in thread B; the night-shift crew in thread C. The braid planner reads those subsets and keeps them apart until you tell it, in the arc notes, that two storylines are allowed to meet.
This is deliberately stricter than a shared cast list. In a long run, a writer’s room drifts: a side character wanders into a plot where they have no business. Keeping the subsets explicit makes that drift visible and reversible.
Characters that look the same in episode twelve
Characters are always generated, never uploaded, because the render model only accepts face references with a known origin. You describe a person once, the library produces a clean front-facing portrait, and from then on every take that includes them is rendered with that portrait as a reference. Add an outfit reference and the wardrobe carries over too; add a prop and it sits in their hands in each scene.
The preset ensemble ships with sixteen people whose ages, silhouettes and faces are deliberately far apart, so a viewer can tell thread A from thread B even on a small screen. You can use them as they are, mix them with your own, or build an entirely original cast.
Voice styles, not voices
Each cast slot can carry a voice style: a short descriptor such as dry and unhurried, or bright and nervous, with an age range and accent. These are prompt descriptors that steer how the render model performs the line. They are not clones of anyone’s real voice, and the delivery will vary from take to take. Think of a voice style as a casting note, not a recording.
- Up to three characters per thread and per take
- Up to nine reference images per take: characters, outfits, props and the set
- Sixteen preset voice styles, each a descriptor rather than a recording
- Outfits, sets and props can be generated for 3 credits or uploaded for free
How an ensemble episode comes together
Ask for a braid and the planner decides which threads carry the episode. Say threads A and B advance while C rests: you get two scenes, each written for its own cast. Shoot them as takes, add a third scene by hand if the episode needs a button, then weave. The public series page lists the episode with the thread letters it braided, so even a first-time viewer can follow who is who.
Takes from the preset ensemble
Each take is one thread's scene, rendered from library references so the cast stays recognisable across episodes.
Questions
- Can I move a character from one thread to another?
- Yes. Edit either thread’s cast subset at any time. Takes already rendered keep their original cast; future drafts use the new subsets.
- Why can’t I upload a photo of a real person as a character?
- The render model only accepts face references it can trace to a generated image. So every character starts as a generated portrait in your library, while wardrobe, set and prop references may be uploaded.
- Will a character sound exactly the same in every take?
- No. A voice style steers the performance but does not fix it, and each render is a fresh take. Expect a consistent character, not an identical reading.
- Who owns the characters and episodes I create?
- You do. Everything you plan, shoot and weave on Series Maker belongs to you under our terms of service. Copyright works differently depending on the country you are in; when ownership matters for a deal or a release, ask a lawyer where you live.
Ready to braid?
The studio opens with thread names pre-filled from this page. New accounts get 2 credits to look around.