Hello everyone,
This week, almost everything happened in the editor. Creating a scenario no longer asks you any questions, creating a character does so much less, the voice styles announced last week are now live, and the AI now knows what you’ve already decided for your project instead of asking you again for every generation.
Voice styles are live
In Settings → Audio → Dubbing → Distribution, Choose a style next to a character’s voice opens styles in four families: Narration, Emotions, Characters, and Effects. Try (1 credit) plays the style on the voice before you pick it, Try in Standard gives the same line without a style so you can compare, and Try on a line tests it on one of its real lines. A style someone has already tried can be heard for free. Styles are selected per character and per language, and dubbing with a style costs the same as without.
The Custom style
The last card gives you four sliders: Rhythm, Expressiveness, Consistency, and Fidelity to the voice. Moving a slider costs nothing; only Try doubles a line, at the price shown. Changing styles works the same way as changing voices: the editor counts the lines that would become muted before letting you confirm.
By the way, dubbing has moved: it’s now in the Audio tab of the settings.
Create a scenario: no more questions
Creating a scenario now asks you nothing: it starts with a provisional title and your writing language (the one from your account), which you can change whenever you want. Three doors at the start: Start from a starter scenario, the guided walkthrough, or Open a blank scenario. Before Create scenario, a review shows what the scenario starts with and where to change it. You can also Save and finish later: the dashboard offers to pick up where you left off. Once created, the scenario opens on its summary; the Editor guided tour no longer forces itself on you—it starts from Help when you want it.
Start from a starter scenario
These are small complete games written by the platform studio. One click, and you’re in the editor with your own copy, ready to modify just like any other scenario. What comes with it is marked Not yet modified until your first edit, and the project summary offers Remove unmodified content to clean everything up in one go. What your own content uses is kept, and the editor tells you what and why.
Create a character step by step
The create button now opens an assistant with five steps: starting point, who they are, their tastes, their appearance, and verification. You can start from one of the ten illustrated archetypes, a character from your packs, a blank page, or click Surprise me so the AI proposes a full one (its cost is shown before). Each field has its own little die to roll another value at random, and the suggested names are real first names, matched to gender. The Sound appearance step is optional: a portrait from your media, an uploaded image, or a portrait drawn by the AI, attached to the character right when it’s created.
A character keeps the same face across portraits
When you generate a new portrait for a character, the AI draws it based on their calmest face in the same outfit (otherwise their default outfit), so it’s clearly the same person. The expression doesn’t carry over: a smiling portrait no longer makes a sad portrait smile. The Draw from this portrait checkbox lets you skip that for an image.
Project image and music settings
The settings gain two tabs: Images (graphic style, framing, quality, visual direction) and Audio (music style, then dubbing). Each value tells you where it comes from (Defined for this project, the starter scenario, Based on World and genre, or the platform) and how many images have already been made with it. Changing a value only affects the next images—nothing gets regenerated.
Every generation tells you what it will use
Just above the Generate button, each AI window shows What this generation uses: style, framing, quality, classification, along with the origin of each value and the cost. A value you change for this generation is marked Modified, with what you can use to go back to the project’s version.
The AI writes in your game’s classification
Generated text and images now follow the classification and content warnings you declare in Publication, rechecked with each generation. A general-audience game stays general-audience, and whatever the warnings don’t announce, the AI doesn’t write.
Conditions read like a sentence
A condition is now written line by line, in columns (the subject, the operator, the value), and then read back as a sentence above the fields: “Charisma at least 30, and Lena at least Friend.” This applies to scene and step triggers, routings, and choice options. Nothing is typed by hand, except the name of a new flag.
Routings read like a set
Multiple routing steps that follow each other are grouped: the first whose condition is met wins, and a Else line tells where the player goes in all other cases. A routing that’s always taken by a broader routing placed before it is marked Never taken, with the responsible step.
Two extra guardrails before publishing
A multiple-choice question always leaves at least one option for each player: the editor refuses the edit that would trap some players without an answer, and explains why. And a condition that no player can satisfy—given the ranges you declared for your game values—blocks publication instead of creating a scene nobody will see.
The steps list
A collapsed step is read in one line: who is speaking and the start of their line, the options for a choice and their conditions, and the list of effects. You can also control it entirely from the keyboard (↑/↓, Enter, Esc, Cmd/Ctrl+D to duplicate, / to search for a step), and you move a step without dragging it: Alt+↑/↓, or the ⋯ menu with Move to position…
Duplicate without redoing everything
A whole scene can be duplicated in one click, from its row or menu. In a multiple-choice question, an option is duplicated too, and Apply to other options… copies its condition or one of its effects to the others.
Small things that make life easier
The scenario title renames directly in the header. The genre and age of a character can be edited on their profile (they were only asked for at creation). And a multiple-choice question keeps its options displayed when you add a step right after.
I hope you’ll like this—feel free to send me feedback.