Docs › Screenwriting › Adapt your novel into a screenplay: what carries over, and what you rewrite

Adapt your novel into a screenplay: what carries over, and what you rewrite

Last updated September 10, 2026 · 15 min read

In one line: turning a novel you already have in Slima into a screenplay — what comes across for free, where to plan the episode split, what the AI can and cannot do for you, and the honest size of the job.

The file tree of one book, Ink Tide (Screenplay). Under Screenplay sit the twelve .script files, one per episode. Below them are Characters, Locations and Worldbuilding — eight, six and four .md files plus Relationship Map.map, folded here so all twelve episodes stay in frame — then Timeline/Main Timeline.timeline, Map/The Chengwen Coast.geomap, and Notes holding Screenplay Outline.md and Story Skeleton.beats. That is what an adaptation looks like: the screenplay and the novel's whole reference shelf living in one book, not two projects.

The one sentence to get out of the way

There is no converter. No "turn this chapter into a screenplay" command, no adaptation wizard, no template, and no AI tool that does it in one move. Pasting a chapter into a screenplay file produces a stack of action lines, not scenes.

What Slima actually gives an adapter is everything around the writing: the same characters and places recognised in both shapes, one board where you decide which chapters become which episode, an AI that can read a chapter and propose scenes into an episode file, and a manuscript table that measures prose in words and episodes in minutes side by side.

The writing itself is yours. Knowing that on day one is worth more than any feature on this page.

One book, or two?

One book, both shapes

  • Character sheets, places and worldbuilding are shared automatically — the same .md links from a chapter and from an episode
  • One relationship map, one timeline, one world map
  • The manuscript table shows both, with a separate total for each
  • Nothing to keep in sync

Two books

  • Cleaner exports and a cleaner shelf
  • A word count that means one thing
  • But nothing at all is shared — every character sheet, the map and the timeline have to be duplicated by hand

Nothing in Slima can reference a file in another book. Not a character sheet, not a map node, not a beat card, not a search. Two books means two copies of everything you want in both, and they drift apart from the day you make them.

If you want the second shape anyway, the shelf's ⋯ menu on the book has Copy. It duplicates every file — chapters, character sheets, the map, the timeline, the world map, the beat board — into an independent book called <name>-copy, re-identifying everything inside so the two never collide. What it does not bring: your notes, your comments, chapter targets, version history, and any image or PDF (those arrive as empty files). Delete the chapters afterwards and you have a screenplay book that starts life with the whole reference shelf.

The sample books are set up this way — Ink Tide the novel and Ink Tide (Screenplay) the twelve-episode adaptation are two separate books, and the screenplay one ships with its own copy of all eight character sheets, six locations, the worldbuilding, the map, the timeline and the world map.

What a mixed book actually looks like

Chapters/
  Chapter 01 - The Copyist Girl.md
  Chapter 02 - The Word Bureau.md
Screenplay/
  Episode 01 - The Copyist Girl
  Episode 02 - The Word Bureau
Characters/
  Shen Yan.md            ← links from both
Locations/
  You'an Scriptorium.md
Worldbuilding/
  Relationship Map.map
Notes/
  Adaptation plan.md
  Story Skeleton.beats

A .script is manuscript, the same as a chapter. Episodes therefore count towards your book's word count, your daily goal, your streak and your habit slime, and the book card's chapter count includes them. That is deliberate — writing a screenplay is writing — but it does mean a mixed book's word count answers a slightly blurred question.

The manuscript table is where the mix reads properly. Press the table icon in the Explorer header. One row per file in tree order, and the unit lives on the row, not the column: chapters in words, episodes in minutes.

Title Target Actual Progress
Chapter 01 - The Copyist Girl 8,000 words 7,412 words 92%
Episode 01 - The Copyist Girl 5 min ~4 min 88%
Total (manuscript)
Total (screenplay)

Two separate total rows appear as soon as both kinds are present — and each one only adds the rows that actually have a target, so set targets before you read the percentages.

One export consequence worth knowing before you commit to a mixed book: Fountain disappears from the whole-book export list the moment the book holds anything that isn't a screenplay. It comes back for a single file or a folder of screenplays. Markdown, plain text and the backup formats flatten a screenplay into prose. To hand over three episodes reliably, keep them in a folder and export the folder — see Handing the screenplay over.

What carries over for free

From the novel Does it reach the screenplay?
Characters/Shen Yan.md Yes. Type Shen Yan on a character cue and the name gets a dashed underline; Ctrl/⌘-click opens the sheet. The folder is irrelevant — it works from anywhere in the book, and aliases in the sheet's frontmatter count too
Locations/You'an Scriptorium.md Yes, the same way, from the scene heading's location field
Worldbuilding .md Yes, as files you open and read; nothing links to them automatically
Relationship Map.map Partly. A map node whose text is exactly a filename in the book links to that file — including an episode. But nothing goes the other way: a screenplay never points at a map node
Main Timeline.timeline In name only. See the warning below
The Chengwen Coast.geomap No links in either direction. A world map is a picture you consult
Book-level notes Yes — a note filed against the book (rather than one chapter) is visible while you have an episode open

A timeline event cannot be attached to a scene. Dragging an episode onto a timeline creates a plain event named after the file — no jump, no back-reference, and it will not follow a rename. The AI is blocked from creating linked events too. Keep the timeline as your own chronology and put the episode number in the event's title.

Plan the split on a beat board first

This is the part worth doing before you write a line, and the board is built for it: one .beats board accepts chapter files and episode files at the same time.

Deciding which chapters become which episode

New beat board Pick a structure, or start blank Drag the chapters in Drag the (empty) episode files into the same beats

Co-locating cards inside one beat is how you say "these chapters become that episode". Nothing else records it.

  • Right-click → More → New beat board. An empty board offers four structures: Three-Act (14 beats), Taiwanese TV (three acts, six beats), Kishōtenketsu, Freytag.
  • Drag chapters and screenplays in from the file tree. That is the only way to put a card on a board — no picker, no menu item, and the AI cannot place cards either. Anything that isn't a chapter or a screenplay is refused.
  • A chapter card shows its word count; an episode card shows its scene count and expands into a scene list you can edit in place — location, interior/exterior, time of day, the target minutes. Those rows write straight back into the .script.
  • Each act column reports beats · files · words. A beat holding no cards says Not written yet.

What the board does not do: there is no "maps to" relation, no coverage check, no percentage, and nothing that notices chapter 7 has no episode. Putting the cards in the same beat is a note to yourself, and it is a good one — it is just not a calculation.

One beat board carrying both the novel and the series. Under the Opening beat sit Chapter 01 - The Copyist Girl (970 words) and Episode 01 - The Copyist Girl (3 scenes) — the same stretch of story in two shapes, side by side on the same beat. The neighbouring Setup beat holds two chapters and one episode. Read the act header: 5 · 6 · 9,069 words — five beats, six file cards, and the words they add up to. While an adaptation is in flight, those three numbers are the progress bar.

Writing an episode

  1. Right-click the screenplay folder → New Script, then rename it in the tree.
  2. Fill in the title and the writer at the top of the paper. Set the target minutes in the toolbar.
  3. Choose the format — Taiwan or Hollywood. See Taiwan format or Hollywood.
  4. Write. The character autocomplete already knows everyone who has spoken anywhere in this book, and the location field already knows every place used in it.

Or ask the AI. It can read a chapter and propose a whole episode's worth of scenes in one turn: it opens the .md, then drives the screenplay's field-level tools — add scene, change heading, add element, update element — in a single batch that either applies completely or not at all.

Every one of those writes is a proposal. They appear on the screenplay as green and amber blocks with Accept and Skip, and nothing touches the file until you accept. Whole-file writes to a .script are refused outright, so "the AI overwrote my episode" cannot happen. How to ask, and how to review: Ask the AI to rework a scene.

What happens when you paste prose in

Pasting into a screenplay line does no parsing at all. The first line lands at the caret; every line after it becomes a new element of the same type as the line you pasted into, with blank lines dropped. Paste six paragraphs into an action line and you get six action lines. Paste them into a dialogue line and you get six lines of dialogue. Nothing detects a scene heading, and nothing detects Name: line.

Do not run a prose chapter through "Paste screenplay text". That importer expects text that is already formatted as a screenplay. Given prose it finds no scene heading, so it wraps everything into one scene (interior, day) with every paragraph as an action line — and it quietly takes your first line as the file's title. There is no warning and no "this doesn't look like a screenplay" check.

The scene's name is the tell. Prose scores nothing either way in the format guess, and a tie goes to Taiwan — so the scene usually comes out called 未命名, in Chinese, whatever language you were writing in. (Prose with enough ALL-CAPS lines to tip the guess to Hollywood lands on UNTITLED instead.) Either name means the same thing: the importer did not understand what you gave it.

So the realistic loop is: read the chapter beside the episode in a split editor, write the scene, and use the AI per line — right-click → AI Actions → Polish, Rewrite, Condense — where the prose gives you the words but not the shape.

Write the plan down as a .md

There is nowhere in a .script to type a logline, and no season-level settings page. So the adaptation plan lives where every other plan does: a manuscript file next to the episodes. The sample book's Screenplay Outline is a good model, and it is six short parts:

Screenplay Outline (Twelve-Episode Micro-Drama)
  In one line          one sentence for the whole adaptation
  Form                 five minutes an episode, one chapter to one episode,
                       three scenes or so, every episode ends on a hook
  How the timing       the words-per-minute the estimate uses, and where the
    is reckoned        per-scene target and the whole-episode target live
  The table            Ep | Scenes | Events | End hook   ×12
  Differs from         interior life becomes action or voice-over;
    the novel          flashbacks only very short; something happens
                       in the first ten seconds
  Sequel hooks         the threads you are deliberately leaving open

The fifth section is the one that earns its keep. Where it differs from the novel is the set of rules you will otherwise re-argue with yourself twelve times.

When not to do it this way

  • Don't wait for a converter. There isn't one, and nothing on the roadmap page will change what you have to type.
  • Don't route a chapter through the screenplay importer. It will name your file after your first sentence and give you one scene, called 未命名.
  • Don't split into two books before you have decided. Nothing crosses a book boundary, and a copy starts drifting immediately.
  • Don't ask the timeline to track your episodes. Its events are text you write.
  • Don't read a mixed book's word count as progress. Chapters and episodes land in the same number; the manuscript table's two totals are the honest view.

Related

Try it in Slima

Open the app and do this with your own book. Free to start, no credit card.

Open Slima
Was this helpful?