Find the story worth telling · 30 min

A strong premise is one character, one want, one obstacle, and one change — plus a theme, the thing your film is really about. If you can state the theme in one sentence, every later decision gets easier.

Ideas are cheap; premises are engines

An idea is a situation. A premise is a situation with a motor: a character who wants something, an obstacle in the way, and a change that will result. "A haunted lighthouse" is an idea. "A retired keeper must spend one last night in her haunted lighthouse to prove to her granddaughter that fear can be faced" is a premise — it already tells you the character, the want, the obstacle, and the arc.

Name the theme in one sentence

The theme is what your film is really about, underneath the plot. Not "a heist" but "trust is built in small moments." Not "a race" but "ambition costs more than it promises." Write it as a full sentence, not a word — "loneliness" is a topic, "loneliness makes people mistake attention for love" is a theme. When a scene isn't working, ask which sentence it serves.

The premise test

Can you see it? Can you say it in one sentence? Does it contain a visible change? If all three answers are yes — and you can name the theme — you have a story worth telling. If any answer is no, keep digging before you write a word of script.

Time to use it. Copy the prompt below, open Muse, and paste it in — replace [paste idea] with your raw idea in plain words, then send. Muse will hand you three premise options, each with a theme, and recommend the strongest. Pick the one that excites you (or steal the best parts of each) — that premise carries you through the rest of the course.

I want to develop a short-film premise. Here is my raw idea: [paste idea]. Turn it into three premise options, each with one character, one concrete want, one obstacle, and one visible change. For each, suggest a one-sentence theme — what the story is really about underneath the plot. Then recommend the strongest of the three and say why. Do not write any script yet.

Build characters who want something · 45 min

Characters are defined by what they want and deepened by what they need. Give your protagonist one clear want, one hidden need, and one flaw — that is enough machinery for a short film.

Want drives the plot; need drives the arc

Ask your protagonist two questions. "What do you want?" — the answer should be concrete and filmable: the trophy, the apology, the keys. "What do you need?" — the answer is the lesson the story will teach them, usually the opposite of their flaw. In a great short, the character gets the need by not getting the want — or gets the want only after accepting the need.

Give them a flaw, not a quirk

A quirk is decoration: she collects vintage spoons. A flaw is structural: she never asks for help, so the story can force her to. The flaw should be the specific thing the plot will test, scene after scene, until the climax demands they overcome it or pay the price.

Two characters are plenty

A short film has no room for an ensemble. One protagonist with a clear want, plus one other character who either helps or blocks them, is a complete cast. If you're tempted to add a third, first ask whether an existing character could carry that function.

Same ritual as Lesson 1: copy the prompt, paste it into Muse, and fill in your premise and theme from the last lesson where the brackets are. You'll get a tight sheet for your protagonist and a slimmer one for a supporting character whose want collides with theirs. Read both over — anywhere a line doesn't feel like your character, rewrite it before moving on.

Help me build character sheets for a short film. Premise: [paste premise]. Theme: [paste theme]. Create a sheet for the protagonist with: want (concrete and filmable), need (the lesson), flaw (structural, testable by the plot), and voice (one signature phrase). Then create a slimmer sheet for one supporting character whose want collides with the protagonist's. Keep each sheet under 150 words. No script yet.

Structure: the shape of a short · 60 min

A short film is one turn: setup, pressure, payoff. Map it as beats before you write a single line of dialogue — structure is cheap to fix on an index card and expensive to fix on page ten.

The three movements

Setup introduces the character, the world, and the want — and plants everything the ending will need. Pressure raises the cost of the want: attempts fail, the flaw gets exposed, the obstacle pushes back harder. Payoff forces the choice the whole film has been building toward, and shows the visible change. One turn, cleanly executed, beats three half-turns.

Beats, not pages

Don't start with "page one." Start with a row of index cards — real or digital — one beat per card. Shuffle them. Cut the weakest. You'll discover the film's shape in an afternoon; discovering it in a finished draft takes weeks. Eight to twelve beats is the right count for a short.

Plant and payoff

Every payoff must be planted. If your ending needs the protagonist to pick a lock, we need to see the lockpicks — or the locksmith past — in the setup. Audiences love an ending they didn't see coming but instantly believe. That feeling is engineered, beat by beat, not improvised.

Write scenes that move · 60 min

Every scene should answer one question: what does the character want right now, and what's in the way? Enter the scene late, leave it early, and make sure something changes before you cut.

The scene question

Give every scene a dramatic question the audience can feel: Will she open the letter? Will he tell the truth? The scene's job is to answer it — ideally not the way we expected. If two scenes ask the same question, merge them. If a scene asks no question, cut it.

Enter late, leave early

Start the scene at the latest moment the audience can still understand: the argument already underway, the decision already half-made. End it the instant the change registers — a look, a door closing, a hand letting go. Everything after that is explanation, and explanation is the enemy of momentum.

Something must change

Knowledge, power, relationship, plan — something has to shift between the scene's first line and its last. Test each scene with one check: could the story skip this scene and still make sense? If yes, the scene isn't moving anything. Rewrite it until the answer is no.

Paste your beat sheet into the prompt below (replacing the bracket), run it in Muse, and you'll get a 5–8 scene list — each with its dramatic question, where to enter late, where to leave early, and what changes. Test every scene against the one-question rule before moving on.

Turn this beat sheet into a scene list for a short film: [paste beat sheet]. For each scene give: location, the scene question (what the character wants right now and what's in the way), where to enter late, where to leave early, and what changes by the end. Merge any beats that ask the same question. Aim for 5–8 scenes. No dialogue yet.

Dialogue that sounds human · 45 min

Real people rarely say what they mean — your characters shouldn't either. Write the subtext first, then let the dialogue hide it. Then cut every line that says exactly what the audience already knows.

Subtext is the real conversation

People talk around what matters: they deflect, joke, change the subject, answer a different question. Write each line twice — first the honest version ("I'm terrified you'll leave"), then the human version ("Take your time. I'll be here. Probably."). The gap between the two is where acting lives.

Give each character a voice

Voice is rhythm plus vocabulary plus attitude. One character speaks in short bursts. Another never finishes a sentence. A third hides behind jokes. Write a "never say" list for each: words and phrases that would betray who they are. Then do the cover-the-names test — if every line could belong to anyone, keep rewriting.

Cut the on-the-nose

If the audience just watched the betrayal, the character doesn't need to announce "You betrayed me!" On-the-nose dialogue states the subtext out loud and kills it. When in doubt, cut the line and see if the scene still works — it usually works better.

Take one scene you're unsure about, paste it into the prompt below, and run it in Muse. For every line you'll see the subtext spelled out first, then a rewritten line that hides it the way a real person would. Keep whatever sounds like your characters; discard the rest.

Help me rewrite the dialogue in this scene so it sounds human: [paste scene]. For each line of dialogue: first state the subtext (what the character really means), then rewrite the line so it hides the subtext the way a real person would — deflection, humor, evasion, or partial truth. Cut any line that states exactly what the audience already knows from the action. Keep each character's voice distinct. Return only the rewritten scene.

Draft the short-film script · 90 min

A screenplay is a blueprint, not literature: sluglines tell us where we are, action lines show only what the camera sees, dialogue carries voice. Write the whole draft fast — momentum beats perfection.

The three tools: slugline, action, dialogue

A slugline opens every scene: interior or exterior, location, time of day. Action lines show only visible, present-tense facts — no thoughts, no backstory, no camera directions. Dialogue is the speaker's name with the line beneath; a parenthetical like "(whispering)" is used sparingly, only when the delivery isn't obvious.

Write fast, fix later

A short-film script is five to ten pages — one page per minute of screen time. Draft it in as few sittings as you can. When you stall on a scene, write "[FIX LATER]" and keep moving. A complete flawed draft teaches you what the film is; a perfect first three pages teach you nothing.

AI as your writing partner, not your writer

AI is excellent at volume: ten loglines, five versions of a scene, a list of what's confusing. It is terrible at taste: which version is yours. Use it to generate options and to stress-test your choices — "what's the weakest scene and why?" — but never hand it the decisions. The script must sound like you on your best day, not like everyone.

The big one. This prompt assembles everything you've built — premise, theme, character sheets, beat sheet, scene list — into a full first draft. Paste each piece where its bracket is and send it in Muse. Expect 5–10 pages with [FIX LATER] marks where scenes need work. That's normal: this is a discovery draft, not a final.

Help me draft my short-film script. Here is my material: PREMISE: [paste]. THEME: [paste]. CHARACTERS: [paste sheets]. BEAT SHEET: [paste]. SCENE LIST: [paste]. Write a complete first draft in standard screenplay format: sluglines (INT./EXT., location, time), action lines (visible, present tense only), and dialogue. Aim for 5–10 pages. Where a scene isn't working yet, write the best version you can and mark it [FIX LATER] — do not stop the draft to perfect anything. This is a discovery draft, not a final.

Rewrite and lock the script · 60 min

The first draft is for discovering the story; the rewrite is for the audience. Read it aloud, cut ten percent, fix only what confuses or bores — then lock it and call it done.

The table read

Read the entire script aloud in one sitting, start to finish, no stopping to fix. Mark three things: stumbles (dialogue no human would say), confusion (you can't picture what's happening), and boredom (your attention drifted). Have AI read it back to you for a second pass — hearing it in another voice exposes what your eyes skip.

Cut ten percent

Before rewriting a single line, cut. The repeated beat, the extra scene that makes the same point, the action line that describes what the dialogue already says. Aim to remove a full tenth of the pages. What's left will be denser, faster, and more confident — and you'll have room to add the one thing the table read proved was missing.

Lock the script

Rewrite in passes — story, then scenes, then lines — and stop when the table-read marks are gone. Perfection is not the goal; a locked script is. Save it, date it, and resist further tweaks. A finished script you can shoot beats a perfect script that stays in a drawer.

Paste your full draft into the prompt below and send it in Muse. You'll get notes only — the 10% to cut, the three highest-impact fixes in priority order, and one thing that's already working and shouldn't be touched. Work the fixes top-down, then lock the script.

Give me rewrite notes on my short-film script draft: [paste draft]. Evaluate only: story clarity (can a first-time reader follow it?), pacing (where does attention drift?), character consistency, dialogue (flag every on-the-nose line), and whether the ending pays off the premise and theme. Then give me: (1) the 10% to cut, specific lines and beats; (2) the three highest-impact fixes in priority order; (3) one thing that is already working and should not be touched. Do not rewrite the script — notes only.