Short Drama Storyboard from a Script (9:16 Shot Table)

A short-drama storyboard is the last cheap draft of an episode. Not a pretty PDF. Not a film-school animatic. A row of 9:16 decisions: who is in frame, how close, what they do, what they say, and which face must match the last episode. Get that row wrong and you will pay to generate a hallway, then another hallway, then a lead who changed coats between shots.

The search box for this job is usually “AI storyboard generator” or, in Chinese, 短剧分镜怎么做. Both phrases want the same thing: a board you can shoot from, not a mood collage. Most pages answer with a generate button. This one answers with a table.

If you are boarding a cinematic short with a crew and a 16:9 frame, read how to storyboard a short film instead. That craft is real. It is also a different bill, a different aspect ratio, and a different audience.

Why a short-drama storyboard is not a film board

A film board assumes you can reshoot. A vertical board assumes you cannot. The episode is one to three minutes, the frame is a phone, and the audience is already halfway to the next title. Omdia puts 2025 microdrama revenue at $11 billion, mostly from people who paid after a free open. ReelShort alone is projected at $1.05 billion in 2026. That money sits behind a cliffhanger, not behind a beautiful wide.

So the board has a different job. It is not “cover every setup the DP might want.” It is “make the first three seconds and the last five seconds undeniable, and keep the same person in the clothes they were wearing yesterday.”

TikTok’s creative guidance found that over 63 percent of the highest click-through videos put the key message on screen in the first three seconds. On a 9:16 storyboard that means shot one is almost never an establishing wide. Shot one is a face, a status drop, or an object that should not be in that room.

Look at a strip of vertical thumbnails the way a feed looks at them: stacked, not side-by-side.

A 9:16 board is a stack of phone frames. If a panel needs a paragraph of explanation, it is two panels.

Practical rule: One panel, one readable idea. If you need three sentences to explain a panel, split it.

How do you break a script into 9:16 shots?

Start from the script, not from a “cinematic trailer” prompt. Paste the episode. Get a timeline of lines and beats. Then cut that timeline into shots the way a vertical editor cuts: on the change that the audience can see.

A usable split:

1. Mark the detonation. The first shot that asks a question. Everything before it is optional and usually deleted.
2. Mark the reversal. The line or action that changes who has power. That is its own shot, usually a closer frame than the setup.
3. Mark the hold. The last five seconds. Mid-action, not a fade. Vertical writers treat an episode as more than one scene, and they cut the cliffhanger inside a scene so the next episode can resume in the same room.
4. Assign a shot size. Wide, medium, close-up, over-the-shoulder — Boords keeps a clean visual list of shot types and angles. Bias toward close-ups. A wide is a location tax.
5. Bind a subject. Who or what must be consistent: the lead, the room, the prop that will pay off in episode four.

If you are still writing the script, stop and fix structure first. The hook / conflict / cliffhanger sheet is the cheaper document. A storyboard cannot rescue a polite ending.

A beginner walkthrough of panels is useful once — then pick 9:16 before you draw, and write the line under the panel instead of dumping dialogue in a separate script the editor re-syncs later.

Walkthrough of storyboard panels. The short-drama version of the same advice is: lock 9:16 first.

Description, reference, still, video: one row per shot

This is the board that actually ships. In DramaSo Studio the storyboard is a table, one row per shot, four columns:

Column	You fill	Why it exists
Description	Timecode, shot size, action, the spoken line	The crew (or the generator) can shoot without a meeting
Reference	The bound character, scene, or prop	Stops a new face appearing in shot 7
Still	One keyframe you can throw away	Motion is expensive; likeness is cheap here
Video	The motion pass of a still you already approved	Last spend

The reference column is the part people skip, then complain that “AI can’t keep a character.” The generator is doing what you asked: inventing a person, because you did not point at one. Bind the lead, bind the room, bind the necklace that matters. Then the still column has something to obey.

The AI storyboard generator is the landing page if you arrived from search. The studio table is where the same job happens on a real episode, after you paste the script.

A paper version of the four columns still works on a cutting mat. Tape the labels. Put polaroids in “still.” Put the phone in “video.” The software is a faster version of that desk.

Description and reference are free. Stills are cheap. Video is the column you unlock last.

Practical rule: Do not generate video for a row that has an empty reference. You are paying the model to guess the cast.

Draw stills cheap, spend on video last

The still is the rehearsal. You are checking three things only: is this the same person, is this the same room, is the emotion readable at phone size. Wardrobe wrinkles, extra extras, a prettier sky — none of that is the still’s job.

A rhythm that does not bankrupt a season:

• Draw stills for every shot on the hook and the cliffhanger.
• Draw stills for coverage you might skip (a cutaway, a hand on a door) only if the episode fails without them.
• Render video for stills you would show a collaborator without apologizing.

This is the opposite of “AI storyboard generator from script” as a one-click film. The search phrase is a tool intent. The page that ranks for it should still tell you to reject. DramaSo’s generator and the studio table both exist so a bad frame dies as a still.

If you came from a premise instead of a script, the AI short drama generator starts earlier in the chain. If you already have pages, stay on paste, then this board.

Decision filter: Would I reshoot this still if it were a photography day? If yes, do not promote it to video.

When AI storyboard generators help — and when they don’t

They help when you already know the shot list. Paste, parse, bind, draw. The generator is a fast pencil for a decision you made.

They do not help when you are using them to discover the story. A collage of pretty 9:16 frames is not a board. It is a Pinterest row. You will not be able to tell a collaborator which frame is shot three, and you will not notice that the lead’s haircut changed.

You have	Use	Skip
A finished episode script	Paste → table → stills	One-click “make a trailer”
A logline and no pages	Write the structure first	Boarding imaginary coverage
A cinematic 16:9 short	Short-film boarding	Forcing 9:16 habits onto a crew day

I would rather board eight ugly, correct close-ups than forty beautiful, uncast wides. The feed never sees the extra forty.

The overseas market does not buy extra coverage either. It buys hook density — the same pressure the board is designed to survive.

Consistency is a boarding problem. If the stills do not match, the video pass will not invent a series bible for you.

From board to a shootable episode

A sequence you can run the same day:

1. Paste the episode and confirm the parse.
2. Split on detonation, reversal, hold. Assign shot sizes. Bind subjects.
3. Fill the table: description and reference first, stills second.
4. Reject stills until the cast is boringly consistent.
5. Render video for the surviving stills. Export 9:16. Do not go back and “fix in generate.”

The board is finished when a stranger could shoot the episode from the rows. If they would have to ask you what happens in panel four, panel four is not done.

Draw the cheap frames. Spend on the ones that survive.

Paste a script and board it in DramaSo Studio

DramaSo Team