“Output Was Cut Off” or “Invalid JSON”: When a Parse or a Storyboard Stops Halfway

Quick answer: Parsing and storyboards are written by a text model that has a fixed budget for how much it can write in one go. A very long script, or an episode dense with dialogue, can run past that budget and the answer arrives incomplete. The site detects this and runs the same job once more. If the second answer is also incomplete, the task is marked failed and its credits stay held: the model was called, so support checks it and then refunds or settles it; it is not refunded automatically. Before running again: for a parse, trim the script or split it into separate projects. In SceneMixer, an incomplete answer is detected and the same job is run once more automatically.
Still from “The Veil at Dawn”, an AI short drama generated end-to-end by SceneMixer (6 characters, 1 episode)
SampleThe Veil at DawnDigital HD1 episodeProduced end-to-end by SceneMixer from the scriptWatch the episodes

After a failed parse or storyboard, open your credit history: that task’s line says Under review until support settles it.

What 'cut off' means

The model writes a structured answer — the cast list and episodes for a parse, the segments and shots for a storyboard. When it reaches its output limit mid-answer, the structure is incomplete and cannot be used; the message reads cut off, truncated or invalid JSON. Nothing partial is kept; your script and any earlier storyboard stay as they were. What each task status means: Task statuses.

The SceneMixer episode editor: asset library on the left, the storyboard in the middle, the video preview on the right, and every segment along the bottom timeline
A cut-off parse or storyboard is marked failed; the script and the episode are untouched and can be run again.

What the site does first

SceneMixer retries every cut-off answer once automatically with the same input; many second attempts complete. For a storyboard, the better of the two drafts is saved if it is usable at all. Only when neither attempt gives a usable answer does the task fail, and then the held credits wait for review instead of coming back on the spot. Refund rules.

A parse that keeps cutting off

Long scripts are read in two passes precisely to avoid this; the threshold is automatic. A parse that still cuts off is usually a script with an unusually large cast and many locations. Trim it, or split it into two projects. Long Scripts Are Read in Two Passes and the caps.

A storyboard that keeps cutting off

An episode is written as segments and shots; an episode whose share of the script is very long, or wall-to-wall dialogue, produces more shots than one answer can hold. An episode cannot be split or its outline edited on SceneMixer, so generate its storyboard once more (a later run often completes); if the same episode keeps failing, write to support with the project and episode. In your next project, keep each episode’s share of the script moderate. Storyboard Checks: Two Drafts.

Rarer relatives

Field type mismatch: the model wrote a number where a word belonged; the site repairs what it can and retries the rest. Wrong language: the model ignored the project language; retried once with a firmer instruction. All of these end the same way — a second attempt, and if that also fails, a failed task whose credits are held for review — and none of them changes what you already have.

TaskUsual cause of a cut-offWhat to do
ParseA script with a very large cast and many locationsTrim it or split it into two projects
StoryboardA very long or dialogue-dense episodeGenerate once more; write to support if it keeps failing
AI CreateA target length past the parse capPick a shorter target

FAQ

Were my credits refunded?

Not automatically. After a second failed attempt SceneMixer keeps the credits held; support reviews the task and refunds or settles it.

Do I lose my script?

No. The script, the summary and any existing storyboard are untouched.

Will running it again just fail again?

A parse with nothing changed often does; trim or split first. A storyboard often completes on a later run.

Why does a longer script cost the same but fail?

Price is by length; the model's output budget is a separate limit, which is why long scripts read in two passes.

Can I keep the half that was written?

No. A cut-off answer is structurally incomplete and is discarded whole.

Shorter input, complete output

Trim or split the script; the second run usually completes.

Open the project

Updated November 15, 2026