Frequently Asked Questions: Story Videos

How do I create a Story Video?

Use New Video on the dashboard, name it, and pick an aspect ratio and art style. A video created this way starts blank, with no scenes.

You then build the script by talking to Sidekick in the right-hand column: describe the story you want, or paste a script. Sidekick writes the scenes, and you edit them from there.

If you're adapting a comic you've already made, don't start blank — use Turn into Video on the issue instead, which writes the scenes from that issue for you. See turning a comic into a Story Video.

Story Videos are web only — you can't create or edit one in the iOS app.

Is my first video scene free?

No. What every new account gets is 100 free credits and its first two comic renders free (a cover or a page) — no card needed. There is no free video scene: every video scene render is charged per second of footage at the rate of the model you picked, and the studio shows the exact estimate before you confirm. See video credits.

Why was I charged for a scene when I expected it to be free?

The most likely reasons, in order of how often support sees them:

• You re-rendered a scene. Every render is a fresh charge, including re-renders of a scene you'd already rendered. Previous renders are kept in Render History on the scene's SCENE tab, and going back to one is free — see rendering scenes.
• You generated a storyboard frame. Storyboard tiles charge 50 credits per frame immediately on click — there's no confirmation step.

Your balance and per-scene cost estimate are shown before you render. See video credits.

How long does a render take, and do I have to sit and watch it?

A scene takes a few minutes. You don't need to keep the tab open — ComicInk emails you when the batch finishes, roughly a couple of minutes after the last scene lands. One email per batch, not one per scene.

Why did my render fail?

Four causes cover nearly all of them:

• Content moderation. A scene's text is checked before you're charged, so a blocked scene doesn't cost you credits; the message names the scene and the reason. Rewrite the scene's narration and description in plainer language and retry.
• The video model's own safety filter rejected the scene. Separate from our check, and it happens after ours has passed: the video model judges the scene's dialogue, description and reference images together. The credits are refunded. Reword the scene's lines and description, then render again — don't retry it unchanged, and don't replace a character image that rendered your other scenes fine. See the next question.
• The model refused your characters' faces. Every model except the default can refuse reference art that shows a face. The scene is retried automatically on Seedance 2.5, so this rarely reaches you as a failure — see Choosing a model and scene length.
• Too many reference images for the chosen model. Each video model accepts a different maximum number of reference images. A scene carrying more references than its model allows is blocked from rendering — either switch that scene to a model with a higher limit, or reduce the characters, locations, and assets attached to it.

A failed scene shows a red tile: click it for the reason, then Retry or Edit Scene. Failed renders are refunded automatically. See When a scene fails.

My scene passed the content check, so why did the video model still reject it?

There are two checks, and they are not the same one. ComicInk checks a scene's text before you're charged. The video model then applies its own safety filter when it renders, looking at the scene's dialogue, description and reference images together — so a scene can pass ours and still be turned down by the model. When that happens the tile reads "The video model's safety filter rejected this scene" and the credits come back automatically.

It is almost always the wording. If the same character rendered your other scenes, the character image is not the problem — the text of this scene is. Look for a line that insults a group of people, a slur (censoring it as "s-hole" doesn't help), or sexual or graphic phrasing; a character who parodies a real public figure makes the filter stricter. Soften that line and the matching sentence in the scene's description, then render again. Retrying the scene unchanged gets the same answer every time.

The quickest route is to ask Sidekick why the scene failed: it can see the recorded reason, point at the likely line, and rewrite it for you. See When a scene fails.

My video is mostly black, with the scene squeezed into a strip at the bottom. What happened?

The clip had been moved off-centre in the preview, so only its edge was inside the frame. Scenes are now locked in place — they always fill the frame and can't be dragged or resized in the preview — and a scene that was already out of position snaps back by itself the next time you open the video. Close the video and open it again (or refresh the page), and it will fill the frame.

You do not need to re-render: the clip itself was always fine, only its position was wrong, and re-rendering never moved it. If you spent credits re-rendering a scene to try to fix this, contact support through the chat widget and we'll put it right.

Why does the dialogue in my scene sound rushed or sped up?

There is too much speech for the length of the clip. A scene renders as one clip of at most about 15 seconds, and the model fits every spoken word into it — so a long speech comes out fast, clipped, or cut off. Each scene has a spoken-word budget of 19 words, narration and dialogue together, and the editor shows an amber warning when you go over it.

The fix is to split the speech across more scenes, a sentence or two each, rather than to re-render the same scene. Fix Audio won't help here either — it re-voices the lines you already have inside the same clip length, so the pacing problem stays. See editing script and scenes.

Where did the Trim and Speed controls go? Where are a scene's tools?

On the SCENE tab of the Tools rail, on the left. Click any scene on the timeline and the rail switches to that scene's tab — it's named for the scene, "SCENE 3" — with Trim, Speed, Re-Render Scene, Edit Script, AI Extend, Fix Audio, Download Clip and Render History. There is no longer a properties panel on the right for a scene; everything scene-level is on that tab. If you haven't clicked a scene, the tab follows the playhead. See the SCENE tab.

Is my scene checked against the content policy before I'm charged?

Yes. A scene's text is moderated before any credits are taken. If it's blocked, the message names the scene and the reason, nothing leaves your balance, and you can edit the scene and render again. Separately, if the video model refuses the scene because your characters' reference art shows a face, the scene is retried automatically on the default model, and a render that fails is refunded. See Content policy and When a scene fails.

Why is my narration being cut short, or why do I see a word-count warning?

Each scene has a spoken-word budget of 19 words maximum, counting narration plus all dialogue together. Go over it and you'll get an amber warning.

Nothing is silently trimmed — the warning is there so you can cut the line yourself rather than have the render mangle it. Scene duration is derived from word count, so a long line is really a request for a longer scene than the model will produce cleanly. Split it across two scenes instead. See editing script and scenes.

Do I have to make storyboards?

No, storyboards are optional. Scenes without them render normally.

What they're for is control. A storyboard frame is a still image of a scene's opening or closing beat, and when one exists it's handed to the video model as the top-priority reference — ahead of your character, location, and asset references. So a scene that keeps coming out with the wrong composition or the wrong character placement is exactly the case where a storyboard fixes it.

Up to 2 frames per scene (opening and closing), 50 credits per frame, charged the moment you click an empty tile.

What resolution do videos render at?

720p. On web there's no resolution picker — every render is 720p.

1080p is a paid upscale, applied in the Publish modal when you're finished. Upscaling is cached per scene, so re-publishing an already-upscaled video doesn't charge again.

On iOS: the app still shows a resolution picker, but 1080p selections are rendered at 720p until its next App Store release. Use the Publish upscale for HD.

The narration audio is garbled or missing on one scene. Can I fix it without re-rendering?

Yes — use Fix Audio on that scene. It isolates the audio, separates the speakers, and re-synthesises the voices over the existing clip, so you keep the artwork you already paid to render. The script is editable line by line in the modal, each line labelled with the narrator or the character speaking.

See extend and fix audio.

Can I put readable text in a shot — a phone screen, a sign, a letter?

Not reliably, and this is worth planning around rather than fighting. Today's AI video models are poor at rendering legible text inside a picture. Asking for a specific sentence on a phone display will usually give you something that looks like writing but reads as nonsense, or shows different words from the ones you asked for.

If the words matter to your story, say them in dialogue or narration instead of showing them. That is the single most reliable change you can make.

My video is in German (or another non-English language) and the speech sounds wrong

Your script is passed through exactly as you wrote it, and the model is told which language to speak. But speech quality in languages other than English is noticeably weaker: words come out mispronounced or slurred into something that sounds close but wrong, and place names suffer most.

What helps:

• Keep spoken lines short. Long sentences give pronunciation more room to drift. Splitting a 40-word speech across two scenes gives a cleaner result.
• Re-render just that scene. Output varies between runs, so a second attempt at a single scene is often better — and much cheaper than re-rendering everything.
• Use Fix Audio to re-synthesise the voices over a clip whose visuals you're happy with.

This is a limitation of the current generation of video models, not a setting on your account. We pick the best model available and revisit that choice as they improve.

Can I make a rendered scene longer?

Yes — AI Extend adds time to the end of an existing clip. It's expensive relative to a normal render (around 900 credits for a 5-second extend at 720p), so it's worth checking whether re-rendering the scene with more narration would serve you better.

Can I add music?

Yes. Pick one track from the built-in music library — searchable by title, artist, or tag — and set its volume. It loops as needed and is trimmed automatically to the length of your video.

Can I still edit a video after I've exported or published it?

Yes. Finished and exported videos remain fully editable. Re-render a scene, change the captions, swap the music — nothing locks.

Can I reorder scenes?

Not by dragging in the script document — drag-reorder there is disabled. Ask Sidekick to reorder the scenes instead.

Can I turn one of my comics into a video?

Yes. Open the issue and click Turn into Video in the Tools rail, or just ask Sidekick to make a video from the comic.

The script is written for you: the video adapts that issue's pages and panels into scenes as soon as it opens, and carries the characters, assets, and art style across so the video looks like the comic it came from. Adapting is free — only rendering the scenes costs credits. Full details in turning a comic into a Story Video.

Can I make a series of videos?

Yes. Convert a standalone video into a series, then add episodes, choosing which characters, locations, and assets each episode uses. You can also convert a screenplay PDF straight into a series of episodes — see converting a screenplay to a series.

Why do my captions look wrong after I edited them?

Check whether you used Recaption. Recaption is a full overwrite that regenerates every caption from the narration and discards manual edits — it asks you to confirm first, for that reason. To adjust individual lines without losing your work, edit them in Words & timing instead. See captions.

Updated on: 20/09/2026