AI Video Generation Options: Every Setting, When You Choose It and Whether It Can Change

Quick answer: Video generation here has options at five moments, and the useful thing to know about each is when you choose it, whether you can change it later and whether it touches the price. Step 1 holds the options that lock when the script is read: type, frame, region, dialogue language and three style lists. Step 3 sets the segment length per episode. Each generation takes a tier, which is the only option that changes the price of a second. Characters, scenes and props have their own options, and a finished clip can be trimmed, repainted, redubbed or sharpened. This page is a reference: one table per moment, with the choices, the lock and the price effect.
Still from “Rocket Pup”, an AI short drama generated end-to-end by SceneMixer (7 characters, 1 episode)
Sample: “Rocket Pup” — Family 3D Animation, 1 episode, produced end-to-end by SceneMixer from the script · watch the episodes

Chosen in step 1: they lock when the script is read

These are chosen before anything is generated. Most of them shape what the read writes into the cast and episode outlines, which is why they cannot be changed afterwards in the same project; for another choice, start a new project with the same script. What can and cannot change after the read.

OptionChoicesChangeable later?Changes the price?
TypeShort Drama (episodes up to 240 seconds), TV Series (up to 300 seconds), Film (no cap)No. The episodes were cut for itNo
Aspect Ratio16:9 Landscape (default, marked Recommended) or 9:16 PortraitNo, once you continue past step 1No. Video is priced by the second, not by the frame
Target RegionEleven choices: ten regions plus Follow scriptNo. Edit descriptions and names in step 2 insteadNot priced separately
Dialogue languageOne of 15; the lines are written in it from the startNo. Edit lines in segments, or start againNo. A segment costs the same in any language
Visual Style21 presets in three groups (live-action, 3D animation, 2D animation); members can write their ownNo. It shapes every sheet and clipNot priced separately
Scene Style74 presets in 7 groups; members can write their ownNot after the read. Edit scene descriptions in step 2 and regenerateNot priced separately
Costume Style90 presets in 9 groups; members can write their ownNot after the read. Edit character descriptions in step 2 and regenerateNot priced separately
Let AI chooseIf styles are left unpicked, AI can fill the rest from the scriptYes, before you continueNo

The preset lists are described one by one in visual style presets and scene and costume styles. The scene and costume cards switch to 3D or 2D pictures when the visual style is 3D or 2D.

Chosen in step 3: per episode

OptionChoicesChangeable later?Changes the price?
Segment lengthStandard 15s (default) or Long take 30sPer episode. Re-parsing re-cuts the episode and discards its clipsNot per second; a long take costs twice a standard segment to redo
Add or delete an episodeAdd an episode at the bottom of step 3; Delete Episode on its cardYesAdding reads only the new pages; deleting removes the episode's clips

Long take 30s can only be rendered by the tiers whose cap is 30 seconds; a segment above its tier's cap is refused before anything is charged. The shortest a segment can be is 4 seconds. 15-second or 30-second segments.

Chosen on every generation: the tiers

A tier is the one option that moves the price. Each kind of generation has its own: a video tier per segment, an image tier for sheets and covers, a text tier for reading scripts and writing storyboards.

KindWhere you pick itHow many tiersWhat changes
VideoThe model button in the editor, per segment; the default is also in SettingsSeveral, from 480P up to 2K, priced per secondResolution, longest segment and price per second. Each segment keeps the tier it was made on; tiers can be mixed in one episode, though the join then re-encodes
ImageSettings, Image ModelSeveral; the default is Qwen-Image 3.0 at 7 credits an imageHow new sheets and covers look and what each costs; existing ones stay
TextSettings, Text ModelSeveral, each with a rate per 1,000 charactersThe price of reading and of storyboards, and how carefully a storyboard is structured

Changing a tier never regenerates anything by itself. The price of a second per tier is in the price list, and which model to pick by purpose says what each family is for. A tier a plan does not include shows Available after upgrade.

On characters, scenes and props

OptionChoicesChangeable later?Changes the price?
Voice, per characterMatch a preset from the built-in library (free, playable), write a voice description, or upload your own (a plan feature)Yes. It applies to shots you generate afterwardsMatching is free; the shots that use it are priced as usual
Looks, per assetA state image for a changed outfit or scene, within caps: up to three for a character, two for a location, one for a propYesA state image is an image, with its own small free allowance on the first project
Own image as a sheetUpload your own (a plan feature)YesUploads count against your plan's storage quota
Reuse from another projectPick existing, in step 2Yes, with a one-step restoreNo, on any plan

Switches: music, captions, badge, interface language

SwitchDefaultWho has itWhat it does
MusicOnEveryonePlays and downloads the episode's score; off gives the dialogue version without music. No track picker
SubtitlesOffAccounts that have bought credits or a plan at least onceBurned-in download for a clip, and for a whole episode an SRT too. Free, with a daily limit
AI badgeOn for free accounts, off for membersMembers can switch it either wayA small mark in the bottom-right corner of new clips; it does not change clips you already have
Interface languageYour browser'sEveryoneOne of sixteen. It changes how text is shown to you, not what the cast says

After a clip exists: fixes and versions

A finished segment can be corrected without making a new take. Every fix is priced on its button before you press it and is refunded if it fails.

FixWhat it changesPrice
TrimCuts the start or the end5 credits each
AI visualsChanges one thing in the picture: clothes, a scar, the weather2 the tier's per-second price, per second of clip
Redub: lip syncRe-times the mouth to a line1 credit a second
Redub: re-performHas the character perform the line againSame as AI visuals
EnhanceRaises resolution one step on a clip you already havePer second, by option; on the card
RegenerateA new take of the segmentThe tier's per-second price, per second

Every new version is kept in Version History, so you can switch back. What each clip fix costs and enhancing a clip have the details.

Other project kinds have their own options

An ad project picks its length (15, 30 or 60 seconds), aspect ratio and resolution in the ad composer, per ad. A remake reads an uploaded clip of up to 5 minutes, fixes its dialogue language when the clip is read and has its read priced by the minute. Both end in the same editor and the same fixes as a story project. Four ways to start a project compares them, and ad type, length, aspect and resolution lists the ad options.

FAQ

Which options cannot be changed after the script is read?

Type, frame, target region, dialogue language and the visual, scene and costume styles. They shape what the read writes. For another choice, start a new project with the same script.

Which option changes the price?

The tier you generate on, and the seconds you generate. Frame, type, language and style lists are not priced separately.

Can I change the video tier for one segment?

Yes. The tier is chosen per segment with the model button in the editor, and each segment keeps the tier it was made on.

Can I mix 15-second and 30-second segments?

An episode is cut at one length, 15 or 30 seconds, but you can split a segment by hand afterwards.

How many styles are there?

21 visual styles, 74 scene styles in 7 groups and 90 costume styles in 9 groups, counted from the Step 1 pickers on November 2, 2026, plus members' own text for each.

How many dialogue languages are there?

15. The language is chosen in step 1 and the lines are written in it from the start; the interface language is a separate setting.

Is there a setting for music?

Music is automatic. A switch turns it off for playback and download, but there is no track picker or volume control.

What changes if I join a plan?

Monthly credits, more generations at once, your own uploads and your own style text, and the badge switch. The price per second of a tier does not depend on the plan.

Choose the options once, before the read

Open step 1 with a short script, set the frame, type, language and look, and read the summary before you spend on pictures.

Start free

Updated November 2, 2026