Creator media, cinematics, and preview

Last verified: 2026-10-04

Story, character, and location assets

Upload only media you own or are authorized to use. Story assets serve the Story draft;

independent Characters own their images in the top-level Character library, while location

asset sections associate eligible Story items with their locations. Choose

representative images that match the published canon and add useful alternative text or metadata

where the editor offers it. Assets retain type, ownership, review, and usage information. Rejected

media cannot become a representative image or published story asset, and deleting a referenced

asset requires removing or replacing its uses first.

Character chat animation

In a character's editor, open Chat animation to arrange image frames into resting,

waiting, and reply sequences. Each frame can select one part of a sprite sheet. Preview

the whole movement, then add ordered branches for emotion, intent, intensity, relationship,

and the character's previous posture. Replies finish in a resting sequence so the character

keeps an appropriate expression between messages. Save the animation, approve its images

and publish the Character's presentation from its editor. Attached Stories use that approved

presentation in character chat without requiring another whole-Story publication.

For a generated transparent character sheet, review the actual poses and choose **Keep for

animation** to approve it as a reusable source. This preserves the character's profile

picture. Review identity, clothes, hands, framing, and transitions before using the source;

approval alone does not establish animation quality. Readers can turn motion off, and their

device's reduced-motion preference is respected.

Cinematic voice profiles

Cinematic voice profiles assign one of Mystrix's reviewed voices, rendered on Mystrix's own

voice box, to characters and the narrator. A character's selected logical voice is persisted and reused throughout the

Story, so episode order and later publication do not silently recast that character. Preview a

profile before relying on it. Provider selection and tuning controls such as speed, pitch, warmth,

intensity, accent, and emotion are not Creator controls: Mystrix applies one fixed consistency

recipe. Voice presentation does not change dialogue text or narrative state. Exact recordings can

be attached to a selected locked script line through the separate performance workflow; this does

not create a custom voice model. Generated clips are cached for playback and matching speech

recipes rather than synthesized again on every listen.

Cinematic music library

The music library uses eligible audio assets and their mood tags, loop and normalization settings,

transitions, and location assignments. Preview a track and check where it is used before changing

or deleting it. Ducking, fades, cuts, swells, and stingers shape presentation but do not create a

story event, choice, or state effect.

Generate a sound effect

In a Story's sound-effect picker or asset manager, choose Generate sound effect, describe the

sound you want, and pick a duration of 10 seconds or less. Mystrix generates one short effect on

its own voice box and saves it to the Story's ordinary asset library, where you can place it on a

cinematic beat and preview it like any other sound. Your description stays on Mystrix's own

servers and is never sent to a third party. The box renders one job at a time, so a sound can

take a few minutes when it is busy. Generation is limited per day, and a generated sound is never

added to Public sounds unless you explicitly share it as described below.

Public sound effects

In a Story's sound-effect picker or asset manager, choose Public sounds to search, preview,

and add an effect another creator has shared. The effect is copied into your Story's ordinary

asset library; you can then place it on the intended cinematic beat. Using a public sound does

not share any of your own sounds.

To contribute an eligible effect you added yourself, uploaded or generated, open that sound in the asset inspector and choose

Share with creators. Review its public title, then accept the reuse statement and share it.

Sounds are never added to Public sounds by upload, generation, metadata edits, or Story publication

alone. Remove from pool stops new discovery and use; copies already added to other Stories

remain there. Do not include private information in a public title or sound. Report a shared

sound from its catalogue row if it may violate the content rules. If sharing, adding, previewing,

removing, or reporting a sound is refused, the message explains the reason in your language: the

sound is unavailable, it changed or is busy (reload and try again), its public title or reuse

statement needs review, or it cannot be shared or reused.

Cinematic scene frames

Scene frames use image assets as location defaults, layered backgrounds, or triggered visuals.

Choose a frame type, duration, transition, and explicit trigger such as event start, beat reached,

choice selected, flag present, relationship threshold, or story time. Keep layers and triggers

consistent with visual canon, and verify that the underlying story condition is actually reachable.

Cinematic events

Cinematic events coordinate camera intent, characters, locations, dialogue direction, visual

assets, and sound around an authored story moment. They can stage a beat or consequence, but they

must not replace a required reader decision or create a contradictory source of canon. Review the

event in story order and keep its linked media eligible for publication.

Beats and speaking order

Open Beats directly from the Story sidebar. Select the event and beat, then arrange the

character and narrator lines in their speaking order. Narration is used only where explicitly

assigned. Use Visual editing to place visual changes against their intended spoken line;

the local preview stays available. Save commits the script and visual composition together.

Image generation and AI Quick Fix use the current beat; review their result before saving.

Beat-level full-frame video

A locked beat can use one finite full-frame MP4 instead of separate voice and layered Stage

choreography. In the ordinary Timeline beat card, choose an approved ready Story video or upload an

MP4, preview it, provide its language and visual description, set the start and end time of each spoken line, name the rights holder, choose your

authority, and confirm the rights statement before selecting Upload and attach video. Reloading the

Timeline shows the same sealed clip identity and media facts.

The clip must be a 16:9 H.264 MP4 with supported AAC program audio (or be explicitly silent). A

beat cannot keep both a video and a separate voice performance, and it cannot bind a video while it

still owns Stage cues. Remove video unbinds the performance without deleting the reusable Story

asset or changing the beat's locked speaking sequence. In Reader playback the clip owns the full visual

plane and its finite clock; buffering pauses progress, the clip never loops, and Retry or explicit

Skip remain available after media failure.

Reviewed Story visual inventory

Upload complete scene images in Mystrix Studio or, where the image controls are enabled, explicitly

request generation or refinement with an OpenAI, Gemini, or Z.ai key entered in the open editor, or

use the Mystrix image service. Z.ai GLM-Image works from the written brief alone and produces full

scenes only, not transparent artwork; its quality check still compares against your references. The key is needed again after the session ends. Review a private

candidate before applying its finished frame to the intended event.

Use Load more in the image source picker to browse the remaining eligible images. Attached

Characters contribute approved images; borrowed Characters contribute only images in their

accepted publication while the grant permits use. Unavailable Character images stay hidden

without hiding later eligible Story images.

The Smagik house account can choose Included house images in the same model menu

or select that same model through the Creator API with `funding: house`. This built-in

connection is exclusive to Smagik. It supports full scenes and transparent source

artwork, with a separate quality check and no cash charge. Independent jobs run

concurrently as worker resources allow. No house-image concurrency setting or two-job

limit is required; API-provider concurrency and request-rate allowances do not cap

this connection. Provider subscription limits still apply.

House-funded images are not subject to paid-account image request-count caps.

They are available for explicit generation and sequences.

A service restart can recover the same completed house-image result without

generating it again. An uncertain outcome remains explicit; review its status

before starting another request.

Publication may also apply an already reviewed house-owned scene image when its exact role and

Story-event scope match.

Where enabled, the beat's compact Image control can turn its saved description into a

sequence from beginning to end. Extra direction, frame count, speed and looping are optional;

the default plays once and holds the ending. Review the full sequence before using it.

Character, location and World editors also have one saved art prompt. Approved examples guide

later generation. Character movement sequences can be saved and selected in another beat.

Generated consistency is checked, not guaranteed; unresolved results stop for review.

Each player receives a bounded immutable artwork snapshot. A session that already started is not

silently switched to a later upload or selection. Uploaded images and reduced video poster frames

may be sent to DeepSeek only for safety verification; DeepSeek returns safety scores and does not

generate or edit the image. That upload-verification input is not sent to OpenAI, Gemini, or Z.ai; an

explicit generation request sends only its prepared brief and creator-selected references.

Visual inventory is presentation only: it cannot create story events, choices, state changes,

overlays, hotspots, or reader interaction targets. Creator-authored cinematic triggers and

interactive overlays remain separate authoring decisions.

Episode preview

Open Preview to screen the complete saved draft episode before publishing. Press Play once;

the player advances through the authored beats and events automatically. Continue buttons, choices,

and other input never block the screening. Conditional beats are included in authored order so you

can review their material too. This screening does not simulate branch consequences.

Use the timeline to seek anywhere in the episode, Restart to return to the beginning, the audio

control to mute or unmute, and the fullscreen control to expand the same player. These controls work

on desktop, phones and tablets, including phone landscape. Browsers without the native fullscreen

API expand the player to the available screen. Captions use the prepared speech or video timing.

The player displays the integrated authored visuals and plays the selected voice, music, ambience

and effects. It does not display separate portrait cards or debugging panels.

Preparation happens inside the player. If media cannot be prepared or loaded, an inline error and

Retry replace playback. Changes to the saved draft during preparation require a fresh load.

An empty episode is identified explicitly. Preview never silently skips an unfinished event.

An episode Preview URL screens that episode; the Story-level Preview link opens the first episode

in authored season/episode order. Back to Beats opens the first event of the screened episode

once it is loaded. Save changes in the editor before opening Preview. Screening may prepare speech

through the normal preview media path, but does not change scripts, Story state or publication.

Documentary and instructional Stories retain their dedicated Review workspaces. Native interactive

preview sessions retain their existing API; this page is the complete episode screening surface.