From script to an edited animated scene without leaving the Premiere timeline. Every voice and every lip-sync shot runs on the ElevenLabs API.
EdMon Producer AB · StockholmDavid Nordén, Co-founderAsk: additional video credits
Live buildRendered anime shots cut in a Premiere sequence
Sc.00 · Sh.01● REC · edmon.ai
Where we are
The audio grant covers us. The video grant doesn’t.
Audio credits granted
33M
Enough for character voices, voice design and dialogue takes across full test productions.
Video credits granted
1M
Spent proving the pipeline: a complete test story, animated shot by shot with Aurora lip-sync.
We’re asking for
+3M
Additional video credits to finish testing Anime Studio end to end before launch.
Why video runs out fast:characters×framings×shots per scene×retakes×languages=video renders
Test storyTillbaka till början
9scenes
108shots, all animated
7frame takes on a single shot
Sc.01 · Sh.02Grant status
What Anime Studio does
Six steps from script to timeline, all inside one Premiere panel
01
Script
Paste a script, start from a template, or let AI write the whole story.
Claude
02
World
Characters in three views plus turnaround. Empty location plates. Props.
ClaudeImage model
03
Voices
A voice per character, designed for them or picked from a library.
11L voices
04
Storyboard
Every shot typed, framed and drawn. Weak frames are redrawn one by one.
Image model
05
Animate
Aurora lip-sync animates each frame to its spoken line.
11L TTS11L Aurora
06
Timeline
“Place in Premiere” lays clips and synced dialogue into the sequence.
Premiere
ElevenLabs APIClaude · image model (Nano Banana)Premiere Pro
Sc.02 · Sh.03Pipeline overview
Step 01 · Start a story
Four ways in, from a blank page to a finished script
Write a Script: paste or write a script and the studio builds the storyboard.
AI Screenplay: pick themes and Claude writes the whole story.
Templates: ready stories such as Cyberpunk Courier and Ghost Channel 13.
Continue a Story: versioned projects with art style and approval status.
PanelStart screen, docked in Premiere Pro
Sc.03 · Sh.04Start
Step 01 · Script & storyboard
The script is the storyboard
Claude breaks the script into shots. Each shot gets a type, an angle, a location and its cast.
Click any name, place or shot type in the script to change it.
Every dialogue line belongs to a character, with an acting direction for the voice.
When a shot changes, its clip is flagged as out of date so nothing slips through.
PanelScript with shot tags beside the drawn frames
Sc.04 · Sh.05Script & storyboard
Step 02 · Characters
Each character is drawn once, then kept on model in every shot
Three views (full body, waist up, close-up) and a front, side and back turnaround, all written from the script. Edit the words and redraw.
CharactersCast with reference views and assigned voices
SheetTurnaround used as the drawing reference
Sc.05 · Sh.06Characters
Step 02 · Locations & props
The same room looks like the same room
Every location is drawn empty and reused as the background plate for its shots, with lighting and atmosphere set per scene. Props the camera sees more than once get their own reference.
LocationsDescription, atmosphere and key features per place
PropsRecurring objects drawn once and reused
Sc.06 · Sh.07Locations · Props
Step 03 · Voices, powered by ElevenLabs
Every character gets a voice of their own
Brand new voice
Designed for the character from their description. Pro tier. Our test cast already uses designed voices.
ElevenLabs voices
Searchable with previews, including the user’s own library.
Studio’s choice
The studio matches a voice to the character automatically.
Every dialogue line is recorded in 3 takes, like on set: one as directed and two variations (softer, slower, and so on). The editor picks the best one.
If no take works, regenerate with creative direction: write a new note, for example “whispered, holding back tears”, and record 3 new takes.
PanelVoice picker: designed voice “Tillbaka Sixten (design C)”
Sc.07 · Sh.08Voice casting
Step 04–05 · Storyboard to video
Every frame is animated to its spoken line
Aurora lip-sync (ElevenLabs) animates each frame to its line of dialogue.
Shots are handled one at a time: redraw the frame, tweak it with a note, or animate it again.
Frame takes are kept, so the editor can compare take 5 with take 7.
Animate one shot, a whole scene, or the entire story in one click.
PanelScene 1 · 108 shots · 108 animated
Sc.08 · Sh.09Aurora lip-sync
Step 05 · Full Dialogue Mode
Each character’s lines render as one continuous take
A scripted scene is shot one camera setup at a time, on one actor. We render the same way. All of a character’s lines in one framing become a single lip-sync video, with handles after each line for listening shots.
Diagram · hatched = handle after each line · vertical marks = line boundaries
Sc.09 · Sh.10Full Dialogue Mode
Step 05 · Coverage
Wide, medium and close-up of the same performance
Each character’s dialogue is rendered in three framings, stacked and in sync on the timeline.
The editor cuts from medium to close-up without ordering a new render and waiting for it.
This is the coverage editors expect from live action, now for animation.
Cost: coverage triples the video renders for every dialogue shot. It’s the biggest reason we need more video credits.
00:0000:0400:08
V3
SIXTEN · CU
V2
SIXTEN · MS
V1
SIXTEN · WS
A1
Diagram · one voice take drives all three framings
Sc.10 · Sh.11WS · MS · CU
Step 06 · In the Premiere timeline
The storyboard becomes an editable sequence
Test sequence“CL test – dialogue takes”: rendered dialogue stacked on V1–V2, synced dialogue audio on A1–A2
Rendered shots arrive as separate clips with synced dialogue audio.
The editor keeps working as usual: trimming, adding listening shots from the handles, then music and sound design.
No export and re-import. Place in Premiere writes straight into the open project, organised in bins per scene.
Sc.11 · Sh.12Timeline integration
Why usage grows fast
One line of script is never generated once
Voice
1dialogue line
×
3takes to choose from
×
2rounds of new creative direction
×
4languagesPlanned
=
24voice generations
Video
1dialogue line
×
3framings: WS · MS · CU
×
4languages, re-lip-syncedPlanned
=
12lip-sync renders, before any retake
Takes come first: the user picks the best of 3 voice takes before any video is rendered, so every line starts as several audio generations.
Coverage multiplies video: each chosen take is animated in three framings for the editor to cut between.
Localization, automated (planned): switch the story to Japanese, German or French and the same storyboard is re-voiced and re-lip-synced. A localized version of the film becomes one button.
Sc.12 · Sh.13Example per line · usage multipliers
What this means for ElevenLabs
Every Premiere editor who uses Anime Studio becomes an ElevenLabs API user
Bring your own key
Users connect their own ElevenLabs account and pay ElevenLabs directly for usage. EdMon charges only for the panel.
Usage multiplies per line
3 takes, 3 framings and every language version on top. One line can mean 24 voice generations and 12 lip-sync renders.
Where editors already work
Around 30M Premiere Pro users. Distribution through Adobe Exchange and edmon.ai.
Sc.13 · Sh.14Partnership value
The ask
+3Mvideo credits
To finish testing Anime Studio before launch. The existing 33M audio credits cover the voice side.
Developer mode: during testing every render runs on the EdMon account. That is where the grant goes.
Test area
What we need to prove
Full Dialogue Mode
Long multi-line lip-sync takes hold sync and stay on model.
Coverage
WS, MS and CU of the same take cut together cleanly.
Art styles
Ligne Claire, Manga Noir and the templates all animate well.
Complete stories
Several full productions from script to cut, as real users will run them.
Takes & retakes
3 voice takes per line and real retake rates, so pricing and guidance are accurate.
Localization
The same storyboard re-voiced and re-lip-synced in a second language.
Sc.14 · Sh.15Request
Thank you
Let’s finish the picture.
We’ll share test renders, a case study and the launch build with the ElevenLabs team as we go.
