AI Story Production Studio
One story in.
One finished film out.
Other tools give you an eight-second clip. Magenica breaks your script into scenes, shots, characters and locations — then produces a coherent 10+ minutes video with the same faces, places and voices from first frame to last.
Up to 60 min runtime
Persistent character bible
Multi-engine orchestration
Clip-level tools
You prompt 90 times, get 90 strangers, and stitch it yourself.
Same prompt: “Israel, the goldsmith, at his bench” · 4 separate clips
shot 12
face A
shot 27
face B
shot 54
face C
shot 71
Director of Louvre
DRIFTthe face changes between shots — no shared identity
DRIFTlighting and grade reset with every generation
DRIFTwardrobe, room and props invent themselves again
DRIFTa new voice per clip, and no timeline to cut on
Magenica — project level
One project: story, scenes, shots, cast, voices, score — assembled and exportable.
One locked project · four linked layers
Story3 acts
I
II
III
26:10
Scenes6 · 108 shots
Characters5 locked
Voices1 + 4 cast
LOCKEDevery layer references the same bible, so shot 3 and shot 101 match
Live walkthrough
From a 4,000-word story to a 26-minute documentary
Follow a real project — “The Greatest Deception in the History of the Louvre”, a documentary short — through all eight production stages. Click any stage, or press play.
magenica / projects / the-louvre-tiara
{{ stageTag }}
DOCX
the-louvre-tiara.docx
4,052 words · uploaded
✓ parsed
THE GREATEST DECEPTION IN THE HISTORY OF THE LOUVRE
What if one of the most famous museums in the world once spent a fortune on a royal treasure — only to discover it had been made a few years earlier in a workshop in Odessa?
Not a thief breaking into the Louvre. Two antique dealers from the Black Sea coast who understood one dangerous truth: museums do not only buy objects, they buy stories...
Paste text
Upload DOCX / PDF
Import article URL
One-line concept
Genre
Documentary
Est. runtime
26:10
Acts
3
Narration
Narrator-led
Structure detected
ACT I
Europe's gold fever and the ruins of Olbia
5:20
ACT II
The commission, the bench, the invented biography
8:05
ACT III
Vienna hesitates, Paris hurries, April 1 1896
5:15
5 characters
6 locations
7 key objects
tone: wry, forensic
Estimated generation cost
Model pricing for the detected 26:10 runtime — compare engines before you render.
Resolution
Video modelPer finished minEst. projectCharacter & speed
{{ m.name }}
{{ m.perMin }}
{{ m.total }}
{{ m.note }}
includes +20% retake budget
{{ runtimeNote }}
billed separately from your plan
Prices are indicative. Generation is metered per shot and charged separately from the subscription plan; the final invoice depends on the engines, resolution and number of regenerations you actually run.
SceneBeatLocationCastShots
01
Europe buys the ancient world
Olbia ruins
—
14
02
Two dealers in Ochakov
Gokhman shop
Leiba, Schapsel
22
03
Seven months at the Odessa bench
Odessa workshop
Roukhomovsky
18
04
A real decree, an invented crown
Study, Olbia
Leiba
19
05
Vienna hesitates, Paris hurries
Vienna salon
Louvre experts
17
06
April 1, 1896 — and the maker's hands
Louvre, Paris
Roukhomovsky
18
6 scenes · 108 shots · every row editable
+ add scene
Character bible · locked references
Israel Roukhomovsky
Goldsmith, Odessa · 34
4 looks
voice #2
Leiba Gokhman
Dealer, the scholar · 40s
3 looks
voice #5
Schapsel Gokhman
Dealer, the closer · 30s
5 looks
voice #1
Location bible
odessa workshop
olbia ruins
vienna salon
louvre gallery · 1896
Storyboard · scene 05 · 17 shots
consistency lock: on
05-01
WIDE · Vienna salon, vitrine open
05-02
CU · the Greek inscription, letter by letter
05-03
INSERT · Austrian press: “may leave Austria”
05-04
MED · French experts around the table
05-05
TRACK · crates carried into the Louvre
05-06
INSERT · receipt, 200,000 gold francs
05-07
CU · label dated 1 April 1896
+ 10 more
regenerate any single shot
Render queue · 108 shots
est. 58 min remaining
Key frames · image model108 / 108 done
Motion · image-to-video81 / 108 rendering
Narration · TTS, voice #126:04 generated
Score & sound designqueued · 8 cues
Every stage can be routed to a different engine. Swap the video model for one scene without touching the rest of the project.
Assembly · timeline 00:00 → 26:10
VIDEO
NARRATION
DIALOGUE
SCORE
SUBTITLES
Auto-cut to narration timing
Scene transitions matched to act breaks
Loudness normalised to −14 LUFS
final cut / the-louvre-tiara / 26:10 / 4K
▶
09:04 / 26:10
Preview limit reached
Free projects export the first 90 seconds. Unlock the full 26:10 cut.
Watermark-free · 4K · commercial rights
Runtime
26:10
Shots
108
Cast continuity
5 / 5 locked
Total time
4 h 05 m
What this export costs
Actual metered generation for this final cut, by engine and resolution.
Resolution
Video modelPer finished minEst. projectCharacter & speed
{{ m.name }}
{{ m.perMin }}
{{ m.total }}
{{ m.note }}
includes +20% retake budget
{{ runtimeNote }}
billed separately from your plan
Prices are indicative. Generation is metered per shot and charged separately from the subscription plan; the final invoice depends on the engines, resolution and number of regenerations you actually run.
The hard part
Shot 3 and shot 101 have to be the same person
Magenica builds a persistent reference for every character, location, costume and recurring object — with controlled variants for age, season, wardrobe and emotional state. The reference travels with the project, so the sixteenth scene matches the first.
roukhomovsky / 1895 / bench
Variants, not rerolls
Ages, wardrobe, weather and mood as controlled looks of one locked identity.
odessa workshop / lamplight
Locations that stay put
The same room, from the same angles, across every scene set there.
the tiara / prop reference
Props and motifs
Recurring objects keep their design so visual callbacks actually land.
Orchestration layer
Not tied to one model. Ever.
Magenica sits between you and the engines. Route each production stage to whichever model wins on quality, price or speed today — and re-route tomorrow without rebuilding the project.
01Regenerate a single shot, voice track or scene — the rest stays untouched.
02Per-stage engine presets saved as house style across your channel.
03Cost estimate per scene before a single frame renders.
Story LLM
Image models
Image → video
Text → video
Voice / TTS
Lip-sync
Music & SFX
Compositing
Built for
Faceless YouTube channels
Documentary
History
True crime
Horror & mystery
Sci-fi
Animated & children's stories
Educational series
Publishers & studios
Agencies at scale
Subscription
Pay for finished minutes, not prompts
Every plan includes the full production pipeline — story analysis, bibles, storyboard, assembly and export.
Creator
One channel, one story at a time.
${{ priceCreator }}/ month
{{ billingNote }}
30 finished minutes / month
Projects up to 10 min runtime
1080p export, no watermark
Character & location bible · 10 refs
Standard render queue
Most popular
Studio
Weekly long-form, full length.
${{ priceStudio }}/ month
{{ billingNote }}
120 finished minutes / month
Projects up to 60 min runtime
4K export · commercial rights
Unlimited references & variants
Voice cloning · 6 narrator slots
Priority queue · 3 seats
Scale
Studios and agencies in volume.
${{ priceScale }}/ month
{{ billingNote }}
600 finished minutes / month
Unlimited project runtime
Custom engine routing & API
Dedicated render capacity
Brand style presets · SSO
10 seats · solutions engineer
Free tier: one project, 90-second export, watermarked. Cancel any time.
Unlock full export
{{ planName }} plan
7 days free, then ${{ planPrice }} / month. Cancel before day 7 and you pay nothing.
In your subscription
Full-length export, watermark-free
4K H.264 / ProRes · subtitles
Unlimited character & location refs
Voice cloning and narrator slots
Priority render queue
Commercial usage rights
Today$0.00
{{ renewNote }}${{ planPrice }}
Subscription active — full 26:10 export unlocked.