Cover
01 / 12
↓ Click · Scroll · Arrow keys

Project
Thalassa

Not designing a toolextracting stories from audiences
PROJECT THALASSA / Beijing Moxun Technology
01 / LISTEN
Audience Story Extraction
02 / SHAPE
Narrative Motif Development
03 / HANDOFF
Storytelling Package Delivery
02

From tool, to living system

Story IP is no longer one-off content. It becomes an asset that keeps growing.

PAST

Product IP

Design it, launch it, exhaust it, then start again.

  • Design
  • Launch
  • Exhaust
  • Rebuild
Emotion
Symbol
Story
Asset
Resonance
Motif
Role
Sample
Living Story IPDATA CULTURE
THALASSA

Species-like IP

A symbolic seed keeps evolving, growing roles, worlds and repeatable stories.

  • Symbol seed
  • Living system
  • Feedback evolution
  • Repeat output
InputAudience data
EngineSymbolic evolution
AssetWorld library
OutputContent production
03

Three walls of current AIGC

CurrentAIGC
now
Problem 01

Emotion wall

Labels do not explain character conflict.

  • Tags only
  • Hidden motive missed
Problem 02

Sameness wall

Aligned models drift toward the average.

  • Too smooth
  • No destiny pressure
Problem 03

Tool wall

Prompt amplifies input, but cannot grow new motifs.

  • Human spark dependent
  • Weak asset memory
Out of reachSoul
new motifs
EmotionLabels miss conflict
ExpressionCorrect but soulless
ProductionCeiling stays input
ResultNew motifs unstable
04

Automated production from audience emotion

InputCrowd
corpus
▸
Step one01

Observe collect

Read desire, anxiety, resonance and motifs.

  • Crowd emotion
  • Cultural motifs
  • Time symptoms
▸
Step two02

Emotion-symbol coupling

Let emotion, symbol and world bind into a story core.

  • Emotional chemistry
  • Symbol reaction
  • Story motif
▸
Step three03

Narrative output

Output roles, scripts, boards and design direction.

  • Character setup
  • Script boards
  • Design interface
▸
OutputProduction
story
↺Delivery calibration · project feedback keeps returning to the system
Crowd emotionReal resonance and hidden conflict
Symbol reactionCoupling in emotion and world libraries
Structure generationProduction-ready narrative skeleton
Delivery calibrationProject feedback sharpens the system
05

Narrative Petri Dish

Audience signals → story motifs
Collection End

Theme + Psychoanalysis

Audience emotion / conflict / motifs.

Evolution End

Chemistry + VSA

Symbols bind, drift, recombine.

Output End

VSM + Logic

Roles, plots, boards, design.

InputAudience · data · motifs
ComputeChemistry · VSA · bonding
HandoffBible · scripts · boards
06

Turn audience emotion into cross-scene storytelling assets

Automated sample output
Affective-field formulaAffective field formula
Audience signalsSocial sentimentCase corpusWorld libraryAnxiety clusterTrauma trace Story motifCharacter arcStoryboard pathVideo sampleScene constraintEmotion trace AffectSymbolbondingGrammarStability TraumaDesireRecallFantasy themeContextEmotion arc
01 · Recall

Read audience pressure

Public or licensed signals become affect vectors.

02 · Bind

Bind emotion and symbol

Motifs, roles and tensions form story reactions.

03 · Output

Produce story assets

Return motifs, character arcs, boards and samples.

RecallAffectMotifRoleVSA/HDCVSMSample
07

AI series: one mother body, many arcs

Automated sample output
Scene-generation formulaScene generator formula
Jiangnan scene moodboard
Storyboard-trajectory formulaShot generator formula

Mother world

Rules, places and history become the shared base.

Scene constraints

Each scene recalls the right place and pressure.

Episode path

Boards follow emotion, rhythm and world logic.

08

Feature animation: make the characters deep

Automated sample output
Micro-action generation formulaMotion generator formula
Cartoon character storyboard grid
World-projection formulaWorld design formula

Inner pressure

Desire and hesitation set the character state.

Micro-action

Emotion becomes pause, turn, touch and glance.

World projection

Props, colour and space carry the feeling.

09

Same automation. We start from audience emotion

Automated ↑Human-drivenCapacity · promptAudience emotion →
ThalassaAudience emotion engine
Lingjing AIAI comics capacity
ShowrunnerAI series platform
Traditional prepManual IP service
AI comics capacityLingjing AI
DoesBoards / images / rendering
StrengthIndustrial speed · cost · efficiency
Thalassaalso automated, but based on audience emotion
AI series platformShowrunner
DoesPrompt to episode
StrengthVirality · interaction
Thalassaless user prompt, more real audience pressure
Traditional IP prepManual creative
DoesWorld / role / insight
StrengthExperience · taste · trust
Thalassaturns audience insight into automation at scale
Everyone is automating production. The difference is what the automation grows from: audience emotion · emotion library · world library · feedback loop
Same automation, but ours grows from audience emotion.
10

A compact early team.

Founder
Dai Shang

PhD, Tsinghua University.

Research
Chunling Wu

PhD, University of London.

FLOW

Turn the emotion engine into a commercial base

Direction one

Data expansion

Expand crowd emotion, motifs, worldbuilding and commercial samples.

  • Crowd emotion
  • Cultural motifs
  • Scene samples
Direction two

Algorithm calibration

Calibrate VSA, VSM and emotional chemistry for sharper output.

  • Symbolic computing
  • Emotion bonding
  • Narrative generation
Direction three

B-side samples

Deliver tourism, film, brand and game samples; turn feedback into assets.

  • Project delivery
  • Client feedback
  • Asset reuse
Run intoCommercial
base
Crowd dataMore emotional samples
Vector librariesWorlds, roles, narratives
Algorithm testsBetter generation quality
Project feedbackDelivery returns to the base
ASK

Thicken the base. Tune the algorithm

Seed · open£2Mpre-money
Need one01

Data purchase + cleaning

Buy emotion corpora, industry cases and world texts; clean and label them.

  • Emotion corpus
  • Industry cases
  • World texts
Need two02

Database build

Build world, audience-emotion and industry-case libraries.

  • World vector library
  • Audience emotion library
  • Industry case library
Need three03

Algorithm + business growth

Improve symbolic computing, emotion bonding and narrative generation; expand B-side directions.

  • Tourism / film
  • Brand / games
  • Education / consumer
GoalThicken base
tune algorithm