LiveLiva world concept art
Real-world discovery
Find a partner through a shared interest and explore a local clue together.
Try a Seek meetup ↗Worlds and vision
World creation concept
Editable worlds and rules
Choose material, define the rules and test a version that other players can visit.
Open the world studio ↗Spatial concept art
Mixed reality and co-presence
Anchor characters in a room or a street so people share the same event. Spatial co-presence is a research direction.
Explore past spatial work ↗
Embodiment research
Persistent identity across devices
Carry identity, relationships and permitted memories across devices. Physical action requires a capable device and its owner’s permission.
Explore identity and runtime ↗LiveLiva Research
Research library (中文) ↗World design and its evidence
What can a companion carry into a different world?
- Same companion
- Check skill compatibility
- Carry a skill forward
Retain identity and the sources of experience. Each world decides which skills it accepts. Bringing a familiar companion back should change at least one choice.
Tolan · Image and source ↗Three cases and the validation method
- Tolan ↗
Shared experiences give the next conversation context.
- Pokémon HOME ↗
Individuals can be stored and transferred. The destination determines compatibility.
- Sony aibo ↗
People recognize their companion through its embodied behaviour over time.
Validation
Enter the same task with a familiar and a new companion. Compare available actions, outcomes and whether players choose the companion again.
What can a remote friend do that matters?
- Shared interest
- Both accept
- Meet in the field
The field player inspects the neighbourhood. A remote partner searches archives. Each acts independently. Sharing a clue or making a promise changes the other person’s available routes.
I’d Hide You · Image and source ↗Three cases and the validation method
- Timeleft ↗
A time, venue and group lower the friction of meeting.
- I’d Hide You ↗
Remote participants have a different view that can influence players in the field.
- Friends & Fables ↗
An AI host and quest rules provide a common setting for improvised action.
Validation
Identify whose contribution changed which action in the replay. Then see whether the same people choose another session.
Who will play the next version?
- Sources and rules
- Playtest and credit
- Publish a remixable work
Pangu turns material, scenes, objects and rules into an editable package. Each playtest targets a version. An adaptation retains its parent, permissions and credits.
LiveLiva world blueprint prototype · Open creator studio ↗Three cases and the validation method
- Astrocade ↗
Moving from prompt to prototype still leaves playtesting, tuning and publishing.
- Roblox ↗
Discovery and invitations bring players. Authors improve the experience based on play.
- Fortnite／UEFN ↗
Cosmetics can travel while each island defines its own rules.
Validation
A second author can publish a rule change independently. New players can feel the difference and both contributions remain traceable.
How can a digital work lead to a real visit?
- Local clue
- Visit and verify delivery
- Reward creators and hosts
Record in-game completion, a physical visit and fulfilment as separate events. Authors maintain route versions. Real services need bookings, capacity, a provider and agreed shares.
CityCipher · Image and source ↗Three cases and the validation method
- Questo ↗
Local authors turn observations and stories into self-guided routes.
- CityCipher ↗
Physical materials, street details and mobile validation deliver one experience.
- Riddle City ↗
Props direct attention to the site. The app supplies hints and tracks progress.
Validation
Compare independent completions, verified fulfilment, maintenance cost and the author’s next release. Cash settlement and repeat purchases require separate validation.
Seven player values in the original research
Companionship, role-playing and autonomy explain attachment. Shared life and local exploration add choices other people and places make possible. Creator economics and portable identity determine what can be maintained, shared or carried forward.
Seven values and their cases ↗Economics: the full cost of one session
Dialogue scales with interaction time, generation with work versions and multi-agent sessions with proposal rounds and retries. Distribution, moderation and author maintenance all count when testing whether repeat participation supports the service.
Inspect workload and cost assumptions ↗LiveLiva Research
Research coverage
196unique research entriesEach bar counts entries in a theme. An entry can appear in several themes.
Open research library (中文) ↗LiveLiva Worlds
Four ways into a world
In the Land design, Liva is for companionship, Town for meeting, World for exploring and Realm for making. The four surfaces form the longer-term vision.
Explore the public prototype ↗- Real-world maps
- Connect a local legend, a familiar shop and real people into an outing. The companion can remember where you went together.
- Generated 3D worlds
- Authors set scenes and rules. Friends and companions play the same version. Later adaptations retain sources and credits.
- Mixed reality
- Bring a digital character to a real table or street so people can participate in the same space.
- Embodied companionship
- Explore how the same Liva might perceive its surroundings and carry out permitted actions through an embodied device.
Works and creators
Play can leave a skill, a keepsake or a new world version. Authors turn local stories into places others can enter. Playtesting and adaptation add contributions. Permission, credits and participation inform how value could be shared.
Creator economy design (中文) ↗How a world runs
Inspect a prototype receipt ↗
Open full diagram ↗Inputs, permissions and outcomes
| Responsibility | Input | Output |
|---|---|---|
| Creation and rightsPangu | Materials, authors and permissions | A versioned, editable world with objects, conditions and rewards. |
| Identity and actionNüwa / A2A | Goals, allowed skills and public proposals | A proposal with the approvals it requires. Each person decides independently. |
| Prediction and executionWorld model / Fuxi | Current version, rules and legal actions | Predictions guide planning. Only validated actions commit state, cost and a receipt. |
| Learning and rollbackNüwa / RSI research | Permitted experience, corrections and outcome differences | Test new tasks and retained skills independently. Reject a candidate or restore a previous version. |
Founder
Huaiyang Liu founded Seengene after studying computer science at Peking University. His work in spatial reconstruction, AR interaction, developer platforms and AI characters informs LiveLiva’s research into people and worlds.
- Arto · Sida · Mira ↗ Spatial reconstruction, positioning and mixed rendering. Arto historically reached over 50,000 developers.
- Artown · Seek ↗ Character memory, behaviour scheduling and AR/VR interaction.
- Seengene ↗ Eight years as founder and CEO, raising close to RMB 100 million, leading a team of around one hundred and reaching tens of millions of RMB in revenue.
These figures refer to the founder’s previous ventures.
Background and projects ↗