Infinite Canvas is not a finished product announcement. It is an invitation for the gaming community to build something together: a persistent, cinematic, browser-native world platform expanded through AI-generated video node fields, modular game systems, and community-created stories.
ContinuumOS already exists. Small-scale prototypes have validated node-based video movement, modular web apps, dialogue systems, scene loading, and map traversal. Now we want to open-source the core and gather the people who can help turn it into a shared world.
Instead of starting with a traditional real-time 3D world, Infinite Canvas represents each place as a field of video nodes. A room, building, village, forest, road, dungeon, or city district can contain a sparse or dense node graph. The video layer is only the stage. The goal is a persistent world with movement, interaction, quests, dialogue, combat, guilds, housing, crafting, economies, and creator-built encounters layered on top.
The foundational pieces are not imaginary. ContinuumOS exists today, along with working prototypes for node-based video navigation, scene systems, dialogue systems, maps, modular web apps, and video movement between nodes. These foundations will be open-sourced as part of this effort.
A place is not a single node. A place is a graph of nodes. The distance between those nodes determines how free movement feels: sparse paths for guided traversal, dense grids for near-free movement.
Every node needs at least one movement video arriving from a previous node, plus one idle loop from that fixed perspective: trees shifting, water moving, light changing, a fountain running, smoke drifting, or crowds passing.
Additional directional videos and metadata can be added to a node over time, allowing players to look, move, branch, interact, and open new paths from that place.
New nodes do not only appear at the end of a movement clip. A branch can be created from any point along a transition path, turning one route into many possible directions.
A room can be a graph. A village can be a graph. A forest can be a graph. A world map links them all, and transitions between graphs can be seamless.
Explore what exists. Generate what comes next. Curate the best paths. Increase the density and fidelity of existing places. Add systems, stories, encounters, homes, guild spaces, and economies on top. You do not need to be a developer, designer, or artist to contribute. If you can explore, generate, vote, test, refine, or help improve a place, you can help build the world.
You do not need permission to be useful. A single better clip, a cleaner branch, a stronger idle loop, a tested path, a useful bug report, or a better version of an existing node can all improve the world.
A place is not a single viewpoint. It is a living node field whose fidelity can grow over time: environment, movement, characters, dialogue, quests, combat, crafting, housing, guilds, economy, effects, and personal expression.
A town can begin as a few connected paths. Later, players add side streets, interiors, alleys, courtyards, a market square, housing plots, combat arenas, guild halls, and local stories. The town becomes more real over time, not just larger.
The first version does not need to pretend to be a conventional real-time 3D engine. But movement is not fixed forever. A place with nodes 100 metres apart feels like a guided path. A place with nodes 2 metres apart feels explorable. A place with nodes 10 centimetres apart can feel close to free movement. The community does not only expand the world's boundaries — it increases the fidelity of existing places.
Characters can run, turn, strafe, jump, climb, emote, interact, and fight as composited video animation over the environment plate.
A fence can know it is climbable. A stream can trigger splash overlays. A doorway can link graphs. A boss gate can load an encounter state.
For water, impacts, boss attacks, cutscenes, special traversal, or dramatic story beats, characters can be rendered directly into the scene or replaced into authored clips.
Use defaults, download approved custom packs, sponsor cloud generation, or use local open-source video models to create personal replacements and map them to nodes.
The first task is proving the shared world loop at community scale. Around that, we want to build the systems that make the world feel alive: places to belong, reasons to travel, things to make, conflicts to survive, and stories worth returning to.
The base operating layer already exists as ContinuumOS, a browser-native architecture for building modular systems as separate web apps with shared state and message passing. As part of Infinite Canvas, the goal is to open-source ContinuumOS and the relevant prototype apps so the community can build on them directly.
Rendering, maps, dialogue, inventory, combat, quests, chat, guilds, housing, crafting, governance, marketplaces, and creator tools can exist as separate web apps communicating through shared state and message passing.
This means contributors do not need to understand the entire engine to improve one part of the world. A team can build a quest editor while another improves character packs, while another experiments with combat, while another creates cartography tools.
The result is faster than a traditional monolithic game stack, easier to test, easier to extend, and natural for community development.
Custom characters are fully supported. They are not meshes requiring a real-time 3D pipeline. They are downloadable media packs: approved animation clips plus metadata.
The world boundary expands through generated paths, but the creator economy should grow through everything layered on top: quests, stories, NPCs, encounters, dungeons, guild halls, housing templates, character packs, emotes, combat events, crafting systems, markets, music, and world events.
Explorers can be credited for opening new paths. Quest designers can be rewarded when players complete their stories. Character artists can sell skins and animation packs. Guilds can sponsor zones, upgrades, events, raids, and cinematic encounters.
The governance model should be community-first: reputation, attribution, transparent contribution records, and shared decision-making. The exact mechanism can evolve, but the principle matters most: the world should belong to the people who help build it.
ContinuumOS exists. Small-scale versions of the underlying approach already work. Node-based video movement has been validated. Several prototype systems are ready to be opened and extended. The larger persistent world still needs builders, testers, designers, writers, artists, AI video experimenters, systems thinkers, and community organisers — but it also needs players. Anyone in the gaming community can meaningfully contribute by exploring, generating clips, improving fidelity, voting on better paths, testing places, reporting problems, and helping decide what the world becomes.