Launched this week
Loom Engine & Mythcore: Omniveil
High-performance, No-OOP 2D engine & MMORPG in TypeScript
4 followers
High-performance, No-OOP 2D engine & MMORPG in TypeScript
4 followers
Loom Engine is a high-performance 2D framework built on a "Sadhaka" philosophy: stripping away the OOP tax to find the performant truth in TypeScript. By utilizing Data-Oriented Design and Struct-of-Arrays (SoA), Loom achieves near-native speeds and zero-allocation physics loops. Mythcore: Omniveil is our pre-alpha MMORPG manifestation of this tech, proving browser-based worlds can be dense and fluid. Built for architects who value raw throughput over bloated abstractions.
Hello Product Hunt,
I’m Misha, Lead Architect and Founder of The World Table.ai
For the past 45 days, I have been in a focused engineering sprint to build the Loom Engine. My goal was simple: to prove that TypeScript can handle the demands of a high-performance MMORPG if we are willing to strip away the abstractions that hold it back.
Most browser-based engines are weighed down by traditional Object-Oriented patterns that create massive Garbage Collection pressure. In building Loom, I made the disciplined choice to ditch the class keyword entirely. By moving to a Data-Oriented Design (DOD) from day one, I’ve developed a pre-alpha prototype that achieves near-native throughput.
Loom Engine is the foundation; Mythcore: Omniveil is the manifestation.
As a solo developer, I’m pushing the limits of:
Custom Physics: A solver designed for zero-allocation loops to maintain a locked frame rate.
Graphics: Pushing WebGL/WebGPU batch rendering to its absolute ceiling.
Architecture: Demonstrating that "lightweight" doesn't have to mean "low-performance."
The World Table is 100% solo and family-owned. I am not looking for Venture Capital or to trade equity. Instead, I am looking for supporters - fellow practitioners who believe in high-performance engineering.
We are introducing Mirror Shards (First 500 Mirror shard supporters will receive Twinned Shard to share with a friend) as a way for the community to back this vision directly. By supporting us, you aren't just buying into a game; you are supporting a pure, independent tech stack built with a "sadhaka" mindset.
I’m looking for feedback from fellow engineers and performance junkies. Is the TypeScript ecosystem finally ready to move past OOP for high-performance apps?
I'll be here all day to discuss the architecture and the road ahead.
- Misha
30-hour engineering update.
The launch went out on Loom Engine v1.7.6 - 3,252 tests, Wave 1.7 networking complete (Presence / Lobby / Matchmaking / AuthorityHandoff / LagCompensation / ChatChannel). That's what the original post launched on top of.
In the 30 hours since, the engine jumped from v1.7.6 to v2.0.0:
Trinity Mainframe - 14 new pure-logic kernels
- SonicSync - acoustic ray-tracer with voxel-DDA occlusion
- LoomVerify - anti-cheat verifier (PASS / RESYNC / REJECT) with Merkle witnesses
- NeuralMaterial - PBR synthesis with F16 / F32 path picker
- InferenceOrchestrator - NPC AI lane router (LOCAL_SLM + CLOUD)
- LoomPulse - player-vibe EMA with consent gating
- LoomFlow - 3-lane packet router with throttle hysteresis
- NeuralAnimationSystem - motion-matching + inertialization
- VoxelComputeSystem - marching cubes mesher
- AetherGrid - N2N authority handoff with split-brain detection
- LoomFSR - temporal upscaler
- SealedAssetRegistry - AES-GCM delayed key disclosure
- LoomForgeBridge - WASM-SIMD physics integration
- GlobalStateLedger - Lamport-ordered spatio-temporal persistence
- LoomStudioOrchestrator - AI Director fact governance
732 new tests for the Trinity surface. 3,984 / 3,984 green at v2.0.0. Promoted to `loom-engine@latest` on npm.
Two new live surfaces showing all 14 kernels running in the browser:
theworldtable.ai/engine/ - playground page with a multi-component sustain dial (drives PhysicsSystem + SonicSync + LoomFlow on 10k entities) and 8 click-to-expand component demos. Real kernels, not screenshots.
theworldtable.ai/mainframe/ - the always-on engineering cockpit grew from 9 panels to 15. STRESS.DIAL (10k entities at 60 FPS sustained, p95 16.8 ms), TRINITY.STATUS (14/14 kernels, 61/61 gates resolved at runtime), SONIC.PROPAGATION (drag source/listener on a voxel grid), PACKET.LANES (live 3-lane throttle), VOXEL.MESH (live marching cubes), AUTHORITY.MAP (AetherGrid two-node handoff with split-brain). The middle band also got refreshed: ECS.WORLD scaled from 28 walking dots to 1,000 entities driven by LoomPulse vibe inference, NARRATIVE.MEMORY swapped from hardcoded persona bars to live LoomPulse bars polling the same shared kernel instance ECS.WORLD drives. One kernel, two views.
This is what data-oriented design pays for in practice. Every panel is a real kernel running in the browser - not a mockup, not a video. 60 FPS sustained at 10k entities on a single canvas with SonicSync + LoomFlow + AABB collisions all ticking simultaneously.
For the engineers asking about the OOP question upthread: the SoA pools are why this scales. No instance allocation per entity, no GC pressure under load. The kernels expose flat APIs (`addEntity`, `tick`, `drainEvents`) and the call sites stay in tight integer loops. The `class` keyword isn't banned for any ideological reason - it's banned because every `new Foo()` is a heap allocation that the GC has to chase later. DOD just makes that cost visible and avoidable.
Demos still active:
- theworldtable.ai/mainframe/ - 15-panel always-on cockpit
- theworldtable.ai/engine/ - playground with the stress dial + 8 component demos
- npm: `loom-engine@latest` (currently 2.0.1)
Mirror Shard $79 one-time, 300 Veil Essence per month forever. First 500 buyers get a Twinned Shard to gift.
— Misha
Today's update of the public pre-alpha. 20+ hour Claude Code
sessions every day to push these:
$100K alpha-launch fundraising goal
Public 8-tier milestone ladder, each tied to a real budget item
(1 week of Claude API runway -> 1 month full ops -> first composer
commission -> concept artist -> RTX Pro 6000 + 3D MMORPG transition
-> 3 months focused dev + Co-Creator funded -> full composer
score across 8 biomes -> 1 year solo dev + alpha launch).
Co-Creator program
$1K-$25K backers literally design game content. $1K = mob variant
brief. $25K = Custom Mythic Boss with iterative design calls. The
Loom AI + dev team build out BehaviorTree, telegraphs, drops,
codex entry. Submissions ship in-engine.
Loom Engine v2.1.0
R11 Loom Director + R12 Render Core (Rust+WebGL2) + R13 Combat
Soul + R14 Audio Director + R15 Itemization + BestiaryKernel.
4,032 tests green across Trinity Mainframe.
Full localization parity
EN / TH / RU shipped simultaneously. Locale-aware currency on
TH means every price renders in ฿ (Mirror Shard ฿2,884, Drop
Packs ฿109/฿365/฿912, goal ladder ฿1,825 -> ฿3,650,000).
19 hand-designed SVG Loom-sigils
Custom line-art glyphs replacing every stock icon and emoji.
8 amber goal ladder + 5 purple Co-Creator + 6 cyan engine state.
Built solo, no vibes coding - spec before code, smoke test before
ship, cache-bust ritual on every release. CLAUDE.md as project
DNA enforced via checklist.
Try the landing: theworldtable.ai/founders.html
PWA demo (single button, 8 biomes): theworldtable.ai/arpg-loom/
#PreAlpha #AIMMORPG #ClaudeCode #SoloDev #GameDev #LoomEngine
#Anthropic #ProductHunt #IndieDev #BuildInPublic
TRINITY OF TECH — the three-mind ingestion protocol behind v2.0.0, now driving the gameplay layer.
🧬 Claude — The Mainframe. The Loom Director. The Reality. Every line of code, every commit, every deploy passes through Claude. (Claude Code, Opus 4.7, 1M context plan.)
📐 Gemini — The Blueprint. The Science. Pre-implementation architectural drafts for the deepest systems. Proposes the shape; Claude decides whether to take it. (Gemini Pro 3.1.)
🔍 Codex — The Verification. The Polish. Post-blueprint audit before code lands in Claude's hands. (GPT-Codex.)
Three minds. One Loom. Zero broken pushes, zero rollbacks across 5 phases in 36 hours.