Hundreds of visual effects creators build remarkable, one-of-a-kind work with AI. But each effect lives and dies on a standalone HTML page with a few sliders. It never reaches the people who need it most – the designers, the agencies, the content creators who work in timelines and keyframes every day. They don’t write shader math. They need production-ready effects.
The demand exists. The willingness to pay exists. The bridge doesn’t. So the work gets lost.
Can't stack 3 effects on different layers with blend modes and opacity.
Can't animate parameters over time. No keyframes, no easing, no seek.
Can't make effects react to music. No FFT, no CQT, no audio bridge.
Can't produce an MP4 at 4K/60fps. Canvas recording is not video production.
Fragment shaders can't do particle physics, fluid dynamics, or flocking. That needs infrastructure no individual can build.
A two-sided market with dedicated AI for each role. Creators build the supply. Consumers generate the demand. Neither needs to understand the other’s domain.
For shader artists, VFX developers, creative programmers
Write WGSL shader math directly or let AI help you get there. Describe the effect, iterate together, preview in real time. The AI focuses on visual logic – the infrastructure handles GPU pipelines, audio bridge, export, and everything else.
For designers, agencies, content creators
Knows the entire effect catalog. Helps you find, combine, and configure effects for your project. Ask for suggestions, get parameter recommendations, iterate together. A knowledgeable creative assistant, not a magic button.
Think of the AI as a mathematician locked in a room. The room has fixed walls (render engine), a fixed window (export), and a fixed door (audio input). The AI can only write equations on a whiteboard – but those equations determine everything the user sees.
let dist = length(uv - 0.5); let wave = sin( dist * 20.0 - params.time * params.speed ); let pulse = 1.0 + params.bass * params.intensity; let ring = smoothstep( 0.02, 0.0, abs(wave) / (dist * 10.0) ) * pulse; return params.color * ring;
Bind group layouts Pipeline creation & caching Vertex shader (fullscreen quad) Uniform buffer alignment (16-byte) Texture management & MSAA resolve Stencil masking for path geometry Per-layer GPU state isolation Audio data injection (bass/mid/high) Parameter UI auto-generation Export frame capture pipeline Double-buffered GPU readback Watermark overlay MP4/WebM/GIF/PNG encoding
GPU buffers are managed by the engine. AI code has no access to device.createBuffer(). Structurally impossible.
WGSL is sandboxed by the browser’s GPU process. Hardware-enforced memory isolation.
Export captures whatever render produces. It doesn’t know if the effect is built-in or AI-generated.
Effect code runs in the render pipeline. The blockchain gate sits before export. Different layers entirely.
Shadow, fill, stroke, effects, precompose, post-processors, blend modes. Effects don’t just stack – they interleave at precisely the right point in the visual hierarchy.
Built & testedFull keyframe system with easing curves, layer timing, time remap, parent-child hierarchy, and per-property audio modulation. Production-grade motion control.
Built & testedImport any track – every effect instantly reacts to bass, mids, highs, and full spectrum. No setup, no wiring. Audio analysis is built into the engine.
Built & testedEvery frame is a pure function of (time, params, audio). Merkle-tree node cache with 80–90% GPU reduction. Cacheable, parallelizable, reproducible.
Built & testedBuilt-in effects, community creations, and Stock additions – all parametric, GPU-accelerated, and composable. The library grows with every creator who joins.
Built & growingType text – the engine breaks it into animatable atoms: individual characters, words, lines, Bezier paths, stroke segments. You get the building blocks. Creators and AI build the animations.
InfrastructureCurves, levels, hue/saturation, color balance, LUTs, color wheels, split toning, gradient maps. A complete color grading pipeline as post-processors.
InfrastructureEvery use of your work triggers a blockchain payment – automatically, verifiably, without exception. No one can use your effect without paying. Secured by smart contract, not by promise.
By designPipelined MP4 at 4K/60fps with zero-alloc GPU readback. Batch spritesheet capture. Not a canvas recording – real video production.
Built & testedThe same effects power both channels. The same payment flows to creators regardless of how their work is delivered.
PNG, MP4, WebM, GIF at up to 4K/60fps. Pipelined GPU readback with batch spritesheet capture. Real video production.
Embed real-time GPU animations on any website. A lightweight player runs directly in the visitor’s browser – no server rendering, no video files, just live visuals.
The second pillar. Particle physics, fluid dynamics, cloth simulation – things a fragment shader can’t do. WASM + SIMD, off-thread, deterministic. Same timeline, same audio bridge, same export pipeline.
GPU fragment shader – math on pixels
Plasma, fire, glow, ripple – each pixel computes independently. 5–30 lines of WGSL. Runs on GPU at full speed.
WASM Worker + SIMD – physics on entities
Particles, flocking, fluid, cloth – thousands of entities interacting. Rust/AS compiled to WASM. Off-thread, deterministic.
Simple simulations run for free in your browser. Need more power? Upgrade to cloud compute with Sanki – cost estimated upfront, no surprises. Your choice, your pace.
Built from scratch. Measured, not claimed.
No framework, no engine — every subsystem is
custom-built, on top of ajv, earcut, mediabunny, opentype.js, webgpu.
A full store with categories, trending charts, creator profiles, and live previews. Browse,
discover, compose – everything works with a watermark.
Burn
Sanki
to remove it.
Create an effect and publish it to Stock. Categories, tags, live preview – your work is discoverable from day one.
Browse by category, trending, or curated collections. Preview any effect live. Compose scenes with as many effects as you want – free with watermark.
At export, you pay per effect in your scene. 5 effects = 5 payments. Creators are paid automatically. The platform takes no commission – its fee is burned.
Pre-composed scenes built by creators using other creators’ effects. Pick a template, customize it, export. One purchase pays every creator involved – effects + template author + burn.
Built-in benchmarking tools for effect creators. GPU time, frame rate impact, resolution scaling, memory footprint – everything you need to craft professional, high-performance effects.
Prices are set by the platform per product type – image, video, live embed. Creators don’t negotiate. Every six months, the community votes on price adjustments. No one raises prices unilaterally.
Pricing, Payments & Creator ProtectionEnpixelment isn’t open source – but its direction is open to everyone. We’re building the engine. The community decides where it goes.
See something missing? Suggest platform capabilities – new layer types, export formats, audio analysis modes, animation tools. If it makes the engine better, we want to hear it.
The roadmap isn’t decided behind closed doors. The community surfaces what matters most – and that’s what gets built next. Real influence, not a suggestion box.
Not just effects – the platform itself. New parameter types, composition modes, simulation algorithms, delivery channels. Your ideas become infrastructure that every creator benefits from.
The flywheel is ready to spin. Be among the first to shape what comes next.
Sign in