How vizcrush processes your data
From raw input to optimized output — the complete pipeline
YOUR DATA
2,000,000
Float64Array points
API · CSV · WebSocket
init()
Auto-detect backend
COMPUTE ENGINE
WASM
Compiled Rust
JS
Fallback
selected
VIZCRUSH ALGORITHM LAYER
Downsample
LTTB · MinMax · M4
Binning
2D · 3D · Hex
Spatial
Quad · Morton · Grid
Streaming
HLL · KLL · Sketch
Transform
Sort · Norm · Log
AI
Anomaly · NLP
MCP
26 tools
CHART OUTPUT
1,920 points
at 60fps — zero visual loss
D3 · Three.js · Plotly · Canvas · ECharts
DATA INTELLIGENCE
Anomalies detected
Trend: increasing ↗
Auto-config: LTTB
Stats · Percentiles · Summaries
AI AGENT ACCESS (MCP)
vizcrush_lttb(x, y, 1920)
vizcrush_detect_anomalies(data)
vizcrush_summarize(data)
Claude Code · Cursor · VS Code · stdio · HTTP
Zoom/Pan → re-crush from full data
9 Rust crates
11 npm packages
23 MCP tools
435 tests
8-27KB WASM gzipped
MIT licensed