OpenLiDARViewer GUIDE // v0.6.2
Launch viewer
RENDER PATHPROBING DATA ROUTELOCAL-FIRST SECTION01 // OVERVIEW
LOCAL-FIRST LIDAR INTELLIGENCE // OLV-06.2

From raw point clouds
to spatial evidence.

ACTIVE MODEVISUALIZEPOINT-CLOUD WORKSPACE

Open drone LiDAR, terrestrial laser scans, mobile captures and streaming datasets directly in a browser tab. Inspect the source, measure geometry, assess terrain and export results with their coverage, provenance and limits attached — while local files remain on your device.

OPEN LOCALLYINSPECT PRECISELYEXPORT HONESTLY
MIT open sourceWebGPU + WebGL 2Local-firstNo install
OLV // LIVE WORKSPACE PRIVATE: ON DEVICE
OpenLiDARViewer displaying a colorized LiDAR scan with measurement and Scan Intelligence panels
RENDERPROBING
POINTS2.4M / 9.6M
FRAMELOCAL ENU
IMPORT PATHS1412 static + 2 streaming paths
NAVIGATION4Orbit · Walk · Fly · Pan
MEASURE MODES97 core + 2 volume workflows
DATA ROUTE0uploads for local files
02 // CURRENT CAPABILITIES

What OpenLiDARViewer does today.

A focused point-cloud workspace designed around the tasks that happen before a full GIS or survey-processing workflow: open, understand, inspect, measure, communicate.

01

Open locally

Drag supported scans into the browser. Header preflight, bounded decoding and memory guards keep large files from becoming a blind all-or-nothing load.

NO UPLOAD
02

See structure

WebGPU rendering with automatic WebGL 2 fallback, EDL depth shading, SSAO, soft splats, adaptive point sizing and multiple color modes.

GPU ACCELERATED
03

Navigate naturally

Orbit around objects, walk through interiors, fly across terrain or grab the scene with Pan. Smart camera presets reframe the scan instantly.

GAME-LIKE CAMERA
04

Measure in 3D

Distance, polyline, area, height, angle, slope and profile tools with editable points, metric/imperial units and session persistence.

SPATIAL TOOLS
05

Inspect attributes

Click or probe points to read coordinates, color, intensity, classification, return data, GPS time and CRS-derived location where available.

POINT-LEVEL DATA
06

Analyse terrain

Ground classification, confidence-aware DTM, DSM, CHM, slope, hillshade, quality scoring and evidence-gated contour deliverables.

HONESTY ENVELOPE
07

Document findings

Annotations, named camera views, inspection sessions, PNG evidence captures, multi-page PDF reports and workflow recordings.

REPORT + SESSION
08

Stream scale

COPC and EPT load progressively by view, octree node and memory budget. The resident set sharpens as you move instead of loading the full cloud.

RANGE REQUESTS
03 // OPERATIONAL FLOW

From raw scan to communicable result.

The interface keeps the first move simple, then reveals deeper tools only when the scan and its evidence support them.

01
INGEST

Open the source

Drop a file, choose one from disk, paste a COPC/EPT URL or use a public catalog entry.

02
ORIENT

Frame the scan

Use the suggested navigation mode, smart views, perspective or orthographic projection.

03
INTERROGATE

Inspect and measure

Probe point attributes, place measurements, annotate findings and save viewpoints.

04
COMPUTE

Analyse terrain

Run ground, surface and confidence stages only when terrain products are needed.

05
DELIVER

Export with context

Save images, sessions, reports, vectors, rasters or contour packages with provenance.

01Input remains immutable

The source geometry is not silently rewritten after load.

02Coverage is disclosed

Full, resident-only or sampled scope travels with terrain results.

03Unknown stays unknown

Missing signals render as an em dash instead of a fabricated score.

04 // SCAN INTELLIGENCE

Description is not the same as assessment.

OpenLiDARViewer separates what the file contains, what can be inferred cheaply, and what a terrain analysis actually supports. Missing evidence stays missing instead of becoming a confident-looking number.

SCAN INTELLIGENCE // LIVE PANELSOURCE-BOUND
OpenLiDARViewer Scan Intelligence panel showing scan metrics and dataset information
01SCAN REPORT
02DATASET CARD
03TERRAIN ASSESSMENT
LAYER 01 // FILE-DERIVED

Scan Report

Reports point count, width, depth, height, density, spacing and detected attributes. When the source records them, it also exposes capture sensor, source software, capture date and georeferenced bounds.

  • Descriptive, not a quality certification
  • Uses decoded source content
  • Unavailable attributes remain “Not available”
LAYER 02 // FAST INDICATORS

Dataset Intelligence

A cheap, informational card derived from headers, bounding-box volume, optional resident-neighbour density and an optional terrain suggestion.

  • Point Density · Terrain Complexity
  • Ground Visibility · Streaming Coverage
  • Terrain Confidence — shown as “—” when no signal exists
iIt does not perform ground classification and does not manufacture a bucket from absent data.
LAYER 03 // ANALYSIS-DERIVED

Terrain Assessment

Appears after analysis and reports two independent axes: Surface Quality asks whether the internal surface is usable; Export Readiness asks whether its reference frame is complete enough for hand-off.

Surface QualityExport Readiness
GEOMETRYExtent + densityWidth, depth, height, spacing and point count.
ATTRIBUTESWhat each point carriesRGB, intensity, classification, returns, time and normals when present.
INTEGRITYAdvanced diagnosticsInvalid coordinates, duplicates, outliers and declared-vs-decoded count.
STREAMINGCoverage is explicitFull, sampled or resident-only — with a caveat when more nodes may arrive.
05 // VISUALS STUDIO

Make point-cloud structure legible.

Color modes, RGB appearance presets, sky environments, point styles and depth cues are concentrated in one visual workspace.

COLOR MODESRGBHeightIntensityClassNormalsHillshade
POINT STYLEClassicSoft splatsInspectionAdaptiveFixed
DEPTH + SKYEDLSSAOStudio DarkBlueprintTerrain
06 // SPATIAL INTERACTION

Measure the scan, not a screenshot.

Tools attach to 3D positions in the cloud, stay editable during the session and can be exported with the rest of the inspection state.

01

Distance

Straight-line length between two picked points.

02

Polyline

Total path length through any number of vertices.

03

Area

True planar area plus horizontal map-projected area.

04

Height

Vertical difference using the scan's natural up axis.

05

Angle

Vertex angle between two arms in 3D.

06

Slope

Rise, run, angle and grade percentage.

07

Profile

Cross-section geometry with a resizable elevation chart.

+

Volume

Cut/fill against a polygon or 3D lasso, with validity guards.

CALIBRATION NOTICE

Measurement is designed for inspection and research. It is not a survey-grade certification unless the complete workflow has been independently validated for that use.

07 // TERRAIN INTELLIGENCE

Analysis that carries its own limits.

Every stage travels inside a shared honesty envelope: coverage mode, source and analysed point counts, confidence, warnings and export readiness.

TERRAIN COMPUTE GRAPHCONFIDENCE-AWARE // EVIDENCE-GATED
STAGE 01Ground filterSMRF core + source classes
STAGE 02DTM rasterMeasured cells stay distinct
STAGE 03ConfidenceCoverage + interpolation distance
STAGE 04ValidationHold-out RMSE + calibration
STAGE 05ProductsDSM · CHM · slope · aspect · hillshade
TERRAIN ASSESSMENT2 AXES
SURFACE QUALITYGOOD
Is the internal surface usable?84
EXPORT READINESSPREVIEW
Is the reference frame ready for hand-off?62

Illustrative state: a clean surface can still remain Preview when CRS or vertical datum evidence is incomplete.

OUTPUT MATRIXPERMIT RESOLVER
  • DTM / DSM / CHM packageASCII GRID + GEOTIFF
  • Analytical contoursGEOJSON + DXF + SVG
  • Printable map sheetMULTI-PAGE PDF
  • Terrain intelligence reportPROVENANCE STAMPED
  • Complete deliverableZIP PACKAGE
CONTOUR STUDIO

Purpose changes presentation, never evidence.

Choose Engineering Plan, Survey Review, Terrain Research, Presentation Map or Custom. The preset can alter defaults, but it cannot elevate a scientific claim.

Contours E3 Slope E4 Aspect E4 Hillshade E4
E4 validates these algorithms on one analytic fixture. E5 field validation, survey certification and the full point-cloud-to-DTM pipeline are not claimed.
CONTOUR EXPORT PROTOCOL

From analysed surface to evidence-aware deliverable.

ONE GATE // ALL OUTPUTS
  1. 01
    Load and verify the frameConfirm projected CRS, horizontal units, vertical units and datum. A geographic CRS or unresolved vertical units caps metric contour support.
  2. 02
    Run Terrain AnalyseThe pipeline classifies ground, builds the DTM, records coverage and confidence, and derives surface products.
  3. 03
    Read both assessment axesSurface Quality and Export Readiness answer different questions; a good surface may still remain Preview when reference evidence is incomplete.
  4. 04
    Open Terrain ProductsThe launcher becomes hidden, unavailable, exploratory or available from the analysis result and reference frame.
  5. 05
    Choose purpose + review settingsEngineering Plan, Survey Review, Terrain Research, Presentation Map or Custom changes presentation defaults only — never the evidence grade.
  6. 06
    Export through the permit resolverAnalytical contours, map sheet, DEM package, report and complete ZIP receive the same decision and provenance stamp.
READINESS GATE // INTERACTIVE MODELTap states to test the logic
EXPORT READINESSREADYSurface is Good and the reference frame is complete.
Analytical contoursGEOJSON · DXF · SVGAVAILABLE · STAMPED
Map sheetPDFAVAILABLE · STAMPED
DEM packageDTM · DSM · CHMAVAILABLE · STAMPED
Complete deliverableZIP + MANIFESTAVAILABLE · STAMPED

Educational simplification of readiness only. The application also checks product-specific evidence and export permissions. Preview artifacts are watermarked; blocked products return a diagnostic instead of a polished deliverable.

REFERENCE RULE

Analytical contours are exact isolines of the grid. Cartographic contours may be generalized for legibility, reference the analytical geometry hash and are never labelled exact. Unsupported or interpolated spans stay visibly qualified.

08 // DATA COMPATIBILITY

One viewer, many acquisition paths.

Filter the matrix by source type. “Streaming” means progressive, bounded-memory loading rather than a full-file read.

LASDrone / aerial LiDARGeoreferenced · coordinate bridge
LAZCompressed LASlaz-perf WASM decode
E57Laser scanner exchangeMulti-scan · RGB · intensity
PTXScanner text formatPer-scan pose applied
PTSScanner point text3 / 4 / 6 / 7 columns
PCDPoint Cloud LibraryASCII · binary · compressed
PLYPoint clouds / meshesRGB supported
OBJMesh scansVertices used as points
GLBSelf-contained glTFMobile and AR workflows
GLTFEmbedded-buffer glTFExternal buffer needs packaging
XYZWhitespace point textChunked decode · optional RGB
CSVDelimited point textChunked decode · optional RGB
COPCCloud Optimized Point CloudLocal / remote octree streaming
EPTEntwine Point TileBinary + laszip tiles
RECOMMENDED FOR SCALECOPC / EPTProgressive view-dependent streaming
RECOMMENDED FOR SHARINGPLY / GLBPortable lightweight workflows
PLANNED, NOT SHIPPED3D TILES / PNTSParser foundations exist; no user-facing open path yet
09 // TECHNICAL ARCHITECTURE

A local-first pipeline with clear boundaries.

Format readers produce a shared point-cloud model. Rendering, measurement, analysis and export consume that model without turning the viewer into a server-dependent application.

INPUT LAYERLAS / LAZE57 / PTX / PTSPLY / GLTFCOPC / EPT
CORE MODELPointCloud + coordinate bridgeImmutable source geometryLocal render originAttribute channels
↙     ↓     ↘
RENDERThree.jsWebGPU / WebGL 2EDL / SSAO / splats
INTERACTSpatial toolsInspect / measure / annotateCamera + session state
ANALYSETerrain coreGround / DTM / derivativesValidation + evidence gate
OUTPUT LAYERImagesSessionsLAS / PLY / OBJ / XYZ / CSVPDF / vectors / rasters / ZIP
01

Browser runtime

TypeScript and Three.js deliver the application through Vite. WebGPU is preferred; WebGL 2 is the compatibility path.

02

Worker + WASM paths

LAZ/COPC decoding uses laz-perf WASM, while large and streaming jobs move work away from the main interaction loop.

03

Layering rule

I/O owns formats, the viewer owns three.js state, and analysis consumes the point-cloud model instead of importing the renderer.

10 // SCIENTIFIC TESTS + ARCHIVE

Tests that can fail. Claims that stop at the evidence.

Version 0.6.2 adds runnable validation, defect records and replay tooling, cross-implementation checks, and reproducibility evidence. The numbers below describe the versioned source package — not a blanket claim of field accuracy.

VALIDATION SUITES ADDED12Reproducibility, scaling, portability, units, backend equivalence, failure recovery, provenance, contours, LAS round-trip, archive portability, seed sensitivity and clean-clone.
DEFECTS FIXED1812 exposed by new suites; 6 found through code review. Four reached statements or outputs published in v0.6.1.
CROSS-IMPLEMENTED PRODUCTS3Slope, aspect and hillshade are at E4 on one frozen analytic DEM; other terrain products top out at E3.
SURVIVING TARGETED MUTATIONS4Four of twelve mutations still survive the full gate. They are published as real validation holes, not hidden as confidence.
GDAL 3.13.1 // FROZEN ANALYTIC DEME4 BOUNDARY
SLOPE11,564 interior cells0.5° preregistered tolerancemax Δ < 0.001°
ASPECT10,932 cells, slope > 2°0.5° circular tolerancemax separation ≈ 0.0002°
HILLSHADE11,564 interior cellsmax 1.00 level vs GDAL0.0000643 vs closed form

i These checks validate the algorithms on this fixture, not the complete point-cloud-to-DTM workflow and not real-world field accuracy. Hillshade also carries a disclosed byte-encoding caveat; its claim relies on the closed-form comparison and exact re-encoding identity, not the GDAL leg alone.

SECOND ARCHITECTUREZERO-TOLERANCE COMPARISON
15artifact hashes+18science scalars

One seeded synthetic fixture produced identical science-scoped output on darwin-arm64 and linux-x64, with tolerance exactly zero. Windows, big-endian hosts and real scans remain outside that result.

A previous gate still passed 6,182 tests with 16 skipped while all 18 corrected defects remained present. Test volume alone is not evidence of completeness.
ARCHIVE SNAPSHOT SCOPE // ABSENCE IS NOT A PASS

The bundled validation snapshot records 11 of 15 collection inputs. Replay results, the generated GDAL cross-check artifact, clean-clone evidence and the aggregate release-gate record are explicitly marked not executed in that snapshot. The E4 statements above follow the versioned claim register and validation report.

Inspect snapshot
RESEARCH ARCHIVE // CITE THE SOFTWARE

Versioned source, DOI resolution and reproducibility documents.

This guide tracks source version 0.6.2. Its citation metadata lists the OpenLiDARViewer concept DOI and a version DOI explicitly described as the archived v0.6.1 release. The guide preserves that distinction and does not claim a version-specific Zenodo deposit for v0.6.2.

11 // VALIDATION + LIMITS

Trust is a feature, not a footnote.

Version 0.6.2 is a validation-and-correction release. The public language distinguishes tested behavior, internal evidence and work that remains unvalidated.

CURRENT RELEASE0.6.22026-07-27
12validation suites added
18defects fixed
E4slope · aspect · hillshade
MITopen-source license

What it claims

  • Local files are processed on the device.
  • COPC and EPT stream progressively.
  • Coverage and confidence accompany terrain results.
  • Slope, aspect and hillshade have E4 cross-implementation evidence against GDAL plus closed-form truth on one frozen analytic DEM.
!

What it does not claim

  • Survey-grade accuracy or standards certification.
  • Cross-CRS reprojection between scans.
  • Full-cloud completeness for sampled or resident-only results.
  • Field validation or E5 evidence.

Operational limits

  • Local scale depends on browser memory and GPU.
  • Wide extents reduce Float32 in-memory precision.
  • WebGPU feature support varies by browser.
  • Remote streaming sources are user-selected network activity.
CLAIM RESOLVER // EXAMPLEOLV-EVIDENCE-01
01source.coverage = "resident-only" 02terrain.surfaceQuality = "Good" 03reference.verticalDatum = null 04export.readiness = "Preview" 05artifact.watermark = true
RESULTProduct remains useful, but it cannot present itself as a fully referenced deliverable.
12 // QUICK START

Your first scan in five moves.

No GIS background is required. Start with viewing and inspection; run analysis only when your question needs it.

01

Launch

Open the viewer in a modern browser with hardware acceleration enabled.

Open viewer ↗
02

Drop a scan

Drag a supported file onto the window or paste a remote COPC / EPT URL.

03

Choose movement

Orbit for exterior inspection, Walk for upright traversal, Fly for terrain and large sites.

04

Read the data

Switch color mode, inspect points and review Scan Intelligence before measuring.

05

Save context

Export a snapshot, report or session so findings, views and measurements stay together.

FIELD CONTROLSPRESS ? IN THE APP FOR ALL SHORTCUTS
1Orbit
2Walk
3Fly
4Pan
W A S DMove
FFocus point
RRe-frame
GHand tool
MMeasure
IInspect
AAnnotate
⌘ KCommand palette
OpenLiDARViewer symbol
READY // LOCAL-FIRST WORKSPACE

Open the scan. Keep the data.

Use OpenLiDARViewer for fast visual understanding, spatial inspection and evidence-aware browser research.

SYSTEM FAQ

Common questions.

Are local scans uploaded to a server?+

No. Local files are read and rendered in the browser. Network access occurs only for remote sources that you explicitly choose to open and for normal site assets.

Does the viewer require WebGPU?+

No. WebGPU is the primary rendering path where available, with an automatic WebGL 2 fallback for compatible browsers.

What should I use for very large datasets?+

COPC or EPT. They are the true progressive streaming paths and keep a view-dependent resident set instead of loading the complete cloud.

Can the terrain products be treated as survey-grade?+

No blanket survey-grade claim is made. The software exposes evidence, confidence and readiness, but field accuracy still depends on the source data and an independently validated workflow.

Can it combine scans in different coordinate systems?+

The viewer does not currently reproject between CRSs. Equal-frame scans can be compared; mixed-CRS alignment belongs in a downstream geospatial tool.

NAVIGATION
Navigate Openesc Close