Workflows¶

Step-by-step recipes for the common tasks. Each workflow is self-contained — start at the top, follow along, end with a result.

Every operation is reachable from the always-visible toolbar (left of the viewer) and from the menu bar. Tools are spread across three menus: Create (build the scene — geometry and scanner placement), Simulate (synthesize scans), and Tools (analysis operations on existing data — pre-processing, segmentation, reconstruction & analysis, registration). The left toolbar shows two of these as blocks — Create and Tools; the Simulate action (Run Synthetic Scan) lives in the Simulate menu and on the Scans panel. These workflows are grouped by the four stages of a typical processing pipeline.

Getting around¶

Import & export — getting data in and out of Phytograph.
Viewer navigation — moving the camera, switching color modes, isolating an object.
Export from RiPROCESS — get RIEGL scans out with the fields Phytograph needs.
Import a RIEGL project — read .riproject / .PROJ scans directly, no RiSCAN PRO round trip.

1. Pre-processing¶

Clean a point cloud — the whole cleaning pass, in the order to do it.
Transform — move, rotate and scale clouds and meshes together, and set the scene origin.
Crop — keep or discard a region of a cloud or a mesh with a box, rectangle, or freeform polygon.
Erase — paint away stray points with a brush.
Filter — keep points by scalar range or by class.
Resample — thin a dense cloud to a fraction of its points.
Cross-section — clip the view to a slab and traverse it.
Compute normals — estimate surface direction, curvature and verticality.
Work with scalar fields — statistics, histograms, and new fields from a formula, over one cloud or several at once.
Backfill misses — recover the sky/miss rays that LAD needs from a scan that dropped them.
Auto-register clouds — align scans that start far apart, validated by loop closure.
Align with ICP — refine cloud-to-cloud, mesh-to-mesh, and cloud-to-mesh alignment.
Stitch clouds or meshes — merge overlapping scans into one cloud, or several meshes into one mesh.
Cloud-to-mesh distance — measure how far a cloud sits from a reference surface.

2. Segmentation¶

Segment ground points — classify and remove the ground with the Cloth Simulation Filter.
Separate leaf and wood — split a scan into woody and foliage points by local geometry.
Segment plant organs — label a herbaceous plant soil / stem / leaf and number its leaflets.
Segment individual trees — separate a multi-tree cloud into per-tree instances.
Label points by hand — assign your own classes with a lasso, to correct a classifier or build ground truth.

3. Reconstruction & analysis¶

Triangulate a mesh — fit leaf surfaces to measure the leaf-angle distribution and G(θ), or reconstruct a surface with Ball Pivot, Poisson, Alpha Shape, or Delaunay.
Fit a crown & metrics — wrap the canopy in a fitted shape and read off height, volume, and width.
Run a tree inventory — a tree list of stem position, DBH, height, crown base and crown size for every segmented tree.
Generate a DEM — build a bare-earth DTM, a top-of-canopy DSM, or a canopy height model.
Extract a skeleton — pull branch topology out of a woody scan.
Build a QSM — reconstruct a dormant tree as connected cylinders with radii, continuous shoots, and shoot rank.
Add leaves to a QSM — place leaves on terminal shoots using phyllotaxis, then match them to a measured leaf-angle distribution.
Estimate leaf area density — invert overlapping scans against a voxel grid into an LAD grid (m²/m³).

4. Scan simulation¶

Generate a plant — produce a procedural plant from species, age, and position.
Tune plant parameters — adjust a generated plant's growth parameters — internode length, insertion angle, girth, curvature, tortuosity — and rebuild it at the same age.
Simulate a LiDAR scan — place a virtual scanner, set its field of view, resolution and beam optics, and synthesize the point cloud it would produce — from a fixed position or along a drone, robot, or tractor trajectory.