Skip to main content

Geometric Reconstruction

WORKFLOW
📖7 min read

Introduction

Geometric reconstruction is the end-to-end practice of rebuilding a scene from visual evidence in Blender. It operates at three scopes:

  • Object-level: Reconstructing and measuring individual objects from photos — anchored by the scale pillar (the source-reliability hierarchy and error propagation).
  • Site-level: Rebuilding a location in 3D from satellite and OSINT imagery — a georeferenced 1:1 baseline plus a matched camera. Operationalized in the Site Reconstruction course.
  • Environmental: Rebuilding the geospatial context — terrain, coordinate systems, and ground control points — the "Ground Truth" stage on which the investigation plays out.

The phases below cover one full pass. Phases 1 and 5 (data handling and reporting) are owned by the Foundation course's Data lifecycle; Phases 2–4 are detailed on this page.


Phase 1: Ingestion & Processing

The data-handling stages are covered by the Data lifecycle — a single, unidirectional path:

  1. Collect — acquire and document sources (registry, provenance, authenticity, technical assessment).
  2. Preserve & Store — sanctuary for originals, hashing, working copies, segmentation.
  3. Process — transform sources into usable assets (temporal alignment, enhancement), with every change logged.

Phase-specific reconstruction decisions continue in Scene Decomposition below.


Scene Decomposition

Asset Strategy & Branching. Scene decomposition involves isolating, categorizing, and preparing every individual element within the frame for reconstruction. This phase turns a complex scene into a structured inventory of assets ready for object and environmental reconstruction.

Inventory

Identify and categorize all static and dynamic entities.

Intelligence

Gather ground truth data, blueprints, and specs.

Methodology

Determine the best creation strategy for each asset.

Asset Identification (Inventory)

  • Macro-Static (Infrastructure): Buildings
  • Micro-Static (Furniture): Street signs, benches, trees, trash cans.
  • Dynamic Entities (Movers):
    • Biometric: Humans, crowds, animals (Agency).
    • Kinematic: Vehicles, machinery (Physics).
  • Ephemeral Events: Explosions, smoke plumes, fire.
  • Environment: Sun position, weather conditions (wet roads/puddles), debris.

Naming & Organization

Assign a unique ID to every item in the Inventory.

Don't use naming conventions that are subject to change (No Car_Lada_Niva_, but Car_Sedan_White). Pick categories, classifications or a taxonomy and visual aspects that are static (a car doesn't change from black to blue, nor does it change from Sedan to Hatchback).

Naming Convention: Category_Description_ID (e.g., MACRO_Building_TownHall_01 or DYN_Car_RedSedan_02).

Folder Structure: Create a directory for each asset containing two sub-folders:

  • /Reference (Blueprints, OSINT photos).
  • /Data (Measurements, specs).

Intelligence (Reference Gathering)

Try to find "Ground Truth" data for each item in your Inventory.

OSINT / Specs: Locate blueprints, satellite maps, vehicle manuals, or weather reports.

Dimensional Verification: Confirm the height, width, and length of objects. Example: Don't guess the height of the lamp post; find the manufacturer spec or measure it relative to a known object.

Alternative Angles: Find existing footage (Street View, News, Social Media) that shows the object from sides not visible in your primary evidence.

Establish Methodology per Object

StrategyDefinitionBest Use Case
SOURCELibrary AssetsMicro-Static & Generic Kinematic. Standard objects (Trash cans, Oak trees, generic Ford Focus) where the specific scratches/dents don't matter. Do not reinvent the wheel.
GENERATEAutomated Creation, ProceduralMacro-Static & Environment. Complex terrain, debris fields, or large city blocks. Use Photogrammetry (Drone/Google Earth) or Scatter tools (Gravel, Grass).
MODELHand-CraftedKey Evidence. The specific building where the event happened, or a unique vehicle with specific damage. These must be built vertex-by-vertex to match the reference perfectly.
SIMULATEPhysics-BasedEphemeral Events. Explosions, fire, smoke. These cannot be modeled; they must be calculated using fluid dynamics based on the visual evidence.

Evidence Assessment

What source material do you have? This determines which reconstruction techniques are possible.

Visual Evidence Density:

  • Sparse (1-5 viewpoints): Limited angles, use camera matching/perspective projection
  • Medium (6-30 viewpoints): Structure-from-Motion (SfM) photogrammetry feasible
  • Dense (30+ viewpoints): Full photogrammetry or Gaussian Splatting possible

Visual Evidence Quality:

  • High: DSLR photos, stabilized video, minimal compression
  • Medium: Smartphone footage, CCTV with moderate resolution
  • Low: Compressed social media video, long-distance footage, heavy motion blur

Temporal Coverage:

  • Single moment: Static reconstruction
  • Discrete events: Keyframe animation between known states
  • Continuous: Full motion tracking required

Supplementary Data Availability:

  • Geospatial: GPS coordinates, satellite imagery, maps, blueprints available?
  • Technical specs: Vehicle dimensions, building plans, object specifications?
  • OSINT: Can you find additional angles from social media, news footage, Street View?

→ Output: Identify which reconstruction techniques are viable (photogrammetry vs. modeling vs. camera matching).


Object Reconstruction (Summary)

This branch focuses on reconstructing key objects from the evidence inventory established in Scene Decomposition.

  • Prioritize objects that are directly tied to the research question.
  • Use reference-driven modeling, camera-constrained modeling, or photogrammetric methods based on available sources.
  • Establish scale and uncertainty before forwarding assets to synthesis.

The full scale pillar — the five-tier source-reliability hierarchy, confidence decision matrix, error propagation, and verification logging — lives on the Scale Estimation & Source Reliability page.


Environmental Reconstruction

Geospatial. Environmental reconstruction focuses on rebuilding the geospatial context of the scene. This phase establishes the "Ground Truth" — the accurate, real-world stage upon which the investigation plays out.

Geolocation

Verify coordinates and establish a relative origin.

Acquisition

Import vector data, elevation models, and satellite imagery.

Verification

Ensure CRS consistency and align with Ground Control Points.

Geolocation

Verify coordinates. If no location is found, proceed with Relative Reconstruction (arbitrary origin).

Chronolocation: Determine the time of day and date to accurately simulate lighting conditions (sun position). See Geolocation and Chronolocation for the supporting techniques.

Terrain & Context Acquisition

Vector Data: Import OSM (OpenStreetMap) footprints via QGIS.

Elevation: Import DEM (Digital Elevation Models) for accurate terrain topography.

Photorealistic Context:

  • Satellite: Google/Bing Satellite imagery (Regular and possibly Historical/Dated).
  • 3D Tiles: Import Photogrammetry Mesh (Google Earth 3D) or Geometry Rip (via RenderDoc).

See Terrain & Landscape Generation for the full terrain pipeline.

Critical Check: When using BlenderGIS ensure CRS (Coordinate Reference System) consistency across all datasets.

Coordinate System Verification

  • Document CRS: Explicitly state which CRS you're using (e.g., WGS84, UTM Zone X).
  • Resolution standards: Specify minimum acceptable resolution for satellite imagery (e.g., "prefer 1m/pixel for urban scenes").
  • Temporal matching: If your event happened in 2023 but Google Earth imagery is from 2019, document what changed (construction, demolished buildings).
  • Vertical datum: Clarify whether elevation is relative to sea level (EGM96, EGM2008) or ellipsoid.

Ground Control Points (GCPs)

  • Identify Landmarks: Find fixed landmarks visible in both your evidence AND satellite/maps (corners of buildings, painted road markings).
  • Anchor Points: These become your "anchor points" for aligning everything.
  • Documentation: Document GCP coordinates in a table with confidence ratings.

Phase 5: Synthesis & Validation

The general validation and reporting is covered by Validate & Report — assembly, spatial/geometric verification, confidence visualization, and the analytical report with honest limitations and boundary awareness.

Phase-specific assembly decisions (which terrain and asset layers to merge) follow from Scene Decomposition and the phases above.


Documentation for Methodology

What to document during Scene Decomposition

In Your Method Section

"The scene was decomposed into [Number] distinct assets. Key evidence, such as [Object A], was modeled manually based on [Reference]. Contextual elements were sourced from [Library Name] or generated procedurally."

In Your Decision Log

Record the following:

  • The source of dimensions for each key asset (e.g., "Vehicle dimensions from 2018 Manufacturer Manual").
  • Why a specific modeling strategy was chosen (e.g., "Used photogrammetry for the statue due to complex geometry and available drone footage").
  • Any assumptions made about object scale where ground truth was unavailable.

Verification

"Modeled assets were overlaid on source imagery to verify silhouette alignment and scale."

What to document during Environmental Reconstruction

In Your Method Section

"The environment was reconstructed using [CRS Name] as the coordinate reference system. Terrain data was derived from [DEM Source] with a resolution of [X] meters. Vector data for building footprints was imported from OpenStreetMap and verified against satellite imagery dated [Date]."

In Your Decision Log

Record the following:

  • The specific CRS code (e.g., EPSG:32633).
  • The source and date of satellite imagery used.
  • Any discrepancies found between map data and visual evidence (e.g., "Satellite image shows a building that is not present in the video evidence; confirmed demolition date via news source").
  • Coordinates of Ground Control Points used for alignment.

Verification

"The alignment of the 3D terrain model was verified by checking the position of [Landmark A] and [Landmark B] against their known GPS coordinates."


Summary

Geometric reconstruction is a layered process: ingest and document the data, decompose the scene into an inventory with a creation strategy per asset, reconstruct objects with a defensible scale, anchor everything in geospatial ground truth, and validate the whole assembly against the source evidence. Each layer earns its keep in the final report — the methodology section is where these phases become proof.

Key Takeaways:

  • Inventory before building: List everything, name it consistently, and assign a strategy per asset.
  • Ground Truth first: Verify dimensions and geolocation with external data whenever possible.
  • Scale before synthesis: Establish scale and uncertainty before objects move into validation.
  • CRS consistency: All data layers must share a Coordinate Reference System.

Further Resources: