Skip to main content

Workflow: Geometric Reconstruction from Visual Evidence

WORKFLOWADVANCED
📖15 min read

Phases​

Data Ingestion & Processing

Important: Do not alter original source files. All manipulations (sharpening, un-distorting) should be done on duplicates/proxies. Log every filter & setting applied for transparency.

Focus: Order and analyze (multiple) source media streams to extract technical data and converting raw footage into the most usable type of assets.

Source Intelligence (Analysis)​

Ingest​

Collect all Video, Photos and other source material.

Master Archive:

  • Create a (protected) directory (e.g., /00_Master_Archive).
  • Save all sources here exactly as received/downloaded.

Fingerprinting (Hashing):

  • Generate a Checksum (SHA-256) for all files (See Hashing)
  • Store the Source Name, Download Date, Download location, Hash in a spreadsheet/text file (See Bellingcat best practices)

Create working_file directory

  • Copy all source material here.

Segmentation​

If one video contains multiple clips, create separate clips.

Naming:

  • Create Directories for each video
  • Extract Frames from each Video

Naming:

Source Grouping: Sort and group files by physical device (e.g., "Source A - CCTV North," "Source B - Witness iPhone").

/01_WORKING_COPIES
  /00_MASTER_INGEST
    [Project]_Master_Compilation_Copy.mov <-- The big file you copied
  /01_SOURCES <-- The result of your "Chop"
    /SRC_A_CCTV_North
      /Footage <-- The segmented video clip (ProRes/DNxHR)
      /Image_Seq <-- The full PNG/EXR sequence extracted from footage
      /Stills <-- Specific keyframes picked for reference
      /Work <-- Undistorted/Sharpened versions
    /SRC_B_Mobile_Witness
      /Footage
      /Image_Seq
      /Stills
      /Work

Technical Assessment (See Camera Fundamentals for Investigation)​

Optics: Analyze footage for Lens Distortion types (Barrel, Pincushion, Mustache) and Vignetting per main source file.

Artifacts: Identify Motion Artifacts (Motion Blur, Rolling Shutter, Compression blocks).

Geometry: Identify sources with optimal Perspective Lines (One-point, Two-point, or Three-point) to determine the best solving method for each specific angle.

Check for Metadata / Metadata Extraction (See Metadata Extraction)​

Action: Analyze file headers (EXIF/XMP) for each main source file (look at the main video, not the videos derived from it)

Key Data: Look for Camera Make/Model, Focal Length, Sensor Size, Date/Time, and GPS coordinates.

Note: If source is from social media, metadata is likely stripped; rely on visual "Technical Assessment."

If no metadata: Estimate if each specific camera uses a Prime Lens (Fixed) or Zoom Lens (Variable focal length). Estimate the Focal Length for Fixed Lens (Wide angle shots, zoomed shots), Possible Sensor Width (mm) for each profile independently.

Temporal Alignment (See video alignment)​

Sync all video footage to a single master real-world timeline (e.g., "Event Time 12:00:00") using metadata timestamps or visual anchors (e.g., a specific flash of light visible in all cameras).

Further Processing​

Image Enhancement (Working Copies):

Create separate "Work Files" for selected images. See if lens distortion can be countered, reduce blur (sharpening), upscale, or adjust contrast to reveal stronger perspective lines.

Last updated on by blender-visual-investigation