Skip to main content

How to understand your Birdi Processing Report

Written by Kayley Greenland
Updated over a week ago

Your Processing Report provides a technical summary of how your data was reconstructed — from raw images through to orthomosaic, 3D model, and elevation outputs.

This guide explains what each section means and when it matters.


1️⃣ Project Overview (Page 1–2)

This section includes:

  • Map Name

  • Company Name

  • Created At

  • Preview image

📌 What it means:
This confirms which dataset and workspace the report relates to. It’s primarily administrative and useful for record keeping or sharing with stakeholders.


2️⃣ Survey Data (Page 3)

This is one of the most important sections.

It includes:

  • Number of images

  • Camera stations

  • Flying altitude

  • Ground resolution (GSD)

  • Coverage area

  • Tie points

  • Projections

  • Reprojection error

  • Camera model details

Key Metrics Explained

📷 Number of Images

Total photos uploaded (192 in this report).

📍 Camera Stations

Aligned images used in reconstruction (182 here).
If this number is lower than total images, some photos did not align.

✈ Flying Altitude

Average flight height above ground (55.4 m).
This influences resolution and coverage.

📐 Ground Resolution (GSD)

1.45 cm/pixel in this report.

This tells you:

Each pixel represents 1.45 cm on the ground.

Lower = more detail.

🔗 Tie Points (239,470)

These are matched feature points between images.
More tie points generally mean stronger geometry.

🎯 Reprojection Error (1.59 pix)

This measures how accurately the 3D model fits the image data.

As a guide:

  • <1 pixel = excellent

  • 1–2 pixels = good

  • 3 pixels = may indicate geometry issues

Your example sits in the good range.


3️⃣ Camera Calibration (Page 4–5)

This section shows how the camera lens was modelled during processing.

It includes:

  • Focal length (F)

  • Principal point (Cx, Cy)

  • Distortion coefficients (K1–K4)

  • Tangential distortion (P1, P2)

  • Residual plots

What this means

All lenses introduce distortion.
The software estimates and corrects this mathematically.

If calibration is unstable, you’ll see:

  • Large distortion values

  • High reprojection error

  • Warping in outputs

For most users:

If reprojection error is low and outputs look clean, calibration is working correctly.


4️⃣ Camera Locations & Error Estimates (Page 6)

This section visualises:

  • Camera positions

  • Error ellipses

  • X (Easting), Y (Northing), Z (Altitude) error

  • Total error

In this report:

  • X error: 8.1 cm

  • Y error: 6.9 cm

  • Z error: 4.7 cm

  • Total error: 11.7 cm

What This Means

This reflects how much the adjusted camera positions differ from their GPS-recorded positions.

Important:
This is camera position error, not map accuracy error.

Without GCPs, camera error often reflects consumer drone GPS limits (~5–15 cm typical variance after adjustment).


5️⃣ Digital Elevation Model (Page 7)

This shows the reconstructed terrain model.

Includes:

  • Resolution (5.81 cm/pixel)

  • Point density (296 points/m²)

What Resolution Means Here

DEM resolution is usually lower than orthomosaic resolution because it’s derived from the 3D surface.

Higher point density = smoother terrain.


6️⃣ Processing Parameters (Page 8–10)

This section documents how the data was processed.

It includes:

  • Alignment accuracy

  • Key point limits

  • Depth map quality

  • Filtering mode

  • Ground classification settings

  • Model reconstruction settings

  • Interpolation settings

  • Ghosting filter

  • Blending mode

Why This Section Matters

This ensures:

  • Reproducibility

  • Audit trail

  • Technical validation

  • Engineering sign-off

For most users, the key items are:

Alignment Accuracy

Medium in this report.
Higher accuracy increases processing time but can improve detail.

RMS Reprojection Error

0.1587 (1.59 pix)
Confirms alignment quality.

Ground Classification

Parameters used to separate ground from vegetation.

This matters for:

  • Volume calculations

  • Contour generation

  • Accurate DEM exports


7️⃣ Point Cloud (Page 9)

This section includes:

  • Total points: 17.9 million

  • Classification breakdown:

    • Ground

    • Low point (noise)

    • Unclassified

What This Means

This is your dense 3D reconstruction before meshing or DEM creation.

More points = more detail.

Classification determines:

  • Terrain-only outputs

  • Volume calculations

  • Vegetation filtering


8️⃣ 3D Model (Page 10)

Includes:

  • Faces

  • Vertices

  • Texture info

  • File size

This is the visual 3D mesh representation of your site.

It’s optimized for:

  • Visualization

  • Sharing

  • Interactive viewing


9️⃣ DEM Details (Page 10–11)

Includes:

  • Raster size

  • Resolution

  • Coordinate system

  • Source classes (Ground only)

This confirms:
The DEM was generated using ground-classified points only, not vegetation.

That’s important for accurate terrain modelling.


🔟 Orthomosaic (Page 11)

Includes:

  • Pixel resolution (1.45 cm/pix)

  • Blending mode

  • Surface source (DEM)

  • Hole filling enabled

  • Ghosting filter enabled

What This Means

Your orthomosaic is:

  • Georeferenced

  • Color-balanced

  • Blended using DEM surface

  • Cleaned for visual artefacts

Hole filling helps remove minor gaps.
Ghosting filter reduces moving-object artefacts.


🔎 How to Interpret Overall Quality

When reviewing your report, focus on:

  1. Reprojection error (alignment health)

  2. Aligned vs total images

  3. Point density

  4. Ground classification quality

  5. DEM resolution

  6. Orthomosaic resolution

If these are strong, your outputs will be strong.


🚁 When to Be Concerned

You may want to review capture conditions if you see:

  • High reprojection error (>3 px)

  • Many unaligned images

  • Sparse tie points

  • Uneven camera error distribution

  • Large areas of low point noise


💡 Why This Report Matters

This document:

  • Validates survey quality

  • Supports engineering sign-off

  • Provides audit traceability

  • Confirms coordinate system

  • Documents processing settings

It turns your dataset from “just drone photos” into defensible geospatial data.

Did this answer your question?