A Comprehensive Guide to Photogrammetry: Principles, Technologies and Workflows

Introduction to Photogrammetry

Photogrammetry is a methodological process used to reconstruct the geometric properties of physical environments or discrete objects from photographic imagery. By analyzing a sequence of overlapping photographs taken from multiple perspectives, advanced computational algorithms calculate three-dimensional (3D) spatial data, resulting in highly detailed digital models and precise measurements. This technique, positioned at the intersection of optics, geometry, and computer vision, plays a critical role in numerous disciplines, including geospatial analysis, civil engineering, architecture, heritage conservation, industrial manufacturing, forensic documentation, and digital content creation for entertainment and simulation.

Photogrammetry represents the convergence of optical science, spatial computation, and digital imaging. As a flexible and continually advancing technology, it empowers users to recreate and analyze real-world environments in detailed digital form. From drone-based architectural documentation to micro-scale artifact preservation and asset creation, its applications are vast and growing.

With increasing computational capabilities, intuitive software, and broader hardware access, photogrammetry is more approachable than ever. Paired with complementary technologies such as LiDAR and machine learning, it is evolving into a critical component of modern documentation, visualization, and analysis workflows.

The Long History of Photogrammetry

Historically, photogrammetry emerged in the 19th century for topographic mapping and aerial reconnaissance. Nonetheless, today digital processing and automation have transformed it into a scalable and accessible tool with a broad spectrum of applications. As with a lot of new technologies, it was first used in warfare in order to get information about the enemy’s fortifications.

While in the past photogrammetry was used for just surveying lands and big buildings, in the present photogrammetry became more than a scanning method, it became a dynamic bridge between the physical and digital worlds, enabling exploration, insight, and innovation (and I don’t even need to mention that it helps you create 3D models in a really small fraction of the time it’d traditionally take to do it, which is the reason why most of us are interested in it!).

---

Principles and Workflow

At the core of photogrammetry lies geometric triangulation. When a single physical feature is captured in at least two photographs taken from distinct spatial viewpoints, its 3D coordinates can be accurately resolved by analyzing the intersection of the sightlines extending from each camera position to the object. This technique, once labor-intensive, is now largely automated through Structure-from-Motion (SfM) pipelines. These algorithms not only estimate the position and orientation (extrinsics) of the cameras but also simultaneously recover the spatial arrangement of the photographed scene

Standard Processing Pipeline

1. Image Acquisition – Strategic capture of high-resolution images with substantial overlap (typically 60–80%) from various angles. This ensures redundancy and facilitates robust feature matching.
2. Image Registration – Automated detection and matching of key visual features across image pairs. Common algorithms include SIFT, SURF, or ORB.
3. Sparse Reconstruction – The initial 3D structure is inferred from matched points, producing a sparse point cloud and estimated camera poses.
4. Dense Reconstruction – Multi-view stereo (MVS) techniques densify the model, creating a more detailed point cloud.
5. Mesh Generation – The dense point cloud is converted into a polygonal mesh representing the surface geometry.
6. Texture Mapping – High-resolution imagery is projected onto the mesh to produce photorealistic textures. Optional outputs include per-vertex coloring and multi-channel PBR texture maps.
7. Georeferencing (optional) – Incorporates control points or GNSS data to align the model with global or local coordinate systems. This is essential for land surveying, construction planning, or any application requiring spatial accuracy.

---

Applied Domains

Note: photogrammetry has a wide range of applications, but in this article we’ll mention the three most important applications: architectural phogrammetry, land surveying and 3D modeling from physical objects.

Architectural Photogrammetry

Architectural photogrammetry focuses on capturing the external and internal geometry of buildings, monuments, and structures using images acquired from ground-based or aerial platforms. Unlike land surveying, which emphasizes uniform terrain mapping, architectural photogrammetry requires varied and adaptive capture strategies based on the building’s size, complexity, and occlusions. Techniques may include manual camera walkthroughs, oblique aerial shots using UAVs, and close-range scans for façades or ornamentation. Outputs include detailed 3D models, elevation drawings, and textured surfaces suitable for restoration, documentation, and digital archiving.

Key Benefits:

• Rapid, high-throughput data collection across expansive or hazardous areas
• Centimeter- to millimeter-scale model accuracy depending on flight parameters and equipment
• Seamless integration with CAD (Computer-Aided Design), BIM (Building Information Modeling), and GIS (Geographic Information Systems)
• Cost-effective alternative to traditional terrestrial surveying

Land Surveying with UAVs

Land surveying photogrammetry typically involves aerial image acquisition using UAVs in systematic grid or double-grid flight patterns to ensure consistent coverage and accurate terrain modeling. This approach is optimized for generating Digital Elevation Models (DEMs), orthophotos, contour maps, and volumetric measurements for land management, construction, agriculture, and environmental monitoring. The workflow emphasizes georeferencing with high-precision GNSS data, ground control points (GCPs), and post-processing in GIS-compatible formats for integration into planning and development systems.

Key Benefits:

• Rapid turnaround and cost savings compared to traditional ground surveys
• Efficient mapping of large or difficult-to-access areas
• High positional accuracy when combined with RTK/PPK systems

GIS and Deep Mapping

Photogrammetry is specially useful as a Data Layer in Geographic Information Systems (GIS) and Deep Mapping (rich-data mapping system which cross-correlate multiple data sources in order to understand the context, patterns, inner workings, history and evolution of a city). For more in-depth articles about GIS and Deep Mapping see the following articles:

• An Introduction to Geographic Information Systems (GIS) and Current GIS Technologies
• Deep Mapping, Creating Maps With Cross-related Context

In modern GIS techniques (that’s kinda a redundancy since GIS is a very modern technique all by itself) photogrammetry can be one of the most valuable Data Layers in the geographic model we’re building.

Object-Level 3D Modeling and Asset Creation

Photogrammetry is equally powerful at micro scales. It is widely used to digitize small- to medium-sized objects with high precision and texture fidelity. Applications include documentation of cultural heritage artifacts, manufacturing quality control, and generation of assets for computer graphics pipelines.

When you’re creating 3D models of physical objects, photogrammetry consists in taking several photos from as many angles as possible. Then your photogrammetry software will use all the positional information in your photos meta data, and use the image data and the positional information to automatically build the 3D model.

The previous image you’re seeing is the photogrammetry software showing you all the angles from where photos of that statue were taken. The software does this automatically.

Visual Asset Applications:

• Complete PBR (Physically-Based Rendering) material sets: albedo, normal, roughness, metallic, and ambient occlusion maps
• Integration into pipelines for digital twins, AR/VR experiences, animation, and 3D printing

Professional studios will not use a single camera, instead, they have multi camera rigs, and these rigs take dozens of photos at the same time, allowing the studio to create life-like 3D models of actual, living people. Hideo Kojima‘s studio utilizes that method regularly.

Of course you don’t need such a setup if you’re scanning static things. A single camera and enough patience are more than enough.

Industrial and Scientific Use Cases:

• Watertight, topology-optimized meshes for simulation or real-time rendering
• Material scanning and surface detail capture for synthetic dataset creation
• Precision inspection and reverse engineering of manufactured components

---

Tools, Equipment, and Outputs

Image Acquisition Hardware

• Cameras: DSLR and mirrorless systems with prime lenses ensure sharp images and minimal distortion. Sensor size, pixel pitch, and dynamic range are critical factors.
• Drones: Multi-rotor UAVs with stabilized gimbals and integrated GNSS systems are ideal for aerial photogrammetry.
• Lighting: Controlled lighting is essential for indoor or object-level scanning to minimize shadows and reflections.

Processing Software

• Agisoft Metashape: Preferred by professionals in surveying, heritage preservation, and architecture.
• RealityCapture: Ideal for game developers and filmmakers needing fast, GPU-accelerated workflows.
• Meshroom: Open-source, user-friendly solution suitable for educational or hobbyist use.
• COLMAP: Research-grade tool with modular design, widely used in academia.
• Cloud-based platforms: Services like Pix4Dcloud provide web-based processing and collaboration capabilities.

Common Output Formats

• Point Clouds: PLY, LAS, E57
• Meshes: OBJ, FBX, STL, GLTF
• Textures and Materials: PNG, JPEG, TIFF, EXR
• Geospatial Outputs: GeoTIFF orthophotos, XYZ point files, KML/KMZ overlasy

---

Strengths and Limitations of Photogrammetry

Advantages

• Non-destructive, scalable, and field-portable 3D data acquisition
• Simultaneous capture of spatial geometry and rich surface texture
• Applicable across various scales and environments
• Survey-grade accuracy achievable with proper control and calibration
• Broad compatibility with hardware and software platforms

Limitations

• Reduced performance on reflective, transparent, or featureless surfaces
• Lighting inconsistencies can introduce processing artifacts
• Processing large or complex datasets requires significant computational resources
• Model fidelity can be affected by lens distortion, motion blur, or poor calibration

---

Complementary Technologies

• LiDAR (Light Detection and Ranging):
  • Captures accurate depth information using laser pulses
  • Performs well in low-light and low-texture conditions
  • Generally more expensive and lacks color data unless fused with imagery
• Structured Light Scanners:
  • Project light patterns onto surfaces to determine geometry
  • Highly accurate for small, controlled scenes
  • Sensitive to ambient lighting and motion
• Multimodal Fusion:
  • Combines photogrammetry with LiDAR or depth sensors (e.g., Intel RealSense, iPhone LiDAR)
  • Enhances both textural and geometric accuracy
  • Common in BIM, autonomous systems, and robotics workflows