This session examines evolving approaches to photogrammetric data processing, comparing Structure-from-Motion and traditional stereo workflows across drone and large-format aerial imagery. Presentations explore software performance, camera modeling innovations, and workflow optimization techniques that improve mapping accuracy, efficiency, and the quality of three-dimensional geospatial products.
1:30 PM – 2:15 PM
Comparing Stereo and Structure from Motion Photogrammetry Processing Workflows Using Small and Large Format Imagery: A Comparison of Different Image Types in Each Software
Presented by Michael Detwiler, Great Basin Geomatics LLC, Greg Saunders, Ampacity, and Drew Knight, Ayres Associates
The geospatial industry continues to expand the ways aerial imagery is processed to produce accurate mapping products for surveying, engineering, and infrastructure projects. While Structure-from-Motion (SfM) software has become the standard workflow for many UAS aerial mapping applications, advances in software and computing power have also made it possible to process traditional large-format aerial imagery using modern SfM techniques. At the same time, softcopy stereo photogrammetry remains an essential tool for applications requiring precise three-dimensional feature extraction and topographic mapping. This session brings together three related presentations that compare these complementary workflows, highlighting their strengths, limitations, and ideal applications while providing attendees with practical guidance for selecting the most appropriate processing methodology for their projects.
Michael Detwiler presents:
The first presentation explores the use of Pix4D to process imagery acquired with a Vexcel UltraCam large-format aerial mapping camera. Although Pix4D is widely associated with drone imagery, the application also has the ability to process large-format aerial datasets. The presentation will discuss project setup, camera calibration, aerial triangulation, dense point cloud generation, orthomosaic creation, and accuracy assessment using real-world production projects. Lessons learned, workflow optimizations, and practical considerations for achieving survey-grade deliverables from large-format imagery will also be presented.
Greg Saunders presents:
The second presentation examines processing the same Vexcel UltraCam imagery using Agisoft MetaShape. While both Pix4D and MetaShape employ Structure-from-Motion principles, each software package offers unique approaches to image alignment, bundle adjustment, dense reconstruction, and project optimization. Using similar type imagery datasets, this presentation compares workflow efficiencies, processing strategies, and output quality, helping attendees better understand where each software package excels and how to select the appropriate platform based on project requirements and desired deliverables.
Drew Knight presents:
The final presentation demonstrates how drone imagery properly captured using a calibrated camera can be processed using traditional softcopy stereo photogrammetry workflows. Rather than relying solely on automated point cloud generation, stereo compilation enables highly accurate three-dimensional feature extraction and mapping. The presentation will illustrate how modern drone imagery can be successfully integrated into established photogrammetric mapping workflows for transportation, utilities, engineering, and surveying applications, while discussing the continued value of stereo photogrammetry alongside today’s Structure-from-Motion technologies.
2:15 PM – 2:30 PM
Higher and Sharper: Camera Modelling Techniques for Aerial Photogrammetry at Altitude
Presented by Andrew Millin, Nearmap
Aerial mapping providers strive for greater capture efficiency: flying at high altitude with long focal length lenses captures more ground area per flight, cutting flight hours and cost per square kilometer while preserving ground sample distance and image detail. Aerial camera systems are evolving, with modern systems utilizing steerable scanning mirrors to capture wide swaths of nadir and oblique imagery. But increasing altitude amplifies error sources that standard bundle adjustment pipelines were never designed to absorb — thermal and autofocus driven focal length drift becomes more visible through a longer lens, atmospheric refraction bends light more strongly over longer oblique ray paths, and steerable scanning mirrors require virtual camera geometry modelling that breaks simple pinhole camera model assumptions.
This talk presents a set of practical camera modelling techniques developed for high altitude aerial camera systems with long focal length lenses. We will dive into the fundamentals of camera projection models and equations, and demonstrate how pinhole camera models can be improved to efficiently model this new reality — dynamic focal lengths, virtual camera geometry, entrance pupil offsets and atmospheric refraction.
Attendees will get an insight into the evolution of bundle adjustment software, camera modelling and aerial photogrammetry in practice, from the trial and error of R&D processes to the discoveries that have enabled step changes in accuracy and capture efficiency.
