Products · VTOL Fixed-Wing

Oblique Aerial Photography Configuration

Oblique capture is bought for facades and for modelling, and it costs more sorties than nadir. Justify it against a named output.

8 of 10Method note

Overview

Define the tilt angles and the number of look directions; five-look is not a default, it is a modelling requirement.
Facade texture needs both sufficient overlap and enough ground resolution; one without the other gives a soft model.
Oblique blocks produce larger datasets — budget storage and processing time, not just flight time.
For city models, check whether the receiving software expects a specific block layout.

Why oblique imagery exists

A nadir image looks straight down and is the correct geometry for mapping the ground plane. It is, however, a poor way to see the vertical faces of buildings, the underside of structures, or anything whose informative surface is not horizontal. On a site with buildings, tanks, bridges or stacked material, a nadir-only survey systematically misses the surfaces that carry most of the information.

Oblique imagery fills that gap by capturing the same area from angled viewpoints, typically four directions plus the nadir pass. The result is a dataset in which vertical surfaces are visible, façades can be inspected, and the geometry of tall structures can be reconstructed rather than inferred. It is the standard approach for urban and industrial site capture for exactly this reason.

The cost is real and should be stated plainly. Oblique acquisition requires more flight lines or a multi-sensor rig, produces substantially more data, and demands a heavier processing chain. It is not a free upgrade to a nadir survey. It is the right choice where vertical surfaces matter and an unnecessary expense where they do not — a flat agricultural block, for instance, gains almost nothing from it.

Rig configuration and the flight pattern

There are two practical ways to collect oblique imagery. The first is a multi-sensor rig: several cameras fixed at different angles, triggered together, capturing all directions in a single pass. This is efficient in flight time and produces inherently consistent geometry between views, at the cost of a heavier payload and a more complex calibration and synchronisation requirement.

The second is a single gimbal camera flown in multiple passes at different angles, or a camera that can be commanded to an angle between passes. This uses a lighter, simpler payload but multiplies flight time by the number of directions required, and it introduces the possibility of geometry differences between passes if conditions change. For smaller sites the simplicity often wins; for large ones the rig usually does.

Either way, the flight pattern must be planned for oblique geometry rather than adapted from a nadir plan. Line spacing, altitude and turn behaviour all change, because the effective coverage of an angled sensor is different from that of a downward one, and because the aircraft must fly the pattern in a way that keeps each direction's coverage complete at the site edges.

Ground sampling distance is not the right specification for oblique

For oblique imagery, ground sampling distance is ambiguous, because the distance from sensor to surface varies across an angled image. A pixel at the near edge represents a different ground distance from one at the far edge. Specifying a single figure is therefore misleading, and comparisons between systems based on a nominal figure are frequently invalid.

Specify instead the resolution required on the surface of interest — typically a stated figure at a stated distance from the sensor, or a requirement expressed as the smallest feature that must be recognisable on a façade. That is checkable, and it maps directly onto what the user of the data will actually do with it.

This also changes how altitude is chosen. Because oblique imagery is usually collected for structures, and structures have height, the relevant distances are from the sensor to the top and bottom of the structure, not to the ground. Plan the altitude against the tallest structure in the area of interest, and check that the top of it is still within the required resolution.

The processing and delivery reality

Oblique datasets are substantially larger than nadir ones, and they are heavier to process. Reconstruction from angled views is more demanding, and the outputs — textured meshes, façade imagery, and three-dimensional structure — are different artefacts from an orthophoto. Budget for both storage and processing time, and confirm that the receiving system can actually consume the output format.

Registration between the oblique views and the nadir pass is what makes the combined dataset usable. If the different views are not geometrically consistent, the result is a set of images that cannot be used together, and the value of the oblique capture largely disappears. Confirm that the processing chain handles this, and verify it on a subset before flying the whole site.

Where the deliverable is a measurable three-dimensional model rather than imagery, the accuracy expectation should be stated and verified against check points, because mesh quality and dimensional accuracy are not the same thing. A visually convincing model can be dimensionally wrong, and downstream measurement will not detect that unless it is checked.

Configuring an oblique programme

The configuration we would propose depends on site size and structure density. For a compact site with many structures, a single gimbal flown in multiple passes is often the simpler and cheaper answer. For a larger area, a multi-sensor rig in a single pass usually wins on flight time and geometric consistency.

Either way the specification should be written against resolution on the surfaces of interest rather than a nominal ground sampling distance, and the flight plan should be built for oblique geometry rather than adapted from a nadir plan. Verification on a subset before full acquisition is not optional.

Send your site area, structure heights and density, the surfaces you need to see, and your output format requirements to our engineering team. We will recommend a rig configuration and flight pattern, and estimate the data volume and processing load so you can size the ground side accordingly. For the mapping baseline that usually accompanies oblique capture, see the orthophoto acquisition configuration note.