Raster guide: orthomosaics, DEMs, DTMs, and more
🕐 Read time: 7 min
Written By Clark Yuan
Last updated 5 days ago
Overview
When you work with drone data, you will encounter several types of raster outputs, each representing a different kind of information about the same site. Orthomosaics, Digital Elevation Models (DEM), Digital Terrain Models (DTM), thermal images, and single-band rasters all look similar at first glance. They are all flat, two-dimensional images stored as GeoTIFF files. But they capture fundamentally different things and serve different purposes.
This guide explains each type clearly, how it is created, and when to use it. It also clarifies what Stitch3D currently supports.
ℹ️ Stitch3D currently supports RGB GeoTIFF rasters only. This includes orthomosaics and RGB aerial imagery. DEMs, DTMs, DSMs, thermal imagery, and multispectral files are not supported at this time. This guide is intended to help you understand the full landscape of raster types so you know which outputs to use for which purposes in your broader workflow.
What is an orthomosaic?
An orthomosaic is a georeferenced aerial image created by stitching hundreds or thousands of overlapping drone photographs into a single, seamless, geometrically corrected map. Unlike a standard aerial photo, which distorts distances based on camera angle and flight altitude, an orthomosaic maintains consistent scale across the entire image, so distances and areas can be measured directly.
The key word is orthorectified. Each individual photo in the mosaic has been corrected for perspective distortion, terrain relief, and camera tilt so that the final composite presents a true top-down view. Traditional orthomosaics are created by generating ortho images using the best available digital terrain model, mosaicking them using the central portion of each image, and calculating seamlines to define boundaries between images.
The result is a single image that looks like a satellite photograph of your site but one you can load into GIS or CAD software, overlay with other spatial data, and measure directly.
How an orthomosaic is created:
A drone captures hundreds of overlapping images with 70–85% forward and side overlap
Photogrammetry software (Pix4D, Metashape, DJI Terra, RealityCapture) identifies matching features across overlapping frames
The software reconstructs a 3D surface model of the terrain
Each image is projected onto that surface to remove distortion
The corrected images are blended and stitched into a single GeoTIFF
Common uses:
💡 Tip: An orthomosaic is the most universally understood spatial deliverable. It looks like a photo, it loads into any GIS or CAD platform as a georeferenced layer, and it gives non-technical clients an immediately readable view of their site. Upload it to Stitch3D alongside your point cloud for a complete client delivery.
What is a DEM, DTM, and DSM?
These three terms are closely related and frequently confused. They all represent elevation as a raster grid; each pixel stores an elevation value rather than a color. The differences lie in what surface each model represents.
Think of DEM as the parent category. Both DSM and DTM are types of DEM.
Digital Elevation Model (DEM)
A DEM is a broad umbrella term for any raster dataset where each pixel represents an elevation value. When someone says "DEM" without further specification, they usually mean a bare-earth surface model. DEMs are used as the foundation for many downstream analyses — watershed modeling, flood risk assessment, cut and fill calculations, and terrain visualization.
Digital Surface Model (DSM)
A DSM is the most general form of surface model that includes all acquired points, representing natural and man-made features. A DSM includes the tops of buildings, trees, powerlines, and other objects. In essence, it is a canopy model, and only sees the ground when nothing else is above it.
A DSM captures the world exactly as the drone sensor sees it from above. It is typically the first elevation output generated from a photogrammetry processing run before any classification or filtering is applied.
Common uses for DSM:
Urban planning and 3D city modeling
Obstruction analysis (aviation, telecommunications)
Vegetation height and canopy modeling
Change detection between capture dates
Digital Terrain Model (DTM)
A DTM represents the bare-earth ground surface where all vegetation, buildings, vehicles, and other above-ground objects are removed. The vertical difference between DSM and DTM at any point gives you the height of objects above the ground.
Generating a DTM from drone data requires classification of the point cloud to identify ground points, followed by filtering to remove non-ground features. LiDAR data is particularly well-suited to DTM generation because the laser can penetrate vegetation canopy to reach the ground beneath.
Common uses for DTM:
Earthwork volume calculations (cut and fill)
Drainage and hydrology analysis
Contour line generation and topographic mapping
Agricultural field grading and slope analysis
Archaeological surveys (revealing ground features beneath vegetation)
Which should you use?
ℹ️ Note: DEMs and DTMs are single-band rasters, meaning each pixel contains one elevation value rather than RGB color channels. They are not currently supported for upload in Stitch3D. If your workflow requires displaying elevation data alongside your point cloud and orthomosaic, the elevation attribute in the point cloud file settings provides this directly in the Viewer.
What is a thermal raster?
A thermal raster is a single-band image captured by an infrared thermal camera rather than a standard RGB camera. Each pixel represents the surface temperature at that location rather than a visible color. Thermal rasters are displayed using a false-color palette — hot areas in red or yellow, cool areas in blue or green.
How thermal rasters are created: Thermal cameras such as the DJI Zenmuse XT2 or FLIR Vue Pro are mounted on a drone and flown in the same grid pattern used for photogrammetry. The resulting thermal images are processed into a georeferenced thermal orthomosaic using the same stitching workflow as RGB imagery.
Common uses:
ℹ️ Note: Thermal rasters are single-band and are not currently supported for upload in Stitch3D.
What is a single-band raster?
A single-band raster stores one value per pixel rather than three (as in an RGB image). DEMs, DTMs, DSMs, and thermal images are all single-band rasters. Multispectral imagery captured by cameras with additional spectral bands beyond visible light is also typically stored as separate single-band files or as multi-band composites.
The most common single-band raster types you will encounter in drone workflows:
Orthomosaic vs. point cloud — when to use each
Orthomosaics and point clouds are complementary outputs. They are often generated from the same drone flight and used together in the same Project.
💡 Tip: Upload both an orthomosaic and a point cloud to the same Stitch3D Project for the most complete and professional client deliverable. The orthomosaic gives clients a familiar visual reference; the point cloud gives them the interactive 3D data they can measure and navigate.