Prevu3D supports processing drone imagery to generate point clouds and mesh models. To ensure a successful reconstruction, it is important to understand the data formats and metadata required by the pipeline. By preparing your dataset correctly before upload, you will obtain sharper textures, better alignment, and higher-quality geometry.
Image Formats
Prevu3D supports standard aerial photos in the following formats:
JPEG (.jpg)
TIFF (.tif / .tiff)
Required Metadata
All images must include EXIF metadata with at least:
GPS position (latitude, longitude, altitude)
Camera orientation (yaw, pitch, roll)
Timestamp
Camera intrinsic parameters (focal length, sensor width, etc., if provided by the drone)
This metadata is essential for image alignment and accurate reconstruction.
Image Quality Requirements
The quality of the reconstruction depends heavily on the quality of the source photos.
Overlap (Very Important)
To align images correctly, your dataset should have:
Front overlap: 75–85 percent
Side overlap: 60–80 percent
Higher overlap is especially important when capturing:
Vertical structures
Complex rooftops
Narrow corridors
Vegetation
Long linear assets
Sharpness & Exposure
Ensure the following:
No motion-blur
No focus issues
Consistent exposure
Avoid extremely bright or dark images
Avoid low-light flights unless using a stabilized camera with proper exposure settings
Tips
Slightly overcast days often produce the most consistent photogrammetry results.
Flight Orientation: Nadir vs. Oblique
Both nadir (camera pointing straight down) and oblique (angled) imagery are supported.
Nadir is suitable for open areas, orthomosaics, and flat-terrain mapping.
Oblique captures façades, vertical structures, and complex assets with greater completeness.
By default, the pipeline assumes oblique capture unless otherwise specified in the configuration file.
Ground Control Points (GCPs)
Prevu3D supports optional Ground Control Points to improve geospatial accuracy and strengthen alignment.
Supported GCP Target Types
We currently support automatic detection of three target shapes:
Square targets
Diagonal (checker-style) targets
AeroPoints
If your GCPs use one of these types, the system can automatically detect them in the images.
GCP CSV File
If GCPs are used, you can upload a CSV file without a header row.
The CSV must follow the expected structure:
GCP name
Easting / X
Northing / Y
Elevation / Z
Here is a template as reference:
Configuration File (JSON)
You may also upload an optional JSON configuration file to specify:
The capture input type (
"default_oblique","default_nadir")The horizontal coordinate system (EPSG)
The vertical coordinate system (EPSG)
Optional geoid model for elevation corrections
Here is a template as reference:
This configuration file allows you to ensure correct georeferencing and alignment when working with survey-grade datasets or mixed sources.
Finding Your EPSG Code
When specifying the horizontal or vertical coordinate system for your dataset, you will need the correct EPSG code.
You can look up the EPSG code for your region or survey system using the following public registry:This platform provides a complete searchable database of all coordinate systems, including:
Horizontal (projected & geographic) CRS
Vertical CRS
Units (meters, feet)
Related geoid models
Area of use
Search by name, region, or keyword to find the correct EPSG code for your dataset.