Drones have made capturing aerial imagery over large areas easier than ever before. We often find many pilots are eager to push the boundaries of just how much data they can squeeze into a single dataset.
At Propeller we have seen an ever increasing number of pilots pushing the limits of their drones and cameras to capture more larger maps at higher resolutions than ever before and we have managed to process it all.
Challenges in processing large sets of drone data
The most obvious consideration here is that more photos equals more work for a computer in order to chew through all of those pixels.
In general, reconstructing 3D models requires a great deal more memory, storage, and processing power to ensure that the machine doing the work can not only hold all of the resources it needs to create your deliverables, but also finish the job in a reasonable time frame.
Better with Propeller
As sites get larger, naturally so do the files containing your final output data. Through Propeller you can provide your team and clients with easily accessible viewed through a browser. This eliminates the need for expensive computing hardware to chew through hundreds of gigabytes of GeoTIFFs and LAZ files or the worry about file compatibility with your GIS software.
Without further ado here are some of the largest sets that have been processed on the Propeller platform.
Goldfields , Western Australia
Covering an area of over 536km2, the Goldfields project represents an area three times larger than the entirety of Washington, DC. It’s been processed and visualized by the Propeller Platform as a single dataset.
This data enabled users to examine land tenements over extremely large areas without the need to access a powerful workstation to open hundreds of gigabytes of image files.
Earlston , Scotland
Pushing the bar for the largest number of photos in a single drone survey, Team UAV captured almost 14,000 images for a client over 17 flights covering an area of 5km2. Processing jobs of this size would be prohibitively time consuming on most computers however we were able to generate all outputs within 48 hours of the survey’s 180 GCPs being matched.
McAlinden , Western Australia
A project by the team at Ninox Robotics, stands out as the project with the largest pixel count we have ever processed. Generally, the computational intensity and size of outputs scales with the pixel count in the provided imagery, to give you an idea of the size of this project:
- The survey contained 10,500 images from 12 flights.
- Each image contained 7360×4912 pixels, over 35 megapixels each.
- The final area spanned more than 246km2.
- The reconstructed point cloud included more than 3.5 billion points.
- The final orthomosaic weighed in at 200 gigabytes.
While a several hundred gigabyte orthophoto would be impossible to open on the average personal computer, ortho images of any size load easily in our Platform.
We are yet to reach the upper limit on how much data can be processed in a single set and we’re confident that we have a way to go before we find it and always welcome the challenge.
It’s worth noting that abnormally large jobs do not fall under our standard pricing model, but if you have a large site you would like flown and hosted in the cloud, then get in touch with our team. We’ll be eager to work with you to figure out the best way to deliver this data to meet your needs.