Have you ever considered initiating yourself in photogrammetry without any experience? Part 2

Now with drone capturing!

As mentioned in my previous article, I come from an aerospace background, with just basic knowledge about drones and photography. I got suddenly involved in Reality Modeling, more specifically as an engineer consultant for ContextCapture, and of course without a clue about this or any other photogrammetry software.

After a couple of attempts, I succeeded to capture my two ???motorbikes with DSLR camera, and I am finally happy with the performance and results, judge yourself at:

Now, I decided now to go bigger and try drone acquisition!

First step? Have a quick look at the ??guidelines and the support kit from the company. Nothing very complex, oblique capture path around the infrastructure with 60% overlap between pictures, adding some survey points and constraints. For reference while planning my capture path I used the following suggestions.

??Batteries charged? Backpack ready? Time for a motorbike trip to the outskirts of Madrid. I was looking for very textured infrastructure, free of trees or obstacles, where I could fly the drone without disturbing anyone. I found a nice abandoned railway bridge, easy and safe to fly by. It was a 130m long bridge, so, I had to plan my capture carefully and efficiently having only one 15 minutes flight battery available. I particularly paid attention to a couple of points of interest to get closer and use these as survey points and scale constraints. Additionally, I set up and fixed my capture settings of the camera, avoiding any inconsistency during the sequence. 3,2,1, takeoff??.

I did 7 parallel walkarounds following the bridge, 3 on each side (at different heights) and 1 from the top, and captured 17 pictures during each path. This gave me a total of 123 georeferenced photos that, once at home, I imported into a new ContextCapture project. 

As the platform identified the sensor size and focal length of the pictures, I proceeded directly to the first Aerotriangulation to have the pictures properly located. I then added 4 survey points and 2 length constraints to make sure my model is accurately scaled.

I then proceeded to the second Aerotriangulation to apply the changes and visualize the results on the generated quality report. This report gives us detailed info about photo position uncertainties, calibration results, scene coverage, and quality measures on Tie Points. 

Being proud of my first results, I could not wait to process the production of the model. For this, I had to select the area of interest and subdivide the work into tiles to make it easier for my laptop’s RAM. 

Selecting the production settings as “standard”, I could easily create a .3mx model to visualize and analyze within ContextCapture Viewer. 

Concluding, I can say the whole process was intuitive and smooth. Without any photogrammetry experience, just considering best practices guidelines, I could easily capture this bridge with my personal drone. In my opinion, the most important part is taking the time to ??plan a safe and efficient flight capture path around the infrastructure, avoiding over capture, and beware of the drone’s battery consumption. And don’t forget to take measurements for scale constraints before leaving the scene!

Next challenge? I’m curious about capturing a telecommunication tower??!

Are you ready to try it yourself?

Feel free to see my latest projects and models at: