Establishing vehicle positions is super important when performing video analyses, and PhotoModeler (the king of forensic photogrammetry IMHO) just gave us all an early Christmas present that makes the process MUCH easier while maintaining the scientific rigor we expect from the team.

Importing and moving point clouds in PhotoModeler

One of the most commonly used methods in my video analysis toolbox is reverse projection photogrammetry. Very briefly, the process typically involves determining the camera location, focal length, and distortion characteristics. From there, you take an exemplar vehicle point cloud and move it around (transform) the coordinate system until it aligns with the pixels that correspond with the vehicle. A good match looks something like this:

Historically, I would solve for the camera in PhotoModeler, then transform the point cloud in CloudCompare, import that into PhotoModeler, check the fit and iterate until I was satisfied. But, I’m not easily satisfied, so that process was time consuming. However, the team at PhotoModeler is always willing to listen to the community and I (and several others) suggested they eliminate the need for the CloudCompare step, allowing us to transform point clouds directly in PhotoModeler. As of this month, the feature is now a reality. This video shows a few ways the new tool can be implemented, and I’ll record a video showing my technique and uses before long.

It's worth noting, programs like 3ds Max and Blender allow for point cloud transformation and can be excellent photogrammetry tools, but they fall short of PhotoModeler in the camera solve (position, focal length, distortion). With this release, we now have a one-stop-shop for scientifically robust reverse-projection analyses incorporating modern technology, point clouds.

If any of the above sounded like gobbledygook, Lightpoint’s running a photogrammetry course for recons in March. We can get you up and running on the above, it’s not that hard, promise. Second, if you need exemplar point clouds for this process, we have ~1,000 beauties scanned with Leica's RTC360 right here.

Happy Holidays, I appreciate you all!

Lou Peck
Lightpoint | Axiom