A colleague of mine just returned from overseas where he flew a UAV as part of an independent exploratory project. He did so with permission and also engaged directly with local communities in the process—as per the guidelines listed in theHumanitarian UAV Code of Conduct. He subsequently sent me this aerial video footage of a camp, which he recorded using a DJI Phantom 2 Vision+.
The analysis of aerial imagery for humanitarian & development purposes is an active area of research at UAViators. He thus kindly gave me permission to share this footage with colleague Matt Shroyer so that we could explore the possibility of creating a mosaic and 3D model from the video. Incidentally, the image below is the highest resolution and most recent satellite image available of the camp on Google Maps. As you can tell, the satellite image is very much out of date.
And here is the mosaic, which Matt kindly produced by taking hundreds of screenshots of the aerial video footage (click to enlarge):
A close up:
We then explored the possibility of creating a 3D model of the camp using the screenshots and SketchFab. The results are displayed below (click to enlarge). The numbers are annotations we added to provide relevant information on the camp. Perhaps in the future we’ll be able to add photographs & videos (captured from hand-held cameras) and other types of data to the 3D model.
Click here to view the 3D model, fly-throughs, and more.
Comments
Yes indeed, thanks Azhar. My colleague only captured aerial videos because he was working on a video documentary. It would certainly have been easier to work from images and use the Pix4D app. On that note, am hoping the app gets updated for the Phantom 3 very soon.
Another option (for Phantom 2 V+) is to use Pix4D's apps (Android and iOS, free).
It will do auto-take-off.. fly in grid... auto-landing... then the images (not videos) are stored in the SDcard with geotag included. We can then feed it to Photoscan as usual.
Awesome thank you!
Hi Max - SketchFab is the repository for models, but the model itself was created using Agisoft Photoscan. Once all the still images were extracted from the video, it took about 2 hours for the software to align the photos, create a dense point cloud, and finally create a mesh and texture for the model.
great results, i apologize for the prior comment.
And what app did you use to make the model? You say SketchFab but that looks like a repository for models, not a modeler itself
THat is amazing! How long did it take you to generate the model?
-
1
-
2
of 2 Next