3D scanning a full size building
The very first time I saw ordinary television working, I was struck by a tremendous idea. "Look here!" I shouted. "If these people can break up a photograph into millions of pieces and send the pieces whizzing through the air and then put them together again at the other end, why can't I do the same thing with a bar of chocolate?
Why can't I send a real bar of chocolate whizzing through the air in tiny pieces and then put the pieces together at the other end, all ready to be eaten?"
-Willy Wonka, Chocolatier
After scanning a few people, some art, and even my cat I was approached by Arland Whitfield and his drone with the idea to scan something BIG.
The building we decided on was Holden Observatory on Syracuse University's campus, one of the oldest buildings in the area and at a little over 2 stories tall it was the perfect sized challenge to try and bite off.
Instead of attaching a depth sensor to the drone we decided to take the route of photogrammetry, which involves photographing the object from all different angles to later calculate the size and shape of the subject.
Essentially the software will attempt to retrace your steps as the photographer and from the photos recreate the objects that were front of you. The easier it is for the software to follow and understand the path of your camera and the objects you were looking at, the better the result of the model.isaac on left, Arland on right in a candid photo between flights
PROCESSFrom my experiments scanning regular sized objects, I knew there were 3 basic angles we needed to capture in order to get the scan.
- Low and angled up
- High and angled down
- Directly above and straight down
PHOTOS FROM SHOOT
Using that process we ended up with a lot of photos, over 400 in total! Here is a sampling of a few of the different angles we were able to photograph.
Software Point of View
After looking at all of the photos we took we narrowed it down to a hardy 274 photos to upload. After uploading it takes awhile for everything to get sorted, analyzed, torn apart and finally pieced back together together.
After we click go the software takes all of those photos, finding common and identifiable elements in each stitches together the objects in the scene.
This photo shows how the software deconstructs the different pieces of the photo before stitching them together. If you look closely you can see different parts of the Observatory!