Ben Farr Independent Learning: Photogrammetry
- by Benjamin Farr
- September 30
- in
Ever since our class last semester got to skype in with the guy who did photogrammetry, I've been thinking about a million different applications that that tech could have. When it came time to work on my final project, I wanted try my hand at using it.
A little bit of Background on me. I'm a Boy Scout certified ecology/conservation director during my summers. One of the largest parts of my job at my camp is teaching about the flora through their leaves. My most valuable teaching tool is passing around leaves for scouts to touch and see. Unfortunately, its very hard to do this when you don't have an instructor or leaves nearby. My project idea was to create a virtual reality environment in Unity that would allow people to learn about leaves without either of those things.
I originally tried to use an app on the iPhone called 123D catch, but ran into problems. My photos wouldn't upload to the site and when they would, the object came back looking more like Frankenstein's monster than a leaf. I visited online help boards to figure out what the problem was and soon discovered leaves just are not meant to be scanned with this app.
The problem with leaves is that they are 2 1/2 dimensional. Flat on two sides, with a little bit of a crinkle and almost no dimension where the two sides meet. I decided to try out using the structure sensor connected to the iPad in the innovation lab instead. This worked a lot better than the iPhone at first, because of its depth sensor, but also had issues. When I tried to rotate around around that point where the two sides meet, it completely lost the object and this time gave me bride of Frankenstein. I visited online help boards again for the app and discovered that I was doing two things wrong.
First, I had the leaf sitting upright on a white table, which was too reflective for the app. Second, even if the table were not reflective, it lacked detail for the structure sensor to latch onto.
I tried again, this time sticking the leaf with tape onto a purple binder of mine, and using sticky notes with different designs on them to help the structure sensor keep track of the area. The sensor had a much easier time, but my scans still broke at that meeting point.
My solution to the problem came accidentally. I tried quitting a scan halfway and accidentally sent it to be constructed. What came back looked like my leaf. So for the other five leaves, I only scanned twenty percent of their surface. The results looked fine.
COMMENTS