top of page

Photogrammetry Research and Practice

  • emersonharrop
  • Nov 1, 2021
  • 3 min read

Updated: Dec 10, 2021

Before I started this assignment, I decided to go out on my own and practice taking photos of different items and putting them in software such as Meshroom and/or Autodesk Recap. I did this to prepare for the assignment so that I already knew stuff that would and wouldn't work.


Below are some of the stuff I decided to practice with and the results that I got.


This was the first object that I decided to use for photogrammetry


Here are the results of after I had put the images into Autodesk Recap, this model overall came out really well for a first attempt. There are a few problems with it like the texture on the side is slightly blurred and due to the lack of images I took underneath the bench, it didn't form the bottom at all. However that is something which could easily be fixed in a 3D software if I were to use it.

Also, another thing is that probably is an issue is that the back is darker that the front so that may cause a bit of an issue if this was going to be used in a project.



This tree stump was another item which I used to practice with.


Here are the scanned photogrammetry results, originally I used Meshroom for it however, it didn't pickup the tree trunk which was the focus of the images for whatever reason. So I decided to use Autodesk Recap which picked up and modelled the trunk that I was trying to get. With this model, there doesn't seem to be too many problems with this one aside from the fact that the top isn't really formed properly so maybe I needed to take more photos of the top as I only had 1 or 2.


This was another item that I saw and decided to see if I could capture its data.


With this I only used meshroom and not Autodesk Recap. This one didn't come out well at all. I think this was because I didn't take enough photos of the post box or that it was too smooth and/or too reflective on some points so that it had difficulty forming the model. The top also didn't come out too well due to the fact that I didn't capture any photos of the top of the post box.


With this one I tried to capture the bench, tree and concrete slab all in the same model.


In this one, the tree came out really really well, the slab of concrete came out pretty decently the top one came out better than the one on the bottom. The bench however, came out awfully which I believe it was due to the surface of the material was too smooth and shiny in the photos for it to be formed properly.


During my research, I tried using both an SLR camera (image below on the left) and the camera on my phone (below on the right) to see if there are any differences between the quality of the captured data. But what I discovered is that there isn't too much of a difference between the two and that the quality of the model depends mainly on what the subject is and the quality of the images that are used. The main difference between these 2 is that the one on the right (photos from my phone) was done through Autodesk Recap and on the Left was done through Meshroom, making the one on the right look better. This means that I might lean towards using Autodesk Recap over Meshroom, only problem with that is that it can take quite a while for the subject to get generated due to the queue to the server unlike Meshroom.



Comments


bottom of page