Generating a #3D#Scan using #Autodesk#123DCatch
Generating a 3D Representation of an Object Using Autodesk’s 123D Catch
I was able to successfully generate my first scan using Autodesk’s 123D Catch and a digital camera. It took a couple of attempts but I was able to get it to create a 3D model (with decent resolution) with a few tricks.
My first attempted scan was an AppleTV remote. In order to generate the scan, 123D Catch needs a series of photos covering the entire circumference of an object. I started by setting the object on a dining table with common central lighting from a typical ceiling fan. I tried to place the object directly underneath the light source in order to get consistent exposure around all faces of the object but otherwise I did not attempt to manipulate the lighting too much. Autodesk’s tutorials recommend a range of photos (around 50 to 70 in total) that capture all sides of the object to be scanned.
To do this I took photos using a simple 16MP Nikon Coolpix at intervals of 15 degrees around the perimeter of the object. I took about 25 photos approximately three inches above the table by simply moving around the table and taking photos about every few inches. I then raised the camera up and rotate the lens down towards the object in order to capture about 25 photos of the top of the object for a total of about 50 images to load into 123D Catch.
After loading the files into an accessible folder on my computer I then opened the 123D Catch application on my desktop. The interface is fairly intuitive to use but there are also a few tutorial videos posted directly on the launch screen once you open the program that can help with a variety of categories (Shooting Tips, First Project, Navigation, etc.). You simply have to click on create project and then load photos. After you have navigated to the folder containing the photos of your target object, you simply select the photos and create the scan. One thing to keep in mind is that Autodesk processes the scan from their own servers. As a result, your firewall might prevent the program from accessing your photos. If it is unable to create the capture, you may need to temporarily turn off your firewall or grant it access in your firewall security settings. I was unable to successfully run a scan until I turned off my firewall while I attempted a scan.
The scan capture process takes a while (roughly 20 minutes for me). Unfortunately, my first attempt did not yield a coherent 3d model of the AppleTV remote on my first attempt. It picked up pieces of the background of the photos but did not yield any 3D model for the target object (the AppleTV Remote).
As a result, I decided to play with the scene to see if I could generate better results. First, I traded the AppleTV remote for a small wooden art figure with more distinguishable features than the plain Apple remote. Secondly, I removed the monochromatic mat that the object was sitting on and set the object on colorful drink coaster. And lastly, I placed three different color markings on the table around the object to help to distinguish the scene (two different colored post-it notes and a piece of blue painters tape). Since 123D Catch attempts to assimilate photos into a 3D representation by aligning occurrences repeating in the photos, the tape markings gave the program some simple color blocks to align when processing the images. I took approximately 50 photos of the art figure trying to approximately the same intervals, distances, and angles that I used for the AppleTV remote.
The result was my first successful 3D scan using 123D Catch. The scan itself is reasonable (but not great) resolution. The art figure was a fully completed mesh with no voids showing in the art figure although several areas were a very rough approximation of the geometry of the object. Areas that were occluded (or heavily shadowed) when taking the photos (such as underneath the arms of figure) were the least accurate representation of the object.
For the most part, it was a fairly simple and easy way to generate a decent representation of an object. The resulting file is an “.obj” file that can then be loaded in most 3D software programs like Rhinoceros and Sketchup.
-B
Source: http://www.123dapp.com/catch