Creating 3D models often requires hours of labor and knowledge of complex 3D modeling software. There is no way around it; if you want to create a 3D model of a specific object, you have to search for the model, download it and tweak it, pay a 3D modeling expert to create it, or spend numerous of hours learning software such as Blender, 3D MAX or Google SketchUp to create the model yourself. Simple models such as Earth, a building, or anything rectangular may not take a lot of time to create, but when creating complex models (e.g. a buffalo), the creation process becomes more grueling.
These three articles link 1, link 2, link 3 which deal with 3D printing and using Microsoft Kinect to scan the physical models to convert them to 3D models, got us thinking about using Kinect to create 3D models for Augmented Reality applications.
Tony Buser explains how to use Kinect as a 3D model scanner to create a 3D model in this video: 3D scan cleanup project. We have followed his scanning procedure to create 3D models. After several experiments, we have identified a way to use Kinect for 3D model scanning more effective and efficient. Below is the breakdown of our procedure:
Prepare the model to be scanned. We used a wooden buffalo (Figure 1) and placed it on a “lazy susan” (rotating circular tray, placed on top of a table to aid in moving food on a large table). Rotating the buffalo with hands would create an inaccurate and deformed scan, so to get the most precise scan, we used the lazy susan. While rotating, you need to pay close attention to the rotating speed. Too slow or too fast will result in a deformed 3D model.
- Before you install ReconstructMe, make sure you satisfy the hardware requirements (see device compatibility matrix for graphics cards) and have the necessary software installed. After you checked for software and hardware requirements and installed ReconstructMe, follow these instructions on how to use it properly.
- Scan the model. Make sure that the model you are scanning is positioned a minimum of 40cm from the device and placed in an area of 1 square meter.
- Once you are done with the scanning, you will be asked to save the model. The file is in OBJ or STL format and will need to be touched up.
Obtaining the scanned 3D model generated via Kinect and ReconstructMe is the first step to creating your own 3D models. There are more steps to follow to have a completed 3D model. For instance, the scanned 3D model may have some missing areas (e.g., holes), rough surfaces, extraneous surfaces, and lack colors. Fixing the model and preparing it for the final use will be the subject of the second part of this tutorial, which we plan to publish by the end of November 2012.
Here is the video we produced to help you visualize the scanning process: