How-To: A Simple Way to 3D Scan an Environment: Image not clear or warped
I am new to RealSense cameras. I have the D435i and have been trying to capture 3D scenes. When I display them in CloudCompare or MeshLab the images are very distorted or missing large sections. Even the article from Intel "How-To: A Simple Way to 3D Scan an Environment" shows what must be an artist's view of a room at the top and toward the bottom the scene that is capture following their procedure. Everything seems distorted or missing sections. Is this normal? What am I missing? This seems a long way from being similar to a colorized LIDAR point cloud.
-
Hi Alphatozeta If you are loading .ply point cloud files into 3D software such as CloudCompare or MeshLab, you may experience problems with the scaling of the cloud if the depth units scale of the program that the ply was captured with (e.g the RealSense Viewer) and the depth units scale of the 3D software that you are importing the ply into are not set to the same value.
The default Depth Units Scale of the RealSense 400 Series cameras such as D435i is 0.001 (m), or 1 mm. The 3D software should be set to the same scale value that the ply was captured at.
It is also worth bearing in mind that a snapshot taken from a static camera may not contain as much detail as a capture where detail has been added progressively in real-time by moving the camera around the scene and passing over the same areas more than once, like in the Simple way to 3D scan an environment guide. Intel demonstrated a similar real-time room scanning procedure from a RealSense equipped handheld device in 2015.
https://www.youtube.com/watch?v=Vpa8vKsapFk
Please sign in to leave a comment.
Comments
1 comment