Describe the bug

When making point cloud from azure kinect, I find a precision problem. You can see it from the picture blow. Notice the color of face from one side was mapped to the wall, while on the other side, the wall's color mapped to the face.

To Reproduce

My approach to make point cloud:

1. using depth_image_to_color_camera
2. depth_image_to_point_cloud

Expected behavior

transformed depth align perfectly to color image

Logs

Screenshots

Desktop (please complete the following information):

• OS with Version:
• SDK Version:
• Firmware version:
  Ubuntu 18.04
  k4a sdk: 1.3.0
  firmware: color:1.6.102 depth:1.6.75 sensor config:6109.7

Additional context