Since DAVID version 2.2, it is possible to scan objects without the background corner (two planes) in the scene! However you can not scan by hand, but you will need a precisely repeatable laser motion, e.g. by using a linear or rotating laser drive or by projecting a moving thin line by a video projector.
From DAVID's point of view, the laser motion must start at the same position each time! This position in 3D space is called the Initial Laser Position. Otherwise the scan result will be distorted (may look funny though).
Since DAVID version 2.3.2, there are two possibilities to synchronize the Initial Laser Position. You can change this in the Advanced Settings, value Laserscanner→LaserPlaneCalculation→LaserMotionEstimation→ScanStartDetectionMode.
In Reference Scan and Repetition Scan, you have to click Start in the moment when the laser motion starts. Of course it has to start at the same Initial Position. This may be difficult to do, but useful when DAVID controls the laser motion over the COM port or receives the Start command from the laser motor controller.
If you choose this mode (default), it is not important at which time you click “Start”. After “Start”, DAVID does nothing but wait for the laser to appear in the camera image. When DAVID sees some laser light for the first time, this is what it regards as the Initial Laser Position! This is true both for Reference Scan and Repetition Scans.
There are two possibilities to do this:
In case 1, the moment when the laser light gets visible in the camera image depends on the objects in the scene! So this is not recommended.
In case 2, you have to make sure that there is something in the scene that the laser light immediately falls on to make it visible to the camera. Also you should not switch on the laser before the laser motion has reached its final speed.
There are two ways for DAVID to “know” the pose of the laser light plane in a Repetition Scan:
If your laser motion speed is not constant, you have to choose option 2. Otherwise, option 1 may give smoother results. Option 2 will work even when DAVID says “No regular laser motion detected” at the end of the Reference Scan.
In both cases, you can choose whether laser poses are to be regarded by their time or frame number. The latter refers to a counter of the frames grabbed by DAVID, and should be used only when you are synchronizing camera frames and laser motor steps. See Advanced Setting
During Reference Scan, both methods are prepared. For your Repetition Scans, you can choose the option in the Advanced Settings
Probably your Reference Scan was too short. DAVID distinguishes rotation from linear motion by the angle between the first and the last laser plane. The threshold is given by Advanced Setting
Anyway your Reference Scan should be as long as possible (covering not just a small part of the camera image) to get a precise estimation of the laser motion.
When you click “Pause” at the end of your Reference Scan, DAVID tries to determine whether the Laser Motion Estimation parameters are acceptable. This is done by comparing the poses of the measured laser planes to the poses where they should be according to the estimated parameters. There are two simple thresholds for the differences between measured and estimated laser planes:
If you switch on the Debug Window (in Advanced Settings), the text window will give you more information about what property of the Laser Motion Estimation is unsatisfying.
Also, you can set
True in AdvancedSettings, then you will see laser planes in the 3D window.
Both thresholds can be changed in the Advanced Settings (
MaxAvrgNormalError) , but rather than increasing the thresholds, you should try to improve your setup!
Probably your laser drive does not perform a smooth, steady motion with constant speed.
In DAVID there are filters that avoid that the calibration panels (and the area around and beyond them) are scanned. When Scanning Without Background, there may be an object where the panels used to be, or even further back. So in that case, in Advanced Settings, navigate to Laserscanner→Triangulation and set AllowPointsBehindPlanes to True and BackgroundFilterFactor to a negative value.
Now you can use Shapefusion to compare the Reference Scan result and the Repetition Scan result. Ideally, they should be identical.
Also, in the 3D-Window of the scanning page, you can check whether the background panels (which have been scanned during Repetition) lie exactly on the grid that is shown there, like in these 2 images:
And not like here:
Here is another example of what effects may occur when the laser motion or the Initial Laser Position is not repeated equally. It shows several scans of the same straight piece of wood, some with extreme deliberate errors to demonstrate the effect clearly:
Green: Reference Scan - Yellow: Repetition Scan without background, slightly imprecise - Light blue: Repetition, deliberate offset in Initial Laser Position (about 1cm, 1 second) - Violet (right): like light blue, but 5 cm - Pink (left): Repetition, correct Initial Laser Position, but moved laser only half as fast as during Reference Scan.