User Tools

Site Tools


Scanning with mirrors

please feel free to correct - the author is no native speaker


since version 2.1 DAVID directly supports scanning using mirrors.

With mirrors placed into your setup, you are able to

A: - capture more than one view of the object hit by the laser‘s lightplane


B: - to reflect the laserplane onto the object to get a almost complete allaround laserline on the object.


DAVID detects the reflection on the mirrors as a second, third or fourth object standing behind the mirror. For the mirror reflects little less light, than captured directly, and the limited optical precission of normal mirrors (with the slight double reflection of second surface mirrors) the meshes resulting from the reflections own less quality than the directly detected ones.

But to solve certain scanning challanges it is one and often the only way. Whenever you are not able to move the object or the background structure (scancorner) without changing the object`s surface it`s the only solution I can imagine.

Creating an allaround lightplane

Most of the current users of DAVID-laserscanner prefer to rotate the laser on a pivot up or below the cam. That means the laser hits the object and the detection planes in an ever changing angle. On the other hand: a linear moved laser never changes that angle (please do not misunderstand this as the „Intersectionangle“ David calculates out of the left an right line on the planes). S

DAVID is only able to detect one enlighted pixel per column of one frame in the campicture. If there are more than one, the brightest is choosen and others are ignored. If you use a mirror and hit it in an not rightangled way, you produce several lightplanes of which some are ignored. So the best way is to hit the mirror(s) rightangled all over the planes and as a consequence to move the laser in vertical: The Laser is reflected into its own plane. Its like using so much lasers as there are mirrors hitting the object from different points on the same plane.

Positioning the cam

In a normal setup the cam owns the position now our laser should have. Well, two things can not own the same position: So let‘s place the cam where the laser was before. This way we can get the intersectionangle of left and right line, that DAVID needs to capture usefull 3D data. So the whole setup should look like ( You might also use only one cam from above for simple objects):


As long as the laser produces more than on single line on the object or the wings, like this


twist and bend the laser till there is only one line as shown above.


To prepare DAVID to scan with mirrors, you have to change some settings available in the advanced settigs dialog, you can open clicking the advanced settings button.


The following assumes that you are using a rightangled closed detectioncorner on which you have fixed the mirrors;

1.Opening the path to


enable detection behind DAVID‘s detection planes.

2. Opening the path to


set the value of the thickness off your mirror on the left side in mm. AND: Opening the path to


set the thickness of your right mirror.


The scanning step works as simple as usual. Make sure, that the camsettings allow a detection of the mirror reflected laserlines. If you´ve switched of the background detection filter (see advanced settings) your scanresult should look like (For normal use it`s better to enable the backgroundfilter by setting its value to 0,2):


Your textured and cleaned result looks like this:



Switch to the SHAPEFUSION window and load your scan. Disable the texture view in the 3D window.


Now go to the list of the loaded files and click your single file with the right mouse button.


clicking “Reflect on mirror planes” DAVID splits your mesh in three parts and reflects the mirrored meshes to the position they were mirrored from.

If your calibration and your mirror-offset (see above) were set right, 2 of the three mirrored meshes should be placed quite to the right position.


The partial mesh resulting from the -X-Z area has moved to the -X+Z area (or +X-Z) Reflect this mesh on the mirror planes once again:


As the result all partial meshes are at their position.


Caution: The overlapping areas are very small. Do NOT use the Free Rotation Registration Mode - this will fail. Use only Fine - or Manual Registration Modes.

The finally fuse result:


Maybe you noticed that the mesh resulting from the -X-Z region is the worst and only very small , for the drectly hit mesh covers most. That´s why I use another angle of mirrorplane/detectionplanes giving back only 2 reflections and of all these possible ones 120° is the best choice, for it gives aspects covering the objects suface best.

But therefore you´ve to go back to your workshop and change some other settings……

to be continued…..

scanning_with_mirrors.txt · Last modified: 2014/06/06 18:38 (external edit)