An open source 3D Scanner made with Raspberry Pi
On the other hand, the solution has some inconveniences:
- the laser’s calibration is a critical factor for the outcome, therefore if the scanner’s structure is prone to get flexed, the centering and the verticality on the rotating plane of the light line may move, with a consequent misalingment and the absence of overlapping among the surfaces detected by the single laser;
- in the case a single laser is used, an absent or wrong calibration may determine deformations in the model resulting from the scan;
- the settings concerning the video camera’s position with respect to the plane’s rotation centre and the correct calibration on the horizontal plane are critical for the attainment of a correct scan;
- if a horizontal surface of the object is found at a height that is greater than the centre of the video camera’s optical system, it is not acquired;
- horizontal surfaces inside the object (as an example, the bottom of a glass) cannot be hit by the laser beam and therefore are not acquired;
- if the object has a surface having different reflection coefficients, the video camera – since it has set a single threshold value – will be blinded in the points in which the laser is reflected the most and will see darkness in the points in which the laser is absorbed;
- with difficulty the STL file of the scanned item will be suitable to be directly used for the printing, therefore the scan will have to be processed by means of 3D processing software.
As for the technique with the video camera only, the advantages are:
- it does not require lasers, nor their calibration, it is therefore insensitive to the lasers’ positioning and alignment;
- the hardware is much simpler;
- it better shoots the surfaces having different light reflection coefficients.
At the same time, it must be said that the software for this kind of scanners is typically born for systems that consider the rotation of the video camera around the object, and not of the item around its axis, therefore:
- a background is needed in order to avoid that the original background (that is a fixed one) is framed as a reference; the white colour seems to be the most suitable one, since it diffuses the light. The part under the scanner’s plate must also be covered;
- a good lighting of the object to be scanned is required;
- it is not suitable for objects having a black surface, or a very reflective one;
- it must be avoided that the object’s shadows are shown in a too pronounced way, since they would remain fixed and the program would understand that the camera is not moving;
- in order to help the scanner, it is needed to put some coloured references on the plate (or on the item, but there it would cover some details), so that they are taken as landmarks; since they are moving, the program will think that it has been the camera to move;
- in the photographs, the object must be shot as a close-up, therefore if it is a small one you need to draw the camera nearer, possibly drawing the landmarks to the centre;
- the photographs have to be shot with a small angular displacement between one and the other;
- it is not suitable for objects having a too uniform surface, or a too uniform colour;
- it requires an image reconstruction software, that is often expensive and a demanding one from the point of view of the hardware resources required from the computer on which to execute the frame processing.
Ultimately, there is no technique being better than the other one, but each one is more indicated for a certain kind of objects to be shot, and for certain scan conditions.