In January 2012 our working group got one of the first Samsung SUR40 (Microsoft Surface 2.0) that were delivered to customers in Germany. We had, therefore, the opportunity to migrate existing Surface 1.0 software developed in the working group of Prof. Dr. Harald Reiterer and moreover test the Surface hardware extensively. What concerned us most was the less reliable touch and tag recognition compared the the Microsoft Surface 1.0. Applications that worked very well on the first version of the Surface led to unexpected application behavior on the Surface 2.0. As consequence, it result in worse user experience when interacting with such applications. To get a insight why the tracking is worse than on a Surface 1.0 unit, we installed the Microsoft Surface 2.0 SDK. The SDK comes along with sample applications such as the RawImageVisualizer. This application, as the name suggests, displays the raw camera image as output to the user (see Figure 1).
As the image exposes the tracking needs to be very sophisticated to separate interaction such as touches or tags from the inhomogeneous background image. The white dots visible in the camera image are grease spots that, furthermore, disturb the tracking. In contrast to the Surface 2.0, the Microsoft Surface 1.0 had a black and homogeneous background image provided that it was calibrated correctly. We couldn’t believe that Samsung and Microsoft are satisfied with such a camera image based on the experiences with our Surface 1.0. Therefore, we decided to call the Samsung Service Hotline to issue a “defect” hardware. A few weeks later, two nice guys showed up with an original Samsung SUR40 calibration board (btw. board is immoderate; calling it calibration poster would be more appropriate, as it is a thin and paper-like material). Nevertheless, they followed the steps proposed by the SUR40 Calibration Tool (see Figure 2).
It took about 1 minute to succeed all calibration steps. First the display needs to be clean and without grease spots. Then the calibration board (white side downwards) needs to cover the display completely. Pressing the enter key starts calibration. An audio signal occurs if this step is done (active sound output). Next the black side of the calibration board needs to cover the display. Again, pressing the enter key progresses with calibration and signals audio when completed. Now comparing the raw camera images before and after calibration (see Figure 1 before calibration and Figure 3 after calibration) one can see that the image changed from inhomogeneous to a nicely black and homogeneous background image.
As a consequence of the Surface calibration all of our applications did benefit from a better touch and tag recognition. The applications are more reliable than before. We have to state, however, that the tracking of the Surface 1.0 is still superior to the SUR40 (Surface 2.0). So if you run into similar problems such as flickering tag recognition or unreliable touch tracking a calibration may put things right.
