FAQs
Read through our frequently asked questions below
When thinking about latency in Vicon real-time data, it is important to remember that small amounts of latency are introduced at every stage of the pipeline. To most accurately track latency, it is important to be able to measure latency end- to-end over a multi-computer pipeline— starting with the physical motion event, capture and processing by the Vicon system, receipt of the 6-dof sample by a client processor, through the rendering processor pipeline, and finally to the data display.
One of our customers involved in real-time virtual reality came up with a clever set-up to measure latency at every stage of their processing pipeline. This testing method is described in general below.
“We built a custom external timing device to capture the start of a motion event and track the resultant motion sample through the system pipeline. Our latency measurement scheme uses the external timing device together with a manually propelled pendulum to correlate the real and tracked motion. The timing device consists of a 100-ns clock, 6 latched data arrays, and 3 serial ports. The pendulum has an IR emitter on the swing arm and an IR detector on the base. Vicon markers are attached to the swing arm so that its trajectory may be tracked in real time. The clock is started when the swing arm passes over the IR detector. Since this is a known point in space, an identifiable event sample will be generated by the Vicon system. Then any of our software, running on any computer, can send commands via the serial ports to latch the contents of the counter as the sample propagates through the system. This allows us to measure the latencies between different stages of a multi- workstation processing pipeline. A photo-sensor attached to the display screen will automatically set latch 6 when triggered, allowing us to measure the end-to-end latency. The stored timestamps may be read back at any time over any serial port to get a list of latencies.
These tests have been run while tracking different numbers of markers and objects to determine how latency increases with the number of objects tracked. (With the Vicon system, the latency appears to increase in a linear fashion as marker count increases.)”