NewTek TriCaster TC1 (2 RU) - Manual (Page 326)

P a g e | 310
D.2.2 LATENCY AND YOUR AUDIENCE
“Latency” – what’s that you ask? In this context, latency refers to the time it takes for the video signal to
transit one part of the IMAG path, from camera lens to final display. Latency is usually expressed either in
milliseconds or in video “frames” (typically either 25 or 29.97 to the second.)
As we said above, really a little video latency is not a bad thing as viewed from a surprisingly short distance
back into the audience. This is just as well, since for all practical purposes a little latency is also unavoidable.
Even so, as long as audio and video are in sync at your seat , only a rather ‘significant’ degree of latency will
be objectionable – unless you happen to be very near the stage. For those in the front rows, a few extra
frames of latency may be rather disconcerting. (It’s true that IMAG was conceived primarily for the benefit
of those further back, but if the latency is too obvious for those nearest the front it can be disconcerting
distraction.)
For this reason, it’s desirable to keep video latency to an agreeable minimum – but put away any notion of
‘zero latency.’ Not only would this require bending the la ws of physics, it would be a bad idea.
Even before considering minimizing latency in the device chain, acknowledging that there is always going to
be some latency calls for some creative thinking with regard to practical staging. For example, if you design
your IMAG layout in such a fashion that those in the front row are unlikely to be able to see the screen(s)
without lifting their eyes from the onstage talent, they are extremely unlikely to notice a small amount of
latency.
D.2.3 LATENCY AND YOUR NEWTEK SYSTEM
NewTek systems excel at IMAG, but inevitably constitute just one (albeit critical) link in a chain of devices. It
is common for each device to contribute a little to the combined total latency for the system.
Your NewTek syste m’s portion of the total latency is well within acceptable standards for devices of this sort
– roughly between 1 and 2.5 frames (the actual amount can vary slightly within this range, depending on
several factors).
For example, suppose a ‘video frame’ su pplied by a camera arrives at an input one millisecond after a different
frame has been sent to the output. Obviously the new arrival must wait its turn; it can’t be transmitted until
the correct duration for its predecessor has elapsed. Thus the newly ar rived frame must ‘wait’ almost one
entire frame, until its turn comes.
How can you achieve the lowest possible latency for the switcher’s section of the IMAG chain? One trick is to
‘Genlock’ your cameras to the switcher output (see Sections Section 3.6, Section 3.13 and 8.2.2). This
bypasses input time-base correctors, ensuring latency is consistent during switching operations. For
genlocked sources, consider disabling
Frame Sync – see Section 8.1.1.
Hint: You can actually assess the latency of your pipeline fairly easily. Run timecode directly to 1) a monitor
and at the same time 2) through the switcher to a second (identical) monitor. Take a photograph that
encompasses both monitors, and compare the timecode shown.