>
Yes: I suspect trickery. Without a common genlock, there is *no way* to
switch that fast, it is simply impossible to define what the card should
do. Or, to be more precise, even if the 4-way switch on the composite
signal could work during the inter-field interval of one channel, that
would not coincide with the inter-field interval of the other channel,
so the next frame on either the old or the new channel will be damaged.
So FWIW my explanation is that the Windows driver duplicates frames
instead of delivering damaged ones. So you need to film 4 (or at least
1) source showing frame-accurate, or better, timecodes.