Interesting question. You might have more luck asking over on the caspercg forum where the devs are active.
Typically you would record ‘clean’ and ‘dirty’ feeds to be able to edit around captions. If you have lots of ISO recording capability and are going to be switching the show in post then you either need to record key/fill of CG as ISO source too, or be able to have CG behave ‘as live’ after the fact.
The former I’ve done – either by recording chroma-keyable single ISO, key&fill as seperate ISO recordings or, where captions are only in bottom half of screen, have the TD build a split – with bottom half of raster being fill and top half being key. (Eg in skiing, where there can be two running clocks as they race close together, this method ties up 2 x record points instead of 4).
The latter I’ve done in the dim & distant past on game shows. The computer generating the GFX would record time of day with its data and, for post, the Gfx company would bring their computer in and we’d feed it timecode from the source VTR. (Back in tape edit days…)
•Should* still be possible – for source TOD you’d probably have to use audio timecode routed to an audio output (Avid has a ‘data’ track which is maybe used for CG data by someone?). But, as I say, Caspar forums best place to ask – do post a link here if you open discussion there so we can follow!
Where this would be especially useful (CG data seperate from pictures) would be all those sport events where only 1 feed is available to the world.
Good luck.