Interlacing
Also known as: interlaced video, 1080i
Splitting each frame into two sets of alternating lines recorded a fraction apart, a broadcast technique no modern camera uses.
Updated
What it means
Interlacing halves the data per field. A 1080i stream carries fifty or sixty half-pictures a second rather than whole frames, which was how broadcast fitted motion into limited bandwidth.
The two fields are captured at different moments. Anything moving appears as a comb of alternating lines when the frame is viewed as a whole, which is the signature of the format.
It is the opposite of Progressive scan, where every frame is a complete picture. Modern cameras and screens are progressive, and the i in 1080i is the only place most people meet it.
How it works in practice
It arrives from old sources rather than new ones. Archive footage, broadcast material and old camcorder tapes are where an interlaced clip comes from in practice, and all three of those are getting rarer every year.
Deinterlacing is the conversion. Done well it reconstructs whole frames; done badly it halves the vertical resolution or leaves combing in the moving areas of the frame.
Converting on import is the simplest approach. Transcoding an interlaced clip to progressive before editing means the timeline never has to think about fields at all, which is the point of doing it early.
What to watch out for
Scaling an interlaced clip produces a mess. Resizing a clip before deinterlacing it mixes the two fields together permanently, and and no later pass separates them out again cleanly.
Field order can be wrong. The two fields have an order, and getting it backwards produces motion that jitters forward and then back again, which looks like a Judder problem and is not actually one.
It survives in metadata without surviving in the picture. A file flagged interlaced that actually holds progressive frames makes applications deinterlace something that in fact needs nothing done to it whatsoever.
How RawClip handles it
Interlaced sources are converted to progressive frames before analysis, since every later step assumes whole pictures rather than fields. Action cameras and drones never produce it, so this applies to archive material and old tape transfers rather than to anything from a current session.