[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

KeyKode data in VITC

I've seen some discussion about various schemes for inserting keykode data
into the VITC and even some observations that it hasn't been used to any
widespread degree. There is a reason for this. It really doesn't have any
practical use if you are generating some kind of log files. In fact, it has
the potential for creating a mess of problems.(I know that this is
tantamout to heresy given the love affair that the equipment manufacturers
have with the idea.) 
Let's say that you insert the keykode into the VITC of all of your dailies.
After on line editing, you then can use an "Afterburner" to make a video
burn-in tape that displays the Keykode. Then you have some poor shnook look
at the first and last frame of every cut and write down the Keykode number.

Okay, let's eliminate the possibility of human error by designing a black
box that can read the VITC directly and generate a cut list. Okay, we'll
even assume that this black box can make sense of the mish-mash of VITC
during dissolves and wipes and that the data can be preserved in computer
generated special effects. You will _still_ end up with an incorrect
negative cut list. 
The reason for this has to do (as so many other esoteric problems do) with
the 3/2 pulldown. When an editor is cutting video, she is free to cut on
any of the 30 video frames in any given second. The problem is that three
out of five of those video frames do not represent a full film frame - they
only represent a fraction of a film frame. So if a given video clip starts
on the second field of a "C" frame (assuming field 1 dominance), it is only
using that film frame for1/60th of a second. The negative cutter cannot cut
a fraction of a film frame so that film frame in the film cut represents a
duration of 1/24th of a second. These errors will gradually accumulate
resulting in a film cut that runs longer than the video cut (even after
allowing for the 59.94 hz video field rate) and a track that drifts out of
When a cut is done on, let's say, an Avid using Media Composer and then the
negative cut list is derived from the EDL using Media Match (or OSC/R or
whatever), part of the algorithm used by these programs includes
compensation for this effect by adjusting the out points of cuts when the
discrepency between the duration of the film cut and the video cut hits
1/24th of a second. By the way, the same thing happens in reverse when a
cut is done with an Avid using Film Composer or a Lightworks and then a
video EDL is produced. Out points must be adjusted to keep the video
running in sync with the film cut. You can see this effect when, in a
series of dissolves such as in a montage, the EDL has incorrect match frame
edits in a certain percentage of the dissolves. 
So even if you only use an "Afterburner" generated cassette to give to a
negative cutter to confirm the accuracy of a computer generated negative
cut list it is only giving them raw data; before the necessary adjustments
have been made. The negative cutter would be better off with a tape of an
Avid or Lightworks dump (but only if the picture had been digitized at
24fps) with a Keykode burn in generated by the computer based on its
database of keykode numbers in the dailies. 
I, myself, use an Evertz 4025 (for all of its shortcomings) to generate
Keykode burn-ins on dailies and I use it to confirm the accuracy of the
keykode in the log files which are generated based on Time Logic's
connection to the Keykode scanner. Either one of these alone is subject to
error, but together they are unbeatable for accuracy. But once the Keykode
numbers are in the log files, having them travel along with the video
doesn't make a whole lot of sense to me.