In part 1 and part 2 on this topic, I said project sizes for edit are getting larger and larger due to never saying “Cut!” I have tried to emphasize that this isn’t just a rant about my needing to manage storage.
This time, I’d like to detail a concrete example of the impact of long takes on a project—how a lack of “cutting the roll” meant that what the director desired didn’t end up in the finished project. (Some details have been changed to protect the innocent.)
This was a project with many moving pieces. The production was shot on location. Post went from offline to grading and then to finishing, all at separate locations. I was doing the finishing.
The director had his hands full both directing talent and appeasing clients during the shoot. For a series of scenes, the camera rolled constantly while directions were given, positions reset and takes redone. Some takes were 10 minutes long.
(After the project was delivered, I was able to watch some of the clips. The recording was stopped probably because the media was almost full. I never heard anybody stopping the roll.)
Offline proceeded as normal. A proxy workflow was used, which certainly helped with the amount of footage being used. The director supervised after the first rough cut and followed all the way through picture lock and on into grading.
As grading wrapped, there was a hint of problems to come. When you render out graded footage, there are two options. You can either render out the full length of each clip or just the length that’s used in the edit plus some extra frames before and after: “heads and tails.”
Heads and tails give the finishing editor leeway to trim shots a few frames here and there if needed. But only as far as the length of the heads and tails—typically a second or two.
If possible, I prefer to get the full clips back. It gives me more options during finish in case there are any fixes I need to do. It’s not a deal-breaker if I don’t get it, but I always ask.
In this case, because of the extreme length of the clips, the colorist wasn’t about to render out full-length clips. The amount of render time would have been extreme—exceeding the budget and the time allotted for the session. Shorter clips with heads and tails were delivered.
And here’s where it all fell apart. The edit included several scenes with time remapping of clips. A clip would play at normal speed, then would run at a high speed for several frames, and then play at normal speed again.
Since picture was locked and mix was already done, my job was to match, frame for frame, the offline. I used all my tricks to verify that the XML I got back from color matched. Everything was perfect, except for most of the time remapping shots.
The clip frames leading into the effect were fine, the frames trailing were fine, but the actual sped-up footage didn’t match. This wasn’t a consistent error.
I noticed those clips that had a constant speed change in the original edit (180 percent, for example) had fractional speeds of 179.7 percent on the XML edit. Try as I might, no amount of adjusting the clip—either by speed or by changing its location in the sequence—would solve the problem.
Next time, figuring out the problem and why it was caused by never saying “Cut.”