Notes on Editing Improvised Video Footage
Working with completely improvised dialogue was probably similar to editing a documentary. The story has to be assembled and rewritten from what was captured.
Adobe Premiere’s sync function works well about 75% of the time. It’s more convenient to use than Red Giant’s PluralEyes but not as reliable.
I always back everything up to multiple devices and at least once to the cloud. If a fire or robbery happens, I’ll only lose a day of work. Same if DropBox or Google Drive loses everything.
Part 1 – Surveying the footage
For each episode, I opened up my outline on a 2nd screen and created sequences in Adobe Premiere for each of the story beats.
Then I went through the footage, cutting clips and filing them away on the sequence related to their beat. This was easier on some episodes than others. As I watched, the story changed and I rewrote the outline and reorganized the sequences, sometimes 7 or 8 times, until I got to the end of the unlogged footage and had everything sorted into a beat sequence.
This was the most mentally exhausting part of the editing because it’s slow, it involves a lot of “writing” and it because at this point I was watching footage that was slow and having to watch it for very specific issues of content, then decide where to put it, and then possibly rewrite everything based on what I was seeing.
When I was done, I had six or seven sequences in Premiere, one for each beat of the story.
Part 2 – Putting together the beats
Once I had the raw footage sorted into beats, I would work through the individual story beat and massage that. These sequences would have anywhere from 30 seconds (for an intro or outro beat) to 15 minutes of material.
I would comb through them and start to piece together a rough story for the beat, moving any material that I didn’t like or didn’t want to use to a “discards pile” sequence so that I could go back to it later without searching through one of the raw clips.
This took a long time as well, and it also triggered some rewriting of the beats. Occasionally it would spawn new beats as I realized that more than 1 major thing was going on in the sequence.
At the end of this part, my beat sequences would still be rough, but they would have a few minutes of the best clips for a given beat and a lot of discarded clips that I left on the timeline (just in case) but separated from the good footage.
Part 3 – Rough Assembly
The next step was to drop the various beats into a master editing sequence, which would eventually become the rough cut. This first assembly would clock in at a more manageable 15 to 25 minutes, depending on the episode.
Then I would start to watch the rough assembly, making bigger cuts as I went through it, and sometimes rewriting the story beats again. Sometimes one beat would have a good transition to another beat, so I would want to get the good transition lined up. Or sometimes the story just played out differently than I imagined, and the order had to be re-arranged.
This part was more fun than the previous part, but still involved a lot of writing work.
After I had the story solidly in place, I would go through the sequence over and over, cutting away the fat and the things that didn’t work. I cut beat repetition (when the same joke or story beat or tactic was used again without any variance in emotion, tactic, tone, physicality, etc.). I cut extra space at the beginning and end of clips. I cut out frivolous words or false starts to sentences (although I left some of these in). And I cut anything that just didn’t work or seem to fit the story.
After doing this for a while (hard to say how many hours this part took, somewhere in the range of 3 to 8), I would export a rough cut.
Part 4 – Getting feedback
I wrote a separate post about getting and interpreting feedback here.
Part 5 – The final polish
This is the most fun part. By this time, watching it 50 times in a day wasn’t so bad because the story was flowing and the only things that remained to be cut were things that I loved but had to go anyway.
I tackled the feedback and made any changes based on that. I didn’t take all of the notes and I paid special attention to notes that came up for more than one viewer – if multiple people are saying the same thing, it’s probably not a matter of personal taste.
Next, I worked on the pacing. I developed a specific style for WFM, using a lot of jump cuts and quick cuts, stacking the lines of dialogue on top of one another. It’s not naturalistic, but I think it works well for the medium and keeps the viewer moving through the story. On the web, 5 minutes is a long time, and I wanted to keep viewers on their toes and make sure that there was enough novelty from moment-to-moment that would tickle their brains and keep them engaged.
Did I succeed in that? I’m not sure, but I did my best. And it wasn’t just about pandering to a web audience. I really like the style I used and it was a good way to manipulate a long story in a short time frame.
Once I was done the final polish, I didn’t lock the picture. Instead, I slept on it for a night and came back to it the next day to give it another pass. If I watched the whole thing without making any changes, I locked picture and sent it off to the colorist.
If I made a lot of changes in the final pass, then I would sleep on it again because I knew that I might look at it differently the next day. If this happened three days in a row, I locked it anyway and shipped it off (otherwise I’d still be tweaking some of them…).