Tuesday, July 31, 2012

Fireworks: The Process of Making a Music Video... Part 11

Let the edit begin!

This is proving more challenging than I first thought.

After Jack Quick at LomoLab worked out a few issues, the footage from Day 1 is looking really great. In order to get rid of the choppy, jumpy nature of the first pass, we came up with the idea of disregarding the half-frames, losing 1 of every four frames in the end. The new workflow is working and the footage is much smoother now, so Big Thanks to Jack.

There was a roll of 160VC that was a couple of stops underexposed. The semi-automated process at LomoLab just couldn’t find the frames. I ended up re-rendering the MP4 from the film scans myself, using (of all things) Canon’s DPP RAW converter software. It has a trimming function that made this tedious process a bit more streamlined. There were a few other rolls that I re-rendered to get rid of jumps when the camera would skip a frame or two.

The first attempt with this was a couple of macro shots of the costume. I had taped a +10 diptor to the LomoKino lens and hoped for the best. The focus was surprisingly sharp. The underexposure, however, rendered the footage unusable, even after the re-render and colour correction in Final Cut. Bummer, but great proof of concept for later.

Another experiment that has proved to be really interesting was a sequence where I had Kyle walk backwards down a set of stone steps. We did 2 takes of this. I reversed the footage, superimposed the shots over each other and synched up the motion. The result is a trippy little shot that will be great in the final edit.

The Portra 400 has (so far) really held up well. The first day’s footage looks great. The film handles over exposure very well. I’m able to dial back the couple of shots that were 2-3 stops over and they still look good.

Day 1 started with the end of the song. No vocals here, just Kyle playing the guitar in a Chapel ruin. The plan of the ever-widening shot is looking great, but the issue is (as I thought it would be) the synch.

The first thought was synching the footage as it came back with the slowed down audio tracks that we shot with. The speed is, unsurprisingly, off. I got the slowed down tracks sped back up to normal speed and onto the timeline. The LomoKino films are rendered at 4fps and around 19 seconds long. I’ve been experimenting with speeding up the films to match the real time audio. 300% proved to be way too fast. I worked my way down to 200% at 25% increments. 200% seems to work, but since there aren’t any vocals to synch to and the instrumentation in the tracks makes the individual guitar strums difficult to discern, it has been a challenge.

The next batch of footage will contain vocal tracks. Then we’ll solve the speed issue.

I still contend that this can work. (Although I’m reminded of a line from The Ghost and The Darkness where Michael Douglas and Val Kilmer are talking about a lion trap… “Did it work?” “No, in point of fact, it did not. But I’m convinced the theory is sound…”)