All, right! The moment you’ve all been waiting for: The Editing Process. This was by far the most ambitious part of this project, and I’m going to take you through the details of how we achieved some of the aesthetics that we’re going for with this video. We edited this project in Final Cut Pro 7.
First, let’s talk about the stop-motion work. We recorded a lot of images. Part of our strategy was to record both Canon Raw and Small JPEGs so that we could use the low-resolution image sequences as a kind of proxy to mess around with until we got the look we wanted. Then it was just a matter of swapping the JPEGs out for the Raw images and we were good to go. On a side note, FCP 7 won’t take Canon’s Raw format, so we used the Canon Digital Photo Professional Batch Process application to convert the Raw images to High Resolution JPEGs.
Once they were converted, we went into FCP 7 and made sure that the Preferences were set so that the still/freeze frame duration was only 1 frame. I cannot underestimate the importance of this step. If we hadn’t done it, we would be manually changing the length of the images from 10 seconds (I think that’s the default), and like I said: we recorded a lot of images.
So, once the images are all lined up, one frame at a time on the timeline, I created a new sequence, and nested the sequence with the single frames inside of the new one. The advantage to using this setup was that we could easily change the length of all the frames in the sequence by doing a speed change to the nested sequence. It was a little render-heavy, but much more convenient.
The look we finally settled on was to do a speed change and triple the length of the clip (33.33%), and added frame blending. That gave us a pretty look with a kind of liquid, dreamy quality to the sequences.
However, doing such long exposures (I think the shortest one was about 1″) left us with an issue: although the rigs were as secure as they could possibly be, this car as a very tight suspension and rides very low to the ground. We were unable to avoid some camera shake. So, it was left to the post production crew to tighten up these sequences in which the shake occurred and make them look solid. I don’t have a “before” version available right now, but I’ll tell you what I did to get the final look. The following steps refer to this video.
We settled on a system involving four layers of interactive matte-ing. Basically, we removed the “V10″ logo from the side of the car, and reconstructed it on top. The bottom layer (V1) was our background. This was composed of one of those nested sequences I mentioned earlier in the post: The original time-lapse images of the car moving through space, slowed down to 33.33% and frame-blended. Then, I opened up Adobe Photoshop. I went through the sequence, and frame by frame marked the ones in which there was insurmountable camera shake. Once inside, I used the Clone Stamp tool to totally remove the “V10″ logo from the side of the car in those selected frames.
The layer directly above this (V2) was the skeleton, or “edge” layer. This is the first step in reconstructing the “V10″ logo. This step used Adobe Photoshop as well. I looked back into the images sequence and selected 5-7 images when the “V10″ logo was lit from different angles. I compiled these images into an HDR image, and then cut it out. Then I took it back into FCP, laid it on top of the background, bumped the contrast and applied an extraction matte to the image. I messed around until I got something I liked, and then turned down the opacity so that the structure was ever so slightly noticeable.
Above this was the meat. V3 and V4 kind of act together to form the “meat” that goes inside the edges established by V2. What I wanted to see was a clean, sharp “V10,” that still allowed the light effects of the environment to play across it. So, V4 is a duplicate of the sequence on V1, the original, time-lapse sequence without any of the camera shake frames removed. We need the light in those frames, so they have to be there. Then, I changed the composite mode of this layer to Travel Matte – Alpha. This means that whatever is on V4 will show through the matte that I build on V3. It sounds counterintuitive, because it is, but that’s just the way FCP 7 does business. On V3 I duplicated the HDR that I used on V2, but the extraction matte I applied was governed by luminance, so that only when the light was strong enough on V4 would it pop through the HDR on V3. I think the effect is very successful! How about you?