How I Edit
This is how I edited
Friend of the Devil
I set up the two Canon A460s on tripods flanking our "stage." Once we were tuned up and ready, I started the left camera then the right. (I will refer to the camera in terms of the camera point of view, so the left camera is the camera on the left from the cameras' points of view, or the opposite of stage right.) Then we played our song. Then I stopped the right camera then the left. It went well enough with the only major flaw being my poor timing at the beginning, so we didn't bother with a second take.
Later I downloaded both the AVIs and named them right and left. Since Bob and I spent more time looking in the direction of the right camera I chose that one as a base to edit from. After copying the files from the camera I opened the AVIs in Quicktime Pro. The fist step I exported the sound track to a WAV. I chose to export the sound at twice the bit rate and twice sample frequency so there would be no losses. Also since I chose exactly double on both parameters there should have been no upsampling interpotation error that can mess up the sound. I called one sound track right.wav and the other left.wav. Then I exported each in the photo-jpeg format which essentially treats each frame as an individual jpeg. This allows better editing with less chance of buffer errors when you export. The exported video tracks were about 3/4 Gb each. Note that there was a bunch of junk at the beginning and end but I left it all intact until the end. I could have cut it out before, but once the sound files were separated from the video I wanted to keep the same start and finish time to ease putting the sound back in at the end.
So then basically I would find a bit of footage from the left camera that I liked and cut and paste it onto the right track which I was using as my base track. The way I did this was be looking for visual cues that could be seen on both tracks that marked a particular instant. I used the transistion from the walkdown to the D chord as one such point. So I went frame by frame until I saw the frame where my finger landed on the fret, and the hit "i." I did the same on the other track and hit "i."
Let me digress for a moment. In quicktime you can mark the beginning of a selction by hitting "i" for in, and the end of a selection by hitting "o" for out. Other programs may or may not follow this convention.
Then I did the same at the end of the piece that I wanted to select from the left camera. I found a frame where Bob's mouth was just starting to open at the beginning of a word. This is the real hard part and the real heart of editing so it is in sync. In the movies they use those boards that say act1 scene25 take14 camera B with the clapper. When they bring down that clapper it marks a point in both the sound recording and on the film and they sync from that point.
Quicktime has a neat feature called "add to selection and scale." If you've copied say 2 minutes of audio and the "add to selection and scale" and you selection was 1 minute of video, then it will crunch that 2 min of audio down to one and it'll sound like the alvin and the chipmunks. If you take 10 seconds of frames from one vid track and add to selection and scale onto the 10 seconds of frames in another video it won't have to do much work to make it scale. If you got it really close everything will sync up, If you added 10.5 seconds to a 10 second section it'll be off at the end of the clip.
Once I was confident that I had both begining and end of the edit marked right on both tracks, I ctrl-C from the left track and then "Edit-Add to selection and scale." then I started the video right before the edit and watched. If the sound tracks from each bit mesh up nicely then you have a good edit. If not undo and try again. Once I got it right to my satisfaction I deleted the extra sound track from the movie properties window. I repeated that basic process for each of the clips fromt eh left camera that I inserted into the vid. I usually get this wrong and have to try a second or thord time to get right.
At this point I made sure all the extra bits of the sound track (from the clips from the left camera) were deleted and I exported the video track to another photo-jpeg qt movie that I called mixed.mov. This flattens the image so you have only one video track instead of multiple overlapping vid tracks. I actually did the export a few times while editing because I have noticed that having too many video tracks increases the chances of an error during export.
When I was done with all of the edits, I went to work on the sound file. For the sound I used the excellent open source program Audacity. I opened right.wav. It was good that I chose to use the righ track as my basis since it is easier to make the two track align by trimming the start off of the one that was rolling first. So I also opened left.wav in another audacity window and did the crtl-a ctrl-c and pasted it into the same window as right.wav. I then set right as the right stereo track and the pasted in left as the left stereo track. Then I zoomed way in to the quiet early part of the sound and found on each track a clap that I had made. I used the timeline to measure the time between the claps on the two tracks and then deleted that same amount of time from the front of the left track. Then I saved the whole mess as a stereo.wav.
Finally I opened stereo.wav in a QT window and crtl-a ctrl-c then opened my exported mixed.mov file hit ctrl-a and then "Edit-Add to selection and scale." I hit play and it was okay at the start and got more out of sync as it went along. I went back to the stereo.wav file and noticed that the left track was about 2 seconds longer at then end and I hadn't cut that off. So I did delete the extra bit of the left track so it ended when the right track did. When I "added to selection and scale" this version of stereo.wav to the mixed.mov it was perfect. After that I cut off the nonsense at the beginning and end, then I exported the finished product using the H264 codec for the video. Note that EVERY time I exported I selected 10 frames per second (that's the speed the cameras record) and highest quality setting, and I let QT choose the keyframes, and I didn't limit the bitrate. The final export for the 5 minute video was about 120Mb and it took about an hour to upload to youtube.
I'd really love it if someone could lead us through the process of laying down multiple tracks so they can be mixed together.