HD Video Editing Ain’t As Easy As It Should Be

I’ve just finished editing the video for a GlobalPress 2009 keynote speech about digital radio given by Tensilica’s CTO Chris Rowen. Now I’ve been editing video so long that I started with Super 8 film (with sound) and a Guillotine tape splicer. That’s a long time. I was editing video on early Pentium machines using Matrox’s Rainbow Runner, a 2-board MJPEG hardware video-editing rig. Things are much easier now, at least for standard-definition video. I use Sony’s Vegas Movie Studio 9, a recent upgrade, and it’s just terrific. Easy to use. Intuitive. A pleasure to work with.

However, I’m editing on an “older” Dell GX270 PC with a 1-core Pentium 4 and an AGP video card and HD video from our Sanyo HD2000 camcorder, which uses a Sanyo-specific H.264 codec, drives the poor computer into apoplexy. There’s just not enough computer to handle the camcorder’s 1920×1080-pixel, 30-fps video—much less the 60-fps video I could shoot. The single-core Pentium 4 and the 5-year-old Nvidia AGP video card with no hardware decoding assist simply can’t play the Sanyo’s video stream, which is kind of amusing when you consider that the $600 camcorder does this task without complaint. But of course, it has a dedicated hardware video decoder so it shouldn’t choke on its own video stream.

Note: If you have any suggestions as to how to make our old video-editing computer more HD-friendly, please leave a comment. I’m interested and am looking at state of the art AGP video cards (or what was state of the art before AGP became “obsolete technology”) with hardware decoding assistance.

For now, because I’m targeting the Web anyway, I transcode the Sanyo HD stream into an SD stream and then edit that. For the keynote shoot, which you’ll find here, there were two cameras. The event organizers had a “pro” SD camcorder in the back with pro sound and I sat in the front row with the Sanyo HD2000—handheld for 30 minutes! The Sanyo’s image stabilization worked really well. So I ended up with a DVD-R with SD video from the event organizers and my HD stream from the Sanyo.

First, I had to rip the DVD-R. I used my old standby, Nero, now in version 9. It made quick work of the DVD, but I didn’t really care for the converted video. I also had the problem of converting the Sanyo HD2000 HD video stream. Sanyo supplies a special version of Nero 8, and truthfully the off-the-shelf version of Nero 9 also handled the conversion, but I wasn’t happy with the converted Sanyo video stream either so I went looking for something better.

What I found is something I’m pretty happy to recommend. It’s a free package called Quick Media Converter from cocoonsoftware.com. Quick Media Converter made quick work of both video streams and, whatever it did, it did a great job of converting the audio. I had not planned to use the audio from the Sanyo HD2000, and didn’t because of the excellent sound from the rig set up by the event organizers, but I could have after Quick Media Converter did its thing.

With two SD video streams, I started editing by laying down both SD video+audio tracks in Sony Vegas Movie Studio 9 and then I aligned the tracks using the displayed audio waveforms as visual clock tracks. This is a decades-old editing trick related to something called “wild” or unsynchronized sound and it works very well in Sony Vegas Movie Studio 9. You can get the two video tracks aligned within a frame of each other. Because the skew between the audio and the frame start is different for the two cameras, you can’t get the alignment exact, but a fractional frame of residual skew at 30 fps isn’t noticeable. Once the two video tracks are aligned, you mute the audio track you don’t want and then simply take care not to screw up the alignment of the video tracks during the rest of the editing process.

One thing I must tell you about Sony Vegas Movie Studio 9 is a plug-in feature called CineScore Studio. It’s slick. You select a stretch of video, activate CineScore Studio, and it makes music based on a selected musical theme with an intro, a middle part, and an ending—automatically and in seconds. Now this is what computers were made for. No more ripping CDs and mixing sounds from clip files. It’s slick, but then I already said that. But it is. Really.

It took a day to edit the two video tracks, add intro and outro graphics, and drop stills of the slides into the mix. Then a “quick” 2-hour render and I had an SD MPEG-2 file of the speech. Unfortunately, with the “quick” Internet connection here at Tensilica, it only took four hours to upload the file to viddler.com and another few hours for Viddler to compress the MPEG-2 file into Viddler’s format. But it’s now online and viewable.