Exploring Creativity and Technology

View and comment on LOLcloverfield

So I just uploaded this video to the “When Cloverfield Hit” web video contest… Top three videos get judged by Cloverfield director Matt Reeves, and grand prize winner receives $4500. My video stars me and my cat, Kenobi, in a parody of both Cloverfield and the whole LOLcats internet subculture.

I just found out about the contest last Sunday, so I had to work fast as the deadline was only a week away. But now that I know the capabilities of Final Cut Express (see my initial review of the software here), I was confident I could complete the video in time. I wrote the script on Sunday night, took some footage with my digital camera on Monday of the skyline from the building I work in, and also shot most of the footage of myself and the cat Monday night. It was a mix of using my handheld video camera for shots where I needed more detail, like of the cat against green screen (in this case a bright green plastic tablecloth from Party City that I threw on the floor), and using my digital camera’s video features, which have a cool grainy quality to them that I thought would add to the hand-held camera look. I wish I had taped down the tablecloth to the floor before trying to coax my cat to play on it, but I wasn’t thinking – so I had a lot of shadows to deal with when keying out the background. Plus the cat kept trying to play under the tablecloth, of course, or scrunching it all up. I did manage to get a few good seconds of him in monster mode, which I looped in the final film.

I think it’s awesome how Final Cut Express allows you to easily mix different video formats into one project. I had no trouble mixing/compositing layers of video cam and digital camera footage – no settings to worry about. As far as keying out the background of the cat, I wound up using a combination of chroma and luma keys, plus I messed with the gamma settings to equalize the green/shadows a bit, allowing the keys to work better. Only had a few minor spots of green left, and those I took care of with the Garbage Matte filter and also hid with some cloud layer generators. Hey, it’s far from perfect, but it will do for a web-quality video.

Spent the rest of the week mostly editing, visual effects and sound, plus a couple extra shots were needed (realized I had to make the room look messed-up a bit after the big shake-up at the end). I used Apple GarageBand to generate the sound effects – some of them were a couple layers of different sounds, and the “meow-monster-roar” sound is my own voice doing the meow with a ton of filters on it. I also used GarageBand to record all of the dialogue, which I did as one track, exported to iTunes, and then imported into FCE. Once in FCE, I put the track into a separate “sequence” and cut it up into individual master clips saved to the browser. Then when I switched back to the main sequence, I could pull the clips down into the timeline and match to the scenes they belonged to. For the helicopter reporter’s voice, I added a vocal transformer effect in GarageBand that shifted my voice deeper. Also added some reverb and echo to the dialogue tracks in GB. I used some of my original dialogue from the actual video I shot for certain parts, and for these I added reverb using FCE’s audio effects.

Editing went smoothly, but then I had a lot of trouble trying to render my final movie. It would get halfway through and then crash, leaving an unplayable file. Even when I tried to do a straight DV render without compression, it would make it about 75% thru. This is where I had to slow down and do some technical research. Busted out the old FCE pdf manual and googled some stuff online. Found out I was basically overloading the renderer with my tons of audio tracks and also I had a couple of separate sequences linked to the main sequence which had a lot of video layers and effects. I had to break down and learn about the orange and red lines across the top of the Timeline… These lines tell you that sections of your movie need to be rendered in order to play in real time when previewed. And these can also be a problem when rendering for output. So I went through, selecting a few clips at a time and doing Sequence>Render Selection>Video, on both my main sequence, and for each of the other nested sequences. (By the way, using nested sequences really helped out a lot with compositing – I used them for the news clip scenes, since I had to distort the entire scene to fit over the shot of the laptop, and also added the BadTV effect). I also did the “mixdown” render on the audio tracks. All of this pre-rendering creates rendered files somewhere on your hard drive that are linked to the reference clip in the timeline. If you then decide to further edit the clip, the associated render file will be deleted, and you may need to render again if you get an orange or red bar above it.

Once everything was rendered (and it didn’t take as long as I thought), I then exported the video as an MPEG for the web. Finally, it all worked smoothly. Then it took a couple hours to get the thing uploaded to the Cloverfield site – the upload kept timing out or something, but it worked in the end.

So please check out my video if you get a chance, and you can also comment and vote (from April 22, 2008 thru May 13, 2008 ) if you register at their site. Thanks!

UPDATE! The contest has ended and my video didn’t win the popular vote, though it was Most Viewed for a while and was also chosen as one of ten Staff Favorites. Hopefully Matt Reeves had a chance to check it out, since his staff seemed to enjoy it! The links to the contest site no longer function, but I’ve posted the video on youtube at http://youtube.com/watch?v=qcHi_OqS6b8, or check it out right here: