This page is designed to answer questions concerning the creation and issues surrounding my piece I Am Sitting In A Video Room. If you haven't seen it yet, please click here.
An homage to the great Alvin Lucier, this piece explores the 'photocopy effect', where upon repeated copies the object begin to accumulate the idiosyncrasies of the medium doing the copying.
Please visit here for more information about Alvin Lucier's piece, and you can listen to the complete piece here.
Full words: I am sitting in a room different from the one you are in now. I am recording the sound of my speaking voice as well as the image of myself, and I am going to upload it to YouTube, rip it from YouTube, and upload it again and again, until the original characteristics of both my voice and my image are destroyed. What you will see and hear, then, are the artifacts inherent in the video codec of both YouTube and the mp4 format I convert it to on my computer. I regard this activity not so much as a demonstration of a digital fact, but more as a way to eliminate all human qualities my speech and image might have.
How long did this take?
I began on May 27, 2009 and completed the 1000th upload on May 27, 2010. It was not particularly labor intensive, but required constant interruption with my other work. I did not do it consistently, and there were entire months that went by where I didn't upload a single one.
Did you upload them by hand?
I did upload them all by hand. I would download the latest video, rename it and save it, then immediately upload that renamed video to YouTube (with pasted information). YT would require between 5-10 minutes to process the new video, then I would download that one. So I guess it was a pretty arduous process. If I were to do this again, I would write a bot in C++...
Why does the video get shorter as iterations progress?
YouTube automatically cuts off the last one or two frames from an uploaded video. While this doesn't make a difference for a single upload, the successive accumulation of these cuts eventually adds up. The first video is 42 seconds long, while the last video is only 38 seconds!
Why does the video get out of sync in the first 100? Why does it get corrected thereafter?
As with removing the last few frames, YouTube also adds a few frames of (silent) audio at the beginning of each upload. Again the accumulation of iterations pushes the audio out of sync. I didn't realize this would be a major issue until about Room 130, where I decided if I didn't fix the audio, it would eventually be removed entirely from the project. So I corrected the sync in Final Cut Pro, and then re-corrected it every 100 iterations. The final version of the audio is worth the corrections!
I thought digital copying was lossless. Why does the video quality get worse?
When copying files, the information is lossless; i.e. no matter how many copies are made they are all exactly identical to the original. This is just bit-jockeying. But what I am doing here is transcoding the video twice per upload. Each time the video gets uploaded to YouTube, it gets translated to the .flac/H.264 video codec -- a process that makes the video smaller but loses some of it's information. Then this loss happens again when I translate it again to mp4 format on my computer. Each time the pieces of information lost are saved on subsequent versions, and accumulated throughout the process.
Oh my god that audio is terrifying! What's going on?!
It gets complicated, but each translation of the audio goes through a process known as "Fast Fourier Transformation", which breaks the audio down into more-easily-parsed tones. Again, however, this process loses a bit of subtle information, which is accumulated. It sounds like water because water actually resonates/fluctuates at many random frequencies, and the FFT analysis unintentionally mimics this as well.
6/1/2010 | 6:02pm | Patrick | liddell lucier alvin codec resolution filter repetition digital noise loss quality copy repeat | Direct Link