Category: Video Sensing

[QUICKTIME (un)dress 320 257]
Inspired by issues of censorship and surveillance, (un)dress is a library of videos that examine the repetitive act of dressing and undressing. Twenty people were asked to perform this daily ritual in front of a camera. The video was processed live during shooting, causing clothing to be the only element recorded from the scene. The result is a series of anonymous yet intimate portraits ranging from the shy to the bizarre.

I moved away from the idea of the puppet emulating you and onto the idea of the puppet responding to you. Rather than communicating between worlds, the two of you come to occupy the same space. I created a series of exercises to work through a few ideas I had.

In “Excuse Me”, the woman becomes annoyed when you enter her personal space.

[QUICKTIME http://itp.nyu.edu/~kh928/archives/ExcuseMe.mp4 160 137]

In “Follow”, she refuses to leave you alone.

[QUICKTIME http://itp.nyu.edu/~kh928/archives/Follow.mp4 160 137]

This is an idea using a setup similar to Vincent’s TouchScreen as a way to control video playback. Touching the screen with fingertips allows you to play small movies at the location where you touch. The movies are small in size, not much larger than the fingertips. They are short in length & looping. The number of times you tap your finger allow you to select which file will be played. Then holding your finger for several seconds in the same spot where you tapped causes the movie to start playing. It plays as long as you maintain contact. If you drag your finger, it follows. Just imagine tiny little videos on screen, following your fingers around.

In terms of content, I have a body of short videos that I made a few years ago based on my subjective interpretation of fragments of other people’s dreams. They are meant to be presented small, in a many screens within a screen sort of way. I originally presented them floating around in a grid-like formation, but I’d love to experiment with a more organic, live presentation. In this case, selection of videos would be random, not user-controlled, so the combination of videos on screen would be constantly changing.

I’d like to use live television as the image source rather than live feed from a camera. TV images would be replaced by preloaded clips of video (of my choosing) with the same RGB value. So I could watch interpreted TV, having a similar color experience as I would with the original, but viewing completely different content.

Uh oh! I feel myself being drawn down a familiar path. So when I’m sitting here trying to understand all this stuff, this “Java” stuff and this “video” stuff, I think of the most basic thing I could possible want to do, figure out how to do it, and then move on from there. So in terms of thinking about tracking, the most basic approach would be to track one thing moving and make something else move in the exact same way. That’s where the puppets come in. I guess my instinct is to personify things in order to better understand them. Track something in real life, make something move on the screen. So I’m basically looking at how tracking parts of the body or the body as a whole can allow you to move characters on the screen. Finger puppets. Hand puppets. Body puppets. So far I’ve been trying finger puppets using the multiple rectangle tracker, but I think I need to switch approaches, because my current setup produces really terrible results. But it’s a start. Perhap I need to try identifying skin blobs. Funny how the what I think of as the simplest thing still ends of taking me forever.

Vision for the future: People in orange jumpsuits jumping around in front of the camera to make their characters move on screen. Full body experience.

Starting with the absolute basics of rectangle tracking, this names the location of the fingertip. Soooooo basic.

[QUICKTIME http://itp.nyu.edu/~kh928/archives/LocationNamer.mp4 160 137]

Another idea for a webcam. The thing I love about snow is how it allows you to visually track movement that has gone on in a given space. Who has been there? Where did they go? How did they move? I’d like to put a camera on a large open space shortly after a fresh snowfall. It could interpret movement of people through space (tracking dark objects) and output some sort of graphic similar to a map with footprints based on that movement. Or it could be set to capture images of the space only when it was unoccupied. (Could this be achieved by observing overall brightness?) These images could then be assembled into a movie, animating the changes made to the snowscape by the invisible forces that move through the space.

Experimenting with removing both background & skin.

[QUICKTIME http://itp.nyu.edu/~kh928/archives/FindSkin.mp4 160 137]

Electronic glass. Looking through a glass, window, viewer. Basically, you are altering the way something is being seen. Right now I see there being two approaches to start out with: doing a treatment of the overall image or working within the image and dealing with content. I keep returning to the idea of selective seeing/censorship, where someone is controlling what you can and cannot see. My idea would be an extension of Dan’s FindSkin code. Rather than coloring skin, I would like to see it removed altogether. How would this look when applied to an entire body? Instead of seeing a body itself, you would just see a cut-out moving through space, kind of like a paper doll. It reminds me of the film “Removed” by Naomi Uman. She takes a short piece of found footage pornography and physically scratches out the bodies on each frame of the film. So you watch the film just seeing these very scribbly figures moving moving around an untouched background. But what I’d like to do would be different in method and in situation. The method would obviously be removing skin through the use of the computer. But in terms of situation, rather than looking at something sensational like a sex act, I’d like to look at something really mundane like someone taking a bath. Because it brings up some interesting questions. Like, it’s not really creepy to watch someone brush their teeth, but is it creepy to watch someone bathe, even if you can’t actually see their body? Is it the exposure of the body or the act that is intimate? Another fun iteration would be to watch someone dress after removing the body & the background so that you see clothing materialize on an invisible body in an undefined space.

A basic idea to start out with for the architectural webcam. A view of the back of a large apartment building at night. A picture is taken each time a light goes out and then the images are assembled so you can see the progression of lights being turned off. Which lights go off then on again? Is there some point at which all the lights are off? What does it make you think about the comings & goings of the people inside? Very Rear Window.