Make Yourself Transparent

by Adam Rokhsar

You are watching a family home movie. The movie is drawn out of points instead of pixels, and each point is colored the same as in the original footage. The height of the point — how much it is above the flat surface of the movie — depends on how bright the color is. The closer to white, the higher the point.

When the angle is right, you can see the images in the video clearly.

The processing works in real-time on movie or camera input. 

I used Max/MSP/Jitter to apply the movie as a texture to an OpenGL plane in three dimensions, and performed a running average on the original movie and on the color data to smooth out the results.

by Adam Rokhsar

A video of Barack Obama’s speech after his election in 2008 is the basis for this visualization experiment. The color is removed from the original video, and the difference between the pixels in each frame is used to map the video into three dimensional space. The larger the difference, the more the pixel jumps from the video plane.

by Adam Rokhsar

I used the Horn–Schunck method to measure optical flow and mapped the pixel displacement to the z-axis of a video playing in three dimensional space.

The results are noisy, but you can see the value per pixel jump when the camera is moved quickly or when there’s a lot of motion.

by Adam Rokhsar

Essentially this is a visualization that when look from the side, reveals information about color intensity, and when looked straight on, allows the viewer to see a version of the original video.

In part 2, the dimensions of the video are larger and I scaled the mapping to height to make it clearer when viewed from the side.

by Adam Rokhsar

I mapped color intensity to height in 3d dimensions, making a real-time, living graph of the RGB values of this family home movie.