This project was inspired by/utilizes the techniques of datamoshing, an indie trend of 2009 that has found a resurgence (with me). The particular effect I am using exploits the nature of video compression. In short, videos are compressed into I-frames and P-frames. I-frames carry the full image while P-frames carry only the motion between I-frames. In this project, the compression is exploited to produce only P-frames, and that motion is then applied to an incorrect starting image. This pushing to the limit of the codec was my starting point- what can we learn when we bend the systems governing video? I also thought the collapse of camera/actor/scene motion to a single type of generic motion was very interesting. To this end, I've used the P-frames from various films and applied them to uniform starting images to better understand the composition and structure of motion. It's akin to injecting a color tracer to better observe cellular action. Clicking any animation below should open up a larger version (possibly only marginally larger, I've kept sizes down for the purposes of this page but I have full size/full color versions available upon request) and I will add more films as I perform more studies!