Augmenting with Optical Flow

by paulshen @ 12:16 am 3 March 2010

1 Comment

  1. Hi Paul – here are the group comments from the crit.

    Basically, your starting point is a re-creation of Body Paint by Memo Akten, which also connects Lucas-Kanade optical flow to the Jos Stam fluid simulation: . I really like your experiments in applying optical flow to cultural artifacts — whether to analyze, or to augment. Good work for a short project; how could you push this into more original territory?

    love the idea of showing derivatives that aren’t apparent, and it looks really great. the connection is very visible. In regard to Body Paint, memo’s project, i think the more fluid implementation that you used is more effective at carrying through the gestural quality of movements, and their transience. I think the body paint project doesn’t quite get that ephemeral quality. His Reincarnation and Golddust projects are similar as well. (He authored the MSAFluid lib you’re probably aware)

    Really mesmerizing. I like the use of the fluid and the colors. One thing I would suggest is to maybe blur the start of the flow particles. Right now you can see the particle dots when/where the movement occurs (in the first and last videos…in the drop video, I didn’t notice anything like that). The thing lacking in the first two for me is that you cannot see the visual input just the output. This is enchanting (especially in the first one), but at the same time, showing the input would make this piece more interesting as well. So, for instance, for the first video, if you projected the flow image on a screen, you could use the shadow of a person to determine the flow vectors.
    But, very, very cool!!! –Amanda

    Overlaying the fluid simulation with the slow-motion liquid drops was really beautiful. I really love the augmentation of existing movement, though I was less inspired by the input from the video feed.

    Cool, I like how you used to to analyze movement in video. Ballet’s particles can be much brighter.

    It’s really nice with the ballet footage, although the dark colors aren’t quite effective. Try making them lighter and more pastel. It might be really interesting to compare different types of dance or movement, and present both the videos with fluid simulation layered, and just the fluid simulation.

    I think the ballet one for me was the most effective because there was more of a link between the real world and the augmentation. The first two were certainly nice to look at, though. I think if you harnessed some of the strengths of each and combined them, you could have a really amazing project.

    I kind of wish you used lines to have a sort of pseudo-mathematical visualization of how they move – like circles being drawn, or something, that provides a bit of analysis about the movement.

    Comment by golan — 6 March 2010 @ 7:13 pm

This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
(c) 2016 Special Topics in Interactive Art & Computational Design | powered by WordPress with Barecity