Movement on the surface of a pool

    Feb 04 2011 | 4:10 pm
    Hi everyone, I'm chewing over a potential project for the summer that would involve extracting information from a swimming pool. The idea is that participants create music and localized visual projects by splashing. I'm thinking that performing motion tracking on the waters surface from above might be one way to achieve this but I'm unsure how effective it would be. The project is aimed at young physically disabled participants and the idea is that their x-y position in the pool is observed from above and that when they create ripples or splashes that this will create audio and localized visual projections that vary depending on their position in the pool. At the moment I just looking for ideas and objects that might be worthy of exploration. Having underwater microphones is also a possibility. Many thanks in advance!

    • Feb 04 2011 | 4:11 pm
      sorry for the duplicated post, I wanted this in both the jitter and max/msp boards! :)
    • Feb 04 2011 | 6:46 pm
      I think you you play around with a combination of frame to frame delta and also a find edges you should be able to get what you want.
      Also, we're all on the same board now. I don't think cross posts are necessary really
    • Feb 04 2011 | 9:37 pm
      Yeah I was thinking frame to frame delta, I've used that before and the jitter cookbooks offer pretty sturdy patches. I worry that the visual projections on the surface of the water will create a feedback loop. I'm guessing that infrared lighting and cameras with the infrared filters extracted might offer a workaround but I have not got a huge amount of experience in that regard.
    • Feb 04 2011 | 9:42 pm
      infrared cameras are virtually the same to use as long as your light source is bright enough and you've lined of the frequency of the optical filter and the emitting frequency of the light source.
      I've never pointed a projector at water before but are you sure you're going to get the effect you want? Wouldn't it end up on the bottom of the pool?
      perhaps you can project a mirror image onto the ceiling or something.
    • Feb 04 2011 | 9:51 pm
      Thats certainly a possibility. In my mind, seeing it from the POV of the participant, you would get colour playing off the surface of the water around you, not distinct images - just streaks of abstract colour localized around you, thats kind of the effect I'm thinking of. This is all very much up in the air at the moment, I haven't even seen the pool yet! ;)
    • Feb 04 2011 | 10:02 pm
      what about dropping out the color of the pool, so only the participants and the water medium are seen?
      also, if seen from above, light sources from the four corners might show the ripples better. Just make sure all equipment is properly protected to insure against accidental electrocution of any kind.
    • Feb 05 2011 | 3:27 am
      wow, that sounds awesome.
      Maybe once you have some color info from the pool, you can specifically eliminate that color from what's coming out of the projector (it sees and tracks blue to bluish-white but only spits out purple, red, green etc.) This should eliminate most feedback problems. With some timed fades of your camera input/colors you should be able to jockey the input/output. And I think that even though a lot of the projector light energy would go to the bottom of the pool, it would still look good, and any local splashing or disturbance would make it look really dynamic and cool.
      Maybe if the participants also had attachments to their arms (like styrofoam "fins" or something) that would give the camera something else to pick up and the projector something else to project upon.
      Really interested in what you come up with, and whether it works well without infrared cameras. Not that they're bad (in many cases absolutely necessary) but if you can use more standard stuff for something, that's always good...
    • Feb 05 2011 | 4:10 am
      the first thing what comes to mind is that there are a shitload of different numbers which you can generate by measuring the _distance between the persons. 3 values from 3 persons, 6 from 4 persons and so on.
      the speed and direction of their movement is also relatively easy to capture and to transform into useful data for audio and video.
    • Feb 05 2011 | 4:15 am
      "I've never pointed a projector at water before but are you sure you're going to get the effect you want? Wouldn't it end up on the bottom of the pool?"
      he will of course stray mirror power onto the water.
    • Feb 05 2011 | 5:05 am
      I'd probably hack a pool cleaner to follow swimmers around and cheer them on.
    • Feb 05 2011 | 7:13 am
      seriously, what about having the water covered by styrofoam balls?
    • Feb 09 2011 | 7:20 pm
      The styrofoam idea is really interesting, thank you all so much for your ideas!
    • Feb 10 2011 | 2:57 pm
      ping pong balls would be better. a matte smooth surface good for projection.
    • Feb 20 2011 | 5:39 pm
      In terms of tracking the participants in the water, I need this system to be as rugged as possible. Predictability is a big thing when working alongside some participants with special needs and want to minimise random actuations of sound and video material. If we are filming from above, in infrared, projecting images onto the pool that are localised to participants, what would be the best ways to extract useful location and movement information? As mentioned earlier frame to frame delta is a possibility, any other objects or techniques that anyone can think of? Ultimately it would be nice to get good sturdy x/y coordinates from each participant and data regarding the level of movement expressed by each participants i.e, you splash a lot, you get a lot of localised video and sound, you don't splash a lot, you don't get much of either. Could cyclops offer good results for tracking? I downloaded the trial several months ago but foolishly didn't have time to use it that much and consequently I'm not that familiar with all of its capabilities. cv.jit? Is frame to frame delta the best way of seeing and evaluating levels of movement? Many thanks for you kind support and brilliant ideas!
    • Feb 20 2011 | 10:27 pm
      Broke: As a noob here, I can't help you. I've still got my water wings on with a slow leak.