# A tool that "traces" the path of a point in motion

**URL:** <https://discourse.opensourcedesign.net/t/a-tool-that-traces-the-path-of-a-point-in-motion/1554>\
**Category:** Interaction design\
**Created:** [February 21, 2020, 2:37am UTC](https://discourse.opensourcedesign.net/t/a-tool-that-traces-the-path-of-a-point-in-motion/1554 "2020-02-21T02:37:49Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![ahawn](https://avatars.discourse-cdn.com/v4/letter/a/22d042/32.png) [@ahawn](https://discourse.opensourcedesign.net/u/ahawn)\
**Post date:** [February 21, 2020, 2:37am UTC](https://discourse.opensourcedesign.net/t/a-tool-that-traces-the-path-of-a-point-in-motion/1554/1 "2020-02-21T02:37:49Z")

</div>

I’m a graphic design student currently working on my thesis project that deals with utilizing motion capture and gestures as a way to create animations and perform video editing tasks. Basically this would be a plugin of sorts that would pair with programs like after effects and blender.

Having little to no experience with coding, I quickly found that this wasn’t something I could easily build on my own , let alone in a few months time. The “real” product I want to showcase will be conveyed through video editing and overlapping animation overtop footage. However, when I display my concept I’d love to build an interactive installation that works as a working prototype of what this could be.

My idea is that I want to build something that “traces” movement in 2D space. Either with the use of movement a body part like an arm (this would be ideal - but as I understand this would probably require machine learning?) or with the use of a marker - like a brightly colored object that the webcam recognizes as the object to follow, like a glove. I’d like to be able to playback the traced animations in a loop - say record 5 seconds of movement and then the screen would playback the animation for a few cycles and then reset.  
I made a crudely drawn diagram to help explain what I was thinking below.

Any ideas on how to start building something like this for a total newbie?

 ![image](https://canada1.discourse-cdn.com/flex029/uploads/opensourcedesign/original/2X/1/1a47f7de777de1cb091bb0f3588b49a2d9761202.png)

---

<div class="post-metadata">

**Author:** ![jdittrich](https://yyz1.discourse-cdn.com/flex029/user_avatar/discourse.opensourcedesign.net/jdittrich/32/42_2.png) [@jdittrich](https://discourse.opensourcedesign.net/u/jdittrich)\
**Post date:** [February 21, 2020, 9:14am UTC](https://discourse.opensourcedesign.net/t/a-tool-that-traces-the-path-of-a-point-in-motion/1554/2 "2020-02-21T09:14:00Z")

</div>

> [@ahawn](#):
>
> Either with the use of movement a body part like an arm (this would be ideal - but as I understand this would probably require machine learning?)

For this I would probably look at VR gaming hardware and/or MS Kinect. They usually have SDKs which make such things easier, e.g. providing a skeleton abstraction of a person in front of it.

> [@ahawn](#):
>
> …with the use of a marker - like a brightly colored object that the webcam recognizes as the object to follow, like a glove.

I remember that friends of mine and I experimented with with several years back. I guess we used [processing and openCV](https://github.com/atduskgreg/opencv-processing) back then; it would be something like [brightest point](https://github.com/atduskgreg/opencv-processing/blob/master/examples/BrightestPoint/BrightestPoint.pde) but for some channel. Or you go for comparing images and draw where there is a difference to other images.