Science SPORE Prize
November 2011

The Open Source Physics Project is supported by NSF DUE-0442581.

## Measuring g in simple free fall (incorrect values from fits) post and replies

Return to the Measuring g in simple free fall (incorrect values from fits) thread

strange fluctuations in basketball throw vx graph - May 12, 2014 at 4:09AM

lookang
190 Posts

hi tracker community, anyone knows why there are strange fluctuations from -4.8 to -6.2 m/s in basketball throw vx graph https://www.dropbox.com/s/gr3ehmk0sgxelha/wangyuxing_312_28_tracker.trz

my current answers are:

Question by wangyuxing 312 28 tracker: for the first 1 second of my video, the ball's velocity in the x direction fluctuates greatly, may I know why this is so? possible reasons are:

possible reasons are:

1. compression has a kind of rounding off error to create a sharp image of the basketball.
2. camera used has some unknown object smoothing causing this effect.
3. unlikely tracker software has a bug in the calculation of vx.

by the way, I made a video tutorial to explain the physics http://youtu.be/M0N5P0zgFaA without addressing the question, just general graph interpreting using tracker

<iframe src="//www.youtube.com/embed/M0N5P0zgFaA" frameborder="0" width="560" height="315"></iframe>

Post edited May 12, 2014 at 4:10 AM EST.

Post edited May 12, 2014 at 4:11 AM EST.

Post edited May 12, 2014 at 4:21 AM EST.

OSP@SG blog
OSP@SG Digital Library

### Replies to strange fluctuations in basketball throw vx graph

Re: strange fluctuations in basketball throw vx graph - May 12 2014 11:48AM

Douglas Brown
159 Posts

I've attached a screenshot that shows the first 12 positions marked with vertical lines so the horizontal positions are easy to see. It looks like the camera is not recording images at regular intervals. I think that many phones and maybe even tablets do this when they can't keep up a steady frame rate. I would try a different camera.

Hope this helps!  Doug

Re: Re: strange fluctuations in basketball throw vx graph - May 12 2014 12:17PM

lookang
190 Posts

Agreed with your analysis.
Thank you!

OSP@SG blog
OSP@SG Digital Library

Re: Re: strange fluctuations in basketball throw vx graph - May 19 2014 1:17AM

lookang
190 Posts

dear prof doug,

based on our student's understanding, it is not the camera but the compression process by iMovie that caused the weird unequal vx.

thought you should know :)

OSP@SG blog
OSP@SG Digital Library

Re: Re: Re: strange fluctuations in basketball throw vx graph - May 19 2014 11:47AM
Paul Nord
15 Posts

I've seen iMovie do some very strange things to videos.  It can make a frame-by-frame analysis impossible.  This even happens in come commercially produced video sources.  Google "Todd Frazier Loses Bat, Hits Home Run".  Those who have analyzed the motion carefully have found that there is a frame missing.  You have to work around that to correctly determine the forces involved.

Suggestion: Don't use iMovie on videos you want to analyze.

Paul

Measuring g in simple free fall (incorrect values from fits) - Feb 9, 2014 at 7:12PM
Zach Kost-Smith
1 Posts

We are attempting to make a simple measurement of g via a free fall for a high school class.  The values given by fits are consistently high (around 11-12 m/s^2 rather than 9.8 m/s^2).

Using the video of projectile motion provided on the website gives the correct value for g. Because of this, we doubt this is a problem with the software itself.

We have tried many things to get this to work including different cameras, orientations (of the camera and meter stick), different balls, locations, backgrounds, lighting levels. We suspect the video source, in particular, we suspect that there is something with the frame rate that causes this issue, but we are unsure. I even went so far as to extract the frames myself (via ffmpeg), mark them via an image editor, and fit the data resulting a value of 12.4 m/s^2 for g.

It looks like the problem is with the video or with the video import, except that the video appears to be fine and looks good after importing into Tracker.

Is there anything that I am missing regarding how to get usable video data into Tracker for quantitative results?

### Replies to Measuring g in simple free fall (incorrect values from fits)

Re: Measuring g in simple free fall (incorrect values from fits) - Feb 09 2014 10:01PM

Douglas Brown
159 Posts

Hi Zach,

One possibility is that the camera uses a CMOS sensor with a "rolling shutter" (see the Wikipedia article http://en.wikipedia.org/wiki/Rolling_shutter). This causes the camera to record the lower section of each frame later than the upper section, resulting in a measured vertical acceleration that is too high.

However, if a rolling shutter is the problem, then turning the camera sideways should fix the problem, and upside down it should give a lower acceleration. So if you tried this and it had no effect, it must be something else :-(

I can think of only 2 other possibilities:

1. the calibration is wrong because the calibration object (meter stick) is further from the camera than the ball.
2. the frame rate is not what you (or Tracker) think it is. You might want to confirm it by capturing a video of a digital stopwatch for a few seconds. Set the video player to display time and see if the time tracks the watch. You change the frame duration in the clip inspector if needed.

I hope this is helpful!  Doug

Re: Measuring g in simple free fall (incorrect values from fits) - Feb 13 2014 2:49AM
Armando Pisani
1 Posts

Dear Zach,

try to analyze the balldrop video you can find at this url:

http://jabryan.iweb.bsu.edu/videoanalysis/

I got a value for the vertical acceleration that is 9.9 +/- 0.5 m/s^2

even if the fps is 30 and the image of the ball degrades quite quickly. The fit is anyway quite good.

However I must admit that in my own video I generally get a larger value for the gravitational acceleration as you can see in the video I recorded when analysing a falling basket ball by tracker in my lab/lecture. You can find it at this url:

http://www.opendante.com/OpenDante/Physics.html

I think this is generally due to misalignment between the plane containing the calibration stick  and the vertical motion of the ball and the plane of the image. This is likely to cause an overestimate of the lengths and of the acceleration as well.

Best regards, yours Armando P.

The OSP Network: