Video Analysis Software: Vernier and Tracker
If you haven’t used video analysis before, let me give you a super quick run down. The basic idea is to look at the location of an object in different frames of a video. By marking the object, you can get position and time data — right from the video. You can use a video you record with your phone — OR find some videos online. It’s great. Oh, I even wrote a book about video analysis.
I normally use Tracker Video Analysis, but Vernier has sent me a free license for their Video Analysis app. What better way to learn this new software than to use it for an analysis — and then repeat the process with Tracker?
Yes, I’m should be biased because I received software for free— but you should know that I’ve been using Vernier stuff for more than 20 years, so there’s that.
Here’s the video that I’m going to analyze. It’s just a plain simple toss of a tennis ball.
Let’s go.
Vernier Video Analysis
Starting with Vernier, it’s actually not an app — it’s just online. The first step is to load the video. I recorded the video on my iPhone and transferred it to my computer to upload it in a web browser. No problems here.
Next, I need to set the video distance scale. There’s a button on the left for “system”. That let’s you both set the scale and the location of the origin. Note that you can indeed “zoom in” on the video to get a better marking on the ruler. However (and maybe it’s just because I’m a newbie), once you zoom in you can’t move around in the video.
In this case, I put a meter stick on the board to use as a reference. Just a quick note about videos. For the best results, you should consider the following points:
- Your reference scale object (the meter stick) should be close to the same distance from the camera as the object of interest (the ball). Otherwise, you will have a perspective error.
- The motion of the object should be perpendicular to the view of the camera. If not, the ball will move closer (or further) from the camera and require a different distance scale.
- Finally, try to be a “reasonable” distance from the motion of the object. If you are too close, the camera distance will be different for the left and right end of the motion compared to the middle. You can fix this by place the camera further away and using a zoom lens. In my case, I had the camera about 2 meters from the board.
Now, I can just mark the location of the object (ball) in each frame of the video. To do this, you just use that “add” button on the left. It’s pretty straightforward. Clicking on the object moves the video one frame forward — this process doesn’t take too long. There is an “autotracking” feature that will find the object in each frame, but this is such a short video I just did it manually.
Here’s what it looks like.
Oh, if you look closely you can see that the ball is blurry due to its motion. Essentially, a video is a bunch of pictures. If the ball moves significantly while the camera is taking a picture, it’s not going to leave a sharp image. Actually, it’s possible to use the spread of the image to estimate the velocity — but that’s another story for later.
With this data, I can look at the horizontal motion of the ball as a function of time. Here’s the data.
Fitting a linear function gives a horizontal velocity of 1.894 meters per second (plus it’s a linear graph, so that’s nice).
Here’s a plot of the vertical position.
This looks very parabolicy, so that’s good. The term in front of t² is -5.184 m/s². This should be 1/2*a which gives a vertical acceleration of -10.368 m/s². Yes, we expect a value of -9.8 m/s² but this is pretty close.
Oh, just for fun — you can also create a plot of velocity vs. time.
The slope of this line is the acceleration with a value of -10.18 m/s². It’s different because I excluded the last two data points — they seem off for some reason (maybe the ball hit something). Also, there’s an option to include the uncertainty in the fit parameters, so I did that.
Tracker Video Analysis
Let’s repeat the analysis using Tracker. Although there is an online version of tracker (to run in a web page), I like the desktop version — even though it’s a java app.
If I use the exact same video in Tracker, there’s a small problem. For some reason, the video gets all wonky when imported as a .mov movie. If I change it to an mp4, it works fine. I don’t know why.
Again, the first step is to set the distance scale and the origin, but that really doesn’t matter in this case.
You can see the blue calibration stick lining up with the meter stick. After that, it’s the same thing. Just click on the ball in each frame of the video. Here’s the horizontal position data.
Notice that I get a slightly different value for the horizontal velocity. But remember, I have a slightly different distance calibration and surely I marked the location of the ball differently. This should be fine. Oh, one quick note about Tracker. In this view, it shows the value for the constants A and B. But be careful. Many times this tiny little window will not show the power of 10 if it’s in scientific notation. You can adjust the size of that box until you CAN see it.
What about the vertical position? Here you go.
This one also gives a larger value for the vertical acceleration at 10.7 m/s² — but it’s reasonably close to the value from Vernier. Oh, just for fun, here’s a plot of the vertical position as a function of time.
Here you can see the problem I mentioned. The value of A (and thus the acceleration) is listed at -1.07 — but there’s that dot dot dot hiding E1. If you make the window bigger you can see that it’s really -10.7 m/s².
Tracker Vs. Vernier
Now for the fun. Which app is better for video analysis? As expected, there’s not a single answer. In short, I think Vernier does exactly what you would need it to do for physics students in a class. It has a simple and clear interface and it matches up with other Vernier software (in case the students already have experience with Logger Pro or Graphical Analysis). Oh, it seems to run fine on a phone (although I didn’t fully test this — you know, because it’s a phone).
What about Tracker? I don’t often recommend the app for physics students — installing Java can sometimes cause some headaches (but it’s really not bad). Of course, Tracker fixed this problem with their online version which seems to work just fine.
Tracker can indeed work for physics classes — especially if the students are using simple videos that don’t have any zooming or panning motion. There might be some issues with video formats — I have to process the video before loading, but this might not be an issue with the online version (I didn’t test it).
Of course, when it gets to more complicated videos Tracker has more advanced tools. This means that there are ways to deal with a camera that’s moving around. Here’s an example of using calibration point pairs to compensate for the motion of camera.
But again, I probably wouldn’t recommend that for normal physics students.