Swing Speed Analysis

I had a couple of questions about using Logger Pro to measure the swing speed in such sports as tennis, golf, baseball and hockey. First, how does "distance distortion" affect the accuracy when measuring from such distances as 10 feet or more away from the subject?

I went into Sports Authority to do a comparative analysis using a high speed video camera and their Radar speed golf simulator. I filmed at 1,000 fps at about 3 feet away( I didn't intend to film that close but the space was tight) and I was surprised to see a blurrrrrrr in the golf ball. Anyway, here are my calculations and I was wondering why there was such a big difference in the club head speed as compared to the ball speed?

Here is the video:


Also, is there an acceptable range (in mph) of error when comparing to radar speed technology?

When doing video analysis it is critical to have your reference distance (that is, the object you use to scale the video) in the same plane as the motion. At the same time, the motion must be essentially in a single plane, perpendicular to the view of the camera.

Now, real motions are rarely confined to plane, so this requirement represents an assumption about the analysis. The farther the motion is from the camera, the less important the effect of this parallax error. When you capture videos, stay as far back as is practical. Zoom lenses are fine.

You can estimate the magnitude of the parallax error with triangles; if you motion is, say, 10 m from the camera, but your reference distance object is only 8 m from the camera, the velocities you calculate will be 20% high.

Comparing video analysis with radar measurements is tough because video gives you a velocity at a particular moment of the motion for a particular part of the object. You don't really know when or where the radar is making its measurement.


Okay, I guess I thought the closer I got to the subject the more accurate the reading. I guess I was wrong. Also, are you saying you should Zoom in with the camera? I wasn't aware of that. I am fascinated at your comment that the farther the motion is from the camera, the less important the effect of this parallax error. I thought that was "Distance Distortion" and that can really throw off the data.

With regard to the radar reading. The Sports Authority device is on the ground shooting up right at the point of contact. I don't think it is reading from more than 2 feet away from the tee. But I have always wondered where those radar readings are made from. Let's say a tennis serve, we know it is not landing in the service box at 130 mph, by the time it lands there it has lost 50% of speed. Yet, they say there serve is 130 mph.

So, do you think the baseball speeds of the kid are reasonable? Do you have any work of other people doing this that I could take a look at?


The geometry of the image is such that you get a flatter field when you're farther away, so go ahead ahead and back off and use the camera's zoom to fill the frame with the motion.

I haven't seen the radar device you describe, so I can't comment on how it might work.

Are the speeds reasonable? Since video analysis lets you pick a very short time interval, and lets you choose the part of the motion of interest, it can potentially give more detailed velocity information than can radar. With radar, you don't know when or where the value describes.

I'm not aware of any other such work, other than http://www.vernier.com/innovate/innovativeuse114.html


You have been very helpful and I appreciate you sharing your expertise. With regard to video vs radar, I like your comment that film can give a more detailed account as it is at THAT moment in time. Like you say, where is the radar picking up these speeds from?

The problem I have with the logger pro program is the element of human error when plotting the points in the program. Take for example the ball. If I touch my cursor on ball at the middle and then the edge the MPH jump dramatically. There seems to be a element of human error that requires several plotting of points and then taking an average. I don't see zoom capabilities in the program for plotting the points.

The same problem happens with the cursor when plotting the other information needed for the program. Let's say there is a 3 feet marker in the horizon, there seems to be potential human error when clicking and dragging the length of 3 feet when it is only an inch long on the computer screen. So, in conclusion, the high speed video doesn't lie, however, the information could get misconstrued if the points are off. Hope this makes sense.


The more points you can mark to get a velocity, the less important any single point becomes. So, choose a portion of the video that gives a bunch of points, and you can get the average velocity across all those points.

When you mark points, be sure to drag the video to fill your screen. That makes it much easier to accurately mark the points, and puts less of a stress on your mousing or trackpad skills.

Your reference object used for setting scale should also be as big as possible.


Okay, that makes sense if I am filming at higher fps. Now, about the scale. Doesn't the subject matter have to be the scale? What I have been doing is using the tennis racket. I know adult size frames are 27 inches so I have been finding a point in the video where the racket is perpendicular to the camera and dragging and marking it from there. I can't think of any other way to find the scale. I don't believe I can mark something off in the distance.

I also realize the subject matter can't move away from where the scale originated. So, my last questions would be if the player makes a step toward the camera, how much is that throwing off the data? 10%, 20%? Golf is easy because they don't move from the point of contact, however, tennis players are always making a move except for the serve.

Thanks again, you have been very helpful


You can use either the object itself to scale the video, or you can use another object in the same plane as the motion.

For example, we have a basket ball shot video in Logger Pro. There's a two meter stick on the floor, just underneath the shot. It is in the same plane as the motion, so it works well for scaling.

The error from the scale object not being in the same plane as the motion is proportional the to the relative difference in distance to the camera.

There aren't that many people who have a high speed camera so they just use a regular 60 fps camera. Any tips on handling the blur of the swing or ball?

There's not a lot you can do without a fast shutter speed, which is not a function of frame rate as long as the frame rate is slower than the shutter speed.

To get the camera to use a fast shutter speed, use lots of light. Daylight is good. Typical interior room lights are not good. If you still have blurred images, use the center of the blur.


There aren't that many people who have a high speed camera so they just use a regular 60 fps camera. Any tips on handling the blur of the swing or ball?




So I am going to to show how much the ball slows down from the point of impact of the serve in tennis to the landing point in the service box. The question is, where do I film from? Is it the middle point between the two points? Obviously, I have to back off quite a bit of distance to capture the entire landscape.

In this case, is the half way point perpendicular to subject matter? Or, do I have to have multiple cameras set up across the court?



Yes, you'll have to back off a long way. Point the camera at the mid-way point. A single camera should be fine.

I was thinking that there might be more biomechanically efficient form that would increase arm swing speed. If there isnt, ill have to try heavy weighted sit ups, and some heavy tricep extensions.

Add a Reply

Please sign in or create an account to reply.

All Forums / Lab Ideas

Go to top