SkyTrak technical FAQ

[davray666]

The idea is to share some technical details about the unit that could be useful in order to figure out how to get the most out of it.
Here is what I have been able to gather (guess) so far.

Unit has 2 cameras and what looks like 2 Xenon flashes and 1 IR LED bank.
There are 2 additional sensor that are probably used to detect the ball launch and trig the camera image capture.
For each shot 3 images are captured. Time between images is
image 1 @ 0us, image 2 @ 4140us & image 3 at 10400us.

The rather long time between image 1 & 2 should produce good accuracy for vertical and horizontal lunch angles. I am more concerned with the spin numbers.
Would be interesting to see some comparison on high spinning shots against some other trusted unit. Especially really bad shots with lots of side spin (or spin axis tilt if we are to be correct)

In order for the unit to catch a high vertical launch angle shot, the ball needs to be placed in front of the red laser dot. The theory is that it will never make it to the second camera field of view if it's too far back.
I have tried with as much as 2". No idea how far it could be placed before it becomes a problem.

[HTTR]

Is there any word on if they will somehow let us view the camera images, I just think it would be interesting to see, I record myself when hitting golf balls using my GoPro and helps see what I am doing wrong so seeing the shots of the club hitting the ball could help users where club face hitting the ball. Also, do any of these technical specs affect the placement of the golf ball when using a tee (as I noticed when using my driver I feel it is the main thing I am curious on the accuracy of.

[TorchRedRob]

This is interesting since they can measure 200mph ball speed. This means that the ball is moving at roughly 300us per inch of forward movement. In order to take the third picture at 10400us, they would need a field of vision that is almost 3 feet wide. I guess that is a reasonable width of view with the right camera? (Granted, not a lot of folks rocking out at REMAX ball speeds...)

[davray666]

The third image is definitely not used to measure any normal iron or wood swing. Not even a soft chip. The ball is long gone by the time that is taken. I would guess they use it for detecting false shots. E.g. you swing your white driver in front of the trig sensor. The unit takes 3 images and sends them to the Ipad. If the ipad can see a ball in the third image then it's not a valid shot and it goes back "detect shot mode". Another thing it could be used for is measuring soft puts. In order to get any kind of accuracy on the horizontal launch angle the ball must travel at least 6", the longer the better. The cameras are 6cm apart. With that and the fact that they can detect 200MPH ball speed. The FOV / camera is probably somewhere around 60 degrees.
That by the way produces a massive geometrical distortion that can be corrected in software. To be accurate you would need to calibrate each lens individually.

I agree with HTTR. Would love to be able to see the 2 first images annotated with dots and lines to show how the software interpreted the image. Pretty much like the vector pro.
This will gives us a chance to detect if the data produced is based on a valid processing or if it got tricked somehow. For the vector pro I almost always used Nike SFT balls.
The ball is crap but the detection was flawless. With a ProV1 it kind of worked but you had to look at the spin line and sometimes make an adjustment based on how the software detected the logo and other times you just had to disregard the shot because it was totally wrong.
With the information you get from the images you can also tune what balls work best and how they should be placed for maximum accuracy.

But above all. Building up trust that the measurement is accurate is key. Without the images as proof you will always have an internal debate between what you felt and what the unit reports.

[DougieMn]

Here's an off the wall question regarding my SkyTrak and SkyGolf360. I love the idea of the shots being recorded and gathering distance info for each club. My concern is about the shots I DON'T want recorded. Do I need to do a direct connection during warmup and then connect via my internet wifi in order to keep my numbers more accurate? What about the shots like knock downs,1/2/-3/4 shots, flops, punches etc? What if I want to keep them out of my data set for that session? Would be nice to be able to turn off the recording of shots into 360 instead of switching connections. (Perhaps this should also be posted in the "Requests and enhancements" as well?)

[Mike Z]

Here's an off the wall question regarding my SkyTrak and SkyGolf360. I love the idea of the shots being recorded and gathering distance info for each club. My concern is about the shots I DON'T want recorded. Do I need to do a direct connection during warmup and then connect via my internet wifi in order to keep my numbers more accurate? What about the shots like knock downs,1/2/-3/4 shots, flops, punches etc? What if I want to keep them out of my data set for that session? Would be nice to be able to turn off the recording of shots into 360 instead of switching connections. (Perhaps this should also be posted in the "Requests and enhancements" as well?)

If you could edit clubs in a reliable way.......... When that happens I will just make a throwaway data club and set the machine to that whenever its something I do not want to capture (warm-up, punch, different club, friend using it with me)...

[DougieMn]

If you could edit clubs in a reliable way.......... When that happens I will just make a throwaway data club and set the machine to that whenever its something I do not want to capture (warm-up, punch, different club, friend using it with me)...

That's a great Idea! I could then just use one of my clubs (ie 3 iron) as the throw away shot club to use as warmup and partial/specialty shots and end up with accurate data on other clubs. Thanks for the idea!

[wilson]

Posted in the General FAQ thread - but relevant here if people are following:

SkyTrakGolf @theguitarman83 SkyTrak is able to track the dimples on the golf ball. However, having some sort of mark on the ball provides best results. 23-03-15, 11:52 AM

[paschmann]

(Not sure if this has been posted) but ... if anyone is interested in reading the details behind the device you can find them here:

http://www.google.com/patents/US20140028838

[davray666]

There has been some debate on the forum on how best to trigger the cameras to ensure a good shot capture.
Here is an explanation on how it works and what it looks like that should help.

Look at the attached images. Both are taken at the same time but with 2 different cameras. The color image is from an Iphone6 cam and the dark one is from a high speed camera.
The difference is that the iphone cam has an IR block filter whereas the high speed cam has an IR pas filter.
Looking at the dark image you will see an IR laser line that is projected in front of the ball. That line can't be seen with your eyes nor a consumer grade camera.
You can regard that line as a curtain that is appr. 65mm in front of the RED laser dot.
As soon as you hit the ball and it passes thru the IR laser line "curtain" the cameras will trigger and take 1 picture each as the ball travels past them.
According to the patent application the laser sensor also measures the ball speed by measuring the reflected laser light intensity that should look something like a sine wave as the round ball passes the "curtain". That could be theoretically used to control the camera trig points, duration and separation.

Among the reported problems we have seen white shafts mentioned. A good golfer normally leans the shaft forward at the impact point. In that case it is highly likely that the shaft passes the "curtain" before the ball does and therefore triggers the cameras based on the shaft and not the ball, resulting in a missed shot since there are no balls in the captured images.

Another problem is missed high vertical launch angle shots. The cameras have a limited FieldOfView (FOV). If the ball is launched at a steep angle while placed far away from the "curtain" it will enter the first cam FOV at a very high position and then fly out of the second cam FOV. E.g. we get an image of the ball from the first Cam but none from the other.
The solution here is to place the ball in front of the ball position RED laser dot and just behind the IR laser "curtain".
So in summary, the ball can be placed behind the RED laser dot in case the vertical launch angle is low as long as you are shooting from the "ground". The ball should be placed in front of the RED laser dot in case you plan to launch the ball higher than some 40 deg. If the ball is on a tee you will suffer from a similar dilemma as for the wedge shots. High launch angle (for a driver that would be somewhere above 20 deg) will cause the ball to fly out of the second cam FOV. The higher the tee the lower max possible VLA. Solution could be to move ball forward or raise the Skytrak unit.

[paschmann]

Cool stuff - thanks for sharing!

[davray666]

Crap, I see now that I wrote tee when I meant to say peg.

[Ironstrokes]

Excellent work davray! That's really cool and interesting to see it works that way, had no idea it had a 'curtain' effect like that. Definitely made it a ton easier for me at least now that I know why some of those high launch angle shots weren't reading. I'm looking forward to testing that out.

[kevin1889]

This is pretty cool. Thanks for sharing.
As I put the finishing touches on my DIY enclosure, this made me think of the cut-out that I'll need protecting the facing of the Skytrak. From what I've read, the Lexan won't work when fully covering the camera, so I'm going to need to cut out a slot in my Lexan sheet. Question is, how big does the cut-out slot need to be so that the full range of the camera is accounted for? I imagine it can't be the same exact size as the glass covering the camera on the Skytrak itself because of this "curtain" and it looks like it expands both downward and upward at an angle from the unit, kind of like this []< ?
I will have my Lexan sheet about 1 inch in front of the Skytrak face, I was thinking of making the cutout about 1/2" wider all the way around. Do you think that would work?

[davray666]

, I will write a post in the protection thread in a minute.

[davray666]

I got tired of guessing the parameters for the camera FOV and sensor resolution and orientation for my optimized protection.
So I opened up the unit and made some measurements. Now that I know exactly what it looks like there are few corrections needed to the explanation above.
1. The camera FOV is way much taller than I assumed. That means that the high launch angle shots will not fall out of view. From the FOV point of view the hardware can actually measure over 70 deg launch angles. The problem is the way the current firmware uses the trigger. That also explains why the accuracy is so bad on HLA and VLA.
2. The distance you can place the ball behind the laser dot is longer than i expected and that includes driver shots.
3. The high VLA shots still need to be placed in front of the laser dot for optimal accuracy but it only helps a little bit since that is not the main problem and it also causes another problem where the club will be in the view with the ball and in some cases in front of the ball.

Moving forward especially when connected to a sim that can inform the ST unit what shot is intended it is possible to change the firmware and improve short game accuracy. If that will happen and if Rapsodo knows how, remains to be seen. I hope they do because as far as I am concerned the short game accuracy is the biggest problem with the current solution.

[Wev]

The way I would understand it is the problem with the VLA is the launch angle will increase from the point of leaving the face with high dynamic loft. ie it might be 40 degrees the millisecond it leaves the face, but might be 45 degrees an inch down the line as the backspin starts to balloon the ball up.

I thought this was called spin loft, but apparently it's not, spin loft is DL - (AOA). So 30 DL and hitting 5 down would be 30 - (-5) = 35 degrees spin loft. Saying that even how trackman describe spin loft isn't actually what they display in their numbers for high spinning shots (the difference probably being a true spin loft).

What Skytrack/ Flightscope and Foresight call launch angle are very different things depending on where and what time they're measured, if you drew a line from impact to 1 inch down the line the LA might be 30 degrees, but by the time it gets to 3 inches the LA from impact to 3 might be 35 degrees as balls curve upwards.

It's all quite confusing and my point with this and the inaccuracy with the LA could easily be 10% depending on where the measurement has taken place. This is just the way it is with camera based systems though, and why doppler would be much more accurate for shorter, high spinning, high lofted shots.