CSS Menu Buttons Css3Menu.com


LX Pro - test results

by Martin Gardiner


LX Pro - July 2 2014 test results

Ball spin and spin axis detection using the ball logo

Best results were obtained using 15 db camera gain and shutter speed of 40 us.

LX Pro - April 14 2014 test results

The new dot method of detecting ball spin has now been applied to spin line or logo balls.

LX Pro - December 31st 2013 test results

Known error fix

More improvements to the image processing. Problem shown below fixed for the most part.

LX Pro - December 26th 2013 test results

Known error

We're still getting errors when players feet are near the unit (usually only when chipping).

More improvements to the image processing is required. Updates for this will be coming asap.


LX Pro - December 23rd 2013 test results

Today we tried a new concept for the LX Pro - eliminating frame 1

To date frame 1 is used to measure ball launch angle and path.

The idea being that the distance the ball image is in frame 1 and frame 3 is far greater than from frame 2 and frame 3 and thus the accuracy of both ball launch angle and ball path would be greater.

The problem with this concept is that when the ball is placed on the ground - ie without a tee - the complete image of the ball can be obscured by the fact that the ball is lying low in the grass fibers and only approximately two thirds of the ball are visible.

Ideally, we would need to see both images of the ball in flight - ie unobstructed from grass fibers in the hitting mat .

In order to do this with just as much accuracy as measuring ball path and launch angle as between frames 1 and 3, we need to increase the delay between frames 2 and 3

Frame 1 - ball on tee

Frame 2 - ball in flight 0.5 ms after impact

Frame3 - ball in flight 2 ms after impact

Increasing the delay between frames 2 and 3 to two milliseconds greatly increases the ball image distances in the frames and so accuracy of ball path and vertical launch angle are maintained plus both images of the ball are of the ball in flight and not obstructed by grass fibers in the hitting mat which could cause errors.

We'll still need at least one frame 1 in order to align the cameras but this could be part of the initialization process and would only have to be done once. Automatically of course.

CP software updates will be published in the coming days that will include these new features.


LX Pro - December 13th 2013 test results

W7 Control panel versions 7.7.6.1

W8 Control panel versions 8.0.2.8

Today's tests revealed a couple of software bugs that prevented the ball launch being detected.

The first was this. The bright club hosel caused issues with the image processing so that the ball location wasn't being correctly detected.

I added new check routines for this so that this problem won't occur again.

The second issue was the left/right ball path speed adjustments that caused the ball speed to go negative. Now fixed in the latest version.

We purposely played a number of negative launch angle shots to see the results. Here the ball is driven off the tee on the downswing. Results were perfect.

Another low drive

Chipping just fine too

So all-in-all - after the bug fixes - we were pleased with the test results. All shots were being registered.

An exception to this will be hitting the ball fat and using the microphone trigger.

In this case, the sound of the club striking the grass turf will pre-trigger the cameras and the results will be unpredictable. i.e. the ball speed will either be too slow or there will be no launch at all.

 


LX Pro - November 27th 2013 test results

W7 Control panel versions 7.7.5.2

W8 Control panel versions 8.0.1.8

Today's tests where about testing the new time delay feature of the CP and using the system with both the line scan camera and microphones active.

This image - of a medium fairway wood shot - shows how much further the ball has moved in frame 3 by increasing the time delay from 1 to 2 milliseconds between frames 2 and 3.

This image - of a similar fairway wood shot to that above - shows how much nearer the ball has moved in frame 3 with a 1.1 millisecond time delay between frames 2 and 3.

Both the above shots were using a combination of both the line scan camera and the microphones to trigger the cameras.

As these shots were quite fast, it can be assumed that the microphone picked up the shot before the line scan camera.

If using just the line scan camera to trigger the cameras - with the Mic 2 microphone switched off - we start to see a problem.

The line scan camera is picking up the ball too late, thus the frames 2 and 3 are very near the edge of the FOV of the camera.

This was just a 86 mph ball speed with time delay just 1.1 milliseconds so much faster shots wouldn't be picked up at all if using just the line scan camera.

That being said, always use the microphone trigger (i.e. Mic 2 should be switched on and it's sensitivity level quite low so that it doesn't pick up other sounds like the whoosh of the club or your voice )

You'll want to keep the Line Scan camera on though for putts and small chips.

Settings on the back panel should then look like this. Switch Mic 2 on via the "M" key in main window of the CP. Blue LED will then light up.

Set Mic 2 level to pick up a full shot impact sound of the club and ball and not too sensitive that it picks up your voice

and set the line scan camera's sensitivity so that it picks up a rolling ball.


LX Pro - November 19th 2013 test results

W7 Control panel versions 7.7.4.5

W8 Control panel versions 8.0.1.2

Today's full drive tests went very well. Below are the 3 frames of a typical full drive shot.

 


LX Pro - November 14th 2013 test results

by Martin Gardiner

W7 Control panel versions 7.7.4.0 - Fixes in 7.7.4.1

W8 Control panel versions 8.0.0.6 - Fixes in 8.0.0.7

Issue 1: System not always automatically aligning cameras when ball on tee is moved or after a shot

Issue 2: Shots not always being detected

The below 4 images show typical shots that I could reproduce time after time without issues

This is frame 1. Note the settings like Shutter Speed (36), gain (20db) and Min ball gray scale set to a high 170.

The main reason for undetected shots up until now was that the players feet or legs were too highly illuminated and this caused the image processing not to detect the ball correctly.

Increasing the Min Ball gray scale from 100 - as in the above image - to 170 resolved the problem

This is frame 2

Frame 3

Composite of all 3 frames.

The tests were done with the LX Auto Run Mode option set On.

Test shots should be made with these settings (soon to be default) and the Control Panel in the Main window (ie not in a camera window - only need to go into camera windows was to check the ball images).

Over 40 various speed and launch angle shots were made consecutively without a single shot not being detected.

When the golf course game software was started, game play could commence.

LX Pro - November 12th 2013 test results

by Martin Gardiner

W7 Control panel versions 7.7.3.9 - Fixes in 7.7.4.0

W8 Control panel versions 8.0.0.5 - Fixes in 8.0.0.6

Issue 1: System not always resetting to ball search mode when ball on tee is moved

Issue 2: It was noted that when ball is placed on a tee position that is too far right of the unit and not inline with the first bank of LEDs then the ball image appears too dark and system won't launch a ball.

Ensure ball is placed as shown above.

Issue 3: The new MIc2 and Line Scan camera trigger mode combination was being switched off after shots or a camera trigger.

When Mic 2 is activated (in the setup window) the blue LED light should light up. This was being switched off after a camera trigger

Issue 4: We noticed that on some units the new Mic2/LS cam trigger mode couldn't be switched off at all. i.e. the blue LED was stuck on.

This was actually found to be a hardware problem.

Technically speaking this was caused by the LVTTL (Low Voltage logic) camera signal lines being pulled down too far for standard TTL logic to always respond to.

LVTTL has a logic high of 3.3volts and this is compatible with standard TTL which has a logic high between 2.8 and 5 volts. However, as these camera IO lines are being used in logic input circuits in the LX, these lines were attached to 47k pull down resistors to ensure a logic level to the IC.s even when the cameras are not plugged into the unit. (CMOS chips don't like floating input lines - i.e. they can burn out - so they had to be either pulled down or up).

Unfortunately some cameras are not putting out a full 3.3v as a logic high and so the pull down resistors are pulling the signal down too far for standard TTL logic to respond to.

On the good side, this new Mic / LS trigger mode feature will probably be standard anyway so it then doesn't matter. i.e. the blue LED light will be on always in this case.


LX Pro - November 7th 2013 test results

Control panel versions 7.7.3.7 to 7.7.3.8

Issue 1: Image processing wasn't processing frame 2 after a shot. Looked like it was processing frame 1 instead which resulted in very high ball speeds and spin rates.

Cause: frame pointer 2 incorrectly set to frame pointer 1. Fix: Now fixed in CP versions 7.7.3.7 and 8.0.0.4

Issue 2: With "Min ball gray scale" set to 100, image processing was picking up reflections from player's white shoes and legs.

Cause: "Min ball gray scale" set too low

Fix: Increase "Min ball gray scale" set to 140

The above image is the composite of all 3 images of the test shot

Issue3: Line scan camera not always triggering cameras which resulted in some undetected shots

Cause: Sensitivity (lower dial on back of unit) set too low.

Adjusting the LS cam's sensitivity was found to be a bit difficult to set just right so that it doesn't pre-trigger from reflections from the players shoes or legs but high enough to always detect high speed shots.

Fix: The LX Pro features two club-to-ball impact detection microphones whereby Mic 2 is software controlled. i.e. it can be enabled or disabled via software controlled signal lines on the cameras.

As this issue was only occurring with higher speed shots, there will always be a very audible club-to-ball impact sound level. If mic 2 is activated then it will definitely pick up the shot and trigger the full frame cameras. We'd then have two trigger sources so that if one fails then the other will be used as a back-up. Note that when mic 2 is enabled, the blue LED will light up.

As you may know, microphones don't pick-up small chips and putts very well but the line scan camera does. On the other hand, not being able to set the line scan camera's sensitivity high enough for high speed shots means that it is best suited for lower speed shots, chips and putts. So the combination of the two trigger sources would be ideal.

This feature will be added from Control Panel versions 7.7.3.9 and 8.0.0.5 which should be available as of Nov. 8 after mid day

Issue4: Manual camera alignment not always functioning

When performing some tests, I like to switch off the "Auto line position detection" and manually set the camera alignment and play shots. Modifications to the underlying functions now make manual alignment and shot testing much easier.

A simple manual function test is as follows:

1. Place ball on the tee position.

2. Switch off "Auto line position detection" in the camera window of the CP

3. Switch to camera 1, frame 1 and click "soft trigger" to capture the ball on the tee.

4. Switch to camera 2, frame 3 and again click "soft trigger" to capture the ball on tee with camera 2.

5. Click the "Align Cameras" button

5. Click "Exit" to go back to the CP's main window. You should now see that the frame 2 has been automatically selected and the system is ready for a test shot.

All following test shots can be made without having to do the above setup