System Updates & Development News
by Martin Paul Gardiner
Electronics and software development engineer and founder of GSA Golf
The GSA Golf range of golf club and ball tracking systems software and electronic hardware is being updated with new features continuously.
New Features are being added and mods being made on nearly a daily basis so please check here regularly for your free updates.
Like all other forward thinking companies, GSA Golf continuously strives to improve their products at any given opportunity.
After all, in this day and age, if you're not moving forwards, you're moving backwards
GSA Control Panel (CP) update V.184.108.40.206 - Windows 7 / 8 / 10 version
February 15 2017
Download standard configuration files hereHere
In case of issues you can manually install the CP update .exe file by following these instructions
Note: The above download is just an update for existing customers and not the full version Control Panel
February 18 2017
Stereoscopic ball track image processing
As an alternative to our standard X,Y camera setup for CX ball tracking camera systems,
new stereoscopic image processing software is in development that will allow you to mount both ball tracking cameras on the ceiling instead of one on the ceiling and the other on the floor .
In collaboration with our Swedish software engineers this software development will also vastly improve the accuracy of the ball tracking system.
First releases of this free software upgrade are expected to be ready later this coming spring.
Note that you can use your existing CX2 ball tracking cameras with this software without any additional hard or software costs.
February 16 2017
Many new customers often miss the camera function assignment instructions so we're repeating them again here
Important ! --- Camera function assignments (CX systems)
In the control panel cam 1 is the ball launch angle camera, cam 2 is the ball path camera, cams 3 & 4 the club tracking cameras and cams 5 & 6 the ball spin cams.
Check that all cameras connected to the PC and go the "Cameras" window of the Control Panel
The control panel is designed to see the cameras in the following sequence
Camera 1 is the Vcam (The floor mounted Vertical ball launch camera in a CX2 to CX6 system)
Camera 2 is the Hcam (The ceiling mounted horizontal ball path camera in a CX2 to CX6 system)
Camera 3 is the Ccam (The ceiling mounted club tracking camera - frame 1 in a CX3 to CX6 system)
Camera 4 is the Ccam (The ceiling mounted club tracking camera - frame 2 in a CX4 to CX6 system)
Camera 5 (bank 2) is the Bcam ball spin camera - frame 1 (floor mounted in a CX6 system)
Camera 6 (bank 2) is the Bcam ball spin camera - frame 2 (floor mounted in a CX6 system)
To identify which camera is which you can click the "Soft Trigger" button to grab a new image or set the camera to "Video Stream Mode".
From the images you should be able to see what camera was being triggered.
If you find that cameras are not being assigned in the correct order then you can manually re-assign the cameras.
To do this, select the camera in the CP cameras window, hold both the shift and ctrl keys down, and press the left or right arrow keys to move the camera up or down the sequence.
Note that color cameras - as used in the CTS club tracking systems - will appear in monochrome if not correctly assigned. i.e. if a color CTS camera appears as camera 1 in the control panel then the images will show in monochrome instead of color. CTS color cameras must always be assigned to camera positions 3 or 4.
February 14 2017
IBS images now stored
IBS (Image background suppression) images are now stored and retrieved when closing and restarting the CP
6 cameras running on an HP Omen PC
We're currently testing multiple cameras on our recommended HP Omen PC
We added the "Rosewill" 4 port USB 3 PCI X1 card to the X1 slot in the HP Omen and connected 3 cameras to this board, 2 to the USB 2 ports on the back of the PC and one camera to the front end USB ports for a total of 6 cameras.
A powered USB hub was added to each camera so that 10ft of USB cable was used in between the PC and the hub and another 6 ft or so of USB cable was connected to the output of the hub to the cameras for a total of 16ft plus USB cable length per camera.
The result was that all 6 cameras worked perfectly.
We'll be testing simultaneous triggering of all cameras later this week.
February 10 2017
Hcam raw angle option
A customer request:
Normally the Hcam measures ball path from a user determined hitting position to the first part of the ball trace, however, some setups with a limited enclosure width may require right and left -handed players to hit the ball to the side of the center line.
In this case switching to the Hcam trace raw angle - which measures ball path from both ends of the ball trace - will be more accurate
Much work is being done on the CX6 system software at present. Today's CP features many improvements for CX6 users.
New: Curved screens
If you'd like a curved screen then you can use the standard frame pipe fittings with bendable carbon fiber rod
Quote from "www.draperinc.com"
Curved projection screens offer several key advantages over flat projection screens.
A curved screen helps offset pincushion distortion from a horizontal expansion lens, and also wraps the image around your periphery more than a flat screen for greater engagement with your projected image.
In addition, the curved image is subject to less ambient reflected light from your room, while your projector’s light is more accurately reflected by the perfectly smooth viewing surface for an ideal picture.
The amount of curvature is determined by the length of the "A rods" whereas the width of the screen is determined by the two "B rods"
A rods would normally be in the 3 ft range and B rods in the 7.5 ft range for a total width of 15 ft
February 7 2017
Linear Left/Right launch angle adjustments
In addition to L/R speed adjustments, LA adjustments may also be required when using the linear method (i.e. when calibration is switched OFF).
This is automatically done when Vcam calibration is ON but wasn't available when using the linear L/R adjustments.
L/R LA adjustments are now done using the up/down keyboard arrow keys and L/R speed adjustments are done with the left/Right arrow keys as before.
Small bug fix:
It was noticed that the ball speed was shown as zero in the Vcam window when an Hcam putt was detected.
The CP still detected the correct putt ball speed but just didn't show it in the Vcam window, only in the CP's main window.
February 6 2017
Calling TGC users
We have a customer reporting that the spin rate shown in TGC is stuck on 4700 rpm although the CP reports the correct spin rate but we can't reproduce this. Does anyone else have this issue?
CX6 - GSA Golf's new flag ship product
The CX6 system uses 6 cameras to capture ball spin as well as club and ball tracking. Note that there are 2 banks of 4 cameras in the CX6 CP
The CX6 - a system with 2 ball tracking cameras (Vcam and Hcam) , 2 club tracking cameras and 2 ball spin cameras - is being tested with our new custom PC at present. Testing should be completed by the end of next week.
Left/Right linear speed adjustments can now be set up to as much as 100% increase/decrease left or right
Previous versions had a +/- 25% limit.
Nicer way of moving camera assignments and improvements to CX6
When moving camera assignments - i.e. using the ctrl/shift left and right arrow keys on a camera displayed in the camera window so as to move the camera from one position to the other -
the selected camera moves with the cursor. e.g. you find that camera 1 - which is normally the Vcam - is the club tracking camera and thus needs to be moved to camera 3 position which is the Club tracking camera position on a CX3 system.
You now just select the camera 1 and - using the ctrl/shift right key - move move to position 3 in two steps and the mouse cursor will follow the steps. i.e. from position 1 to 2 and then from position 2 to 3
More Ball speed tests: comparing camera motion trace method to freeze frame method
The 70mph ball trace to dual freeze frame speed comparison tests also show very good accuracy and consistency.
February 4 2017
Using two IR lamps to illuminate the ball from both the side and top
The above is a screen shot from customer "Jim" showing a perfect ball trace achieved by placing one IR LED lamp next to the Vcam as well as the ceiling mounted IR LED lamp
February 2 2017
More Ball speed tests: comparing camera motion trace method to freeze frame method
The above two images are camera images of the same shot. The image on the left is from the Vcam camera using the ball trace method - i.e. long exposure time - to measure ball speed.
The image on the right is from the dual camera CTS system using the freeze frame method to measure ball speed. This system uses dual aligned cameras to capture two high speed frames of the ball in flight timed at precisely 1 millisecond apart.
At first we didn't think it would be possible to capture both the ball trace image and the freeze frame images at the same time as it was assumed that the flashes from the CTS would flood the Vcam camera with light but
as the flash duration is very fast this wasn't an issue. (Note that the bulge at the end of the trace was from the flash)
The purpose of these tests is to confirm that the ball trace method is as accurate as the dual freeze frame method. Previous low speed tests of a ball being dropped from a known height proved that the two methods produced the same results.
The tests done today was of a higher ball speed and as can be seen also produced identical results. In the coming weeks we'll be doing the same tests but with balls launched at over 100 mph using a golf ball launcher.
January 31 2017
Linear Left/Right ball speed adjustments back on
The Left/Right calibration adjustments can now be switched off (and stored) so that the standard linear L/R adjustments can be used if preferred.
The percentage of adjustment is now displayed for every shot too
Left/Right ball speed calibration adjustments have been extended to +/- 20 degrees
If ball path is towards the Vcam camera then 3% - of the max calibration - for every degree over 10 degrees is subtracted from the trace length
If ball path is away from the Vcam camera then 3% for every degree over 10 degrees is added to the trace length
All speed factors displayed in the Vcam window
1: the current Left/Right path trace length calibration factor used for the shot
2: The current LA trace length factor used for the shot
3: If club carry factors are switched on then the factor used for the shot is displayed
4: Global ball and club boost (speed adjustment) factors
January 24 2017
Launch Angle (LA) speed adjustments - Drives too long - short irons too short
In addition to left and right ball path speed adjustments, LA speed adjustments are also required as the higher the ball flies, the further the ball is from the floor mounted Vcam camera and thus the trace length will appear smaller.
The above image is of a white tube of a known and constant length that represents a ball trace. The trace length is 153 pixels and there is no LA speed adjustment.
The length of the trace here equates to a ball speed of 77.83 mph.
This image is of the same white tube only it's launch angle has been increased from 8.56 degrees to 31.97 degrees.
Notice that the ball trace length - due to the ball being further away from the camera - has decreased from it's original 153 pixels to 143 pixels.
Without any LA speed adjustment the ball speed would thus be calculated at a lower rate of 73 mph (from 77.83 mph) and if - in addition the LA speed adjustment is negative - the ball speed will be even lower (68.60 mph here).
To compensate for the decreasing ball trace length with launch angle the LA Speed should be set to around 21%.
This percentage is a percentage of the 60 degrees and starts to kick in at the user adjustable starting LA (in this case from 10 degrees).
Note that the current speed increase percentage is now also shown - in this case the ball speed was increased by 6.29% to bring the speed back up to it's correct speed of 77.8 mph.
Testing showed that the decrease in measured trace length ratio to ball height is quite linear so a smooth and gradual linear LA speed increase works well.
Beeps driving you crazy?
Sometimes the CP will sound a beep sound if the game software is not reporting that the next player or next shot is up and the system times out.
You can now switch these beeps off in the CP's setup window (bottom left hand corner).
Beep when a new image is captured by the camera
For test purposes it may come in handy to hear a beep when the cameras grab a new image.
You can use this to ensure that the line scan camera detected a ball and sent a trigger to the camera when adjusting the sensitivity.
Even though the line scan camera features a bright LED trigger light, it may not be easy to see when swinging a club so a temporary acoustic signal will be useful.
You can switch this "Beep on new frame capture" on in the CP's setup window (bottom left hand corner).
Ball speed tests: comparing camera motion trace method to freeze frame method
In order to test a measured ball speed we need a reliable reference. i.e. a method of launching a ball at a known speed.
As this is not so easy to accomplish we first started by simply dropping a ball from a certain height and measuring it's speed at near impact to the ground.
Established physics calculators were used to determine what the speed of the ball will be at near ground impact when dropped form 2.5 meters which turns out to be 7m/s or 15.6 mph.
The CTS was converted from a club tracking system to a ball tracking system and - turned on it's side and connected to a line scan camera - setup to measure the ball speed.
As expected the results were very accurate - at least considering the ball is still accelerating during the 1 ms time spacing taken to grab the two images.
Using the trace method on a Vcam (i.e. using the comparatively long exposure time of 10ms to grab the image) we found that the measured speed was too fast.
The scaling factor used to convert pixel distances to real world distances was calibrated using a white pole of a known length so it wasn't this causing the error.
Studying the images it could be seen that the calculations weren't taking the ball's diameter into account.
i.e. the distance the ball traveled was being measured from one end of the trace to the other and not from ball center to ball center.
Subtracting the ball's diameter from the distance corrected the error.
However, as this is just a low speed test we still haven't found the cause of the high speed measurement discrepancy which requires that the scaling factor - derived by the white pole calibration - be increased by some 50% to correct.
i.e. using a scaling factor of just 0.175 instead of the default value of 0.286 will probably result in measured ball speeds being less than they really are.
High speed testing will start next weekend using a golf ball launcher.
January 20 2017
Ball speed tests: comparing camera motion trace method to freeze frame method
This weekend we're comparing the motion trace (or motion blur) method of calculating real ball speed with a dual camera/dual freeze frame method of calculating real ball speed.
The motion trace method is the standard method we currently use to measure ball speed. i.e. we're measuring the length of the trace of a ball in flight with a known camera exposure time.
The length of the trace is converted into a real world distance in meters or feet and then - using the formula v = d/t - we calculate the speed where t is the 10 ms exposure time of the camera.
Oscilloscope tests yesterday confirmed that the exposure time of the camera is indeed exactly 10 milli seconds and
- using a white pole of a known length to simulate a ball trace and setting the scaling factor so that the calculated length is exacltly the same as the white pole's real length -- we know that the pixel to real distance scaling factor is spot on.
So - in theory at least - the ball speed measurements should also be spot on but tests show that the measured speed is always around 15% less than expected.
In order to correct this we have to increase the scaling factor to bring the measured ball speed up to what we would normally expect.
The dual freeze frame method involves grabbing two very fast images of the ball within a known time, measuring the distance it has traveled within the two frames and again using the v=d/t to calculate the speed.
For this we're using the CTS Pro with its dual aligned cameras and dual flash units. As this method has to be absolutely correct it should be very interesting to see where the dependency in the ball trace method is coming from.
Results will be published later this weekend.
January 20 2017
The new PX3 and PX4 systems will be available within the next few weeks
January 18 2017
Customer License update only
Camera IR filters
False alarm on this. Low brightness images are more due to the IR lamp. The IR filter makes no difference.
We're currently comparing IR illuminators
January 11 2017
Camera IR filters
We've recently noticed that IR filters of the same pass wave length of 720nm - as used by our cameras - vary greatly in visible light pass from one manufacturer to the other.
i.e. while one 720nm IR filter blocks all visible light, others let a fair amount of red visible light through.
This - we presume - would explain why some trace images with some customers look bright and full while traces with other customers seem rather dull.
We've identified the best IR Filter manufacturer now so if you think that your traces images are not as bright as they could be, let us know and we'll arrange for a filter exchange.
Fix: Optical sensor mat Loop Count occasionally being reset to 2
Fix: Club face angle shown in the "Club Fitter" mode window was reversed
Vcam trigger delay
In addition to the Hcam trigger delay, there's now also a Vcam trigger delay.
The trigger delays are used when moving the line scan camera nearer the ball hitting position so that ball path range is greater and for systems not using line scan cameras.
January 7 2017
CX3 / PX5 without Line Scan camera
We're currently testing the CX3 and PX5 without using the line scan camera.
Trigger is then sources from the PX2 club tracking mat (PX5) or the trigger mat (CX3).
The main advantage of using these systems without the line scan camera is that no black carpet scan area is required.
Disadvantage is that every shot must be played from the same place on the trigger mat or PX2.
CX3c to CX3
CX3c customers can now switch to a CX3 setup using the same CX3c color camera.
CX3 uses the standard club tracking dot method to capture club face angle, club speed and club path whereas CX3c only captures club face angle.
January 2 2017
As some customers prefer Vcam putting speed calculation to Hcam putting, we've re-introduced this method.
The Launch data time-out has now been reduced from 10 seconds to 4 seconds.
The time out is normally there for when the game software is not reporting back to the CP that the ball has landed and is ready for the next shot.
If a ball is played during the time out period then there will be no ball launch in the game.
E6 club selection
The CP now shows the club selected by E6
Improved camera alignment process
CX3 / CX4
A warning message is now displayed in the CP's main window when the measured club face angle exceeds the user defined max face angle.
PX2 / PX5
Double clock frequency on new PX2 sensor mats causing double club speed
The new sensor mats (i.e. since fall 2016) are fitted with a double speed system clocks but the CP was still using the old clock speed to calculate club speed which resulted in double measured club speed.
This is now corrected.
Vcam calibration percentages can now be applied to the Vcam ball speed and ball launch angle calibration tables.
i.e. you can set the ball speed calibration adjustment for example to 50% and the ball LA calibration to 80% .
To set, move the mouse cursor over the "Calibration ON/OFF" button and use the arrow keys as instructed to adjust.
Update for PX5 / PX2 users
Ball launch will now still occur even if club face angle could not be detected due to one of the two face timers still running.
i.e. when the ball is struck more towards the toe of the club so that the entire club face didn't pass over both timer sensors.