|
Post by imperfectgolfer on Dec 2, 2011 10:43:40 GMT -5
Consider Mike Finney's comment in this BM-forum thread. www.brianmanzella.com/golfing-discussions/16446-just-thought-3d-motion-analysis.htmlIt is probably the first time that I agree 100% with MF - the graphs produced by 3-D machines cannot be accurate because they use different axis of rotation for their measurements, and the axis also changes constantly throughout the swing (and the 3-D machine operators/interpreters choose to ignore this fact). The most inaccurate of all the graphs is probably Teeace's 4D graphs - see the 3jack forum thread. richie3jack.proboards.com/index.cgi?board=general&action=display&thread=2984Teeace even admits that he measures shoulder rotation around the vertical axis. That's crazy! In the late downswing the right shoulder is moving downplane and nearly parallel to a vertical axis and its angular velocity would be under-recorded during that time period. Note the dip in the red graph in the late downswing. Do you think that's "real"? Do you really think that a skilled PGA tour golfer's shoulder rotation slows down before impact and then suddenly accelerates markedly after impact. Jeffy posted this graph: He then stated-: "As you can see, the shoulders are decelerating through impact, consistent with a flip." He has a perspective of golf biomechanics that is very different to mine - causally connecting flipping with a shoulder deceleration phenomenon at impact. The shoulders are not decelerating at impact (and that graph is inaccurate because it measures shoulder rotation around a vertical axis), but the left shoulder is moving up-and-away allowing for the full (unimpeded) release of PA#4 and there should be no flipping (in the absence of a right arm-induced push-force that flips the club) if a golfer can maintain left hand speed through impact. My biggest gripe about these 3-C machines is that they only measure individual golfers, and they don't solve the problem of understanding golf swing biomechanics because there is nothing in common between different individual measurements eg. Bubba Watson versus Brian Gay, or Ben Hogan (whose pelvis rotates non-stop through impact) versus Dustin Johnson (who stalls his pelvis at impact). Jeff.
|
|
|
Post by nmgolfer on Dec 2, 2011 11:47:56 GMT -5
I disagree completely. TeeAce's system is the best I've yet seen. Locations of all sensors i.e.cameras are accurately known and stationary. They use an inertial coordinate system not something attached to a moving body. Provide the image processing software is accurately validated I'm of the opinion the at the 4D will provided the very best motion capture data sets we seen thus far.
|
|
|
Post by natep on Dec 2, 2011 12:51:21 GMT -5
|
|
|
Post by imperfectgolfer on Dec 2, 2011 13:21:19 GMT -5
nmgolfer,
I agree that Teeace's 3-D system can produce accurate 3-D data showing the motion of body parts, and the club, in space.
Natep has shown how it can accurately reflect the 3-D motion of the hands and club in space in that diagram.
However, one still has to interpret that data in an useful manner. Teeace has produced graphs, but those graphs are "intrepretative" in the sense that they have re-interpreted a 3-D motion into a 2-D graph showing either angular speed or linear speed. That requires the use of reference points - reference axes. I would take his graphs seriously if you could demonstrate that his choice of reference axes are very appropriate for every single moment of the downswing/followthrough - remembering that all body parts and the club are moving according to the complexities of the issue that MF brought up-: "instantaneous screw axis theory basically says that it is very difficult to measure linear speeds (across the video screen) and angular speeds (rotational) because the body parts are CONSTANTLY changing their relationship (angles between their corresponding axes of rotation) to each other during the swing which can potentially throw off the measurements..... "
Jeff.
|
|
|
Post by nmgolfer on Dec 2, 2011 13:39:12 GMT -5
Welcome to the forum natep! It looks to me like it holds incredible promise. I've done a lot of work with sensors (accelerometers strain gauges microphones ect.) and I'm pretty aware of what can and cannot be done with them. Suffice to say a great deal of care in the prep. calibration and data acquisition must be taken in order to get accurate results. Even then an uncertainty analysis will be... shall we say ... disappointing and the notion of ever having something commercially viable for mass usage should probably be forgotten. These sorts of factors undoubtedly drove 4D to develop what they have. While I'm certain there will be some learning curve and growing pains, the thought of using high speed cameras with digital image processing and computation for data aquisition I find very appealing. I hope their company succeeds and these things become ubiquitous.
|
|
|
Post by nmgolfer on Dec 2, 2011 13:49:38 GMT -5
If the position of any location be it hand clubhead shoulder what have you can be extracted from the multiple camera time-coded camera images, and I believe it can, then from there is is easy to calculate velocity as it is merely the change in position with respect to time. Same for acceleration. Now take two points instead of one. We can draw a line between two points and using the same data for the individual points we can extract angular velocity as well .... in all three dimensions (yaw pitch roll... axial or what the BM calls gamma... more difficult). It do not see where interpretation becomes the problem that concerns you. I do find it interesting that Rob Neal has changed the way he reports linear velocity and that it now agrees what Nesbits measured. But I'm not surprised. As I mentioned above I've done a lot of experimentation an seen data misinterpreted many times. As my old boss once said: if it was easy any trained monkey could do it. Well in the 4D system I see the makings of a measurement system any trained monkey could use nmgolfer, I agree that Teeace's 3-D system can produce accurate 3-D data showing the motion of body parts, and the club, in space. Natep has shown how it can accurately reflect the 3-D motion of the hands and club in space in that diagram. However, one still has to interpret that data in an useful manner. Teeace has produced graphs, but those graphs are "i ntrepretative" in the sense that they have re-interpreted a 3-D motion into a 2-D graph showing either angular speed or linear speed. That requires the use of reference points - reference axes. I would take his graphs seriously if you could demonstrate that his choice of reference axes are very appropriate for every single moment of the downswing/followthrough - remembering that all body parts and the club are moving according to the complexities of the issue that MF brought up-: "instantaneous screw axis theory basically says that it is very difficult to measure linear speeds (across the video screen) and angular speeds (rotational) because the body parts are CONSTANTLY changing their relationship (angles between their corresponding axes of rotation) to each other during the swing which can potentially throw off the measurements..... " Jeff.
|
|
|
Post by imperfectgolfer on Dec 2, 2011 16:46:01 GMT -5
nmg,
You have not addressed my criticism - that the graphs are plotting the angular velocity of shoulder rotation with reference to the same axis throughout the entire downswing - a vertical axis. How can that produce an accurate reflection of angular velocity if the shoulders are not rotating at 90 degrees to that axis, and where in the late downswing the shoulders may be rotating at <30 degrees angle relative to that same vertical axis?
Jeff.
|
|
|
Post by nmgolfer on Dec 2, 2011 18:13:51 GMT -5
Its quite possible I don't understand you Jeff. I thought it clear once the motion was accurately measured its quite a simple matter to transform it into any coordinate system the analyst desires see: en.wikipedia.org/wiki/Coordinate_system#Transformations_between_coordinate_systemsIf you want to know what shoulders are doing you look at the two points (or centrolds if you are using the locus of multiple points to define ends of the line segment) and them plot the motion (both linear and angular of that line segment. Its doubtful that such a body part is ever rotating about just one axis however. so you look at all then use vector algebra to find THE "axis" of instantaneous rotation and magnitude. Does that make sense? Think of a hologram... once you've captured the event in data (digits) you can look at it any which way you want.
|
|
|
Post by imperfectgolfer on Dec 2, 2011 21:18:01 GMT -5
I agree that one can look at a hologram anyway one wants. I can also agree that you could theoretically measure linear/angular speed using some system of instantaneous axis calculations.
See this journal article - which I would like to read if I knew how to get a copy.
The Future of Instantaneous Screw Axis Theory to Golf Instruction and Beyond
J.P. Carey, A.S. Vena The Journal of Applied Golf Research - June 30, 2011
Abstract
Many golf instructional programs and institutes analyze 3D golf swing kinematics through software packages. Our group applied a promising technique, instantaneous screw axis (ISA) theory, to provide a better simulate segment rotation and a simple and practical means of looking at the kinematic golf sequence. Our work focused on identifying ISA location and orientation, and segment angular velocity of major body segments involved in the golf swing to determine ISA applicability to evaluate kinematic sequence. The method successfully demonstrated that magnitude of maximum angular velocities increased from the most proximal segment (the pelvis) to the most distal segment (the left arm), in accordance with the summation of speeds principle.
Furthermore, in most of our test cases the maximum angular velocities occur in the desired kinematic sequence, where the first maxima was achieved by the most proximal segment and followed by the more distal segments in the kinematic chain – demonstrating a range of player abilities. Furthermore, the ISA method provides clear visual representations of segment rotations, which we believe would be beneficial to instructors and players alike. Based on these findings, we believe that there is significant potential application of ISA for golf instruction and golf biomechanics analysis software. Furthermore, this approach opens a large range of potential future areas of golf research, instructional means and an additional tool for equipment design and tailorability opportunities.
Do you believe that Teeace is doing that in his measurements?
Where is your evidence?
Why does he state that his graphs of shoulder rotation are based on a vertical axis?
Jeff.
|
|