|
Post by chipitin on Sept 6, 2013 12:18:21 GMT -5
Kelvin's reference to "shoulder" readings on graphs is incorrect if AMM is the system he references (and has readings from some of his elite players). AMM has the Thorax, not shoulders. Once again, imprecise language being used.
|
|
|
Post by imperfectgolfer on Sept 6, 2013 12:21:10 GMT -5
Learning "something" about pelvic/shoulder motional biomechanics from Moe Norman.
Note that he rotates his left thigh counterclockwise continuously/non-stop during his downswing/followthrough, but he cannot quite get his right pelvis around enough by his finish position so that the entire pelvis is square to the target at his finish position. That's not surprising because his post-impact shoulder rotation is far less vigorous and far less complete - compared to Gary Woodland - and therefore there is less available energy moving down the spine/torso from the upper torso to the lower torso. However, that doesn't mean that MN is intentionally slowing down his pelvic motion after impact.
Note that he obviously slows down the club motion between P8 and his finish position where both arms are extended towards the target. However, the slowed-down arm/club motion (incomplete followthrough/finish phenomenon - relative to Gary Woodland) doesn't speed up his shoulder rotation after impact. In fact, his arms are always roughly in front of his rotating upper torso between impact and his finish position, which means that his upper torso rotation is not faster than the rotation of his two arms/club.
Jeff.
|
|
|
Post by chipitin on Sept 6, 2013 13:02:00 GMT -5
Quote Originally Posted by Jeff Martin View Post I also asked Tapio why patterns like Noren's have not been seen in AMM/TPI reported data:
I don't really know how AMM is doing that, but all devices I've seen got one sensor at lower back, not at hips. By that they assume how hips are working [but] there is flexible part between those points, so it can fool the measurement.
One other reason seems to be filtering. Now if they got, as it seems to be, long and strong gaussian filtering, also things that happens after impact affects to measurement before impact. In other words, if one got really low rotation speed at hips way after impact, that filtering starts to pull down the graph line even before impact. We got few frames before and after impact where is no filtering at all, so we avoid that mistake. You can see the same when looking those hand speed and club head speed graphs, where others got smooth deceleration around impact and we got sharp line going straight down just at impact.
So, there are differences in assumptions and methodologies between systems that could account for the discrepancy. As far as I know, none of the biomechanists have seriously looked at those differences. Sound scientific?
Mike Duffey's response:
Jeff, I hope you see that it is a bad idea to ask someone who doesn't use or apparently understand a system to defend it. Kelvin was put in this position and I think took some unfair hits when asked to defend it. Tapio doesn't use AMM and has not displayed a good understanding of the system (or at least a good ability to discuss it in English. I wouldn't rule out a language barrier as part of the issue). It would be much better for him to address the strengths and weaknesses of his own system.
As I said above and will say again now - we are trying to sort out those differences. it is challenging because Tapio does not seem to want to reveal much. I understand he feels it would give away what her perceives as valuable technical information, but that does not make it easier to make the comparison.
About his filtering comments, which are quite interesting. AMM says they do not filter. Assuming they are telling the truth (I see now reason to lie), then this is not an issue. I will mention that the mass of the sensor may act as a mechanical damper, so there might be something similar to a small amount of physical filtering going on, but teh type of error Tapio is discussing ONLY applies to mathematical filtering. It is not an issue with AMM. The high frequency content of impact can create issues with data collection, and that is especially true with optical systems. Therefore, I believe all of us optical system users do filter. What we do with my system to specifically avoid the issue Tapio is talking about is filter pre and post impact separately, therefore there is not cross-over noise that. This method is pretty common in biomechanics data collections where there is a single, high-frequency event, like hitting a golf ball. It has been shown the be a valid approach. Further, in no way would filtering cause what we are seeing in the graph. IT IS NOT EVEN CLOSE. You would likely see a small, sine wave-like bump just before impact - and it would look very strange. Those errors are easy to pick out visually in this kind of movement. It would have a miniscule effect, especially on the pelvis and torso graphs. Those segments have a lot of mass and the frequency content is much lower in those signals. It is like saying a brief, 20mph gust of wind made a jet an hour late on a 3 hour flight. .
|
|
|
Post by chipitin on Sept 6, 2013 13:06:13 GMT -5
The graphs in question are one's Tapio posted on facebook and Jeff re-posted on his forum if you want to see them you can look there.
There are two very important questions here: 1. What is shown in that graph as compared to traditional graphs? 2. What evidence is there that the pattern shown is good?
Regarding question 1: Given all of the currently published data on the topic thus far, markerless video-based 3D capture is not accurate and reliable. I believe some day it will be, and maybe Tapio has a found the right methods, but he has not published validation data. He mentions that it is accurate for wood and metal products, but those are not deformable objects, so achieving accuracy in those settings is much easier than in the golf swing where we have overlapping segments that are deformable. It is appropriate to provide evidence of accuracy with new technology. Let's assume, however, that Tapios data collection is good. The next question then is, does his data presentation match what we typically see? Just because it does not match doesn't mean it is wrong, but it does mean we shouldn't directly compare the two (at least not without understanding the differences) Tapio has avoided thorough explanation of his system, but he has confirmed that his pelvis rotation is relative to the room, not to the pelvis. As such, that data should probably not be used to compare because any side or forward tilt of the pelvis (which are virtually guaranteed) will alter the speed shown on his graph event thought the pelvis is has not slowed or sped up.
Regarding question 2: Looking at the graph, at the time that the pelvis is seen speeding up into impact, the shoulders are slowing and the hand speed is either flat or slowing slightly. So it does not seem that the increase in pelvis speed is contributing to increased velocity as you get closer to the club. My opinion is that club head speed at impact is more important than pelvis speed.
Quote Originally Posted by Jeff Martin View Post Tapio's explanation of Noren's pattern is more or less identical to Kel's concept of the "second fire" and "push-pull coupling" of the hips in elite golf swings. Tapio also indicated that it was rare. In other words, if you hadn't studied a lot of elite swings, you might never see a pattern like Alex's. We know from Prof. Duffey that Kwon and the rest of the biomechanists are not expert in elite swings. Jeff, You are right that I have not collected data on tour-level players, and that I believe Dr. Kwon has little or none. But you and I have specifically discussed the fact that Phil has, you even confirm this in your August 31st post in the "scientific bias" thread. I don't know the members of your forum, but my guess is that the number of tour players he has in the AMM data base is greater than the number of tour players coached by all of the members in this forum combined. I am obviously stating this as a guess not as fact, and I ask you to not state things that you know are incorrect as fact.
|
|