1 Topic by obevz

  • obevz
  • Member
  • Offline

Hi All...long time no post...I trust you are well as well can be...doing' OK in Australia at momt...
Apart from some 300fps USB2 cameras, my main camera via USB3 is a Basler 525fps camera - indoors and don't seem to get enough frames in the downswings...just wondering if it's my settings  or the camera itself. I don't need "pretty" images just need to capture the downswing part of the shaft to impact. Generally I only get 3-4 frames from the 9 o'clock to 6 o'clock positions. Club head speeds are sub 115mph.

I upgraded my lights facing the golfer...from 2 x 150W flicker free LED lights to 4 x 200W flicker free LED cameras. Plus 3 x 200W LED behind and above the golfer...with no (minimal) light on projector screen. (Have also purchased [yet to arrive] a 25,000 lumens Imalent tactical torch MS06 - plan to run at 4000 lumens for 2 hours) I used my phone to check the lumens...and it was around 1000 at 2metres. Don't know how accurate that is tho.

What would be the better settings to capture more frames...should I go mono? The Basler camera runs at 520+ fps (and my setting is at max 550fps) and does not seem to catch the 9 to 6 frames. I've messed with various light settings as well as recording types - just doing my head in at the moment. I'm running K 9.2 methinks.

Thanks in advance...

2 Reply by lex_81

  • lex_81
  • Member
  • Offline

just let´s have a thought experiment
a quarter circle swing (approxm.2,5m @ clubhead)
assuming the swing needs 1 sec, that would be 2,5m / 525 frames/s = 4,7mm distance per frame ~ 9km/h
but:
swing starts slower and accelarates at the impact i.e. 150km/h > that means if the clubhead comes to the impactpoint the distance per frame increases significantly.
in theorie to get the above ~4,7mm distance/frame you have to increase the framerate by approxm 15.
~8000fps/ but that´s highend and needs a lot of special light /8-10Kw HMI/
basically it depends what you really want to see: if the exact point where the club strikes the ball in clean slowmotion, to my experience at least 10-15000 frames/s are necessary
but TBH in my private setup with USB slomocams, for me much more important is the shuttersppeed/exposuretime. Working with 100 or 150fps with clean edges it gives enough information to see what happens and it helps a lot to put some markers (points and strokes) on your club

3 Reply by obevz (edited by obevz 2021-11-30 23:11:28)

  • obevz
  • Member
  • Offline

Hi Lex...appreciate your inputs. Good info.
My science approach  is that if the camera records at 550fps...I know a golf swing from top of swing to strike takes around 0.2 ~ 0.3 seconds, lets say 1/2 second...so I would expect about 200 frames...when I do frame/frame...I do not get that many frames.
I'm not looking for "clean" info...rather the stages of shaft/head position and where the release occurs - and if possible the face angle of the club (open/shut).

Also, I'm trying to identify the best setting of the "little" spanner in Kinovea for the camera. ATM I'm just doing hit and miss (pun re golf). Hence  the "help" request -

BTW -club head speed of 150mph is beyond all but 00.1% of golfers...head speed more like 100-115...and old people like me...around 90mph...

4 Reply by lex_81

  • lex_81
  • Member
  • Offline

:-) here in europe we use the simpler metric system so 150km/h ~ 90mph
the higher the speed and shorter the distance the more frames you´ll need
Kinovea has such good drawingtools which could help alot if you try it with markers on your club, maybe with reflecting or fluorescent one, than you also don´t need crazy spotlights, which do more harm to your eyes than benefit to your swing :-]

5 Reply by obevz

  • obevz
  • Member
  • Offline

Thanks to Reiner...I was able to change my settings and I'm all good wrt the frame capture issue...really pleased with getting a true 555 fps from my Basler USB3 camera...

2020-Aug-14 05:09:00
Re: Kinovea 0.9.3 (beta) (39 replies, posted in General information)
It rather sounds like a bottleneck in your system.
I would go step by step to evaluate it.
- Use only 1 camera with your recording parameters. 1,5 seconds delay and 2,5 seconds recording time fits perfectly to grab a golf swing.
- Set the recording mode to „retroactive“. Kinovea then grabs the video into memory and writes it to disk AFTER recording. So you can eliminate slow disk speed.
- During recording, look on the amount of dropped frames DURING recording. Simply click on the red button to get your video.
- If dropping occurs , try to use uncompressed formats like MKV or AVI, reduce the video size or reduce the framerate, if your camera supports it. I don’t know the options you have with your camera. Lowering size and fps will reduce load and dropping. Your camera still is monochrome, so the amount of data to be transferred is still low than using a color camera.
- Check your CPU-load to see, if your system (processor and chipset) is capable of transferring the data to the system. To my knowledge, direct measuring the USB-load is somehow difficult. At this stage, try to avoid any elongation cable to connect the camera. Plug it in directly.?Basler has a tool to measure the maximal bandwidth, but you have to connect a Basler camera. By the way, they have a new camera line released this year at about 290 Euro that is very interesting.
- Once you get a stable run with 1 Camera, you could add the second camera. The best option is to have a second USB-Port on your system. However, a lot of systems have more than one USB-connectors, but NOT a real second independent port. It depends on your Chipset of your motherboard. if you have 1 USB2 and 1 USB3, connect each camera to one of these ports. Be sure not to have other demanding devices connected parallel on this ports.
- If you have connected the second camera (related to a separate Instance of Kinovea), check your CPU-Load again. On a motherboard having one of the newest Chipset Z390 with 2 independent USB-Ports (1 USB2, 1USB3.1), using an Intel i5/9600K one instance uses about 35% of CPU-Load during delayed recording. Upgrading to an i9/9900 it decreases to about 15-20% of CPU-load.
Hope that helps.
******************
Sure did...! thanks again