You are not logged in. Please login or register.
Kinovea - Forums → Posts by Chas Tennis
I don't believe that there is an output on the FH100 for live video but I'm not 100% certain.
See Reply #3 and the ones at the end of the thread.
http://www.kinovea.org/en/forum/viewtopic.php?id=435
Suggest also that you look at the User's Manual for computer or TV operations.
(I have tried to determine if the FH100 with Eye Fi card could be used to immediately view the camera recording on an iPad but have not found that information.)
That is exactly the information that I have been looking for!!
It is easy to see why I could not see forearm pronation at the same time as internal shoulder rotation(ISR). Forearm pronation occurs while the elbow is still bent before the semi-straight arm position, internal shoulder rotation and impact. Forearm pronation is difficult to see without well placed markers on the wrist and elbow.
Thanks!
Some considerations for your golf set up-
Standard Approach: For most golf swing analysis the Casio FH100 and earlier Casio cameras with manual shutter speed control (at about 240 fps) do a completely adequate job in outdoor light. For tennis I have put black tape markers near my elbow bones to indicate upper arm rotation (internal shoulder rotation). One early test indicated that Kinovea could track the black tape on my arm. I did not pursue farther to optimize the tracking. Other convenient markers might be, for example, circles on dark background clothes or sweat bands with bright circles. For golf swing tracking in strong sunlight illumination this seems like a feasible tracking approach.
Retro Reflective Tape & Illumination: To illuminate the few square meters of a golf swing in direct sunlight or with overcast skies seems very difficult. To send a strong beam onto a much smaller area might be feasible. ? Using a very large mirror behind the camera lens to reflect the sunlight onto the golf swing might also work to illuminate retro reflective tape. But a large mirror is a cumbersome set-up and the reflection would disturb the golfer.
LEDs on the Golfer: On an overcast day directional LEDs placed on the golfer and club with output beams aligned toward the camera lens should be visible in the video. In direct sunlight? But if the LEDs change pointing angle too much the signal will go down or be lost. Directional LEDs worked for the bicycle application indoors. Probably also workeable for runners and other applications where alignment is maintained. (For tennis applications where everything turns this alignment issue seemed a big problem to me.) Diffuse LED's don't have as much sensitivity to alignment but they will be much less bright for outdoor applications.
Camera Lens: For accuracy avoid extremely wide angle lens set-ups where the magnification at the center of the image is considerably greater than toward the edges. The more telephoto lens with the camera farther from the object the more 'flat' and accurate the image will be.
.................Did you end up getting Stosur's kick serve?
I'm contemplating going down the 3M retroreflective tape route, on top of 20mm spheres, to measure 10-15 places on a golfer (and one on the club head) via multiple cameras. ..................................................
Idea is to add one or more IR LEDs to PS3 Eye cameras and replace their IR filter with a visible light filter and work solely in the IR domain (I'm only after tracking information leading to data, not image). No idea at this stage how many IR LEDs will be needed to work outside or even if I've have enough power via USB to drive them all.
I do have a Casio FC100 but will go with the PS3 Eyes because they are cheap, they can be synced and give me shutter control. They can do 125fps but only at 320x240 which I'm thinking will give insufficient accuracy, so I'll stick to 60fps with 4 times the pixels. Frustratingly the FC100 doesn't have shutter control in its high-speed modes, unlike your FH100. Unfortunately the PS3 Eye cameras also have a rolling shutter which give noticeable distortion at golf swing velocities.
Stosur's kick serve
Update: 9/30/2012 - https://vimeo.com/40449544
I have other serve videos and can easily see internal shoulder rotation but I'm having trouble clearly seeing pronation.......the term 'pronation' is badly misused in tennis discussions - outside its proper definition. My current high speed video objective is to understand the timing of pronation relative to the main driver of the serve - internal shoulder rotation. A fun project.
Search "motion capture" and see what these professional motion capture systems are using. I believe that many retro reflective systems use rings of many LEDs on each camera with the camera lens in the center (optimal for the retro-reflection). I believe that nearly all motion capture systems work indoors where the lighting can be controlled. One had some outdoor capability. See Vicon link. Maybe there was another. ?
I posted a thread for information on high speed video cameras and considerations.
http://www.kinovea.org/en/forum/viewtopic.php?id=435
See Reply #1 for frame rate, shutter speed and Jello Effect (especially the golf examples)
See Reply #11 for a discussion of the very large difference in illumination levels - 100X - between indoor lighting and outdoor sunlight.
See reply #12 for more on the Jello Effect.
Look at my Vimeo video tests for LED source strength and also Jello Effect tests. These tests are simplified simulations of actual high speed video conditions. I'd rather be able to calculate Jello Effect under all conditions, camera settings, etc. but don't know how so I simulate them. These tests are very easy to do and tell you if sources are bright enough or if you need to worry about the Jello Effect. (Unless you test, I'd worry.)
For example, this bar on a disc is rotating at about twice the speed of a 100 MPH golf swing viewed from the side and filling the frame.
http://vimeo.com/21529497
Using the Casio FH100 at 240fps, the bend in a golf club would be about 1/2 the bend seen in the video simulation.
Recommend doing calculations in Reply #1 for your PS3 for various frame rates and shutter speeds. I assumed a 100 MPH golf club head speed for my examples. Set up some Jello Effect tests and test for your marker and lighting approach.
...................................................
Following is a photo of frontal knee tracking for road bike fit using 5 mm LED markers on each knee:
.........................................................................................
Nice result.
Could you provide some of the photographic details -
1) Output wavelength/color of the LED?
2) Was the LED directional or diffuse? If directional, any issues aligning the LED emission cone (10°?) to include your camera lens?
3) Camera model, frame rate, shutter speed, ISO, etc. Estimate of the resulting motion blur.
4) Did the LED appear to saturate the signal levels in the video?
If you try some outdoor measurements in direct sunlight, specular reflections from bicycle surfaces and background objects will compete with the LED. On a completely overcast day these solar specular reflections would be absent and would make tracking of the LEDs easier.
Have you tried placing retro reflective tape on the foot and blocking other interfering specular reflections such as from chromed metal parts, painted surfaces, etc? Place the light source next to the camera lens and illuminate the retro-reflective tape on the foot. Or copy the illumination techniques used in 3D motion capture. Some discussion and references are in the marker thread.
Kinovea tracked the ping pong ball marker on my tennis racket well when the background was dark or featureless.
Your application may be favorable for the use of markers especially since you can view from the side and the markers probably would move in a plane. Is the bike moving or a stationary bike?
Thread with marker discussions:
http://www.kinovea.org/en/forum/viewtopic.php?id=404
My experiments with markers for tennis stroke analysis were brief. Kinovea tracked a brightly painted ping pong ball marker on my tennis racket until it got near a white line in the tennis court. It always lost track approaching the line.
I did not spend much time looking for improvements in the marker approach for tennis stroke analysis partly because tennis strokes are a complex 3D motion in a cluttered background. I believe that 2D motions such as peddling or running are more favorable to using simple markers with a single camera.
If you describe your experiment from the optical standpoint I'll see if I can give you some considerations for a marker approach.
Interesting discussion on the term pronation and I think that we can agree something is still missing in how we both see the usage of the term.
In USA tennis I believe that most average players don't have any idea of what the term 'pronation' invloves. To make matters worse many informed tennis parties, instructors, etc. incorrectly use the term 'pronation' to describe a combination of pronation and internal shoulder rotation because both add to turn the forearm/wrist. Since internal shoulder rotation uses entirely different muscles than pronation and contributes much more power to the serve, it is a very detrimental misuse of terms. I was misled for decades............
As for now, to me pronation is CCW rotary motion [between elbow & wrist] when looking at my right hand regardless of the angular position. I have a big uncertainty as to the absolute angle to be assigned. ROM (usually around 90°) is measured with the thumbs up arrangement as described with a "0" degree at the top as a reference. That reference "0" may possibly only be used for ROM measurements. ? ( To me there is a term inconsistency if the wrist is in the supination side of the circle but is rotating CCW, still pronation? For example, 'The wrist is in supination and pronating.'?)
I hope that the OP does not have any similar issues with foot supination. With a small ROM and no markers that measurement from video seems very challenging. (Would attaching a laser pointer on the bottom of the foot and videoing the pool scattering of the beam be a useful marker after the fact?)
Chas Tennis wrote:[Term Usage Pronation. ..............................
.............the anatomically neutral position being full supination - so standing with the elbow flexed at 90 held in to the side and thumb straight up is 90 pronation rather than zero. ..................................
Thanks so much for the information. You clarified many points.
There is still one point that I am unclear about. It might involve conventions in certain countries - definitions of pronation and supination that could be different between the US and your country. It also might involve a difference between the absolute angle of pronation that you have discussed and angles discussed for measuring range of motion for pronation.
You say above that with the thumb straight up pronation angle is 90°. The usage in the videos below for measuring range of motion of pronation & supination I would interpret to indicate that the pronation angle is 0° for the thumb straight up. For measuring ROM the difference between two angular reference points is measured so the absolute scale might not play a part. ? Still, they set their goniometers to 0° to begin ROM measurements with the thumb up.
http://videos.rehabstudents.com/measuri … upination/
Beginning half of the video as the last part covers very limited ranges of motion for injuries.
http://www.cdc.gov/NCBDDD/video/JointRO … index.html
My issue is to simply measure and time pronation in videos of tennis serves like this one.
I can see internal shoulder rotation very clearly by movement of the elbow features but the quality of the video for seeing the wrist position without markers makes me very uncertain about the angle of pronation.
I’d appreciate any observations, comments or suggestions.
The OP wants to measure foot supination. The ROM for foot supination is probably much less than the ROM for wrist pronation in the tennis serves that I’m having difficulty with. Measuring foot supination angle seems like a very challenging goal.
I don't know how to measure foot supination nor have I successfully measured similar rotatory body motions.
I have a similar problem - to measure wrist pronation during a tennis serve. To see pronation (between the elbow and wrist) and to separate it from upper arm rotation (internal shoulder rotation) the angular difference between the elbow and wrist must be measured (in this case from a single viewpoint camera). I have videos that show internal shoulder rotation very well but it is difficult to get the wrist position from the videos without markers because of the quality of the videos. I cannot place markers.
If I could place markers and duplicate the set up after the original videos were taken - and maybe you can? - I might use markers as placed in this report on baseball pitchers.
Look at the markers used to indicate shoulder angular positions for baseball pitchers, photos page 51.
http://www.jssm.org/vol7/n1/7/v7n1-7pdf.pdf
Place the swimmer and camera in the pool as originally taken and place very clear markers on the swimmer's foot, leg, etc.. Of course you can do most work to develop and try out the technique without the pool. Calibrate marker position vs foot supination in the video. Maybe you can get a reasonable estimate using the markers and then use that as a guide/template to estimate the foot supination in your data videos. Could Kinovea lines be used to construct a calibrated scale for supination given the set up is nearly the same.
[Term Usage Pronation. I recently realized that my understanding of anatomical pronation was uncertain for the hand. I believe, but am not certain, that pronation for the hand is defined as the angular position in degrees from an anatomically neutral position. Bend elbow, thumb straight up/vertical. Rotation of thumb in medial direction gives degree of pronation but in lateral direction degree of supination. In other words, I'm still unclear if by definition pronation is only an angular position (degrees) and if used for a direction of motion (counter clockwise) the term is not being correctly used. Uncertain of usage..?.. The sloppy usages of the term "pronation" in tennis is ridiculous.]
Sometimes videos can be quite dark and I find the fact that the background of the user interface is white it can be quite difficult to see details.
If the user interface could be reversed with a dark background and light text, lines etc.
There is a reason for what you have noticed.
When light enters the eye and travels through it some light may scatter and then be spread across the retina. All eyes would scatter, but especially if the viewer has some cataract in his eye lens the scattering would be greater. In the case of a bright screen surrounding a dark video some of the light from the screen scatters into the area of the retina that the video is imaged on. With a dark video the scattered light may make the video harder to see as you pointed out. You can block out the surrounding screen light with dark paper - cut a rectangular opening and hold the paper somewhere between your eye and the screen - and see what difference it makes to you.
Likewise, if light from a desk lamp, overhead lighting, or window, etc., is shinning into the eye from any direction, that scattered light will have the same effect. You can quickly check for the effects of stray light by holding your hand up to block it and see the difference it makes when viewing darker videos.
A dark screen/light screen option might be a plus in some cases.
Re: [Discuss] Saving dialog - simplification project
(Please excuse my Quote method. I somehow often seem to lose the site Quote method when copying & pasting to Word. Probably my error.)
...............................................................................
Clarification of terms –
As soon as you add a key image, a stopwatch or a track, ……………..
“key image.” You use “key image” is a specific way that is still unclear to me. I’m not very knowledgeable of computer terms and Googling "computer key image" gets pictures of door keys.
Whenever I hear the term “key image” I am trying to interpret what it means and to what it applies. Is it a single identified frame of video perhaps with something added such as a track or label? How does the term “key image” - when used for a track - relate to a series of video frames that shows both a developing track and also the complete track? Are all frames with tracks, etc.“key images”? Or is it the image itself only of the track, label or stopwatch with information on where it is placed in the video frame?
The export menu would always be enabled to let you save the video as an image sequence or to save it slow motion even if you did not enrich it.
“an image sequence" has no special meaning in terms of “key images”? Correct?
This is definitely possible. When you save a KVA file, you get to choose the file name.
- If you keep the suggested name that matches the video file name, this particular KVA file will be opened automatically when opening the video. (companion file).
- If you choose a different name, you will have to import it explicitely later, using "Load key images data" menu.
Assigning the desired KVA must be done before opening the video or the “companion” KVA file will be applied in opening the video? Correct?
"Load key images data" will list the available KVA files for the selected video?
edit:
Some caveats,- Contrary to the current default option, the original file would not be backed up in any way, so it must be clear that it shouldn't be deleted.
Hopefully this will be clear since you will only be able to use .kva as format target.
Terms. Not familiar with the term “a format target”. The term“file extension” is very familiar to me.
http://en.wikipedia.org/wiki/File_format
But the user still has to understand that this is not a video format, and that this file will only "work" when imported onto an existing video.
- Currently the observational references, (.svg and images), are not saved into the KVA.
(It is certainly technically possible though, as SVG is XML, and raster images could be stored either in CDATA tags or encoded in Base64.)
I guess by "observational references" you mean things such as the cross hair coordinates that are useful for work secessions but that you would not want to display? Images? Is there a list of the observational references?
This would leave only two options on the dialog, and this could actually be handled differently:
1 - There could be a simple "save" menu that would always save analysis to text file. (KVA). (KVA = Kinovea Analysis text files?) Do nothing if there is no key images or tracks, etc. ?
Please clarify. It would be important for me to save KVA files with tracking information. Also more than one KVA file might be associated with a single video. I am unclear whether the KVA saves the tracks.
2 - And a separate "export" menu that would save the working zone back to a video. Now this menu could yield a dialog allowing for various kind of exports (diaporama, time freeze, maybe image sequence, etc.)
Oh, that could even be designed as a tabbed dialog for extensibility (youtube export, dual export with vertical/horzontal option, etc.)
I have no experience with these video options. I used Vimeo a few times to test earlier.
What I'm doing may be typical for an individual trying to develop an athletic skill with limited and optional needs to communicate display videos.
I want to analyze my tennis strokes using high speed videos & Kinovea. I don't yet have much experience using Kinovea version 8.15.
Description of my use –
1) Take high speed videos of my strokes, friend’s strokes and pro strokes. Target specific stroke motions such as internal shoulder rotation (total arm) and forearm pronation (elbow to wrist) on the serve. For pro servers I have 150-200 high speed videos of this compound axial arm rotation at 240fps.
2) Next, practice my serve trying to incorporate what I see in the pro videos (making adjustments for safety, the skill level difference, etc.) I still don't have the conclusion from step one that I need on forearm pronation timing relative to internal shoulder rotation. That's because the pros don't have markers on their arms and I'm unsure of the timing of both pronation and internal shoulder rotation prior to ball impact. Pronation is very difficult to see with certainty while internal shoulder rotation is very easy to see. The video would work with markers to clearly see pronation.
3) Future. Using Kinovea compare my practiced serve to the pro-serve regarding the timing of both pronation & internal shoulder rotation. No practice yet until I understand pronation timing. When I use Kinovea I would want to save more than one analysis for a given video especially for tracking different objects. There would be one video and sometimes several KVA files associated with that one video. Save new or modified KVA file for each work secession with a given video.
4) Finally, in the future communicate what I have using Kinovea videos with analysis tool results permanently in the video using the option discussed here.
[edit joan : added quote tags for clarity]
................
I want to be able to measure the position of a single player on a pitch ( so distance to goal-line and distance to side-line )
on a single frame.
Is it possible using the perspective grid ? and using the measurement line ?You can measure the length of the side-line and goal-line.
Can you derive the position of a certain player from this ?The camera is placed high up in the stands and is perpendicular to the field. ..................
I don't know work very much with that type of measurement or know the answer to your question. Generally, you need to have a scale and the field seems ideal.
To approach it try -
1) Place the 4 corners of the grid on the corners of the field. The grid lines should show the player's position, x,y on the field. ?
2) If only 3 corners of the field are visible maybe placing 3 corners of the grid on those 3 points and making grid lines parallel to the field lines might properly place, orient and scale the grid. ?
3) Using 8.15 I could align & size the grid using 3 known points on the tennis court (corner, singles side line & baseline, one another point) and court lines. I used more that the grid corners for alignments, I used some inner grid lines.
I have not used a measurement line.
Not sure if this is a similar issue or not but we are trying to use the 'Image Capture' / 'Delay display' option on a pretty high spec. PC with > 60% free RAM (4GB) with a Panasonic NV-GS330 min-DV camera connected using Firewire. We are trying to use this to display time-delayed routines of competitive trampolinists, something we have seen done with Timewarp software, it works but the imagery doesn't seem clear enough at the extremities (e.g. arms moving fast through a somersault) although it was on the Timewarp we tried as a demo version.
It sounds as if it might be motion blur if the extremities are blurry and the other parts of the image are not.
I'm not familiar with Timewarp, or the shutter speed control for the Panasonic NV-GS330 (likely to be AUTO). Did you view the same video on both Timewarp and Kinovea with different results?
View your video in Quicktime or other player to determine if the blur is in the video or a software artifact. If slow objects are sharp and faster objects, like the extremities, are blurry then motion blur sounds likely.
If two videos were compared, what shutter speeds were used in each of the videos that you refer to? If AUTO exposure control was used then the shutter speed is unknown and so is the motion blur.
Some AUTO control cameras might adjust the shutter speed to short times when the light is bright and to long times when the light is less bright. For those AUTO control cameras the motion blur will vary with the light level. For example, in outdoor sunlight the shutter is very fast with little motion blur but indoors it very slow with large motion blur. If you took your videos indoors the light levels are low and the shutter speed has to be slow to get useable videos.
I am not very familiar with video file compression or its artifacts- I just avoid it. Make sure that your camera has not applied video compression in saving the video file.
If you think that motion blur is an issue for you, see the other recent threads discussing it.
Jello Effect Measurements for the Casio Ex FH100 Camera
The Jello Effect discussed earlier in this thread, Reply #1, is a distortion of the image produced because differently lines of the video image are recorded and read out at different times. For example, a line at the top of the frame might be recorded & read out at time, t, but a line at the bottom of the frame will be recorded & read out at a later time, t + delay. (The shutter speed for a given pixel is unaffected.) This effect is caused by the operation of the camera sensor’s electronics. The shutter operation that causes the Jello Effect is called a Rolling Shutter. Rolling Shutter operation is used in most/all CMOS cameras like the Casios and other affordable high speed video cameras.
An example of the Jello Effect is a false bend in the image of a golf club that is moving very fast. See reply #1. I believe that the more lines of video or still pictures that are involved, the greater the Jello Effect is likely to be. Therefore, be aware of the Jello Effect especially for still pictures and HD video pictures that have many pixels and lines to read out.
There is also another shutter type, called a Global Shutter, where the image is recorded, recording is stopped and then the lines are read out. Global Shutters do not display the Jello Effect.
It is easy to search the internet for: 1) Jello Effect, 2) Rolling Shutter, 3) Global Shutter, 4) Youtube Jello Effect, and 5) Youtube Jello Effect + Camera Model to find lots of information and many videos on Youtube. Often the model of camera can be listed and examples of videos and stills can be found for that model.
Still, it is difficult to find quantitative information on the magnitude of the effect for a given model of camera. I have not found useful information for the Casio FH100. The manufacturers do not quantify it in technical notes for you because it is not a plus to aid sales. It is another thing to worry about when deciding to purchase a high speed video camera.
That being said for most athletic motions captured with better quality high speed video cameras the Jello Effect will probably be small or negligible. I have not yet noticed JE in FH100 high speed tennis videos because the tennis racket velocity has not been high enough.
Test for Jello Effect. If you place a rotating disc with a line across the disc in front of the camera so that the line fills the image from top to bottom you can easily measure the JE for your camera for a given frame rate, shutter speed, etc..
Here are three Casio FH100 Jello Effect measurements for very fast objects:
1) Rotating bar at 8-9 revolutions/sec, 240fps.
2) Rotating bar at 21 rev/sec, 240fps.
3) Extreme Jello Effect at 330 rev/sec, 240fps.
As discussed earlier and in referenced links, a golf club at 100 MPH rotates at about half the rate of the first video above - a small but noticeable effect with the Casio FH100.
UPDATE 4/12/2014 - AIPTEK Action HD camera, 60 fps, 720 p. Cylinder rotates 8 rps. The bar is parallel to the cylinder side but it falsely appears to be at an angle because of the Jello Effect.
https://vimeo.com/30281692
Casio FH100, 240 fps. Cylinder rotates 21 rps. The bar is parallel to the cylinder side but it falsely appears to be at an angle because of the Jello Effect.
https://vimeo.com/29160003
UPDATE 10/9/2012- Stills Taken in Burst Mode
Jello Effect Tests for still photographs using the Casio FH100 in 'Burst Mode'. The FH100 in electronic mode will save up to 30 high-resolution stills taken at a rate up to 40 fps. In mechanical shutter burst mode it can record at a reduced rate up to 10 fps. Click "Gallery Page" to get descriptions.
http://forums.dpreview.com/forums/post/39393614
Kinovea - Forums → Posts by Chas Tennis
