We live in an era where everyone has a video camera at their fingertips, making video analysis both affordable and available for sports coaches and athletes. Overall, this is a very positive thing. But a lack of understanding of some important fundamental principles of technique analysis has led to misuse and, more worryingly, the viral proliferation of misinformation via social media. It makes me laugh, at times, when I’m told, “Anybody can be a biomechanist nowadays.”
Well, if only being a biomechanist was about pressing the red button to start and stop a recording.
On a few occasions I’ve been presented with a video and asked to “do my magic,” but the quality of the video is poor or it’s missing some key elements a biomechanist would need to extract anything meaningful from it.
Yes, anybody can press the red button, but there is definitely more to it than that.
Biomechanics can leave the lab but must bring along best practices
Sports movements characterised by fast moving limbs, impacts and short contact times typically require frame rates over 100 frames per second (fps) for biomechanical analysis.
Back in the 1990s, the only technology that could deliver this were cine cameras, which could exceed 100 fps. Video technology at the time peaked with VHS and SVHS tapes, recording at 25/30 fps. Experts would apply specialist software and processes to de-interlace each frame, which, if your deinterlaced each frame into two fields, you increased the video speed to 50/60 fps. Using our cine cameras, we would have to wait several weeks to get the footage processed before we could start analysing the film.
Tweet ThisEven with access to those sport-specific devices, my go-to is still video. It is the most engaging medium to assist in the coaching process, and it is available immediately for athletes and coaches to review the performance
@PhilipGrahamSm2
Smartphones and digital cameras exceed even the best specialist gear of 25 years ago. They capture at 120/240/480 fps in slow motion modes and the video is available immediately for analysis, a game changer for biomechanists and coaches.
The beauty of video is that it is accessible. Not everyone has the luxury of having motion capture, timing gates, speed guns or OptoJump, but most of us have smartphones with these capabilities “off the shelf.”
Even with access to those sport-specific devices, my go-to is still video. It is the most engaging medium to assist in the coaching process, and it is available immediately for athletes and coaches to review the performance. The question, then, becomes one of the quality of the information. By adopting a few basic practises, we can generate more accurate and consistent data that we can use to compare and evaluate our athlete’s progress.
Video biomechanical analysis starts with a trip to film school
Before we talk about the reliability of drawing lines and angles on video, we must first consider some aspects of recording that give us the best quality of footage.
Planning how to capture video will make the difference between it being usable for analysis and or just another memory in the family video collection. Practitioners should also remember that you don’t necessarily need to measure anything from video. Creating a photosequence of the positions at key instants and over phases of the skill provides a great source of feedback to athletes and coaches. However, even for more qualitative use of video, the following considerations are paramount.
Avoid blurring
Have you ever noticed that when you pause a video the image of the athlete is blurred? If this happens, you can make the image very clear by using a high speed shutter, a feature that most camcorders have. The shutter determines how much light enters through the aperture, which affects the image’s crispness. For “high speed,” set the shutter speed to 1/500th – 1/1000th of a second. Or, you can use “sports mode,” which does the same thing.
Unfortunately, smartphones generally do not have this function, which often means that recording video in normal video mode – even in 4K or full HD (FHD) – still results in some blurriness.
I currently use a Samsung Note 20 5G Ultra, and I can get really clear video when I pause while using the FHD at 60 fps. WIth my old iPhone 6, the only way I found to avoid blurred images was using the slow motion high speed video at either 120 or 240 fps. That allowed me to play the video back in slow motion or adjust it to normal speed. The downside was the reduced resolution of the image.
Be sure to check how the fast moving parts of the body – the arms and legs – come out when you pause your video. If the video is still blurry with normal video functions, record in the slow motion function.
Set a frame rate that can capture sporting movements
Frame rate can help reduce blurriness but you may want greater accuracy in your observations of key athlete positions.
Filming in slow motion at 120/240 fps gives you more frames in which to define critical instants, such as foot contact with the ground or the moment of impact while kicking a ball or swinging a golf club. Higher frame rates are particularly useful if you want to use the video to measure time-based metrics: contact time, step frequency, phase time.
I often hear coaches say that they prefer to watch the recording in normal speed, as this gives them a better feel for the timing and rhythm. I can totally appreciate that. They make coaching decisions everyday without necessarily reviewing video. However, the beauty of slow motion functions on smartphones is that you can adjust where the slow motion kicks in. You can then watch in normal speed and have the option to slow the playback to review a critical part of the movement, such as the take-off in the long jump.
Location is everything: Setting the vantage point
As a coach or biomechanist we need to be consistent with our filming position, particularly if we want to compare our athlete’s technique and assess whether he has made improvements.
To decide on the best filming position, consider the technical model of the sports movement you are analysing and in what plane(s) the most important movements occur.
In the long jump, the main movement is in the sagittal plane and the main technical considerations come in the last part of the approach and take-off. Consequently, I position myself in line with the take-off board. When filming the 100m, I would line myself up at the start line to look at the movement in the block; or put the camera in line with the 60m mark to look at the movement at maximum speed.
Steady cam vs. tracking shot
Depending on your intended use for the video, you have a choice of two modes of recording: fix a camera on a tripod and record a small section of the skill, or pan the video to capture the whole event.
For the long jump, again, I typically record the whole movement from the start on the runway all the way through the landing in the sand pit, standing in line with the take-off board. I zoom in at the start and zoom out as the athlete approaches, but always maintain an approximately 7 meter horizontal frame of view. That framing gives me a good image size of the athlete.
I also set up a fixed camera on a tripod to capture in slow motion the penultimate step, final step and take-off. Set he camera to manual focus (on infinity setting) so the athlete is clearly in focus when she comes into the view.
Additional considerations
If you’re filming a movement where the athlete covers a long distance, like the run up on the long jump, set up further away from the runway. This will provide a better perspective of the approach run (sagittal rather than frontal), and make it easier to keep the athlete in the centre of the frame. The zoom function can capture a larger image size, as necessary.
If you’re using your video footage to record split times, position yourself in line with the split markers, and make sure you capture the flash of the starting gun at the start.
More than anything, practitioners need to develop a recording strategy and implement it consistently in order to obtain video that permits valid comparisons.
Video analysis preparation: Error checking, calibrations and calculations
Many mobile apps and computer software packages are available for biomechanists and coaches to enhance their observations, reviews and analyses. Practitioners can play, pause or advance the footage frame by frame; capture screenshots of key instants; and quantify aspects of movement like time, distances. or angles. But just because users can generate a number doesn’t mean that the number is accurate.
Tweet ThisCompetence in using a technology is not just about connecting it and getting some numbers
@PhilipGrahamSm2
Measurement errors
Figure 1 shows an action figure with a fixed knee angle. I’ve recorded it from several different vantage points, starting from a sagittal position and coming around to a frontal position. The true angle in the sagittal plane is constant at 118.1 degrees, but my measurement increases to 169.5 degrees purely because of the camera orientation. Correct and consistent vantage position can make that much of a difference.
Another consideration is the frame rate. When Jonathan Edwards broke the long jump world record 27 years ago, the live TV broadcast replayed the jump and the commentator informed the viewers that Edwards planted his foot directly under his body. That, the commentator said, was the key to breaking the record.
Well, in all the analysis I’d done on Jonathan and other jumpers, this was never the case. The foot is always planted slightly in front of the body in order to execute the take-off.
The difference was that we used a higher frame rate than the TV broadcast, which was 25 fps. The lower frame rate gave the impression his body was directly over the foot at touchdown, but in reality they missed the detail because of the lower frame rate.
I’m pleased to say that TV broadcasts and livestreams use high speed video more often nowadays.
Setting the conditions to measure the right metrics
Understanding the mechanics and the technical models of the sports skill will guide you to knowing what metrics to quantify from the video. If you want timing data alone, then your major consideration is the speed of the movement you’re analysing.
For sprinting, you ideally need to record at 240/250 fps to be accurate to 0.004 seconds. In comparison, if you record at 25 fps the interval between successive frames is 0.04s. Putting this in context, at 25 fps you may get 2-3 frames of ground contact for an elite sprinter running at max speed. At 250 fps you’ll get around 25 frames, which makes the key instants of foot contact and toe-off much more defined and the contact time more accurate.
If you want to simply count the number of foot contacts in a 100m race, normal speed video will suffice. But if you want to determine step frequency at maximum speed, you’ll need to count the number of frames in a gait cycle, i.e., the instant of toe-off from right foot to the instant of left foot toe-off in the following step. Normal speed video at 25 fps would capture six frames. The step time would be 0.24s and the step frequency 4.17 steps per second (step frequency = 1 / step time). Recording at 250 fps you may find that the step cycle was 55 frames, giving a step time of 0.22s and a step frequency of 4.55 Hz. Now this may not sound too bad, but we know that running speed is the product of step length x step frequency. Let’s assume that the athlete has a step length of 2.5m. The speed estimates would be 10.43 m/s and 11.38 m/s, respectively.
Clearly, the error in the measurement of step time due to a low frame rate results in a very poor estimate that you could not rely on to infer any real changes in performance.
Similarly, if you determined there were 5, 6 or 7 frames in the step cycle, then the velocity estimates would be 12.5 m/s, 10.43 m/s and 8.93 m/s, respectively – variations too large to have any meaning. A one frame error at 250 fps results in speed measurements of 11.57 m/s, 11.38 m/s and 11.16 m/s. The error per frame is about 0.2 m/s, and this feeds into our assessment of whether the athlete has improved.
We can more accurately determine the key frames of foot contact and toe-off by having sufficiently sized, clear images taken at the most appropriate vantage point.
Different measurements require different visual calibration references
Measuring distances, such as step lengths, require a reference or calibration object to scale the video.
Calibration is not just putting an object like a hurdle in the background and using that as the scale. That approach is highly inaccurate due to errors in depth and perspective. We must, instead, calibrate in the plane of movement.
If you’re using a fixed camera, recording a hurdle in the middle of the lane at the beginning or end of the session is sufficient as long as the zoom and camera position do not change during recording. Calibrate the video with the hurdle dimensions in the software, and import the horizontal and vertical calibration values into the videos of your athlete performing the skill.
This type of calibration is appropriate for a detailed quantitative analysis where you manually digitize a full movement sequence using a segmental model to analyze technique. The result would be instantaneous positions of the centre of mass, horizonal and vertical speeds, joint angles and joint angular velocities for each frame of movement.
For measuring step lengths, the easier option is marking the floor before the session, giving you a “permanent” horizontal reference scale within the view.
The white tape in Figure 2 is at one-meter intervals on either side of the lane. The tape provides a scale across the ground that we can use to determine the position of the toe relative to an origin – in this example, the front of the take-off board.
However, we must calibrate through the plane of movement, not just along the edge of the track. In Figure 2, proper calibration through the line of the foot indicates that the foot position is 2.10m from the board. If, instead, we calibrated along the near side of the track, this toe position would come back as 1.95m; at the far side of the track it would 2.23m; and, if you used the hurdle in the background as your calibration point, your output would be 2.88m.
Be aware that with this method you need to recalibrate the video for each foot contact – always in the plane of movement.
Synchronizing video with other technologies
Synchronising video with other technologies has a multiplicative effect. The combination gives us the extra piece of visual information that helps us understand the phase of a movement that another signal, e.g., force or EMG, relates to.
We tried many years ago to incorporate a video feed into the ForceDecks platform. We had some initial success with GoPro’s, but it wasn’t the seamless and user-friendly solution were hoping for. Understanding something like the countermovement force-time graph is not easy for first time users and, despite breaking the graph into phases (unloading, eccentric, concentric, etc) and overlaying the displacement graph, it is still helpful to visualise the athlete and see the positions they’re in.
However, earlier this year ForceDecks released ForceDecks Vision, which captures high-speed video concurrent with force data. This helps practitioners see how and where asymmetries occur and what area is the root cause.
But again, we must be mindful of some important considerations here.
When collecting signals from other devices, you may have to question the usefulness of synchronising a 1000 Hz sampling rate with video filming at 25/30 fps. Extracting useful, time-scale relevant insights from the data requires us to use the high-speed video (slow motion) function to marry the data streams. By increasing the frame rate from 25/30 fps to 120/240 fps, you improve the ratio of frames to force data points from 1:40 to 1:8. When analyzing a drop jump contact time of about 0.2 seconds, extra frames make the difference between a spurious conclusion and a valid observation of how the athlete accepts and generate force.
Practitioner competence is the ultimate limitation of any technology
Accessibility of technology is no longer a limiting factor. The lack of knowledge on collecting accurate and meaningful data to inform decisions in sport is.
Tweet ThisAccessibility of technology is no longer a limiting factor. The lack of knowledge on collecting accurate and meaningful data to inform decisions in sport is
@PhilipGrahamSm2
More than ever, the fundamental measurement processes I learned at university and developed over my career are essential to the work I do with athletes, students and coaches. Sadly, these practices are not common knowledge among sports performance practitioners. Consequently, practitioners may not have the scientific underpinning of the measurements they are taking and making decisions from. At the same time, many sports scientists are “pluggin’ and playin’” without fully understanding the finer details of the quality of data.
We are constantly reminded by the athlete monitoring system companies that “the quality of what you put in is what you get out,” so we have to go back to basics here.
To make our data more impactful, we must take a step back and address the fundamental questions of what information we need, and then what technologies are available for us to obtain that information quantitatively. After choosing a sufficient product, you must understand its limitations and the quality of the data you’re generating.
Competence in using a technology is not just about connecting it and getting some numbers. Practitioners should understand the errors and where in the data collection and analysis these errors arise. This requires an understanding of the scientific principles that underlie the technology and its measurements. Only then can we discern when either the tech companies or we ourselves are stretching the limits of its capabilities.
It’s good that companies innovate and push the boundary of their tech, but the practitioners have a responsibility to understand the devices’ limitations and let those limitations determine the boundaries of application and analyses.
Developing sports practitioners who can make technology useful
The sports science curricula of university degree programmes have changed alongside the practical techniques and methods in the field. More experienced practitioners have to understand the effect that the changes in education have on the mindset of younger practitioners entering our environments.
Education tends go directly to application. The speed with which we can access information – whether basic principles or real-time athlete performance data – has displaced a full understanding of the basic principles that underlie everything we programme and measure. How many practitioners are using force plates without really understanding Newton’s Laws?
Applied content and “evidence-based” sports science publications can benefits the field and our athletes greatly, but not if they come at the expense of less thinking time, less preparation and less thorough analysis.
With that, some parting guidance relevant to all technology integrations for sports performance.
- Before incorporating any technology into your practice, read up on the fundamental principles that govern the quality of the data. Go so deep that you reach the point where you can find faults, identify and correct data collection errors, and understand the limitations of the equipment at hand.
- Find yourself a mentor. A degree gives you some knowledge, but don’t be so eager to think you know it all. A mentor can save you more time than you can imagine because they’ve probably tried to solve similar problems, and have the experience and knowledge of what worked and what didn’t.
- Don’t go directly into the “we can measure that” mentality. Try to work out answers to problems by first applying fundamental principles. You’ll be surprised how useful this process is in developing your understanding and saving time.
- Develop your ability to critique and question your “evidence.” Simply because an article is in a journal does not mean you should automatically rely on it for “evidence-based practice.”
We are in an era where there is far too much noise in the sports science and S&C world. Publications and social media content often serve other personal and academic agendas, and therefore are not always aligned to developing athletes. Your first filter is your understanding of the fundamental principles of science and your area of expertise. From that foundation, question how you can apply the findings or concept. You’ll be surprised just how much falls into the gap between theory and practice. “This was the plan” is the going-in position for a lot of reporting, but the reality may have been very different. High-performance sport is a chaotic and changing environment that, by its very nature, compromises experimental control. Remember that you are working with athletes, not subjects, and their development far outweighs your desire to “science” them.
