Bike Fit App rebuttal

Recently I did a review of a popular bike fit app developed for the iOs system where I tested it against an infrared motion capture system to see how accurate it could be.  It didn't fare very well and the developer wrote a comment explaining why he thought I was wrong to be so tough on his app.  Initially the comment got caught in my spam filter so I just saw it and took 10 minutes to respond to each of his points.His points are in blue while mine are in black:As the developer of Bike Fast Fit, you can probably guess that I am disappointed in this review. You implied that Bike Fast Fit software itself responsible for the difference in your measurements. I believe you are wrong as there are 5 main sources of error in measuring a bike fit with videos:1. Parallax – As you mentioned in your previous blog post, getting the camera plane in line with the bike plane is critical. Bike Fast Fit has great alignment aids to minimize this as much as possible, so I’m assuming this isn’t the source of error.Agreed that this can be a major cause of error in video bike fits.  I like the alignment aids on the app however I can't help to notice the difference between iPhones/iPads (and all mobile devices really) show when you have a spirit or bubble level app.  Lay four or five of them on the same table they often will all have different readings -- they're not off by a lot but they're still off.  Can't help but think that since your app is using the same technology for it's alignment tools that this plays a role in the differences between devices.2. Identifying Points on Anatomy – As you will be covering in a future blog post, identifying the points on your anatomy is critical to getting accurate and repeatable measurements. Usually this is done with placing dots or other markings on your body. Bike Fast Fit has detailed instructions for placing each point and being a PT, I’m assuming you were consistent in your placement. But it only takes a difference of 1/4″ in location on the Knee Dot to get more than 1/2 degree difference in Knee Angle. So if the stickers were moved at all from measuring with your infrared this could be a source of error.Agreed.  Setting landmarks is very important.  I've been setting these landmarks for bike fits for about 15-20 years -- pretty sure I got them in the same/right spots.3. Identifying the Dots in the Video – Putting your measurement lines in the exact center of the dot is critical to getting an accurate measurement. Bike Fast Fit has an offset handles with a magnifier to allow precise centering on the dots.Ditto.4. Camera Frame Rate – As you correctly mentioned in your previous blog the frame rate of the camera is one of the most critical things for repeatable accurate measurements. Unfortunately the devices you used (iPad and iPod Touch) both likely had a maximum rate of 60 fps. For Knee Angle this really isn’t fast enough. When recording at 60 fps and pedaling at 90 rpm (1.5 revs per sec), you are getting 45 frames per revolution or 8 degrees of rotation per frame. So it is unlikely that a frame is exactly at the bottom of a cycle. Knee Angles at the bottom of the rotation with videos at 60 fps change by 1.5 degrees per frame. The infrared system you were using likely is recording at 180 fps or 120 frames per revolution or 3 degrees of rotation per frame. This speed difference is the likely cause of most of the error you saw. Starting with the iPhone 5s the camera could record at 120fps and with the iPhone 6 we record at 240fps.Again, agreed.  Higher frame rate, better video.  I will stipulate that the devices I was using were not the best for this app.  I didn't do this for any malicious reason.  I did it because this is what I had access to.  And some of the people you are selling this app to will also be using these devices as this may be all they have access to.  So would my review be "app might be useful with an iPhone 5s or higher only"?5. Foot Angle – The last possible source of error that I will mention is the Foot Angle or how much you point your toe down at the bottom of the stroke. Changing your Foot Angle by a degree can change the Knee Angle by a half degree or more. Unfortunately this is a source of error that can’t be controlled by the software or hardware. Comparing Foot Angles across videos is critical to ensure repeatability, but they aren’t mentioned here.Yes, ankle and foot position does affect the knee angle.  But if we're looking at the same rider with the same bike, the same landmark dots, that's in the same gear and at the same cadence, the knee and foot position will remain fairly consistent from one pedal stroke to the next.  All of these recordings (video and infrared) were done in this manner.  Plus I went through and measured multiple pedal strokes to establish some consistency in the measurements.  Even if I took 5-8 measurements of the same angle in different pedal strokes the variance was low -- below the threshold for the accuracy of the app.Plus, I'm not sure you want me to dig into the ankle and foot angles too much.  Since I mentioned that the smaller measurements (from my article: "Of note was that on the smaller measurements, like trying to measure knee lateral travel, hip vertical motion, and especially ankle motion, the margin of error was much greater.") had some of the consistently worst measurements, I left them out because they didn't add appreciably to the discussion.  I wouldn't be confident using them as a reference for anything happening at the knee.I would bet if you ran the videos you took with Bike Fast Fit through Kinovea you would get the exact same angles.Yes, if I put these videos through Kinovea (the free PC-based software I recommend in my DIY bike fit series) I might get the same numbers.  This only reinforces the limitations of the app.  Because when I recommend the use of Kinovea, I also recommend that the best data will be obtained by breaking the video sequence into two recordings -- the upper body and the lower body, or something similar.  Essentially getting closer to, and focusing in on the joints you're wanting to measure.  Yes, this takes longer, but yields better, more usable data.  You don't make this suggestion in the instructions on your app, and if you did, you'd be unable to use the alignment tool on the app.Plus the Kinovea app is still free -- your app is not.Getting accurate and repeatable bike fit measurements from video is challenging and we continue to work hard at trying to make Bike Fast Fit as accurate as possible.I understand you arguing in defense of this product you developed, but you can't tout the greatness and accuracy of the app, and then blame any errors on the limitations of the video technology and the end user.  Video has definite drawbacks as a means of measuring something as fast (in cadence) and detailed as a cyclist on a bike, and your app doesn't do enough to offset these drawbacks.  There are two ways to overcome these problems and get accurate data.

• You can spend a lot of money on high frame rate cameras, in-line self-leveling lasers, lighting, etc.
• Or you can spend a lot of time by taking slightly above average technology and focusing it on a smaller portion of the cyclist -- essentially getting closer to the rider.  This requires multiple shots to visualize the whole rider though so therefore is more time intensive.

And I haven't even addressed the "normal limits" ranges you've recommended for the measurements, which as I mentioned in the review, are just wrong for a dynamic bike fit.  Following these ranges will not lead to a good outcome.You referenced a few of my articles and my DIY bike fit series, but I've never put it out there as something to use or be emulated by professional bike fitters -- in fact I make clear that if you'd like to be taken seriously as an expert bike fitter, you need (besides education and experience) to invest in technology that shows you are committed to this practice.   If your app was marketed as an entertainment app -- something fun to play around with -- I wouldn't take much issue with it.  But by touting it as built for professional bike fitters ("We listened to your requests and developed Bike Fast Fit Pro with the professional bike fitter in mind.....you can add business name, address, contact information, and logo to bike fitting reports and send PDF bike fitting reports right to your clients." -- right off the app itself) you infer a level of accuracy and usefulness you can't deliver.