DIY Bike Fit - Motion Tracking

[thatmdee]

Hi all,

I was wondering if anyone has done a DIY bike fit -- and if so, what was the outcome?

I've been watching some of bike fit adviser's videos (http://bikefitadviser.com/ and YouTube) who suggests some basic motion tracking using software like Kinovea. So, palpate and place some tracking dots on (I have a role of bright green sticker dots from Officeworks), film yourself in 60fps or higher, and then put the resulting video through Kinovea to track angles, etc for the purposes of adjusting your bike fit.

Has anyone tried a method like this and had any success? Seems like for high frame rate recording, you need good lighting. Any tips for a cheap DIY softbox or similar to give some nice light so high frame rate recording allows me to accurately place and track dots on the ankle, knee, hip etc?

I haven't been able to find a lot of information out there on how people have approached it, and what success (if any) they had. I started delving into the possibility of writing my own tracking app using something like OpenCV in Python or C++. But I'm a little reluctant to go deep down the rabbit hole if it's not really going to provide any benefit.

I have a trainer (Neo), phone capable of 120fps recording, a small tripod.. So I'm most of the way there (bar some additional lighting -- i.e. DIY softbox). I'm more curious to see if others have taken a similar approach, and if, with minor adjustments over time, improved their bike fit.

I suppose I could always spend a couple of hundred on a bike fit, but I'm not sure who to see locally, and there seems to be so many bike fit tools/systems out there, it's hard to know which to go with. I like learning and self-sufficiency, and if I could get a reasonable result at home, it would be good -- as I could slowly tweak the fit over time to suit.

Thanks!

[g-boaf]

You can just use any portable lights.

Other way is use any modern video capable DSLR and bump up the ISO. Use F5.6 or F4.0 too and that should give acceptable results.

[thatmdee]

Thanks.. I was originally looking at LED worklights, as they don't put out any heat but seem to throw out a tonne of light, and seem to have a colour temperature of around 5000K (I know a photographer will scoff because it's probably not very accurate).. A basic tripod LED worklight is around $59 at Bunnings.

The main hurdle I've run into, is that if it's AC, it's likely to cause flickering (50Hz AC) -- unless the frame rate is a multiple of the AC's cycles per second. Unfortunately, the couple of worklights I saw seemed to be 50Hz AC input.. I couldn't see anything indicating they were transforming to DC though, which I gather would be fine.

We have a glamour mirror which uses LED bulbs, and is obviously AC.. When filmed at a high framerate, the flickering is quite bad.. So I'm working from the assumption of if I buy an LED worklight, array, or whatever -- it should probably be DC.

[g-boaf]

Yes, forgot about that. You can also just set yourself up outside? Put one of those photo backdrops behind you if you must have a featureless background.

[CaffeineAU]

Thanks.. I was originally looking at LED worklights, as they don't put out any heat but seem to throw out a tonne of light, and seem to have a colour temperature of around 5000K (I know a photographer will scoff because it's probably not very accurate).. A basic tripod LED worklight is around $59 at Bunnings.

The main hurdle I've run into, is that if it's AC, it's likely to cause flickering (50Hz AC) -- unless the frame rate is a multiple of the AC's cycles per second. Unfortunately, the couple of worklights I saw seemed to be 50Hz AC input.. I couldn't see anything indicating they were transforming to DC though, which I gather would be fine.

We have a glamour mirror which uses LED bulbs, and is obviously AC.. When filmed at a high framerate, the flickering is quite bad.. So I'm working from the assumption of if I buy an LED worklight, array, or whatever -- it should probably be DC.

The power supply in really cheap and nasty LED lights will not filter the rectified AC well at all, so it will flicker dramatically. A properly filtered supply will give quite clean DC and the LED's shouldn't flicker.

If there's any sort of dimming function, avoid it as it will use PWM to do the dimming and potentially cause more flicker.

Not sure what camp the Bunnings one fall in to though...

[inds27]

Hi @thatmdee

What was the outcome of your trial and error? Did you end up delving down the rabbit hole of trying to code anything in OpenCV or C++? I'd be really interested in having a look if you did! I've been trying to as well, also in OpenCV but I'm getting a bit stuck choosing which angles to specify (knee min or knee max at different parts of the pedal stroke for example). It works (I think), it's just a bit long winded!

Thanks!

[Calvin27]

I was wondering if anyone has done a DIY bike fit -- and if so, what was the outcome?

I did a retul bike fit which worked somewhat (fitter said I had gotten 90% of the way there via trial and error lol). I was intrigued at the process and attempted to DIY my on bike fit. My main approach was opposite to yours. instead of high fps shooting, I used low shutter speed long exposure. I can't find the files as it was about 2 computers ago but went something like this:

1. Setup trainer in a darkish spot. doesn't have to be pitch black but darker is better.
2. (optional) In the background, place the trainer and bike as close to a grid wall as possible. Then setup the longest lens and shooting distance your room/space can afford. I used a 55-200mm lens and zoomed in on the leg area (you can do the whole body as well and he retul did tell me some interesting things about the upper body).
3. Get some reflector stickers, ideally in small 1-2cm circles and stick them on the bits you want to measure. For me, side of knee, side of ankle. front of knee, side of hip.
4. Play with the exposure settings - I ended up with 5s and a dim light source to feed the reflectors. (optional) You can obviously add things like shutter delay to get skip the initial spinning up of wheels, longer/shorter exposure time to adjust the amount of revolutions etc.

What you should end up with is a plot of the movement of where you placed the stickers. It was quite close to the retul test and the common things were noticable - for me it was one right knee taking a somewhat strange motion when viewed front on. The difference with the retul is obviously the whole thing is calibrated and you get exact measurements. There are ways you can do this (i.e. grid at the back and a very long lens), but the general gist of it will be there.

Tips:
- Use a camera that have a phone app,laptop that you can fully control the manual control settings. You don't want to keep getting off the bike, it's annoying trust me.
- I prefer red dots but depends on the background and lighting conditions.
- Small dots are harder to position but give you much better resolution.
- Wear dark clothes and shoes etc. A lot of cycling stuff has reflectors which stuff the exposure.

*Edit: coffee has been consumed and I've now realised I respinded to a 3 year old post lol.

[g-boaf]

I was wondering if anyone has done a DIY bike fit -- and if so, what was the outcome?

I did a retul bike fit which worked somewhat (fitter said I had gotten 90% of the way there via trial and error lol). I was intrigued at the process and attempted to DIY my on bike fit. My main approach was opposite to yours. instead of high fps shooting, I used low shutter speed long exposure. I can't find the files as it was about 2 computers ago but went something like this:

1. Setup trainer in a darkish spot. doesn't have to be pitch black but darker is better.
2. (optional) In the background, place the trainer and bike as close to a grid wall as possible. Then setup the longest lens and shooting distance your room/space can afford. I used a 55-200mm lens and zoomed in on the leg area (you can do the whole body as well and he retul did tell me some interesting things about the upper body).
3. Get some reflector stickers, ideally in small 1-2cm circles and stick them on the bits you want to measure. For me, side of knee, side of ankle. front of knee, side of hip.
4. Play with the exposure settings - I ended up with 5s and a dim light source to feed the reflectors. (optional) You can obviously add things like shutter delay to get skip the initial spinning up of wheels, longer/shorter exposure time to adjust the amount of revolutions etc.

What you should end up with is a plot of the movement of where you placed the stickers. It was quite close to the retul test and the common things were noticable - for me it was one right knee taking a somewhat strange motion when viewed front on. The difference with the retul is obviously the whole thing is calibrated and you get exact measurements. There are ways you can do this (i.e. grid at the back and a very long lens), but the general gist of it will be there.

Tips:
- Use a camera that have a phone app,laptop that you can fully control the manual control settings. You don't want to keep getting off the bike, it's annoying trust me.
- I prefer red dots but depends on the background and lighting conditions.
- Small dots are harder to position but give you much better resolution.
- Wear dark clothes and shoes etc. A lot of cycling stuff has reflectors which stuff the exposure.

*Edit: coffee has been consumed and I've now realised I respinded to a 3 year old post lol.

Coffee makes everything better. interesting reply for DIY bike fitting.

I can do only limited changes to my bike fit, just saddle and position of the shifter hoods. One piece bar/stem is a limiting factor. I don’t need to do any changes fortunately, but for those looking at a new bike, that’s something to consider.

[MattyK]

Have done succesful DIY fitting.
Step 1: record a decent quality video, (well lit, landscape mode, full bike and rider in view, camera at hip height and perpendicular to saddle, etc)
Step 2: upload to Youtube
Step 3: submit to slowtwitch forum for expert review and feedback.
Step 4: if necessary, follow recommendations and repeat from Step 1.

[AUbicycles]

Is 60fps necessary... wouldn’t a regular frame rate still give you the same basic info to identify body movement?

A power meter tracking each pedal is a useful addition to a DIY setup as you want even up the power on both sides but also have fairly smooth and constant output. In theory it could mean adjustments to you setup (like shims in the cleats) to compensate or even set you in a position that isn’t 100% technically perfect (as per maths) but better suited to your body and comfort.

For a DIY setup a tip is a laser. For the front view you can set the laser on your knee (top of the stroke) and while filming can see both the total alignment during the pedal revolution and sideways movement.

Back to my DIY fit (before a professional fit), I was very close so on the seat height and position it was only millimeters but I got a different saddle and stem plus shims for the cleat. It meant that there wasn’t a big adjustment so it was natural and effective.

[trailgumby]

You get what you pay for.

There is so much more to a bike fit than people realise. Shoes are critical for me. There's a guy I use who is a wizard with shoe setup. AUBicycles knows him as well. There is absolutely no way I could have done the lengths of ride I have done in the last couple of years without incurring major niggles that would likely have put me off the bike for weeks without the work I have had done with him.

Until you've had a bike fit done by someone who knows what they're doing, you simply don't have the experience to know what you need to achieve.

[g-boaf]

You get what you pay for.

There is so much more to a bike fit than people realise. Shoes are critical for me. There's a guy I use who is a wizard with shoe setup. AUBicycles knows him as well. There is absolutely no way I could have done the lengths of ride I have done in the last couple of years without incurring major niggles that would likely have put me off the bike for weeks without the work I have had done with him.

Until you've had a bike fit done by someone who knows what they're doing, you simply don't have the experience to know what you need to achieve.

Folks, pay attention to the above, that’s absolutely right. Proper bike fit is very important for riding in comfort over long distances. That’s how I clock up 500+km weekly. I had a bike fit done by a very clever bike-fitter who knew me for a long while and had seen the way I ride, how I pedal, etc. It made a huge difference.

[warthog1]

Never had a bike fit done. Will pass 100k on strava this year. Rode for years without strava.
Done multiple long rides.
I paid attention to my body and adjusted cleat position, reach, seat height, set back etc to suit. Got my top tube length right and used handlebar stem length adjustment to suit.
Yes it was an evolutionary process but so was position as fitness, strength and flexibility improved.
Latest bike I just parked beside an older one and copied the meaaurements. Came out fairly close and has been fine.
Never had any sort of ride reducing injury that hasn't been a result of crashing.

Just to add it was initially done whilst racing so position was aimed toward speed. Pretty easy to tell what works when you are able to test your performance against the same variables(other riders). I don't race now but done enough ks in the same position that it still basically works.
I bunch ride with some very good riders so hanging on when they are going for it is enough, for an aging never was like me.