Tuesday, 15 May 2018

Image stabilisation

The big problem with recording your own videos for cycling (for the interactive trainer) is quality.

Resolution: This is a perennial issue and the current GoPro Sessions camera is an economical place to start. The current model supports 4K or Super HD.

Image stabilisation: This is the big issue and can be resolved by throwing money at the problem but I am not recommending that. The drone community has a lot to offer in terms of add on products. The Feiyu WG2 is a good camera stabilisation device but is not directly mountable on the bike. It is alow cost device and the motor strength and response time is too slow.

You can wear the WG2  as recommended by the manufacturer for this application, however there are a number of products that can help with directly mounting on the bike. The issue with wearing the device is you get extraneous bits of bike and you in the shot.

I therefore decided to use a handlebar mount and solve the stabilisation issues.

 
It all looks a bit clumsy but I have tried all sorts of solutions before settling on this one. The design has 6 parts as follows:

1) Camera: GoPro HERO5 Session Action Camera Camcorder
Otainable through varios on line sources
2) Feiyu WG2 Waterproof Wearable Gimbal for GoPro Hero6/5/4 Session Action Cameras
Otainable through varios on line sources
3) Anti Vibration Damping Rubber & Anti-drop Pin for DJI Phantom 3 2+ Camera Gimbal
Based in China, lover-world has been an eBay member since 12 Mar, 2012
eBay item number: 391261888114
4) STO Universal wire rope anti-vibration mount
https://www.facebook.com/AlturnUSA
eBay item number:273120516399
5) 3M EAR Classic Foam Ear Plugs - PP-01-002
Based in United Kingdom, rubber_soul_01 has been an eBay member since 31 Mar, 2002
eBay item number:330844252101
6) Carbon Fiber Camera Gimbal Mount FPV Damping PTZ for DJI Phantom Drone GoPro 4
Based in China, xubolucky has been an eBay member since 17 Jul, 2017
eBay item number:152967290599

Here is a recent video with the new stabilisation strategy: https://videos.kinomap.com/en/watch/4927rp

I am well pleased with the result, here is a video describing how it was done:https://youtu.be/natWF6QINtM







Monday, 2 April 2018

Kinomap

Kinomap is another interactive cycling app. It has the virtue of uploading geo located videos and playing them back with realistic resistance on your trainer. It has the current ability to run on either the Apple I Phone or Android platforms. They have ambitions to bring out a MS-PC version.

I decided to give Kinomap a try as my ambition is to record videos of my favourite rides in Scotland upload them to Kinomap or equivalent  then ride them in the winter.

As an experiment I uploaded three videos on a trial basis:

* Inneos to Linlithgow                      4.94ml    150m         26:30
* Bathgate Alps from Linlithgow   17.06ml    544m      1:43:27
* Torphician 29-1-18 (b)                  9.92ml    318m          57:57

My first attempt was: * Torphician also listed  but the geo located data file had an error. The first try always has mistakes I guess.

The main problem recording videos for Kinomap is attention to detail and preparation. They take up a lot of memory and you have to take care of any stops (reducing the file size) by switching the camera off/on and GPS at the same time. Who wants to look at a man eating his sangwages or worse. I have a Garmin VIRB camera with built in GPS. However when first powered on it can take up to 30 seconds for the GPSto lock on to available satellites. You must therefore not start up the video till lock has been achieved. If you do screw up there are a number of GPS editors available that can bail you out.

* FitFileRepairTool       http://fitfilerepairtool.info/
* On line GPX editor      http://www.gpxeditor.co.uk/

If you record large (long) videos, you will find that you have numerous video files. A number manufactures of cameras limit the maximum file size and you also need to stop sometimes mid ride. The GPS will sometimes start a new file after being powered off. The FitFileRepair Tool is good for combining the GPS files and Kinomap provide a very efficient concatenation app for combining Video files. I use the concatenation tool and a PC app called Filmora to edit unwanted footage and reduce the size for uploading. Care should be taken when editing the video footage as the GPS has to be edited at the same point in time by the same amount. I would therefore recommend switching off the video when getting ready or stopping for any reason and make sure you switch off the GPS (very important). I keep two GPS's going at the same time, one for uploading to Strava and one for Kinomap videos. Your video may be shorter than the total ride length. It makes little difference that the GPS is built into the camera.

Quality

The other problem is vibration and horizontal stabilisation. The Garmin VIRB has rudimentary stabilisation via software but no horizontal stabilisation. This limits the quality of a lot of image taking on Kinomap.

I have invested in a Sports Camera 3 axis stabiliser but I have not resolved the application on a bike yet as vibration is still an issue. The alternative is to wear it which I think is a bit of a hazard having had some accidents on a bike. This will be the subject of a further blog.

Resolution

I take my videos in 1080P resolution in an attempt to future proof them in case Kinomap should support that resolution in the future.

As an example the Bathgate Alps video in MP4/1080p takes 14.7Gbytes of disc space and rendered down to 720p takes 3.7Gbytes in other words a doubling in resolution takes 4x more memory. A 4K video on that basis would take 16x more memory or 240Gbytes.

The upload time is restricted by the broadband pipe that you can budget for. My standard package from Virgin media has an upload speed of 5 Mbit/s. So for the examples given above the following upload times at best will result.

720p        3.7Gbyte         1.64hrs
1080p     15 GBbyte        6.66hrs
4K        240Gbyte           107hrs (4.4 days)

Engagement

The thing that keeps me going on Kinomap or Zwift is a large screen and something interesting to look at and listen to for at least an hour. I  find rides of about an hour at my physical limit of climbing is what I am looking for. Kinomap has lots of rides in that category. I don't crave human company or competing with avatars on Zwift for an hour. That is why I have not listed any rides so far more than 1.5hrs. I think my future videos will not exceed 2hrs riding time for an average rider.

The music issue can be solved easily technologically and I will not dwell on that but the large screen is an issue on Kinomap. They only support smart phones and tablets which do not trip the threshold of engagement for me. I need a 22" display at the end of my handlebars or an even larger screen further away.

I have purchased Chromecast for $25 and a good quality graphics display for $80 from Amazon. As my Samsung S7 supports Cromecast problem solved...........well not quite. I have a technical issue outstanding with Kinomap support. I will update this blog if resolved.

The main issue with engagement is resolution. In this day and age 1080p is the minimum. Kinomap only supports 720p HD, even if you upload 1080p. 720p HD is fine on a 10" tablet but not very good on a 22" screen. Some of the scenes become very blocky (sometimes).

The main issue is quality and variability, some peoples view of a good quality video differs from mine.  I think a star system should be invented by Kinomap to help with selection. You have to try before you ride otherwise you could be disappointed. The Kinomap web page is very good, the filtering mechanism works very well and you can run them on a large PC screen in straight video at 2x or 4x speed recorded. Some videos are recorded from a vehicle. There are two issues that come to mind so I avoid videos recorded that way:

1) They tend to ride faster than you can cycle so quality can be impared when you slow them down.
2) They will not be doing any scenic cycle tracks without breaking the laws in most countries.

I upoloaded one of my videos to YouTube to compare Kinomap quality vs YouTube, try it yourself. Make sure you select the correct resolution:

https://youtu.be/GZegp4hXaEs

Gps to resistance

This is Kinomaps star feature in my opinion. Rides that I have recorded the realism is spot on.

















Monday, 8 January 2018

Real life ride adpopted as training workout

Zwift has the ability to import custom workouts using a web based tools that simulate your favourite "in real life " (IRL) ride. My favourite IRL is a hill just outside Linlithgow called Kingscavil Castle ride. It is a hill about 1000 ft and has a 17% section as part of it. I have a Strava record (IRL.gpx) of one of my attempts and I used that as a starting point.

One of the reported problems with using an IRL.gpx record is; there are a lot of inconsistencies in the record of altitude vs time, so you get a number of spikes and dips which result in sudden changes to resistance on the trainer which is unpleasant to ride. You can use a route gpx instead and assign an average speed to each record but that gives an unrealistic effort pattern to compare against.


The above graph was produced by a tool called gpxsmoother. The blue line is the original IRL.gpx file from Strava and the green line is the workout friendly  IRL smoothed.gpx

You then take the IRL smoothed.gpx and load it into a tool called gpx-to-zwift-workout, which produces a a IRL.zwo file (Zwift workout file format) you then insert this file into  Documents/zwift/workouts/.




The method of doing this is also explained in real-world-to-zwift-workout.

After loading it into Zwift and trying it out on the trainer I observed a couple of things that are worth noting.

1) This example IRL ride is two hours long, and although this representation  is realistic it is a bit long for my attention span on Zwift. There is a lot of idling below 150W for my liking. However there is a large chunk in the middle of the workout in green (<150W) that can be removed and still preserve the essence of the ride. The problem if you remove this is you will not return to the same height as you started.

2)  The intensity is normally set by default to your ftp which is stored within Zwift. The first time I ran this the hill peak was a bit too easy.  It normally takes me all my short term watts to get over the summit about 225W at least for 4 minutes.
 



The results from Strava before adjusting the ftp to 211W


 
 

 
 
 

Monday, 6 November 2017

2017/18 Training

Lama's 5 x 4 TT Intervals - Shane Miller

Shane miller is a professional cyclist and journalist he writes for the likes of  Zwift and TACX. He has a video https://youtu.be/Apnwg1TAsIU that is the basis of my focus for this training plan.

Googling "Lama": It is a competitive 100K route in England around lake Lama. However he is pictured with Lamas so it could be that connection. His signature is GP Llama. It was possibly some of these reasons Shane constructed this training regime. I have decided to adopt this as my winter ftp builder.

My current ftp is 200W or 2.63W/Kg. It is not bad for my age (67) but improvement is what my road cycling is all about. Shane put together his regime presumably for competing in a Lama event so it should be good enough for me. I would like to come out of the winter this season at 3W/Kg or 228W or a 14% improvement. For my declared ftp (200) here is the suggested starting point. P1 - P5 are the yellow and Orange blobs on the screenshot diagram below.
Start point for me.


Intervals
% ftp in W
W
W/Kg
P1
106
212
2.79
P2
108
216
2.84
P3
110
220
2.89
P4
112
224
2.95
P5
114
228
3.00

I am a little concerned with my heart rate so I initially back off a couple of percent at the beginning to ease myself into the regime. Here is the original screenshot from the video.


All set I then conducted my first training ride on Zwift. The scenery is just the same for a normal ride except the trainer is in ERG mode. It apparently stands for Ergo Mode. Ergo mode adjusts the trainer resistance for a given cadence to match your target power chosen (The yellow, orange and grey power bars). It is like a servo system where you (the human) are the motor and Zwift is the rest of the control system. See ERG mode Some people don't like being controlled by a computer so it is not for everyone. You can switch off ERG mode on the workout from Zwift.


The results are shown above are my second attempt and are 1% down on my start point. The power curve shows how ERG mode maintains the target power during the 4 min interval. Also you should note the heart rate climbs to 164 bpm at the end of the session so I am obviously not pushing it beyond reasonable for my age.

Best fit MHR (Maximum Heart Rate) formula

 After years of searching for an arithmetic formula that's more accurate, the authors of this article (see link) have developed the "best-fit" formula and believe it to be the most accurate to date. You can use this formula in conjunction with the sub-max tests (see link) to determine your more accurate maximum. When you average all of these sub-max and arithmetic formula's together, you'll be close to your true maximum heart rate. Here's the formula:

210 minus 50% of your age minus 5% of your body weight (pounds) + 4 if male and 0 if female = Estimated MHR.

In my case 210 - 33.5 (50% of 67) - 8.36 (5% of 167) +4 = 172bpm.

After telling Strava my calculated Maximum Heart Rate (MHR). Strava sets a limit at  >166 for a red line Z4 giving me a small margin of warning when going into the danger zone Z5.

Start

 

Suffer Score

The Strava Suffer Score is an analysis of your heart rate data. By tracking your heart rate through your workout and its level relative to your maximum heart rate, Strava attach a value to show exactly how hard you worked. The more time you spend going full gas and the longer your activity, the higher the score. In this case it was 54

Points in the Red

Points in the Red is the portion of your Suffer Score that you earned in your Zone 4 and Zone 5 heart rate zones. Points in the Red shows you the intensity of your workout – if your Suffer Score and Points in the Red are the same, you were hammering the entire time! In this case the points in the Red was only 18.

Cadence

 My cadence for the workout is very low at about 66, it would be better if I increased this in future as it is more efficient, however in ERG mode I find this very difficult, just a case of getting used to ERG mode.  I choose today the London Red Hill course, and even set some records, but this is not the object of the session. I have chosen a hilly course so at least you think you are achieving something during the ride when the load is high.

Conclusions

The training ride does what it says on the tin. It definitely pushes me especially at P5. I have a hill near where I live that takes me 4m:46s to the top called Kingscavel brae. Thanks to Strava I recorded 250W average power  on my last PB time. Unfortunately I have no knowledge of my heart rate when I did this as I had no strap on. If I want to improve on that hill I need to expand my horizons in the training session. To this effect I have constructed a spread sheet increasing the difficulty by 1% each time. Before doing a session I will fiddle with the numbers to try and achieve at least what I do on Kingscavel brae. Hopefully this will train my Cardio system to stay within the Z4 range while I eventually put out my 14% more effort (228W for 20 mins) and give me a bit more margin for the climb at (250W) for 4 minutes.

Monday, 2 October 2017

Introducing the TACX Neo

Time for an upgrade



I have spent roughly a year with the TACX Vortex doing over 1000 miles, shortly after purchasing it I had a bad accident and subsequently used it for recovery from my injuries (see earlier posts). The Vortex was my introduction to interactive trainers and I found a few things about it less than ideal:

* Does not simulate the freewheeling (energy storage effect) of a real bike on the road, if you stop pedalling it stops although the Zwift avatar does keep going your knees and legs know better. It does not smooth out the or absorb the power strokes. I am not convinced, long term, this type of trainer is good for you.
* It does make a bit of a racket completely unlike the outdoor experience. I have a separate room in a bungalow so it is not much of an issue for me.
* It is extremely sensitive to tyre pressure. Unlike a road bike when the tyre pressure is down the thing just slips on the tiny roller and you have to get off and start again after inflating the tyre. To find out the correct pressure you have to calibrate it and remember the ideal pressure. Pressure in a tyre is temperature sensitive.
* You need a training tyre or it will shred a road tyre. You cannot take the bike out on the road with a training tyre, but I only use my old steel racer for training anyway I guess.

For the above reasons I decided to buy a TACX Neo, about two levels above a Vortex costing the best part of £1000.

As you can see it has a rather futuristic design and folds away quite neatly. However watch your back while moving it as it is quite heavy.

Cassettes

One of the reasons I chose the Neo is that it is direct drive, so I was looking forward to missing out all the faf with tyre pressures. Direct drive means you need to load the Neo with a compatible cartridge however and bolt it down hard.

Before I started down this road I emailed numerous suppliers and checked on the compatibility of my training bike with Campagnolo 9 speed drive train and the TACX EDCO freewheel adapter. EDCO is an freewheel adapter that takes Shimano, Sram and some Campagnolo Cassettes. None of the suppliers  gave me a definitive answer on compatibility except Wiggle. Unfortunately that information subsequently (after purchase) turned out to be wrong.

TACX says "The Shimano/Sram 9, 10 and 11 speed will fit all NEO's. 8 speed Campagnolo will not fit. Click here for more information." In the section labelled Suitable cassettes it says Campagnolo 9 speed Centaur 4 splines is compatible unfortunately after much correspondence with Wiggle I could not find a supplier that could supply one. Wiggle supplied the wrong cassette with 8 splines in the first place and eventually returned my money (and let me keep the cassette). Here is my solution using bits of a Shimano and Campagnolo Cassette: https://youtu.be/TNSCLAzjrt4. It was my first venture into bike cassettes and they hold no fear for me now.

I believe you can make various adapters spacers for 8 speed cassettes if it is Shimano compatible.

After  installing  the cassette I had to change the chain and jockey wheels which were worn.  I did not want to compromise my new cassette with an old chain.

First Impressions

Once I got the drive sorted out it was time try it all out. The software did what it said on the tin "plug and play" no issues what so ever, all I had to do was tell Zwift what trainer I was using today as it kept the Vortex as a selection, and off we went. I did expect to get "Road feel" out of the box, however10 minutes on Google/TACX support told me the firmware needed to be updated. It is updated from an app on your phone and again this was trouble free and sure enough the "Road feel" appeared with a selection button on Zfift to turn it off. I did not particularly like the road feel as it made more noise and kind of negates having a Neo in the first place. The firmware came with other upgrades to features I had not found fault with yet.

I have done three sessions of about 18miles each and I have to say I am pleased with my purchase so far and the effort to get going was very worth while. It makes a bit more noise than I expected, it kind of introduces itself above 2.2W/Kg and low cadence. It appears there is a resonance between the motor load unit in the Neo and the Campagnolo drive train. It is not a particularly bad noise (more like a rumble) but unexpected given the marketing blurb for the Neo. I am fairly sure it is not a fault. Perhaps when I get to 3W/Kg it will be ??? Perhaps it is telling me to increase my cadence. I find if I ride outside for months without using the trainer my cadence steadily goes down.

The future

The reason I bought the trainer is to improve my ftp (functional threshold power). I joined a road club a year or so ago and they all appear to get fitter and fitter every time we go out. I want to use the winter to improve. I satred out at an ftp of 1.7 when I joined the club and I am now up to 2.7 I would like take that to 3 or greater. I achieved that previous improvement in my ftp with the Vortex. It also helps with the Scottish weather in that I don't care too much if I can't get a cycle outside now I have a more realistic trainer. As you get older the issue is muscle wastage I want to stave off the effect as long as possible.

I would also like to do a few races on Zwift, one day I may get the courage to do a real race.

The TACX films look quite interesting, however they need a i7 compatible lap top.











Sunday, 12 March 2017

The results so far 12-03-17

I have had the trainer now since mid November 2016 a total of at least four months. The story is complicated as in December just after Xmas disaster struck. I had a major accident on my outdoor bike resulting in a partially collapsed lung and multiple broken ribs. I was facing two months of recovery before taking to my bike outdoors again. However having just purchased the trainer it gave me an ideal opportunity to accelerate my return to fitness.

Zwift reports it's results directly back to Strava exactly similar to the reporting of Garmin results for outdoor activities.

The key concept to keep in mind to track a recovery situation is ftp (Functional Threshold Performance). ftp is nothing complicated, it is basically the amount of energy you generate over a 20 minutes period. Currently I generate about 194W and my weight is 78Kg so my ftp is usually around 2.5W/Kg.  Zwift reports ftp to Strava however you need the Premium package to dispaly the results in terms of graphs alternately just take an image off the Zwift results page and note the numbers down.

The following graph shows my recovery. Less than two weeks after my accident I was able drag myself on to my trainer. The X axis is days (as I did 30 days continuous every morning for an hour).  The Y axis is ftp as previously described. The red line is ftp as calculated by Zwift and the blue line is the average (normalised for terrain) normally reported by Strava without the Premium reporting package.
 
 
 I hope you can see from the results I improved my ftp from 1.24 to 2.5 an improvement of 2x in performance over 4 weeks. The purple line is my average ftp before the accident.



 
The graph above is when I started mixing trainer and outdoor activities. The interesting stats from Strava of average power can vary quite widely but howevr the Strava computed ftp is much more consistent. I offer no explanation for this as Strava don't explain the exact method. The lesson here is if you want to track your progress out doors the the Premium results are necessary.


Thursday, 17 November 2016

My first club ride

The first Lothian Cycle group run I did was with Alison Johnston and Alan Hardie. Other members Kevin and Philip could not make it otherwise we would have had five virtual cyclists. A lot of time was spent getting Skype to work as you need the correct handle to find the person you want to conference with. We jointly worked out the best route to suit our needs would be the London Classique which is a 3.3 mile route and further decided we would do several laps of that. The other routes were considerably longer and higher climbs.



My conservatory, where my event was taking place, is colder than the Baltic at this time of year so I did a couple of laps just to warm up arriving just on time 8pm at the start line. Each lap was 3.3 miles taking about 10 minutes. It is remarkably difficult to stop exactly on the stop line as you don't have brakes, at least I have not found a control for that yet. We fafed around trying to locate a phone box unfortunately there are lots of phone boxes in London. Eventually we all connected via the leader board within a few yards. It is then easy to cycle together and we did 3 laps from that point cycling at a Monday Madness type of pace.

After the very convivial ride I received the results in Strava as normal. Instead of plugging my Garmin into the computer it just arrived like magic certified by Zwift as a genuine virtual ride. https://www.strava.com/activities/778337815
https://www.strava.com/activities/778293078

Strava presentation

Conclusions

The ability to join people you know in real life on a virtual cycle is awesome. I have tried cycling on my own and it is no where near enough fun. The purpose of all this is to keep doing it thought the winter months so the added motivation of meeting up with like minded enthusiasts cannot be understated. The addition of a Skype conference call just adds spice to the meal.

I think the meeting up on track is a difficult issue without brakes especially if the start line is on a bit of route going down hill. I have read the forum on this issue and there is lots of discussion.

The published map for London could do with some enhancements it is a bit devoid of reference points like Marble Arch etc

However the whole experience is well worth while and I may try and join an event.

FTP stands for Functional Threshold Power and represents the most watts you can expect to average over an hour. This can be used to improve your performance by joining training shedules that push your limits or you can use it to gauge which events you are suitable for.

On my Watopia flat route I averaged 132W and I currently weigh 77Kg so my FTP is  1.7W/Kg I can now use this to gauge where I go from here.

Checklist (before starting off)

* Open the windows
* Wear a hat to collect most of the sweat
* Water bottle
* Mobile
* Jersey to start with till you warm up
* Make sure windows/Zwift is up and running
* Old towel