CARV Digital Ski Coach - used by a carving snowboarder - Review

[SunSurfer]

This review has been updated after additional information was received from the CARV data team. The significant changes are in the "Tech stuff" section.

https://www.facebook.com/share/p/ow5tShXg4DCPBooG/

(see above Facebook post for some images from my early runs with the CARV app.)

I bought a Version 2 CARV Digital Ski Coach setup (boot sole sensors, Hintertux app version) in mid 2024 for use during a 3 week carving snowboard trip in New Zealand in August. I knew that CARV is planning to release new tech in the Northern hemisphere Autumn 2024 but that was going to be too late for me, and the set I bought came with one of the few remaining lifetime licences.

For the purposes of the review I rode UPZ RC10 boots with mondo 29 Intuition Pro Wrap liners. I bought the large CARV senors and these fitted the boot shells without problems. I remoulded the liners after fitting the CARV sensors and boot fit and performance were unaffected by the CARV units.

Binding angles ranged from 45 rear / 50 front with 2 degrees inward cant on a wide twin tip, to 75 rear / 80 front with 3 degrees outward cant skwal style, but most riding was done with a fairly conventional 60 rear / 65 front with 6 degrees of heel / toe lift and no cant on both feet.

Riding was tracked mostly at Roundhill, with single days at Mt Dobson and Cardrona ski fields. All fields were in New Zealand.

Setup

The boot top transmission units charged quickly, the supplied USB-A to proprietary connector cables are the only option for charging. Charge levels are shown in the phone app and CARV say they are able to do 3 full days riding without recharge. I charged mine every couple of days and it looked like they could have done 3 without problems.

I tried the CARV app on a Nokia 6.1 phone and a Samsung Galaxy A51 phone. No matter how I tried I couldn’t get the A51 to provide GPS data to the app so missed 2 days of location data, whereas the Nokia worked without problems. See the CARV website for an up to date list of compatible devices.

On your first use the app calibrates the boot sensors for motion and force. Motion calibration is done with the boots off and still. Force calibration has the boots on, buckled, and lifted off the ground one boot at a time. Over the 3 weeks I was using the system I had to recalibrate one boot once for motion, and one or both boots maybe 3 or 4 times for force. Most days however, a check of the indicator in the app showed that the boots were ready to use from the start, even if I had charged the transmission units the night before.

Did CARV use improve my riding? YES.

Because I was snowboarding the coaching hints from the app weren’t going to be relevant so I rode in Free Ski mode without audio feedback and reviewed my data in the app at days end.

Of the 12 metrics that the app tracks the following were useful to me as a carving snowboarder

Edge Angle
Progressive Edge Build
Turn Shape
G Force

There are also curved turn graphics for Edge Angle and Pressure that show how these change over the course of your Right and Left turns. These were probably the most useful and clearly showed from my first day with GPS data that my Right (Regular stance toeside) turn was weaker than my Left. Both angle and pressure levels were lower on the Right, as were the individual turn SkiIQ values.

In my carved turns pressure levels are highest on my inside boot at the point where the board turns away from the fall line on the exit path from the turn. This is the point where G-forces are highest. Pressure levels in the outside boot are low throughout the turn, reflecting that the outside foot is almost lifting up the outside edge, rather than pressing down on the inside.

Over the course of my trip I was able to improve my Right turns and more frequently match the Left turns for Edge Angle and Pressure on the curved graphics. This was by both consciously angling over earlier on the Right, and by accentuating medial knee movement of the knee on the outside of the turn i.e. Left knee moving across for a Right turn.

My (for what it’s worth) best SkiIQ rose from 129 on the first day of the trip to 136 by then end. I say “for what it’s worth” because SkIQ is a metric designed for skiers. According to the app my “skis” are not parallel, my edge similarity is poor, I put pressure on the wrong ski, and my fore/aft balance is all wrong.
Despite that, SkIQ still tracks upwards as the quality of my carved snowboard turns improves. I tried a series of 4 runs on the same Green rated, freshly groomed slope, with incremental improvements in turns. I went from skidded turns, to gentle carves with low edge angles, to higher edge angles and tighter turns, to full effort. Each step produced a significant rise in the rated SkiIQ for the run.

My Progress tab

To the right of each metric score in the app is a small arrowhead. Tap it and a new page opens up to tell you about the metric. There is a My Progress tab here, tap that and your best 5 day scores for that metric are averaged and your progress since you started presented.
I made progress for all 4 metrics I noted as useful above. But I have room to improve.
My Progress average edge angle was 56 degrees for the trip while I recorded run max. edge angles as high as 73 degrees. My Progress trip average G-force was 1.9G, close to my best run average of 2G. My best run SkiIQ was 136, but I found individual turns with SkiIQs over 140.
In general terms, turns with higher speeds, edge angles, and G-forces generate higher SkiIQs.

Tech stuff

The CARV app needs at least 10 turns to make a “run” and calculate the metrics.

From what I can glean from my data, the app and website, and my conversations with the CARV team, in general terms the accelerometers in the sensor track edge angles and sensor positions relative to the overall direction of motion. The Edge Angle and G-force run metrics are probably run averages, rather than mode or median values.

------------- The CARV data team replied to some questions after this review was first written.

Edge angle is measured by the absolute tilt of the insert around its long axis. G force is also measured relative to the change in direction of the long axis of the insert. The overall working assumption, not unreasonably, is that the inserts are in ski boots aligned with the long axis of skis.

Edge angle and G force are measured  for each turn in a run.
Max edge angle for a run is the maximum measured turn edge angle of the "outside ski".
Edge angle for a turn is the average of the maximum edge angles of both inside & outside "skis".
Run edge angle and G force are the median values of the run set of each metric.
--------------
The effect of the binding angle, lift and canting on the edge angle measurement can be worked out, and I confirmed what follows with bench testing with a board, wedges, and a phone runnning a clinometer app.
measured edge angle / cosine([90 - binding angle] degrees) = actual edge angle
e.g. for my binding angle of 60 degrees and a max measured edge angle of 73 degrees
73 / cosine(30 degrees) = actual edge angle
73 / .866 = 84 degrees

The 6 degrees of lift I use on each foot makes about a +/- 1 degree change to the measured edge angle. This is an effect that could be neglected except by the most obsessive data nerds!

Any binding canting will be change the baseline of the edge angle measurement for that binding.

A physics grad could probably explain how to calculate the corresponding compensation for the G-force calculation. That's beyond my maths knowledge.

The curved graphics in the app for Edge Angle and Pressure are coloured based on relative values rather than absolute values. So dark areas represent where the high values were for that run rather than whether the values were absolutely high or low.

After the more skilled of my riding companions crashed into me while shooting GoPro footage I elected not to try out the CARV Video Coach feature, where a mobile phone video can be linked to run data.

Overall Impression

This is written after some time and reflection on the data, and email conversations with the CARV team, after my trip finished.

CARV data helped me improve my snowboard carved turns. The techniques used to improve came from elsewhere.

If I was back on snow now I would probably try Monitor Mode for each of the 4 metrics I found useful. The Edge Angle Challenge would be fun too.

CARV collected useful data at all the binding settings I used (see above).

CARV would, in my opinion, be most helpful to beginner & intermediate riders wanting to progress, because it would enable them to track progress. However, the lack of widely accepted carving coaching tips to help them make progress is an issue. CARV data could potentially inform the development of coaching materials.

What would I like to see in CARV?

Max Edge Angle in the run summary data.

A computer version of the app, much like how Strava have both a phone app and computer views of users data. This would allow easier individual turn analysis.

Users able to view edge angle graphs for their turns. Examples of these are currently shown on the CARV website. The examples shown appear to split apart the inside and outside ski data to help see differences, but the extent of the offset in the data is not made explicit (looks like +/- 10 degrees).

And, pigs might fly, that there would be enough interest from the carving snowboard community to make it worthwhile for CARV to develop a coaching version for snowboarders who want to carve.

What about the new tech?

Some months ago I got the chance to test the new website for the new tech available Northern hemisphere autumn (see CARV website for the waiting list). It’s likely to be a boot top unit only with no sole pressure sensor. This likely to be more robust than the current setup where the reinforced flat cable between the sensor and transmitter is a potential point of failure. No boot sole sensor also means absolutely no effect on boot fit and no need for 3 different size sensors. A boot top only unit would be easily swappable between boots. What the effect would be of the greater range of motion in use of hard snowboard boot cuffs on data capture is unknown. Pressure data would not be available, but this may not prove to be a major loss. It appears that edge angle and G-force are some of the best indicators of a quality carved turn.

[SunSurfer]

Friday 6th Sept 2024 evening (NZ time)
The review has been updated after additional information was received from the CARV data team. The "Tech stuff" section has been significantly changed.

[st_lupo]

Thanks for the update.  I was wondering about that edge angle measurement given there was no calibration that could correct for binding angles.  

Lots of potential if they would ever do some simple SW mods to allow it to measure snowboard stuff (don't really care about the coaching feedback).

There's two types of people this would appeal to 1)people that want to improve their technique.  2)people that want to endlessly adjust their equipment to find an optimal setup.  I hesitate to think about which category I would fall into...

[SunSurfer]

@st_lupo Laughing with you because I love fiddling with my setups and trying different stuff!

I'm pretty sure I'm more in the first group (improve) after my experience this year, because I'm sure I have room to improve.
But the ability to measure that improvement in edge angle, G force and turn symmetry is something I really value.
That also allows you to actually tell if a change made to riding setup or technique actually improves the carve.

I not sure it would necessarily help extreme carvers build technique.
Racers might need to actually see edge angle and G force graphs to work on their stivot / carve / early finish turns.

Edge angle calculations at low angles (duckfoot stance end of the range) are likely to be less accurate given the way this CARV tech iteration works.

[SunSurfer]

CARV 2 boot top only sensor units have now been released for sale and the updated analysis app will apparently be available sometime in November.

The boot top sensors are easily swappable between boots. AFAIK only day or season subscriptions are available for new purchasers. Existing lifelong licences continue.

New app apparently just analyses the relative motion of the fixed points where the sensors are. The older boot sole sensors will work with the new analysis app and there is no absolute need to upgrade. This should mean that snowboarders get pretty accurate edge angle measurements without any correction for binding angles required.

Thinking of the future, the technology should be capable of analysing rotations in pipe and park riding if the app was developed in that direction.

[fluxgame]

This seems pretty interesting and (at the early bird pricing at least) less of investment than I was anticipating. I think I might pick one up. Are you open to assisting with analyzing data from a snowboard perspective, @SunSurfer?

[SunSurfer]

@fluxgame Hopefully minimal geometric compensation required now for being a snowboarder, certainly for edge angle.

We ride differently from skiers so the app coaching is unlikely to be particularly helpful. The data from the new app will need to be thought about from a carving snowboard technique perspective. Happy to be part of that kind of discussion.

[ShortcutToMoncton]

Interesting stuff, and cool to play around with. 

Feels like it’s in a similar bucket as many other day-related projects I see: everyone always wants “more data”, and it’s interesting to analyse, but it’s often unclear whether there’s any meaningful outcome/takeaway to be made. We’ve seen with many sports that data analytics can lead to a revolution, but it feels like the critical mass is whether there’s enough data. 

I played around with one of those ShockWiz mountain bike shock devices a few years back. It was really cool. It gave me lots of interesting data that ultimately told me that my mechanic was pretty good at setting up my suspension. They said “I told you.” But it wasn’t terribly useful beyond that. 

I’d be really interested to hear whether you’re able to achieve any meaningful takeaways. My gut says that more data points throughout your whole body would be required: for example, if you could put sensors in a base layer skinsuit (helmet and both shoulders, arms, hips, and legs), then it feels like that freaky skinless-monster demon @crackaddict recently debuted on YouTube could model your movements and track body positioning throughout your turns (along with measuring G-force, angles etc). Now that sounds like a incredible data revolution!! 

[SunSurfer]

@ShortcutToMoncton 
"I’d be really interested to hear whether you’re able to achieve any meaningful takeaways."
In the review I posted earlier in the thread I talked about the improvements I personally made in my riding.
I improved the matching of my turns, getting higher and earlier edge angles on my toesides, tightening those turns up to match my heelsides. This was a direct result of looking at the data CARV provided.
I managed to make my measured G-forces higher. As far as I can tell G force should be being measured directly down through the boot sole so should not need any geometry compensation. (A theoretical calculation of G-force relies on a) nil, or a constant amount of snow compression in the turn, and b) the correct picking of the angle between rider COM and the board edge)

I could already confidently and cleanly carve dark blue runs, even the top section of Revelstoke's Pitch Black that runs beside the top of the chair. Beginner and intermediate carvers have far more change to make than someone riding at my level.
But at present because CARV coaching is designed for skiers, you have to be able to look at the data and develop your own insights to improve. Not everyone will be able to do that effectively.
If more carving boarders start using CARV then as a group we may be able to develop our own coaching tips to go with various data patterns.
For skiers CARV is cheaper than taking even just a few lessons with an instructor. As carving snowboarders, we have YouTubers like James at the end of the day, but almost nothing on the actual slope to help us improve.

[philw]

I guess I'm a sceptic, just because I think it's a sort of skier concept, that there's some sort of perfect turn thing which exists and all the little skiers and the skier teachers try to copy that perfect thing. I reckon if it feels right it's probably right. Or: if you need a machine to tell you it's right, then it's wrong. But then I think I'm somewhat past the phase of learning that this type of thing is aimed at, and maybe I'd have used it when I was still trying to work it out.

 

[SunSurfer]

@philw It's a tool originally aimed at getting people to carve turns. It is aimed at people who want to deliberately improve rather than those who are satisfied with where they are. It is potentially an instructor you can have with you anytime. Some instructors are using it as a tool to add a layer of detailed assessment and feedback. In the original version the coaching is working towards unskidded, carved turns, with high edge angles, generating and requiring higher G forces to be handled and balanced.

That sounds to me like the kind of turn that carving snowboarders aspire to make. There are precious few instructors to teach how to do that.

I find it hilarious to read the gripes of people who because they have skied for many years fancy themselves as excellent skiers only to have CARV give them a low SkiIQ on groomers.

The new version of the sensor and upcoming app is advertised as being able to coach bump skiing and powder skiing as well. How well it does that remains to be seen.

[SunSurfer]

 

[Spoon Shao]

Shocked to find this thread! I have used Carv during 23-24 season for 14 days (skiing), thanks for the info, will definitely try it out!

 

Edited October 28, 2024 by Spoon Shao

[SunSurfer]

Updated my CARV app to the new Nevado version this morning, and surprised to find completely new numbers for my season just gone. Looks like they have reanalysed the raw data. My SkiIQs have pretty consistently jumped into the upper 140s, and my best is now 150. CARV Grim Ripper status! Not too shabby for a snowboarder using a skiers app.
I've attached screenshots from my highest scoring run on the steepest consistent trail at Roundhill Ski Area, Ozzi's Way to illustrate some points about the data and interpretation.

The first of the 3 turn shape and assoc. metric graphic looking at whether my "skis" are parallel had me puzzled to start with, given that it shows "ski" convergence (toeside, I'm regular stance) then divergence (heelside). The only sense I can make of this is that the analysis is picking up the changing relationship as the board bends in the turn, and that the board is bending tighter on the edge on the snow, and less on the upper edge. The board becomes like a section of the upper sidewall of a cone in shape. That would bring my toes towards each other on toeside, and my heels closer together on heelside, and produce the motion the app is seeing. On this run I scored 8% for Parallel skis, and I found another run where I scored 0%! If my interpretation is correct then alpine snowboarders should be aiming for low scores on this metric.

2nd Turn shape graphic looks at edge (boot sole angle with my CARV inserts) angle through the turn. By experimentation I've moved to a technique where I emphasise the "edge" angle of the boot on the outside of the turn. This seems to give me great edge grip. And that can be clearly seen in the darker colour on the outside track. Am I twisting the board, or am I just seeing the natural lateral flex in the bindings? I suspect I am producing some board twist since I use standard Intec bindings which are stiffer laterally than TD Sidewinders or bail bindings. Would a relatively rigid isolation plate setup reduce both this edge angle difference and the "ski" convergence/divergence issue? Things to experiment with next season.

3rd turn shape graphic: G force through the turns. Surprise for me here! I'd felt/thought my max G-force would be exerted on the exit from the fall line in the 2nd half of the turn. Not according to the sensors.

Looking at this overall, looks like next season I could work on getting the board on edge earlier in the turn, moving my centre of mass forward and towards the centre of the turn earlier, and unweighting more in transition. Previously I've tried to be as smooth as possible, a kind of minimalist turn where everything inessential is stripped away.

Edited November 6, 2024 by SunSurfer

[SunSurfer]

The Ski Orientation curve graphic recording my "ski tips" as converging on my Right (toeside) turns then diverging on my Left (heelside) turns above is a Bend It! image.
It was generated by a 2023 build Coiler Contra with a 10.5 m SCR and appropriate for my weight being bent to make 6m radius turns.
The pictures demonstrate the concept of the effect on binding alignment when the board bends and twists. The black lines are parallel when the paper is flat.
The converge/diverge signature will be there for any snowboarder carving turns while using CARV as long as there is some degree of twist in the board.

100 - my Parallel Skis metric score = my Bend It! metric score.
 

[SunSurfer]

Tech Update: Nevado app motion analysis assumes CARV 1 sensors are aligned along the skis length, while for CARV 2 sensors it makes a calibration on orientation after the first few turns, presumably based on the riders direction of motion.
For a snowboarder CARV 2 sensors should give accurate edge angle information, as long as the setup is not capable of any significant side to side flex. For softbooters that may well be a significant issue. Finding a mounting location which is fixed relative to the board surface will be necessry to get accurate data.

[nextcarve]

Just stumbled across this 15$ device:
 

But of course the big work on software had to be done. At least there is an API:
https://github.com/WITMOTION/WitStandardProtocol_JY901

[SunSurfer]

On 12/23/2024 at 10:11 AM, nextcarve said:

Just stumbled across this 15$ device:
 

But of course the big work on software had to be done. At least there is an API:
https://github.com/WITMOTION/WitStandardProtocol_JY901

Fascinating! Yes, looking at the website there is room for substantial further development for individualised use. CARV have created the added value with their analysis, data presentation and, coaching for skiers.

But this would be a good basis for anyone wanting to build their own.

[craigjlu]

Came across this in hopes this application also worked for snowboarders. Even with not all the data (and no coaching) available, it does seem to be helpful. Do you know how well this would translate to someone who carvs on a regular soft boot snowboard set up? (Japanese and Korean style carving) 

Picked up technical carving not to long ago and riding and Ogasaka with 39⁰ front and 33⁰ at the rear with a traditional binding set up. From what I understand, Il lose some of this data due to the boots having more flex than a hard boot alpine set up. 

I also noticed with the "skis parallel" data point. Wouldn't that almost always be off for the most part since the front and back boots are set up at different angles? Even if the board does flex causing it to change, I assume the base point would be the difference in your boot angles? Eg, mine are 39⁰ and 33⁰ so my base would be 6⁰. Feels like a lot of math to work out the ideal board flex during a turn based on the degree of parallel change. 

[SunSurfer]

@craigjlu All my data has been gathered using CARV 1 boot sole sensors. Being in New Zealand I haven't had the chance to try the CARV 2 sensors.

The big difference between the two is how the reference direction of motion is calibrated. The user has to tell the app which sensor is in use. When the Nevado app looks at CARV 1 data it assumes the sensor is aligned along a ski, so if a boarder uses them then the binding angle geometry definitely is an issue.

With CARV 2 sensors labelled L and R, the sensors could be anywhere on the appropriate boots. So CARV has a calibration process to work out the reference frame for its calculations. That should work the same for skiers or boarders and the edge angle measurements therefore should be accurate.

What will happen to the parallel skis score is unclear. The actual metric is irrelevant to us as boarders, we know where our feet are! But the current app data is interesting. The most recent update to the app gives the actual amount of divergence of the "skis". And it tells me I produce roughly 25-30 degrees of snowplough (toeside) and tip divergence (heelside)! Now I only have 5 degrees of binding splay in my setup, so it's measuring something else altogether. I'm assuming it's a reflection of the changing angles of my feet as the board bends in the turn, but I haven't managed to work out the principles of the calculation the app is making. My first attempt assuming a 30 degrees of arc across my 56cm stance distance produced an effective turn radius that was so short that it was ridiculous.

But how that will play out with CARV 2 sensors I can only guess at this point.

[craigjlu]

@SunSurfer It's all pretty interesting stuff. I really hope that they do make the software more applicable to snowboarding in the near future. Not to sure how it is for hard boot carving but if you were to pay a good instructor for Korean/ Japanese style soft boot carving, you would probably be paying 2-400 just for the hour. And that's if your even able to find someone who isn't booked out. Tech like this can really help people learn and improve on their own. Hope to see it in snowboarding soon.