Sooth Ski: Comparing Skis with Measured Data (2100+ skis!!)

[alude]

Hi all! I am a co-founder of Sooth Ski. We just launched what we think is the most advanced and extensive ski comparator. It is free, it is available online for self-service and it is based on real independent measured data. We spent the last few years building a new measurement machine that can collect the most important informations about a ski, in a way that is more precise than anything else we have seen. The machine is also real fast, so that we can measure all-the-skis-in-the-whole-world, each year, for all models and all lengths, without wanting to kill ourselves.

The database currently includes more than 2100+ skis. It has skis from as far back as 2010-11, and we just added 300 skis from 2021-22. We should add close to a thousand more new skis within the next few days/weeks...

The machine measures the geometry (width and height), as well as bending & torsional stiffnesses, at each mm along the full length of the ski. We also measure the mass of each ski. We think that these raw measurements can be used to describe the behaviour of skis with high accuracy (please, let's try not to go into damping for now!). We included torsional stiffness, even if it has been mostly neglected everywhere else. We found that it is one of the parameter that varies the most across models, and that it is also one that skiers are the most sensitive to (i.e., related to edge grip, precision, playfulness). From these measurements, we calculate a bunch of values that are available in the comparator, including surface area, surface-to-weight ratio, average stiffnesses, rocker length/height, setback, tip taper, effective edge, etc.

The comparator is available here: https://compareskis.shinyapps.io/compare. It should be easy to use, but the idea is to select a few skis from the list on that page by clicking on their name, before hitting the "View/compare skis" button to get to the comparaison page. You can also search for skis by using specifications (e.g., waist width, mass, stiffness), brand, year and categories. On the comparison page, you will get a few graph displaying the geometry of the selected skis, zoomed views on the tips/tails, a table of the most important parameters and a few graphs that we find useful. You can also customize your own graphs. At the bottom of the page, we have "distribution" graphs that display how the selected skis ranks with respect to all similar skis (currently roughly defined as all skis that are +-5cm long and +-5mm wide... not perfect, but we needed something and find it useful). Think percentile here! We find it super useful to figure out if you are getting a lot (or a little) of something. Once you know your preferences, it is great!

We have a short tutorial here: https://soothski.com/compare-skis-wi...ication-guide/

There are a few ways to use this:
1. Compare skis that you know to figure out your preferences
2. Compare skis that you know, or that were evaluated in reviews, to figure out what these numbers mean
3. Compare skis that you know to skis that you are considering
4. Search for skis by specs and discover many new products / brands
5. Etc.

It is particularly easy to compare very similar geometries. When you have such a case, you can look at stiffnesses and mass to get a good idea of how a ski will behave. If the geometries are wildly different, then you will need to think a little bit more about the interplay between all the variables (e.g., a ski could have a ton of torsional stiffness, but if the tip/tail don't touch the snow it won't have much edge grip).

We would love to hear about what you think of this comparator / database? Anything missing? Do you find it useful?

We hope this will trigger a few discussions about ski specs and what they mean! We will try to jump in a few thread with data-based comparisons, but feel free to ask questions here. The numbers don't lie, but they also don't speak by themselves. Even if we think they provide a good picture of each ski, we know it is not the full picture. We want to hear about what doesn't fit right!

Tell us if things need fixing, if you need more explanations, etc!

We are happy to measure the skis of brands that are not represented so far in the comparator. We just measured what was somewhat conveniently located around us...

Hopefully all ski nerds are happy now! Don't waste too much time on it! :-)

Stay tuned for more data/features and pray for snow!

[ZomblibulaX]

Been watching you guys for a bit and I would just like to say that I think this is fucking rad.

Thanks for putting all the work in, very cool project, very useful.

[detrusor]

Definitely a cool project. First 2 skis I searched for weren’t there.

Dp pro
Supergoat


Sent from my iPhone using TGR Forums

[Core Shot]

Very cool

But there are no long skis!

[cat in january]

That does look to be a problem for the tgr crowd. Great concept and looking forward to seeing your growth

[kid-kapow]

this is so awesome!

Thanks.

Kinda surprising to see that MFree108s are relatively stiff compared to the Atris and that the 2022 MFree108 182s are stiffer than the 2021 versions. It kinda makes me curious if there are just tolerances within the production that is shwcased, if the measurement method changed or if the one is a mislabeled 192

[motogreg]

actual data vs some yahoo on the internet hand flexing and saying "yep, pretty stiff"? what is the world coming to.....

[schindlerpiste]

IDK. Sounds gapic. It's all foreign to me. I like to actually feel (hand flex, etc.) to get a feel for the ski. I'm all for the collection of data, but there's nothing like fondling a ski

[singlecross]

I see a partnership opportunity with Blister adding graphs and shit to their reviews. NTTAWWT.

I’d like a 1% introduction broker fee. Can accept crypto.

[toast2266]

This is great. Pretty interesting, and I like the scatter plots that allow for easily comparing other skis that are in the same neighborhood.

One issue I see is that stiffness (both torsional and by length) are averaged. But a ski that's stiff under foot and soft at the tips / tails will feel a lot different than a ski that's moderately stiff everywhere, even though on average the two skis might be similar. Same could be said for a ski that has a soft tip and stiff tail, vs a ski with a stiff tip and soft tail. That said, that adds a whole lot more variables into the discussion, and I have no idea how to effectively present that data.

It would be nice, for each ski, to have a graphic that shows the stiffness values over the length of the ski. It wouldn't be something that'd be conducive to a comparison scatter plot, but it'd at least allow for toggling back and forth between a few skis to see where they're stiff / soft.

[skibrd]

Super cool. I’ve had sketches to do something similar for a while, but good to see you got around to it!

I’ll have to play with it later.


Sent from my iPhone using TGR Forums

[abraham]

Very cool project, I've looked around a little when you posted before. Reminds me of the Endre Hals stuff.

It would be nice, for each ski, to have a graphic that shows the stiffness values over the length of the ski. It wouldn't be something that'd be conducive to a comparison scatter plot, but it'd at least allow for toggling back and forth between a few skis to see where they're stiff / soft.

If you could tilt the chart after selecting a handful of skis you might be able to represent the stiffness over the length with bars that change size, color, something to give more context.

I forgot, but I'm pretty sure the Hals system used normalized flex numbers over a normalized length. So a ski might be a 56765 flex or a 27777 flex. I don't remember what the normalizing was, but it would be interesting to measure the skis at different lengths to check.

[river59]

This is incredibly fucking rad! Thanks for taking the time to do this and share it with the world.

Word to the wise - clean your computer screen before looking at the scatter plots. Just spent way too long trying to click on an outlier that was piece of lung butter.

[frorider]

Cool tool. Data base is missing the freeride skis tgr tends to ski, so for me it's not yet useful but I assume the database will grow.

Longitudinal flex profile would be useful.

[EWG]

This is super cool. Issue is sizes - in order to really compare you'd need to have the entire size run of every ski so you can be looking at the same sizes. For example I looked up Jefferys, Marksman and Mordecais, and the sizes you had available for each were 191, 177, 186/177. Makes it hard to compare.

Otherwise, this is could be really useful when the data is there.

[cat in january]

Or just focus on one size say 177-180. I know manufacturers vary specs across sizes, but it would get you close and work to compare

[alude]

Trying to address most comments...

Regarding size of skis : We are really trying to measure all lengths. It is not that hard or long if we have access to the skis. Scaling is not uniform across brands, and that is specially true for the long sizes that TGR crowd like. We measured skis in the past few years mostly in local shops on the East Coast, so long sizes are not that common. We are planning to add most lengths for 2021-22 skis within the next few weeks.

Small brands & old skis : We really want to represent them. We know that a lot of cool skis are coming from them. We think such database can help people discover new products. We are happy to measure anything that comes our way, so if you have contacts let us know. We could even think of a road trip, to measure all the cool skis in your quivers and stop by small manufacturers. Any idea are welcome.

Full flex/torsion profile : Yes, that is a good suggestion and it is on our radar. We have that data and we know it is important to have it to get a better idea of how a ski will perform. Average values only get you so far. If you are dying to see the full comparisons, just let us know here and we will do our best to post the full graphs. Please find the skis of interest in the comparator first and post the url so that we have the ski ID numbers.

Disclaimer --> Ski length and related measures: Prior to 2020-21, we were not measuring the full length of the ski. We were measuring the full geometry within a few cm of the tip/tail, but the exact position where we would stop would change from one ski to the other. We felt that you could still get a good understanding of how a ski would perform by looking at the geometry data. However, this makes extracting the full length of a ski difficult, as well as many properties that are referenced with respect to the end of the tip/tail (e.g., tip height, rocker length, taper, etc.). That is why we are currently using the nominal length of the skis in the comparator (for backward compatibility with older measurements), and you should expect a few cm of error for other measurements. Sometime the value can be off by quite a bit (e.g., a few cm on tip height can be 20-40%). We improved the machine last year to measure the full length and the measurement should be pretty much spot on starting in 2020-21. We will add a note about this directly in the comparator.

If you could tilt the chart after selecting a handful of skis you might be able to represent the stiffness over the length with bars that change size, color, something to give more context.

Very cool idea! Will take some time to implement, but very cool! Thanks.

I forgot, but I'm pretty sure the Hals system used normalized flex numbers over a normalized length. So a ski might be a 56765 flex or a 27777 flex. I don't remember what the normalizing was, but it would be interesting to measure the skis at different lengths to check.

Hals stuff was awesome. I remember using his pdf to make my first pair of ski! The SFI is normalized for "normal" value for that length. Conceptually it is similar to us, but I was never able to find the formula used. We normalize with "similar skis" (by length and width), which, if the database contain enough skis should be stable. The values used for normalization depends on length and width. I could see that we might want to better define "similar skis" in the future (e.g., considering the intended purpose of the ski).

Kinda surprising to see that MFree108s are relatively stiff compared to the Atris and that the 2022 MFree108 182s are stiffer than the 2021 versions. It kinda makes me curious if there are just tolerances within the production that is shwcased, if the measurement method changed or if the one is a mislabeled 192

Our measurement tolerances are within 3-5% for the stiffnesses. We also validate the measurement quite regularly while we measure with a reference ski (i.e., every 100 skis or so). Calculating an average value for stiffness is however tricky cause you need to define the start and end points. If you get more of the tip/tail, the average will seem lower. Happy to take inputs here about how to do this best?

That being said, the most different 2022 and 2021 MFree108 are within 13% in their bending stiffness and almost the same in torsional stiffness. This is not a huge difference, but it can look like a huge difference on the torsional vs bending stiffness graph, specially if you look on a wide computer screen. Our experience is that a production batch (i.e., within a year) is fairly consistent. However, from one year to the other, manufacturer might source their material from different places (wood is rarely consistent) and the tuning of their machine might be a little different (13% difference in stiffness could be explained by a 4% core thickness difference... this is quite small). But who knows what really happens during manufacturing!

In any case, you shouldn't try to read too much from the numbers if the stiffness differences between the skis are smaller than 10-15%. Some ski designer told us that it is hard to feel bending stiffness differences of less than 10-20%, even for experienced testers. This tool is particularly useful to do a rough selection, when there are significant differences between the skis, to get you to 80%. The final 20% should probably still rely on on-snow demo. Maybe we should round numbers or make everything relative to other similar products to prevent digging too deep into the numbers?

[kid-kapow]

the reason the 13% number is of interest between 21 and 22 MF108 182s is that the 192 MF108 is supposedly 15% stiffer overall than the 182. As such, a 13% increase in stiffness could indicate a change in the layup to have the 182 more resemble the 192. Or it could just be variation within production tolerances / natural variation with or across production runs, or down to differing methods of measuring the ski - I do not know. Regardless, the feedback on the 192 vs 182 is pretty universal in that the 192 is noticably stiffer and more potent, while I have found the 182 to be a bit on the soft side in the shovels especially. As such, having precise measurements can give us tangible clues to how a ski evolves and infer changes on how it will ski.

in closing, I would strongly urge you standarize how you measure all measurements on a ski - be it straight pull for lenght, torsional stiffness in the front/aft part of the ski by locking a certain section in a standardized bsl underfoot or whatever, or how measure torsinal rigidity by where you afix the measuring device (contact points or effective edge or whatever, as long as it is consistent across all ski). I kinda understand from your post that you have already done this and have refined your procedures over the last few years. The idea is to prevent you ending up instilling a with a kinda false sense of accuracy, but where manufacturing tolerances and measuring margins of errors end up with a quantitive description of a ski that is if more accurate than an subjective hand flex still only an approximation of varying degree of accuracy of the objective truth so to speak (fantastic sentence I know).

I do not mean to come across as overly critical or unappreciative of the awesome work that you have and are continuing to put into this database - I think it is fantastic, especially when seen together with other sources of precise measurements such as Endre Hals' graphics published through FriFlyt. Keep up the stellar work!

[abraham]

Very cool idea! Will take some time to implement, but very cool! Thanks.

Looks like you're using plotly, I haven't needed to do anything similar but I bet you could make a quick test if the concept with streamtube-plot. I'm not certain it's the right idea, I suppose you could add the stiffness to the highlight labels but then it's hard to compare between skis and you've obviously put a lot of thought into making comparisons easy.

Hals stuff was awesome. I remember using his pdf to make my first pair of ski! The SFI is normalized for "normal" value for that length. Conceptually it is similar to us, but I was never able to find the formula used. We normalize with "similar skis" (by length and width), which, if the database contain enough skis should be stable. The values used for normalization depends on length and width. I could see that we might want to better define "similar skis" in the future (e.g., considering the intended purpose of the ski).

Could the similar ski normalization be turned off? I could see using a tool like this to say 'I've used the following skis, some I like and some I didn't' and the trying to see what's similar or if a new ski is closer to one group or another. But you might have 4 pow skis, some carvers, and a couple touring skis which might make the similar grouping less useful?

Our measurement tolerances are within 3-5% for the stiffnesses. We also validate the measurement quite regularly while we measure with a reference ski (i.e., every 100 skis or so). Calculating an average value for stiffness is however tricky cause you need to define the start and end points. If you get more of the tip/tail, the average will seem lower. Happy to take inputs here about how to do this best?

I'm curious if you've tried measuring skis after some amount of use? Not that it's necessarily helpful, or easy to represent/quantify, but it might be interesting to see if something like weight/square meter has a connection to how the flex changes after use.

[Core Shot]

One other thought.

As you build the database, it’s harder to drink from the firehose.

Maybe start adding a filter for type of ski.
Race. Carver. All mountain. Powder. For example.
Like the old ski mag ski reviews.

Although it might be easier just to add a waist width and or side cut radius filter?

[HukuTa_KydecHuk]

I have questions to stiffness measurements. I tried comparing Bibby'12 190 vs Candide 5.0 '19 183, because I own both. Site says Bibbys are significantly stiffer while bending, however when I'm flexing them by hands, it appears to be the opposite. Even if you compare blisters reviews on both, Bibbys tips and tails stiffness were always rated 6.5-7, while Candides at 7.5-8. I mean, difference is not subtle, it's very significant.

[alude]

the reason the 13% number is of interest between 21 and 22 MF108 182s is that the 192 MF108 is supposedly 15% stiffer overall than the 182. As such, a 13% increase in stiffness could indicate a change in the layup to have the 182 more resemble the 192. Or it could just be variation within production tolerances / natural variation with or across production runs, or down to differing methods of measuring the ski - I do not know. Regardless, the feedback on the 192 vs 182 is pretty universal in that the 192 is noticably stiffer and more potent, while I have found the 182 to be a bit on the soft side in the shovels especially. As such, having precise measurements can give us tangible clues to how a ski evolves and infer changes on how it will ski.

I get all of what you are saying. I which for more accurate numbers. We are putting a lot of effort so that our measurements are within 3-5%, but I can't really speak about what is expected about manufacturings tolerances and ski design (there are ton of different stories). Our machine gives us numbers, but we can only speculate on where the difference is coming from. I think we shouldn't worry too much for differences of less than 10-15% on bending and torsional stiffness. I think right now this is the level of accuracy that we should expect from our measurements and manufacturing tolerances, and this is also pushing the level of sensitivity of most skiers.

If you can wait 2-3 weeks, we should have measurement for all 2021-22 M-Free.

in closing, I would strongly urge you standarize how you measure all measurements on a ski - be it straight pull for lenght, torsional stiffness in the front/aft part of the ski by locking a certain section in a standardized bsl underfoot or whatever, or how measure torsinal rigidity by where you afix the measuring device (contact points or effective edge or whatever, as long as it is consistent across all ski). I kinda understand from your post that you have already done this and have refined your procedures over the last few years. The idea is to prevent you ending up instilling a with a kinda false sense of accuracy, but where manufacturing tolerances and measuring margins of errors end up with a quantitive description of a ski that is if more accurate than an subjective hand flex still only an approximation of varying degree of accuracy of the objective truth so to speak (fantastic sentence I know).

This is deeper than you might think!

Why do many, including us, measure/display the mass of a ski to within 1 gram when manufacturing variations are easily on the order of 50 g? Why do we measure tip height and camber very carefully when it can change with temperature (if the laminate is not well balanced)? Is it better to show a rounded number or just tell people what the expected error is on the measurement?

For all geometrical measurements, we are very accurate since 2020-21 from one end of the ski to the other. I can dive deeper is you guys are interested, but we get sub-mm width and height accuracy.

For bending and torsional stiffness, it is more complex. We measure everything in between the supports of the machine, which we typically place at the points where the tip/tail start to rapidly curve up (not at the point of early-rise or at the start of the rocker, that would be too short in most case). Right now, the comparator displays the average value between that. Having a little variation about where you put the support is not critical. Take for example a ski that has an average value of 250 Nm^2 over a length of 150 cm. If we are missing 10 cm in the tip/tail @ 50 Nm^2, that means that we should have had (250*150 + 50*10)/(150+10) = 238 Nm^2, which is only 5% off.

But I get your point about standardization and we want to move toward that. What do you think about the average value for a section that is in between the tip/tail positions minus "X" cm on each side? Can we fix "X" for all skis without creating problems?

If we can agree on a section length, we could easily subdivide it in a few sub-sections.

A very deep side note is that the average that we are providing is not quite the same as a hand flex test. A hand flex test does a triangular scaling of the stiffness and will put more emphasis on the area where your hand is applying the bending force is. Our average has a uniform scaling. We could provide a triangular scaling, but neither is perfect. That being said, what you should think about is the pressure distribution on the snow and the twist of the tip/tail. This will vary depending on the force that you are applying on the snow, the ski geometry and the snow itself. I will stop here for now, but we can calculate that.

I do not mean to come across as overly critical or unappreciative of the awesome work that you have and are continuing to put into this database - I think it is fantastic, especially when seen together with other sources of precise measurements such as Endre Hals' graphics published through FriFlyt. Keep up the stellar work!

We appreciate the constructive feedback and the discussion.

[alude]

Could the similar ski normalization be turned off? I could see using a tool like this to say 'I've used the following skis, some I like and some I didn't' and the trying to see what's similar or if a new ski is closer to one group or another. But you might have 4 pow skis, some carvers, and a couple touring skis which might make the similar grouping less useful?

I got confused with something else. We do some normalization in other contexts, but not in this online comparator. In the comparator, we just calculate an histogram of each spec for similar skis and display where the skis of interest are falling on this histogram. A better definition of "similar skis" would change the histogram.

We could calculate a mean and standard deviation from this histogram to calculate something like the Ski Flex Index, but we don't. We still display raw values. I personally think a percentile rating would be more useful as not all distributions are normal.

We had the idea of letting the user choose or define the "similar skis" used in the comparison. If many people feel it would be useful we could work on that. Otherwise, it should be pretty simple to use the categories that we defined in the similar skis comparison.

I'm curious if you've tried measuring skis after some amount of use? Not that it's necessarily helpful, or easy to represent/quantify, but it might be interesting to see if something like weight/square meter has a connection to how the flex changes after use.

Yes, multiple times and using different methods (fatigue testing in the lab, measuring a ski throughout the season of U18 racers). The initial idea of this project was that all shops would own our machine, measure your ski when you ask them for a tune/wax and tell you that you ski life is about X% over. We were never able to measure any difference in bending or torsional stiffness. At least, nothing above our 5% quoted accuracy.

We all experienced that skis are "ageing" or becoming "dead", but I highly doubt, given our measurements, that it is through changes in bending and torsional stiffness. Maybe we missed something or maybe it is something else! Finding nothing is never a good proof that something is never gonna happen.

[alude]

I have questions to stiffness measurements. I tried comparing Bibby'12 190 vs Candide 5.0 '19 183, because I own both. Site says Bibbys are significantly stiffer while bending, however when I'm flexing them by hands, it appears to be the opposite. Even if you compare blisters reviews on both, Bibbys tips and tails stiffness were always rated 6.5-7, while Candides at 7.5-8. I mean, difference is not subtle, it's very significant.

The bending stiffness difference is on the order of 30% (378/290).

The Bibby that we measured is the Bibby Pro 190 (https://blisterreview.com/gear-revie...ibby-pro-190cm). Is yours the same?

The Candide 5.0 183 that we measured was the 18/19 model. It weighted 1997 g. Blister reviewed the 19/20 (apparently unchanged in 20/21). They say that the core changed from 18/19 to 19/20. The weight of both CT5.0 ski in the 19/20 pair were 2081 & 2115 g (https://blisterreview.com/gear-revie...on-candide-5-0). Which year do you have?

The weight difference between both year is large and it seems like Faction did a redesign of these skis. If you attribute the weight difference exclusively to a core thickness variation (2115/1997 -> 5% thickness variation throughout the ski), then you could expect the CT5.0 19/20 to be about 20 % stiffer in its bending stiffness (i.e., (2115/1997)^3 because stiffness is proportional to thickness cubed). Faction could have done anything to these skis, including making the tip/tail much thicker/stiffer or vice versa.

Here is the raw data about these skis (note that the tip and tail are reverse left/right with respect to the online comparator):



You can see from these measurements that the bending stiffness of the tips and tails of both models are quite similar (@ [-0.6 -0.5] and [0.5 0.7]). The Bibby Pro is much stiffer in its center-back section. If the increased weight of the 19/20 CT5.0 also increases its stiffness, then I would expect the tip/tail to feel stiffer than the Bibby Pro as you describe.

Makes sense?

[alude]

One other thought.

As you build the database, it’s harder to drink from the firehose.

Maybe start adding a filter for type of ski.
Race. Carver. All mountain. Powder. For example.
Like the old ski mag ski reviews.

Although it might be easier just to add a waist width and or side cut radius filter?

Totally agree and it is all already there!! Many people miss these features apparently. We are open to any comments on how to improve the UI...

You can search by name / brand / model in the search bar. If you input something and press enter, the list of ski will get filtered by the keywords.

You also have the advanced search options. You can filter by specs. See below:


You can add as many spec filters as you want. Just click the "add filter" button. You can also filter by year, brand, gender, and category using they other boxes. The categories are very inclusives as of now and only based on geometry and mass. They are just preprogrammed filters. If you don't like them, build your own! :-)

One good way to use the comparator is to add filtering criteria until you get a short list of 10-20 skis. Once you have that list, then you can go look into the detailed comparisons, read reviews, find on-snow demos, etc.

Enjoy!