A modest proposal

[kallend 1,647]

A Method for Evaluating the Perfection of 2-D Formations of Unlinked Skydivers

Proposed by John Kallend

Introduction:

A method is proposed for the evaluation of 2-D formations of unlinked skydivers, such as those that occur in wingsuit formation jumps. The method:

1. Can be used to evaluate ANY pre-declared formation of 4 or more jumpers, regardless of symmetry of design. No constraints are placed on the design of the formation, allowing organizers freedom to create new and interesting formation designs. It can be used to evaluate formations based on any regular grid or plane lattice (square, diamond, hexagonal, triangular, rhombic, etc., or formations that are not based on a lattice, such as rings, semicircles, ellipses, stars, letters of the alphabet, whole words, etc.

2. Requires no special software beyond standard packages, but is readily amenable to implementation by computer.

3. Allows arbitrary relationship between neighboring jumpers as long as they are pre-declared

4. Has only one arbitrary variable, used to specify the error tolerance permitted.

5. Is objective. All evaluators would reach the same conclusion.

6. Does not allow cumulative errors.

7. Does not give special status to any jumpers (such as a leader or base).

8. Can be adjusted to handle lens distortions.

9. Is conceptually simple – anyone can understand it. It simply measures how far the jumpers are from where they are supposed to be.

The premise:

Any method of evaluating a formation should allow complete freedom to the formation designer to create any formation of their choice. The evaluation method should not place constraints on formation design.

The design will specify where each skydiver should be relative to the formation as a whole (not just relative to the nearest neighbors).

Given that absolute distance measurements between jumpers are not possible with current technology, all distances within the formation will be relative to the overall size of the formation.

Given that the line of flight is difficult to determine from still photos, the method provides a way of orienting the photo with the pre-declared design.

The degree of perfection of the formation as flown will be determined by the distance from the jumpers' actual positions to their slot positions in the pre-declared design.

How it works:

1. Organizers create an accurate image of the intended formation, either graphically (a picture) or digitally. The skydiver slots in the formation are numbered to identify each slot. If the formation is based on a regular plane lattice, the lattice image can be used with the slots marked.

2. The two slots in the design with the greatest separation from each other are identified. The straight line between these slots will be used to orient the formation Let this line be called “L”. The distance between the slots defines the size of the formation which will be used to scale the photograph(s) of the actual formation as flown. Let this distance be called “D”. If there is more than one pair of slots having the same greatest separation (for example, the diagonals of a square), the designers will pre-select one. Use of the greatest separation minimises potential measurement errors.

3. The two slots in the design having the least separation are identified. The distance between them will be used to specify the error tolerance for the formation. Let this distance be called “Q”.

4. The skydive is photographed, with the videographer directly above or below the formation.

5. The correspondence between the jumpers on the photograph and the slots on the design is identified. If no one-to-one correspondence can be found, the formation clearly fails.

6. The photograph is oriented so the line “L” on the photograph is parallel to the line “L” on the formation design (see step 2 above).

7. The photograph is scaled so the distance “D” on the design (see step 2 above) is the same as the distance “D” on the photograph.

8. The design is placed on the scaled and oriented photograph (either pictorially or digitally if a computer used). The deviation of the actual positions of the jumpers from their designed slot positions is noted This represents the error. The maximum error is recorded)

9. If the maximum deviation from the assigned position exceeds the required tolerance (expressed as a % of the distance “Q” in step 3 above), the formation fails.

10. If the maximum deviation is less than or equal to the required tolerance, the formation succeeds.

...

The only sure way to survive a canopy collision is not to have one.

[mccordia 73]

Quote

1. Can be used to evaluate ANY pre-declared formation of 4 or more jumpers, regardless of symmetry of design. No constraints are placed on the design of the formation, allowing organizers freedom to create new and interesting formation designs. It can be used to evaluate formations based on any regular grid or plane lattice (square, diamond, hexagonal, triangular, rhombic, etc., or formations that are not based on a lattice, such as rings, semicircles, ellipses, stars, letters of the alphabet, whole words, etc.

check..

Quote

2. Requires no special software beyond standard packages, but is readily amenable to implementation by computer.

You are looking for advanced methods of judging, yet hope to find it implemented in standarddized software.
And software IS NOT part of the judging METHOD. Its a simplified/easier way of doing so.

Quote

3. Allows arbitrary relationship between neighboring jumpers as long as they are pre-declared

4. Has only one arbitrary variable, used to specify the error tolerance permitted.

5. Is objective. All evaluators would reach the same conclusion.

6. Does not allow cumulative errors.

check

Quote

7. Does not give special status to any jumpers (such as a leader or base).

Who does everyone fly towards on a bigway (FS/Freefly)
Who sets the speed/Fallrate?

and how do people call that person.....'base' by any chance?

Quote

8. Can be adjusted to handle lens distortions.

Could you name the software package that has this feature available (in a 100% correct and easily acces-able feature?).
Though as mentioned before, something we can (and will) include at a later point in time. Though these are practical issues, that dont have anything to do with the actual measuring method. They just limit the cameraman in terms of distance he/she flies and what (non fish-eye) lense to use, to minimize distorting.

But again, a more than possible feature for later inclusion.

Quote

Any method of evaluating a formation should allow complete freedom to the formation designer to create any formation of their choice. The evaluation method should not place constraints on formation design.

check..

Quote

The design will specify where each skydiver should be relative to the formation as a whole (not just relative to the nearest neighbors).

check.. as thats what actually happens when all flyers fly to a neigbour, but also have a minimum and maximum vallue/distance they are allowed to fly. Though (and hope you care to think with us, instead of rebelling) we are more than welcome to also ad additional measuring parameters to the flex-grid on top of a formation. Measuring most extreme corners/angles at the edges for additional validation of overal for and shape.

Quote

Given that absolute distance measurements between jumpers are not possible with current technology,

Is it?
Its not that hard to get a close to accurate distance with a simple (digital) tape-measure on top of a known proportion of an object. be it a rig, helmet or arm-length. Even displaying that vallue next to all jumpers.

And refering to your point number 2 again, have fun measuring that one for all jumpers in image editing software.
Dont forget to switch of the lights when done...

Quote

1. Organizers create an accurate image of the intended formation, either graphically (a picture) or digitally. The skydiver slots in the formation are numbered to identify each slot. If the formation is based on a regular plane lattice, the lattice image can be used with the slots marked.

check..also possible for multiple points.
In future, we could even see if judging on video (moving) is possible. though thats still a long way of for now.

Quote

2. The two slots in the design with the greatest separation from each other are identified. The straight line between these slots will be used to orient the formation Let this line be called “L”. The distance between the slots defines the size of the formation which will be used to scale the photograph(s) of the actual formation as flown. Let this distance be called “D”. If there is more than one pair of slots having the same greatest separation (for example, the diagonals of a square), the designers will pre-select one. Use of the greatest separation minimises potential measurement errors.

Good one (elliminating the rest of the quote). one thing to play with.
In essence, its the same method we now use. Only getting the base distance/angle from the whole formation (it shouldnt just be distances you judge, as that permits a V formation to fly as a straight line or vice-versa, and still count as a succes. Angle also counts.)

Lots of good stuff to think about. And once you get passed the scare of creating a custom application to assist in doing 5 weeks of work in an image editing application, give a shout, and we'll gladly have you think along. Thats the whole intent.

Happy to see this post.
As its highlights things to work on, to improve. A process that should have been initiated a year ago in this 'assertive' way....but a good one..

JC
FlyLikeBrick
I'm an Athlete?

[kallend 1,647]

Quote

Quote

2. Requires no special software beyond standard packages, but is readily amenable to implementation by computer.

You are looking for advanced methods of judging, yet hope to find it implemented in standarddized software.
And software IS NOT part of the judging METHOD. Its a simplified/easier way of doing so.

Incorrect. Although computer implementation of the whole thing is possible and easy, all that is really needed is software that will scale and rotate a picture. Picasa would do it just fine (free download). The rest can be done by hand quite quickly.

Quote

And refering to your point number 2 again, have fun measuring that one for all jumpers in image editing software.
Dont forget to switch of the lights when done...

Doesn't have to be done for all of them, just the two that are farthest apart.

Ever hear of a ruler?

...

The only sure way to survive a canopy collision is not to have one.

[DSE 3]

Please send this to MaryLou? It's concise, well-thought, intelligent, and although it's not perfect either, it's "more perfect" than what we currently have going before the FAI from the USA, and better than what the USPA rushed to accept because they were told that "wingsuiters support this as evidenced by the 71 way."

Thank you John, for constructive and open discussion points.

[SuperGirl 0]

hmm. that's a very interesting proposal.
woohoo! very happy to see the ideas flowing. keep em coming, folks!

I am trying to think of various cases in which your system allows a formation that the judging software doesn't, and evaluate whether the formation is actually worth considering , i.e. would it make sense to try to fly like that or would it be even harder to get results than a different shape with the same number of people.

first we've got the case of the random mess...
people may say oh but with your system you could fly ugly shit, if planned... but in reality flying a bunch of shit would make it harder to actually be able to match the shit on the ground with the shit in the sky, without sightlines and proximity helping you stay in your slot.

Then there is the case of orderly flocks that are separated into connected components (and my definition of "connected" here is that you can put lines on it in the software without breaking it) like we see in some of the airplane pictures, or like Scott Campos had planned for that bigway event in '08 if I'm not mistaken (correct me if i'm wrong, wasn't there)
Basically a formation of several rather orderly big chunks...
This might be a good case to consider.
Though again harder to fly such that the distance separating the chunks is exactly as planned on the ground... but who the fuck knows, if we lie on the painted dots long enough maybe we can really dial in those bigger distances :)

So finally we come back to one big orderly chunk which is right now covered pretty well by the software. (slightly different measure but I suspect pretty similar in ultimate results achieved)

It appears that for this method the smaller distance you use between flyers, the easier it will be to minimize the error so that the formation is achieved. right? not that records should be easy or anything... but once you get to really big numbers you try to plan for the design that is most likely to achieve success, no?


Now aside from all that, the real beauty of this method here might lie in its extensibility to 3d formations. That would require a minimum of 3 photographs if I'm not mistaken. Hmmmm. Now THAT would be interesting.

[mccordia 73]

Quote

Ever hear of a ruler?

Its funny that the initial comment was that placing dots on each flyer would take to long, and now its measuring the distances between each one by hand. Thats even more work.

I hate to break it to you (respecting your age and all).
But its 2010...not 1910

But regardless of method. A ruler in combination with one of those school geo-triangle thingies (measuring the angle, which you seem to forget, and IS an important part in formation shape), and just doing those 2 vallues on each jumper, will work just fine.

But if you have a choice of 5 hours per formation judged, or 5 minutes. Ill go with the times, and create some actual software for the porpoise. But again, its not mandatory. Depends on how you prefer judging to be done.

Fast, to the point is my prefered method.
Much like the judging on 4 way formation often using a specialized system in competitions. Wonder how ritch the people are now that designed that one...

JC
FlyLikeBrick
I'm an Athlete?

[mccordia 73]

And again...this kind of thought IS what we need more.

In short, we have a mesh/dots over a bunch of flyers, and the equation used to measure the flying is what we really need to work out.
And I see some promise in several things mentioned.

And (promise) will also mention this in the FAI proposal as one of the areas to further explore.

JC
FlyLikeBrick
I'm an Athlete?

[SuperGirl 0]

Quote

one of those school geo-triangle thingies (measuring the angle

you mean a protractor?


By the way, I now see clearly what Kallend wants to do. He might be the only one left with enough patience to use a ruler to measure formation distances and compute record measurements by hand.
If so, he will then be the key in judging any formation ever made... and he will charge an insurmountable amount of cash to be hired to judge every single record in the future.
Guess someone is secretly planning to get rich, huh?

[kallend 1,647]

Quote

Quote

Ever hear of a ruler?

Its funny that the initial comment was that placing dots on each flyer would take to long, and now its measuring the distances between each one by hand. Thats even more work.

No, you do NOT have to measure all the distances. Only 2 are needed (I called them "D" for scaling and "Q" for for tolerance), and only the biggest error need be considered. It could be done with a circle of the radius of the tolerance drawn on a piece of tracing paper.

Sometimes computers are NOT the easiest solution to a problem.

Quote

I hate to break it to you (respecting your age and all).
But its 2010...not 1910

But regardless of method. A ruler in combination with one of those school geo-triangle thingies (measuring the angle, which you seem to forget, and IS an important part in formation shape), and just doing those 2 vallues on each jumper, will work just fine.

If people are in the right place, the angles will be correct. If they are in the wrong place, my method will show it.

Quote

But if you have a choice of 5 hours per formation judged, or 5 minutes. Ill go with the times, and create some actual software for the porpoise. But again, its not mandatory. Depends on how you prefer judging to be done.

That's nonsense. I could do it in 5 minutes for a 100-way using nothing more than Picasa, a printer and some tracing paper.

Since the algorithm is well defined, it could readily be programmed for those who never learned to use a ruler.

Quote

Fast, to the point is my prefered method.
Much like the judging on 4 way formation often using a specialized system in competitions. Wonder how ritch the people are now that designed that one...

Which is more important - fast or accurate?

My proposal asks the question "How far are you from where you are supposed to be?"

Seems pretty obvious to me that this IS the question we should be asking.

...

The only sure way to survive a canopy collision is not to have one.

[kallend 1,647]

Quote

Quote

one of those school geo-triangle thingies (measuring the angle

you mean a protractor?


By the way, I now see clearly what Kallend wants to do. He might be the only one left with enough patience to use a ruler to measure formation distances and compute record measurements by hand.
If so, he will then be the key in judging any formation ever made... and he will charge an insurmountable amount of cash to be hired to judge every single record in the future.
Guess someone is secretly planning to get rich, huh?

Well, Andreea, I think I could train you to do it in just a five minutes using only a picture of the planned formation, a picture of the actual formation, Picasa (free software), a printer, and a piece of tracing paper.

Scale and orient the photo, print it, overlay the formation diagram, look for the largest error, and measure it. That is no more work than what Mark did with "the grid" in Elsinore, maybe less because there is no best fitting to be done.

...

The only sure way to survive a canopy collision is not to have one.

[kallend 1,647]

Quote

Quote

one of those school geo-triangle thingies (measuring the angle

you mean a protractor?


By the way, I now see clearly what Kallend wants to do. He might be the only one left with enough patience to use a ruler to measure formation distances and compute record measurements by hand.
If so, he will then be the key in judging any formation ever made... and he will charge an insurmountable amount of cash to be hired to judge every single record in the future.
Guess someone is secretly planning to get rich, huh?

OK, a REAL LIFE example, NO SECRETS to be revealed at a later date.

Attachment 1, the 25 way diamond formation design. The Line L and distances D (longest design dimension in the formation) and Q (shortest interjumper distance) are identified.

Attachment 2. Mark Harris's photo

Attachment 3. The locations of the jumpers' heads identified, and scaled so the formation sizes D correspond.

Attachment 4. Overlay of actual head positions on formation design, with largest error identified.

The largest deviation from the correct position was 21.4% of the distance Q (the smallest defined spacing between jumpers). The average error was 11.8%


I did the analysis entirely by hand except for scaling the picture, in under 10 minutes.

...

The only sure way to survive a canopy collision is not to have one.

[mccordia 73]

Cool...shows promise.

Maybe you should somehow see if there is anything you could do to actually try and have some room for further development and testing.
As there is this FAI thing next week, where some people are trying to get a different system accepted. Its nice to see you have it all worked out.

If only there where some people going to the FAI meet to show there ARE alternatives. *sigh*

JC
FlyLikeBrick
I'm an Athlete?

[mccordia 73]

That aside, it would be good if there was a way you could incoperate/allow for any skew or slight off-heading aim in one wing of the formation.

Take the FS 300/400 way records, and try and copy/paste that one on top of a photo.

You will see some of the wackers are rotated quite a bit of their axis.
If it wasnt for their relative positions to eachother (grips) and the ability to judge them on that account, it would for sure not match any predrawn/scaled map/image you could overlay.

What if only one (say the left) side of a 100 way 'scales' out, because of people flying 30 cm too far from eachother.
Would that put the person on the far left corner so far away from his/her slot, it would be deemed 'out'. Even though the overal picture could maybe look incredible?


This really is serious interest, but could you take a diveplan such as this one and use your method to judge the actual positions and succes of the formation (ignoring docks).

Would it be a good thing to judge the formation as a hole, or would it be better to judge people on their own position relative to other flyers?
And is there maybe a hybrid way of doing this?

JC
FlyLikeBrick
I'm an Athlete?

[kallend 1,647]

Quote

Cool...shows promise.

Maybe you should somehow see if there is anything you could do to actually try and have some room for further development and testing.
As there is this FAI thing next week, where some people are trying to get a different system accepted. Its nice to see you have it all worked out.

If only there where some people going to the FAI meet to show there ARE alternatives. *sigh*

Unfortunately I only had the idea on Wednesday (2 days ago) at about 3:00a.m. CST, worked out the details over breakfast, and posted it here straight away. It's rather short notice to go to Europe! I have submitted it to USPA.

Today was the first opportunity I had to test it on a real formation.

...

The only sure way to survive a canopy collision is not to have one.

[mccordia 73]

picture

And maybe take one whacker (all facing the same way, roughly), and judge it like it was a wingsuit formation with your system.

JC
FlyLikeBrick
I'm an Athlete?

[kallend 1,647]

Quote

This really is serious interest, but could you take a diveplan such as this one and use your method to judge the actual positions and succes of the formation (ignoring docks).

Hey, I'm in that formation!

I think my method can judge ANY predefined formation, although there's no point for docked formations.

The concept is easy. Either you are in the right place, or you are not. If not, measure how far you are from the right place. The biggest error on the Summerfest 25 way was around 22% of the defined spacing between jumpers. Most think that was a pretty good looking formation.

I'm still trying to establish a reasonable pass/fail tolerance.

WHAT TOLERANCE ARE YOU USING? Austin Powers was not very enlightening.

...

The only sure way to survive a canopy collision is not to have one.

[kallend 1,647]

Quote

picture

And maybe take one whacker (all facing the same way, roughly), and judge it like it was a wingsuit formation with your system.

It would require that I know where they were supposed to be, which I can't tell just from the photo.

The whole point of the method is that the formation is defined ahead of time, and what the jumpers do is compared with what they were supposed to do.

...

The only sure way to survive a canopy collision is not to have one.

[mccordia 73]

Quote

I think my method can judge ANY predefined formation, although there's no point for docked formations.

Why not? Is there no point, because it doesnt work?

It was a serious question. Only in a mathematic world, does everyone fly like pieces of lego.

If we fly a formation with V (60 people).
10 cm spacing.
But the left wings is a few degrees off, and the last person because of that (according to your system) ends up being 15 meters from where he should be. Yet the picture looks incredible, and as tight as could ever be.

What would your system say?
Again, asking for a way to include these kind of (everyone would say) succesfull formations in the validation.

Quote

WHAT TOLERANCE ARE YOU USING?

Funny..I wanted to ask you the same thing about your hearing-aid.


Your top post in this thread reads> A modest proposal (though having difficulty with the 'modest' part, but lets ignore that one for now)

You posted a proposal for feedback.
"Scotty dont..."

We will. You'll be able to comment. Give input. Anything you could want.
But untill we actually post a proposal. And publicly show it. And ask for feedback/input.

But at this point in time, completed and flawless proposals are not important, if nobody is there to SHOW there ARE alternatives to the one system that IS being presented for acceptance, WITHOUT input.

As a professor, I may hope you also dont grab any piece of paper that you see a student walk around with, and start rating and commenting on it before they even finished a first draft.

You also wait untill somebody hands it to you for feedback (given they promise to actually do that)

JC
FlyLikeBrick
I'm an Athlete?

[kallend 1,647]

Quote

Quote

I think my method can judge ANY predefined formation, although there's no point for docked formations.

Why not? Is there no point, because it doesnt work?

There's no point because the criterion for judging a docked formation is already well established. My proposal states very clearly that it for judging unlinked formations.

The very first line says:

A Method for Evaluating the Perfection of 2-D Formations of Unlinked Skydivers

Quote

As a professor, I may hope you also dont grab any piece of paper that you see a student walk around with, and start rating and commenting on it before they even finished a first draft.

Your document is posted on the Internet for all the world to see. No-one "grabbed" anything, it was presented to us.

...

The only sure way to survive a canopy collision is not to have one.

[mccordia 73]

Quote

Your document is posted on the Internet for all the world to see. No-one "grabbed" anything, it was presented to us.

No it wasnt. Someone linked to it. Even though its only there for the FAI, and showing a proposal for alloted time to present a concept.
Its not a hidden thing, or problem for anyone to see/read.

But there are reading materials, that function as a lead-in to presentations held next week at the FAI. Nothing more, nothing less.

Quote

A Method for Evaluating the Perfection of 2-D Formations of Unlinked Skydivers

Thats just your statement, and you ignore the question:
What about skew in a formation? your method doesnt allow for this, even though its something that exists in any real world flock/formation.

Id be more than eager to include this method of yours in software form, and maybe even see if some hybrid is possible. Taking the best from any method.

but I hope you have the patience, and decency to also allow for some actual time to work on things....as you may be at home with a ton of time on your hands. But trying to make sure we can develop this further, some of us have other things to attend to (while jugling a day-job and social life)

JC
FlyLikeBrick
I'm an Athlete?

[kallend 1,647]

Quote

Quote

Your document is posted on the Internet for all the world to see. No-one "grabbed" anything, it was presented to us.

No it wasnt.

Funny that we can all read it there, then. Even Google can find it.

Quote

Quote

A Method for Evaluating the Perfection of 2-D Formations of Unlinked Skydivers

Thats just your statement, and you ignore the question:

We don't evaluate a Ferrari on how well it hauls 20 ton steel beams. We don't evaluate an Airbus380 on its ability to dogfight.

My proposal was not designed to evaluate linked formations, and I stated as such very clearly.

...

The only sure way to survive a canopy collision is not to have one.

[kallend 1,647]

According to the method described in post#1 of this thread, the maximum position error in the formation in the following link is 52% of the inter-jumper spacing, and the average position error is 25%.

news.flylikebrick.com/2010/01/florida-state-record.html


Since the method has been described in complete detail, anyone who wishes can verify these numbers.

...

The only sure way to survive a canopy collision is not to have one.

[bofh 0]

Quote

Quote

I think my method can judge ANY predefined formation, although there's no point for docked formations.

Why not? Is there no point, because it doesnt work?

It was a serious question. Only in a mathematic world, does everyone fly like pieces of lego.

The proposed method could easily be extended to handle the case of wackers by partitioning the original formation and judge each partition on its own and also judge the angle between the partitions. How it is partitioned should of course be specified before the jump and I think the judges would have to use their judgement to cover all cases.

In the case of the V formation, each leg would be a separate partition and the angle between the partitions' L-lines would be compared.

[kallend 1,647]

Analysis of the Elsinore 68-way:

The formation is skewed, with a skew angle on average of 2.5 degrees. The formation is also stretched slightly along line of flight (by 6%) giving the formation as a whole a distortion strain of 7%.

If the formation is evaluated against its design, the maximum position error of any jumper is 48.5% of the inter-jumper spacing and the average is 24.9%

If the jumpers are evaluated against the formation's actual overall (slightly skewed and stretched) shape the maximum position error drops to around 30% of the interjumper spacing.

...

The only sure way to survive a canopy collision is not to have one.

[Costyn 0]

John,

I've read through this thread again and I like the simplicity of your idea.

Like Jarno, I'd like to see it added into our software to and see what it does with various formations. The interface shouldn't need too much tweaking, we only need to add some code to do the calculations. Yes, you can do it with a ruler and some tracing paper, but I think if we want more people to play with it, we should have something where you can click around and have it come up with a number.

So yea, those are my thoughts.

Cheers,

Costyn.

Costyn van Dongen - http://www.flylikebrick.com/ - World Wide Wingsuit News