Merged Discussion of femr's video data analysis

In fact you DO make assertions about NIST details further down in THIS post, but appear unable to criticise it in the slightest.
Wrong. I believe the shape of your graphs is more accurate than NIST's linear and curve fits. I thought I had said so in the same message and some messages ago.


Double-standards son.
With a basis on the above? Yah, sure :rolleyes:


So, where might I find your posts moaning and whining about lack of NIST error analysis, lack of perspective correction, lack of camera motion extraction, lack of... ? :)
I don't see any point on commenting anything on that. I also don't comment on mistakes by Millette, for example, even if it seems clear he made at least a couple. I'm not discussing with Millette. I'm not discussing with NIST. This is a forum in which you made claims, and I'm discussing them with you.


During the earliest near-release moments, indeed.
Not only.

SGerrors.png

(cropped only so the width isn't so annoying)

There will certainly be a lower error band during periods of no/minimal motion, for numerous reasons.
Which is obviously not the case for the period you claim to be over g.


ROFL. Why have you not produced an error analysis for your plots ? Why is the behaviour of your curve different to that of the 1-Pass SG plot around the release point ?
I have not used my graphs to claim over g. You have.


ROFL. Not confused in the slightest son. You referred to one image whilst THINKING about a different one.
:boggled:

That's false as I already explained. I talked about a segment of the blue horizontal line in the graph as being the acceleration that NIST claimed, in the context of discussing the utility of your graph. My statement was correct.


Or are you unaware that NIST did two fits
I'm more than aware of 12.5.4, as well you know. They only did one acceleration "fit" btw :rolleyes:
Where do I say "acceleration" fit? :confused: You introduced it when you pointed to the green line. :boggled:


So ? All after the fact dear boy.
:confused: After what fact? I referred to the blue line since the beginning. The blue line is a straight horizontal line that corresponds to the derivative of NIST's linear fit. I referred to NIST's fit. The reference should have been obvious since the very beginning.


Could you explain to me exactly why I should waste any of my time with you ?
Because you can't resist answering a single sentence addressed to you, I guess. It's not that I'm putting a gun in your head for you to reply.


Exactly what is it that you are trying to prove ?

Exactly what purpose do you have for the data ?

What exactly are you trying to achieve ?
1) I'm now more convinced about early motion. The trace data was not enough for that; movement seen in the video was clearer.

2) You have claimed that the acceleration went over g. I don't think you have proved that.

3) I am defending a claim I made and you have attacked, namely that the issue with perspective is distance from the camera (depth).

4) I think I am in a position to undo perspective effects from Cam #3. That would enable usage of that data which currently can't be taken enough advantage of.


When are you going to stop wandering around between early motion and acceleration derivative ?
You don't get to dictate the limits of discussion here pal :)
Ditto.

So is Dan Rather, which is equally fully affected by all perspective effects
Er, no. Whilst, of course, the same range of differing perspective distortions affect the Dan Rather footage, the magnitude of several of those effects are much smaller for such than for the cam#3 footage.
So, Dan Rather is also fully affected by all perspective effects, is it not?

Equally is another word for also. Or it is in Spanish.

Prove pgimeno wrong version the umpteenth? :rolleyes:


I don't have problems with such in the slightest...it simply isn't...so discussing it is a pointless waste of time. Might as well have a pointless discussion about what the situation would be if it was filmed with a fish-eye lens, or a telecentric lens :rolleyes:

It wasn't. Pointless waste of time.


ROFL. The WTC7 footage suffers from numerous perspective distortions, in all three dimensions, non-linearly. Cam#3 especially so. Correct perspective correction goes far beyond your primitive and repeated assertions about "depth". "The isue is" ?! ROFL.
If you suspect perspective issues, evaluating the distance of the measured points to the camera will give you an idea of how important the perspective issues are.

In the case of Dan Rather, the distance is too big and the variation of distance too small for the effect of perspective to be of any significance. The movement is also predominantly vertical so the angle of the fall with the camera plane is also quite small to be significant. For Cam #3 the angle is probably even less than for Dan Rather.


Depth "data". Per sample ? Ew. If you want to go generate that data, go ahead. Quite what you want to do with it is your own affair.
I might give it a try. As a first step I need the Dan Rather and cam #3 location. What information do you have?

My best shot at cam #3 location was this: http://www.formauri.es/personal/pgimeno/xfiles/11-s/Cam3Location.png but it still has some issues, namely the wall of the building on the left is not well aligned with the picture, the windows don't match very well (though that might be normal), and the corners of old WTC7 and new WTC7 don't match perfectly. Not to mention that I've seen issues with the heights of the buildings given by Google Earth in past. ETA: Also, it assumes 1:1 aspect ratio, which may or may not be correct.

My aim is to calculate the 3D path to measure the depth and the angle with the camera plane for each data point. With that, I can give perspective-corrected data for the view in real world units. My idea is to apply minimum and maximum estimations of the error to the data, in order to produce two datasets that determine an error band.


and it doesn't need to be extremely accurate, as small variations in depth make a very small difference
For what purpose ?
For the purpose of removing the perspective effects from the data without adding much noise/error.


How did you remove lens distortion ? :rolleyes:
I won't, just as you didn't for Dan Rather. Lens distortion is not perspective distortion.


How did you negate adding noise to the sub-pixel fidelity data ?
I hope to be able to give an error band and prove the fidelity of the manipulation of the data.


You can't ignore other perspective effects and call the result an orthographic projection.
Nor was I suggesting that :confused:


No, it's the point. You don't have "the depth", you don't have information to be able to generate "the depth".
Moving your goalposts again? "The depth" was totally not the issue before. It was completely unrelated to the issue. I was so wrong about it being the issue. Now the issue is that I can't generate it. :rolleyes:

I just need the camera positions to generate "the depth". That and:

- Position of the NW corner.
- Height of the building.
- A reference point at a known height, in order to match both videos. Possibly two, depending on whether the aspect ratio is not 1:1
- The traces from both cameras.
 
Last edited:
The error bands are bigger than the error bands at some points? Can some mathy folks help me with how that works please?
 
I believe the shape of your graphs is more accurate than NIST's linear and curve fits.
Still avoiding actually crisicising them I see.

I don't see any point on commenting anything on that.
Convenient.

I'm not discussing with NIST. This is a forum in which you made claims, and I'm discussing them with you.
NIST made claims, such as "2.25s at gravitational acceleration". Oh dear :)

Not only.
And ? The useful output from all the various trend views is simply trend shape and magnitude. All methods presented show similar trend shape and similar magnitude. Peak acceleration between both poly and both SG plots are within a few ft/s^s.

Which is obviously not the case for the period you claim to be over g.
My assertion stands.

I have not used my graphs to claim over g. You have.
So ?

If you are concerned, perform an error analysis.

That's false as I already explained.
No, it's not. Can't be bothered to requote it all again, but can do if necessary :rolleyes:

Where do I say "acceleration" fit? :confused:
ROFL. The acceleration value derived from the linear regression. They didn't bother differentiating their non-linear velocity function, much to W.D.C's consternation.

The reference should have been obvious since the very beginning.
Are you really going to make me quote you again on such a pointless and trivial point ? You messed up on graph referencing, that's fine. Accept it and move on son.

Because you can't resist answering a single sentence addressed to you, I guess. It's not that I'm putting a gun in your head for you to reply.
When you talk crap, it has to be addressed. Otherwise the peanut gallery could think some valid point has been made.

1) I'm now more convinced about early motion. The trace data was not enough for that; movement seen in the video was clearer.
Dear me. What a joke.

2) You have claimed that the acceleration went over g. I don't think you have proved that.
Prove me wrong :)

3) I am defending a claim I made and you have attacked, namely that the issue with perspective is distance from the camera (depth).
I haven't attacked. I've told you you're missing a whole heap of additional perspective effects.

4) I think I am in a position to undo perspective effects from Cam #3. That would enable usage of that data which currently can't be taken enough advantage of.
Go ahead.

So, Dan Rather is also fully affected by all perspective effects, is it not?
It's real world footage...of course it is affected by perspective, just to a lesser extent than for cam#3 in terms of extracting vertical motion of the NW corner of WTC7. Catch up ! This is like cutting teeth, repeatedly.

If you suspect perspective issues, evaluating the distance of the measured points to the camera will give you an idea of how important the perspective issues are.
Ye gads man. Have already looked at perspective in depth a long time ago.

If you are interested, do it.

As a first step I need the Dan Rather and cam #3 location. What information do you have?
Not a lot. You'll have to work it out I'm afraid. Then I can spend a couple of years complaining that your results are useless as you do not know the exact camera locations.

My best shot at cam #3 location was this
See above. I'm afraid I'm not going to be assisting you much. I can get within a couple of feet(ish). Once you're inch-perfect by all means begin the process of perspective correction for whatever purpose it is that you have in mind. You will of course also need to provide a full error analysis...:rolleyes:

See that above ? Irritating, isn't it.

Howsabout this...

Hey, if you want to look at perspective correction for the cam#3 viewpoint, great. I didn't go there as it is a whole can of worms, and imho is likely to add noise to existing data rather than reduce it. Good luck.

My aim is to calculate the 3D path to measure the depth and the angle with the camera plane for each data point.
You can't determine depth during descent with one camera view. You'd be better off determining camera-object (wtc7) orientation during static motion period and performing a simple geometric correction.

With that, I can give perspective-corrected data for the view in real world units. My idea is to apply minimum and maximum estimations of the error to the data, in order to produce two datasets that determine an error band.
Good luck with that. Remember to include all the other inherent noise sources on top.

For the purpose of removing the perspective effects from the data without adding much noise/error.
Hmm.

I hope to be able to give an error band and prove the fidelity of the manipulation of the data.
Good luck.
 
Last edited:
You have claimed that the acceleration went over g. I don't think you have proved that.
Tell you what, as I am bored to death of this pointless word salad...

Why don't you look at displacement data between, say, 12.6 and 13.2s (the period of maximum acceleration indicated by the various acceleration trend plots). Or why not widen the duration to, what, 12.5s to 14.5s...smooth out the bumps ;)

1) Start with displacement/time data as provided (or generate your own)
2) Determine velocity using symmetric difference. Use an appropriate window width. You'd have to use a 13 sample window to match NIST spacing.
3) Perform a simple linear regression.
4) State the average acceleration that pops out the other end.

If you get a value over-g, you're in the right ballpark.

;)

NIST didn't bother with error analysis, using the same method and with their data tainted by many issues, so you shouldn't have a problem with your resultant value. Ta da.

(Are you going to post or not ? You've been at that for an hour already)
 
Last edited:
No, it's not. Can't be bothered to requote it all again, but can do if necessary :rolleyes:
I'll do. I introduced the topic of NIST's linear fit here:

And if you had paid attention to the context of my post, you'd have seen that my suggestion regarding using Cam #3 acceleration plots included ignoring the vertical axis units and using the graph for what the Dan Rather based ones are good for: looking at the overall shape of the graph. Which, beyond the fact that it reveals a segment not as horizontal as it would be expected when comparing it with NIST's analysis, I don't see the point of.
Highlighting and bolding added. True, I left NIST's curve fit out. I was trying to be brief. Ironic that that snippet was the target of another series of rants from you. You quoted the highlighted part and asked:
My response clarified what line I was referring to, though it left the specific segment somewhat undefined, hoping you'd be familiar with it enough as to not need further clarification:
NIST basically say that the acceleration roughly corresponds with the blue line in this graph for a certain segment:

http://femr2.ucoz.com/_ph/7/513801604.png
(Again, bolding added). I knew all that time that you had another graph isolating NIST's segment for the SG plot only, but that would miss the variety of graphs all showing the same rough shape. This was all the time in the context of having said that the utility of your graphs didn't go to the point of quantifying the acceleration, but it merely served to compare the shape with others, in particular NIST's. I commented on that following the posting of that graphic, in the same post:
The shape of your graph reveals the behavior is a bit different to an horizontal straight line. That's all the utility I see in your acceleration graph without an adequate quantification of error. And I don't see the shape deviating that much from the horizontal line within the degree of precision they used.
All that being clarification of my point. My comments certainly applied equally well to NIST's acceleration plot. You quoted the highlighted part in your response and added, besides asking about the error band:
Did you notice the green line on the graph you linked ?
Perplexed by the question, I asked:
Your reply:
Jeez...

[qimg]http://femr2.ucoz.com/_ph/7/2/513801604.jpg[/qimg]
You said - NIST basically say that the acceleration roughly corresponds with the blue line in this graph for a certain segment.

I pointed you towards the green line, which is the NIST formula for velocity differentiated to acceleration.

The green line is somewhat a little differnet to the blue line.

Now what can you see out of the round window :rolleyes:
OK, not what I was talking about but since it's true that the green line also differs in shape, I conceded anyway:
Yes, I agree that the shape is somewhat different to NIST's in that case too. That's about the utility of your graphs, to compare shapes.

I was referring to this one:

[qimg]http://femr2.ucoz.com/_ph/7/2/5930538.jpg[/qimg]
"this one" being "this line", "this segment", "this fit"... You interpreted it as "this graph". And now you have the guts to accuse me of changing my words "after the fact" when I was referring to the same fit and segment since the very beginning? But you continue digging your hole:
You said..."NIST basically say that the acceleration roughly corresponds with the blue line in this graph for a certain segment."

Which part of the green NIST curve would you say fits your statement ? :)
That's just as nonsensical as asking, Which part of the NIST curve fit corresponds with NIST linear fit?

Really? I mean, really? And wasn't I clear since the beginning that a segment of the blue line corresponds to their linear fit?

Puh-lease. :rolleyes:


Are you really going to make me quote you again on such a pointless and trivial point ? You messed up on graph referencing, that's fine. Accept it and move on son.
Clearly, I referenced a valid graph since the beginning, even if I pointed to another valid graph (although less rich in stuff to compare to) later. And I still have to see you concede anything, despite the number of times I've proved you wrong. :rolleyes:


2) You have claimed that the acceleration went over g. I don't think you have proved that.
Prove me wrong :)
"Wrong" as in "it is false that the acceleration went over g" is yet again the same fallacy you've been falling into countless times. I don't claim that. My claim is that you haven't proved it, which makes it a different kind of wrong. And you have proved yourself wrong in that sense when you stated that you don't know the error. You later gave an error band and I showed that the error is bigger than that error band, therefore the band can't be correct. There, proved.


I haven't attacked. I've told you you're missing a whole heap of additional perspective effects.
There are no additional perspective effects other than depth. It's the only variable that differentiates orthographic (non-perspective) projection from perspective projection, as I noted before. Now there are a lot of variables that can have an influence on depth, sure. But it is the ultimate variable that is into play here, and the one you dismissed.

You mentioned FOV and orientation as examples. FOV doesn't alter depth at all. Orientation does, as it changes the coordinate system and therefore the camera plane with respect to which depth is measured. FOV does have a role in converting the data to 3D, as the process involves casting rays from the camera's focal point through each image's pixel, therefore the distance to the image (focal distance, more properly speaking) influences that. But there are indirect ways to measure it if you have location information of certain 3D points in the image.


It's real world footage...of course it is affected by perspective, just to a lesser extent than for cam#3 in terms of extracting vertical motion of the NW corner of WTC7. Catch up ! This is like cutting teeth, repeatedly.
Which makes it silly for you to contend my assertion. Proves you didn't pay attention to my words despite you being
Always paying attention.
:rolleyes:


See that above ? Irritating, isn't it.
Not as much as your games with refusing to provide the SG data or other data you've refused to provide, such as location info.


You can't determine depth during descent with one camera view.
I know. That's why I said:
I just need the camera positions to generate "the depth". That and:

- Position of the NW corner.
- Height of the building.
- A reference point at a known height, in order to match both videos. Possibly two, depending on whether the aspect ratio is not 1:1
- The traces from both cameras.

(Clarification: 1:1 alludes to the pixel aspect ratio, not to the image aspect ratio.)
 
You can't be trusted. Swamp the trivial point with waffle.

Here y'are...

.I was referring to this one:

5930538.jpg
No, you weren't (at least not to those unblessed with the power of mind-reading), as you repeatedly cited this one...
...when making all statements prior to you saying...
.I was referring to this one
Simple as that son. You can try and shoe-horn something out of it, but it's trivial, boring and inconsequential. Quite why you'd bother trying to make a meal out of it is beyond me :rolleyes:

I knew all that time that you had another graph isolating NIST's segment
Pity you "were referring to that one" whilst citing a different one then :rolleyes:

And I still have to see you concede anything, despite the number of times I've proved you wrong.
LOL. Believe what you please good sir ;)

"Wrong" as in "it is false that the acceleration went over g" is yet again the same fallacy you've been falling into countless times. I don't claim that. My claim is that you haven't proved it, which makes it a different kind of wrong. And you have proved yourself wrong in that sense when you stated that you don't know the error. You later gave an error band and I showed that the error is bigger than that error band, therefore the band can't be correct. There, proved.
ROFL.

I'm "soooo" not interested in your wandering definition of "wrong".

If you don't think there is a period of over-g, prove it ;) (And good luck with that)

There are no additional perspective effects other than depth.
Oh dear.

Now there are a lot of variables that can have an influence on depth, sure.
Not just "depth". Spatial position in all three dimensions..or in this case four dimensions (though very slight).

But it is the ultimate variable that is into play here, and the one you dismissed.
I haven't dismissed "depth" in the slightest.

You mentioned FOV and orientation as examples. FOV doesn't alter depth at all.
Are you quite sure about that ? :) Perhaps you'd care to highlight why a point TL of frame and a point BR of frame are not affected by lens field of view effects. Perhaps you'd care to highlight why narrow field of view lenses and wide field of view lenses are identical in terms of perspective correction. Etc.

Orientation does
Who are you trying to "educate" ?

It's clear you are googling as you go.

Which makes it silly for you to contend my assertion.
No, it doesn't. I have made my views on the benefit (or not in this case) of attempting perspective correction upon the available trace datasets clear numerous times.

Proves...
Nope.

Not as much as your games with refusing to provide the SG data
What "SG data" have I refused to provide ?

or other data you've refused to provide, such as location info
ROFL. If I provide you with such data you'll simply spend another year whingeing about how it was calculated. Do it yourself.

What "other data" have I refused to provide you with ?

Splendid. Now, how are you going to go about synchronising multiple videos, managing differing scaling metric, feature resolution, differing noise levels, differing camera position accuracy, ... ? :rolleyes:

(Clarification: 1:1 alludes to the pixel aspect ratio, not to the image aspect ratio.)
That would be a wonderful "clarification", except the pixel aspect ratio isn't 1:1.

See my previous post :)

Also, a quick point from my earlier post you decided to ignore...

The useful output from all the various trend views is simply trend shape and magnitude. All methods presented show similar trend shape and similar magnitude. Peak acceleration between both poly and both SG plots are within a few ft/s^2.


(How long does it take you to write a post ?! Approaching 2 hours...)
 
Last edited:
As I have several times challenged those opposing femr 2 - there are three questions current on the thread:

The first one is "Are there errors in femrs work?" This one has been pursued with some vigour as if it was relevant and significant. It isn't. It is a strawman argument based on a truism and a lie by innuendo - the truism being that all such measurement techniques have errors - due to "noise" and perocedural artefacts" to put it simply. The false implication - strictly the as yet unproven implication - is that the errors are "big enough to matter". Which I doubt but I'm not the one making the claims and I try to avoid "reversed burden of proof" trickery.
Those of us who are capable of speaking plainly can answer that first question more straightforwardly: Yes, there are errors in femr2's work.

The second is the question which no-one opposing femr wants to address. And it is the one that matters and the only one on this topic that is significant to numbers of members. "Are the errors big enough to matter?" Status - no one has shown that they are. (Therefore they aren't BTW)
femr2, Major_Tom, and possibly a few of their enablers are pretty much the only folks here who think femr2's video analysis matters. Even they have gone to some trouble not to explain why it matters. Those of us who are still waiting for someone to explain why femr2's video analysis might matter (even if it were entirely correct) are unlikely to think femr2's errors matter with respect to 9/11 conspiracy theories.

Asking femr2's critics to explain why femr2's errors matter is reversing the burden of proof. It's up to femr2 to state a purpose for which his errors would not matter. He hasn't done that.

On the other hand, femr2's errors might matter in this sense: By repeating and defending his errors, femr2 may confuse those who would like to understand the benefits, techniques, and limitations of video analysis.

I comment on femr2's errors for much the same reasons I would comment upon errors made by someone who's denying relativistic length contraction or saying Gauss's law for magnetism forbids magnetic reconnection. My comments, like femr2's, don't have any apparent connection to 9/11 conspiracy theories, so it would be fair to say they might be more appropriate for the subforum devoted to Science, Mathematics, Medicine, and Technology. For whatever reason(s), femr2 and Major_Tom want us to discuss their errors here.

The third question is of specific interest to pgimeno. At the risk of getting him wrong I understand it to be that he is not convinced that femrs explanations of his methods and data support the methods.
More plainly and accurately: femr2's methods support only some of the meager conclusions femr2 has drawn from his acceleration graphs. Several of femr2's detailed conclusions are unsupported by his analysis and are probably wrong.

That doesn't matter if 9/11 conspiracy theories are your only concern, but if that's your only concern then femr2's video analyses don't matter at all (to you).

The error bands are bigger than the error bands at some points? Can some mathy folks help me with how that works please?
femr2 claimed a degree of accuracy that's contradicted by his own graphs.

When given clear evidence of errors in femr2's graphs that exceed femr2's claimed error bound, femr2 airily dismissed the criticism, apparently hoping to bury pgimeno's entirely correct criticism under another femr2 gallop.

You knew that, of course.

Here are some other recent examples of femr2's doublespeak:

One of the "funniest" recent "turns" (of which there are many as silly "angle" after silly "angle" is refuted) is the repeated decision from posters within the thread (primarily pgimeno and W.D.Clinger) to totally ignore the 1-pass Savitzky-Golay acceleration derivative curve provided over a month ago.
femr2 posted his true Savitzky-Golay acceleration curve the day after I had posted part 1 of my occasional series on Savitzky-Golay smoothing, in which I showed that femr2's alleged Savitzky-Golay curve of almost two years ago had not been generated by Savitzky-Golay differentiation using parameters that were anywhere near the parameters femr2 claimed to have used.

As I explained in part 2 of my occasional series on Savitzky-Golay smoothing, the Savitzky-Golay algorithm is always 1-pass, and the curves it produces always have several properties that femr2's alleged Savitzky-Golay curve does not have. In the paragraphs quoted above and below, femr2 is putting the "1-pass" in front of Savitkzy-Golay so he can pretend his alleged Savitzky-Golay curve was the result of using Savitzky-Golay differentiation on his data. This is another example of femr2 redefining words for deception rather than communication.

There was never any reason for anyone to comment upon femr2's belated discovery of true Savitzky-Golay differentiation. His curve added nothing to the Savitzky-Golay curves I had posted the day before, and my part 2, posted the following day, offered a more complete and accurate description of Savitzky-Golay smoothing/differentiation than can be found in all of femr2's posts combined.

Given that their alleged "issues" with the older acceleration curve, provided about 2 YEARS ago, are based upon them not knowing exactly how it was produced, I find it very "interesting" that they continue to completely ignore the 1-pass SG curve for which they do have full details for complete replication.
femr2 did not provide "full details for complete replication" when he posted that curve. femr2's true SG curve resembles the SG curve I posted for smoothing polynomials of degree 2 and windows with 61 data points, but there are too many small differences for me to believe femr2 used those parameters.

So far as I know, femr2 has never stated the parameters he used to generate that Savitzky-Golay curve. femr2 has admitted he's been withholding details needed to replicate other curves.

femr2's acceleration curves deserve to be ignored. When I use Savitzky-Golay or some more sophisticated algorithm to generate an acceleration curve from femr2's data, I know exactly what methods and parameters I'm using, and it's very easy for me to explain my methods and parameters to anyone who cares. Why should I (or anyone else) bother to guess at femr2's parameters or methods?

But then again, how could such continue to complain/reject/... when they have...

a) The raw numerical source data
b) The numerical data in real-world units and full details of how they were computed
c) Full disclosure of acceleration curve generation methodology
d) Ability to replicate acceleration curve themselves and perform any amount of "analysis" of both error propogation and derivative parameter variation.
We have a) and b). So far as I know, femr2 has never given an adequate description of the methods and parameters he used to generate his alleged Savitzky-Golay curve of almost two years ago, so I think he's basically just lying about c).

We certainly have the ability to produce acceleration curves ourselves and to analyze their artifacts and errors. Some of us have done so.

femr2, on the other hand, has allowed himself to be misled by smoothing artifacts in his acceleration curves. femr2's inability to acknowledge the obvious contradiction between his claimed ±3 ft/s² error and his own curves was fairly typical of femr2's video analysis.
 
Those of us who are capable of speaking plainly can answer that first question more straightforwardly: Yes, there are errors in femr2's work.
There is certainly an amount of unavoidable noise in the raw data, but I sense you are trying to infer some other kind of "Error". Care to elaborate ? ;)

It's up to femr2 to state a purpose for which his errors would not matter. He hasn't done that.
Incorrect.

By repeating and defending his errors
And what errors may they be ?

I comment on femr2's errors
Again, what errors ? ;)

For whatever reason(s), femr2 and Major_Tom want us to discuss their errors here.
ROFL. I respond to posts where they are. This is tfk's thread btw.

femr2's methods support only some of the meager conclusions femr2 has drawn from his acceleration graphs.
Would you like to make some grandeur conclusions ? :rolleyes:

Assertions from me have been pretty minimal, indeed.

Several of femr2's detailed conclusions are unsupported by his analysis and are probably wrong.
Dear me. A handful of clearly stated approximate "stages" and the irresistable urge to try and "prove femr2 wrong version the umpteenth" kicks in. Splendid :)

femr2 claimed a degree of accuracy that's contradicted by his own graphs.
Nope. I stated a simple error band based on a simple wide window symmetric difference.

femr2 airily dismissed the criticism, apparently hoping to bury pgimeno's entirely correct criticism under another femr2 gallop.
ROFL. See above.

femr2 posted his true Savitzky-Golay acceleration curve the day after I had posted part 1 of my occasional series on Savitzky-Golay smoothing
ROFL. Nope. I posted it...
Rather than waste more of my time engaging in fairly tedious ping-pong with opinion based assertions from others, and as I'm a thoroughly super chap, have a peek.
8th February 2013, 01:38 AM GMT, over 17 hours before your post of 8th February 2013, 06:50 PM GMT
. :rolleyes::D

in which I showed that femr2's alleged Savitzky-Golay curve of almost two years ago had not been generated by Savitzky-Golay differentiation using parameters that were anywhere near the parameters femr2 claimed to have used.
ROFL.

1) Where have I claimed that the old SG curve was generated by savitzky-golay differentiation ?
2) Why can you not make up your mind whether I have withheld "parameters used" or provided "parameters used" that you disagree with ?

the Savitzky-Golay algorithm is always 1-pass
The Savitzky-golay filter can be applied as many times as you please. Multiple passes are quite "real". Your "spherical cow" tendencies cropping up again Will.

In the paragraphs quoted above and below, femr2 is putting the "1-pass" in front of Savitkzy-Golay so he can pretend his alleged Savitzky-Golay curve was the result of using Savitzky-Golay differentiation on his data.
ROFL. Nope. You can pretend you know what I'm thinking, or make up your own version of my intent, for your own purposes, but I'm afraid you're quite wrong.

There was never any reason for anyone to comment upon femr2's belated discovery of true Savitzky-Golay differentiation.
ROFL. If your dumb assertion were true I wouldn;t have been discussing it for these many, many long and tedious months :rolleyes:

Do you really believe the crap you write Will. Whilst I find your intensely rude, false and self-serving assertions rather unpalatable, you don'alf make me chuckle at the same time.

Unadulterated "address the arguer" :)

His curve added nothing to the Savitzky-Golay curves I had posted the day before
ROFL. You were late to the party.

Let me see how this "feels"...

W.D.Clinger's curves added nothing to the Savitzky-Golay curve I had posted 17 hours ealier.

Nah. Simpering smarm is "icky".

femr2 did not provide "full details for complete replication" when he posted that curve.
ROFL. Full details were provided to pgimeno when he asked :rolleyes:

So far as I know, femr2 has never stated the parameters he used to generate that Savitzky-Golay curve.
I'm afraid I don't care what you don't know Will. Why not ask ? Or is it your preference to make incorrect assertion after incorrect assertion ? ;)

I prefer it that way. It's entertaining.

femr2 has admitted he's been withholding details needed to replicate other curves.
CurveS ? Ooh, which ones ?

I've certainly withheld, from you, the process by which the old SG curve was generated, and it has been very enlightening to see your long-term, many and varied responses.

What other curve(s) are you referring to ?

femr2's acceleration curves deserve to be ignored.
You seem to have one hell of a job doing so... :rolleyes:

You've even written your own SG implementation, just for me. Ah, fanx :) Shan't be using it thouh. It's imcomplete, written in an obscure and pretty redundant language and coded to behave in a ridiculous manner (simply for you to try and prove a point). But thanks anyway [thumbs-up]

When I use Savitzky-Golay or some more sophisticated algorithm to generate an acceleration curve from femr2's data, I know exactly what methods and parameters I'm using
So do I ;)

Why should I (or anyone else) bother to guess at femr2's parameters or methods?
Indeed, yet you have done so on very many occasions, to much personal mirth.

Have you decided whether I have withheld details, or provided details you disagree with yet ?

So far as I know, femr2 has never given an adequate description of the methods and parameters he used to generate his alleged Savitzky-Golay curve of almost two years ago, so I think he's basically just lying about c).
ROFL. Neat. As is perfectly clear if you read the full post you so blatantly omitted context from the list refers to the 1-pass SG plot, as I am sure you know.

Tut. Tut. Such underhand deception. You've done it before, but never so blatantly and obviously. A new low.

And as a reminder, full details have indeed been provided, regardless of what you might choose to believe, or who you may have a desire to accuse of lying (how ironic).

femr2's inability to acknowledge the obvious contradiction between his claimed ±3 ft/s² error and his own curves was fairly typical of femr2's video analysis.
ROFL. See above.

REALLY losing your touch Will.

Keep up the good work matey, and, you never know, I might tell you all about your mostest favoritest plot. Actually, that might be very entertaining indeed.

With my name in bold so many times in one post, I hope people don't start getting the wrong idea ;):blush: x
 
...
REALLY losing your touch Will.

Keep up the good work matey, and, you never know, I might tell you all about your mostest favoritest plot. Actually, that might be very entertaining indeed.

With my name in bold so many times in one post, I hope people don't start getting the wrong idea ;):blush: x
This is your technical stuff. Impressive:reflective of the technical papers of Ross, and Kuttler; failure.
 
femr2 spotted two errors in my previous post.

First, my timeline overlooked two one-word links femr2 had provided in an entirely different thread:

Rather than waste more of my time engaging in fairly tedious ping-pong with opinion based assertions from others, and as I'm a thoroughly super chap, have a peek.
8th February 2013, 01:38 AM GMT, over 17 hours before your post of 8th February 2013, 06:50 PM GMT
. :rolleyes::D
The Savitzky-Golay acceleration curves shown in part 1 of my series on Savitzky-Golay smoothing included the curve for smoothing polynomials of degree 2 with 51 points per window, but I hadn't noticed that femr2 had linked to a graph of that curve in his post above.

femr2 had not yet stated the parameters (2,51) of that curve. On the day after I posted part 1, femr2 got around to posting those parameters, which I overlooked because they were (once again) posted in an altogether different thread.

(ETA: Overlooking that second post in the other thread was my second error.)

Speaking of part 1:

in which I showed that femr2's alleged Savitzky-Golay curve of almost two years ago had not been generated by Savitzky-Golay differentiation using parameters that were anywhere near the parameters femr2 claimed to have used.
ROFL.

1) Where have I claimed that the old SG curve was generated by savitzky-golay differentiation ?
Applying Savitzky-Golay smoothing (without differentiation) to femr2's position/displacement data yields a smoothed version of that position/displacement data. Without differentiation, it does not produce a velocity curve or an acceleration curve. Iterated smoothing (without differentiation) never generates a velocity or acceleration curve. If femr2 did not use Savitzky-Golay differentiation to generate his alleged Savitzky-Golay acceleration curve, then the methods he used to create that curve are even more of a mystery than ever.

The rest of femr2's long response to what I wrote yesterday was intellectually vacuous.
 
Last edited:
Applying Savitzky-Golay smoothing (without differentiation) to femr2's position/displacement data yields a smoothed version of that position/displacement data. Without differentiation, it does not produce a velocity curve or an acceleration curve. Iterated smoothing (without differentiation) never generates a velocity or acceleration curve. If femr2 did not use Savitzky-Golay differentiation to generate his alleged Savitzky-Golay acceleration curve, then the methods he used to create that curve are even more of a mystery than ever.
That's what the post I deleted was about. It went like this:

1) Where have I claimed that the old SG curve was generated by savitzky-golay differentiation ?
What kind of childish game are you playing?

You claimed that here:

I chose to apply two single order derivative SG passes rather than 1 single order derivative SG pass and 1 second order derivative SG pass, as I wanted to minimise window width at all times, and second derivative SG passes were pretty darn noisy still.

My first pass resulted in velocity data from displacement data.

My second pass resulted in acceleration data from that velocity data,

On second thought, I deleted it because I interpreted that he meant one-pass SG differentiation, even if he didn't say so. Just one of the blunders he got us used to.
 
I'm afraid I don't care what you don't know Will. Why not ask ?
So you can play your childish game? I have asked explicitly and you didn't reply:

What degree and window size did you use for 1 pass? And for each pass in the 2 pass version?

What degree and window size did you use for 1 pass?
:)

2, 51

I suggest "others" don't make assumptions about what parameters they "think" may have been used on other graphs. Assumption is the mother of all...

(Though it can be entertaining)
 
the methods he used to create that curve are even more of a mystery than ever
To you, no doubt.

The rest of femr2's long response to what I wrote yesterday was intellectually vacuous.

Interesting. You've "fessed" to two errors, however, there are quite a few others you've "glossed over". Nay mind, though I recommend those reading the thread have another look through my response to see what I mean.

A few points you class as "intellectually vacuous", and have chosen not to respond to...

femr2 said:
W.D.Clinger said:
there are errors in femr2's work.
There is certainly an amount of unavoidable noise in the raw data, but I sense you are trying to infer some other kind of "Error". Care to elaborate ?
It would seem you are disinclined to elaborate.

femr2 said:
W.D.Clinger said:
femr2 has admitted he's been withholding details needed to replicate other curves.
CurveS ? Ooh, which ones ?
It would seem you are disinclined to list these "other curves" you refer to.

The next point requires additional quoting I'm afraid.

during discussion with ozeco41 I posted the following...
One of the "funniest" recent "turns" (of which there are many as silly "angle" after silly "angle" is refuted) is the repeated decision from posters within the thread (primarily pgimeno and W.D.Clinger) to totally ignore the 1-pass Savitzky-Golay acceleration derivative curve provided over a month ago.

Given that their alleged "issues" with the older acceleration curve, provided about 2 YEARS ago, are based upon them not knowing exactly how it was produced, I find it very "interesting" that they continue to completely ignore the 1-pass SG curve for which they do have full details for complete replication.

Another point which makes the decisions to completely ignore the acceleration curve that they have full details to replicate themselves even "stranger".

But then again, how could such continue to complain/reject/... when they have...

a) The raw numerical source data
b) The numerical data in real-world units and full details of how they were computed
c) Full disclosure of acceleration curve generation methodology
d) Ability to replicate acceleration curve themselves and perform any amount of "analysis" of both error propogation and derivative parameter variation.
Bolding mine.

It's utterly, blatantly clear that points (a) to (d) refer to the 1-pass Savitzky-Golay acceleration derivative curve.

All of (a) thru (d) are correct.

So, W.D.Clinger, who has "graciously" accepted two errors on his part clearly indicates that his response to that post is correct, as he has not responded further. Hmmm...
We have a) and b). So far as I know, femr2 has never given an adequate description of the methods and parameters he used to generate his alleged Savitzky-Golay curve of almost two years ago, so I think he's basically just lying about c).

Now, W.D.Clinger has conceded his error about details for the 1-pass SG curve being provided, which is fine, however, his clearly deliberate quote-mining to suggest the points (a) thru (d) refer to the old SG curve remain "ignored".
 
So you can play your childish game? I have asked explicitly and you didn't reply:

pgimeno said:
What degree and window size did you use for 1 pass?
femr2 said:
2, 51

I suggest "others" don't make assumptions about what parameters they "think" may have been used on other graphs. Assumption is the mother of all...

(Though it can be entertaining)
Oh dear.

What degree...2

What window size...51.

Get a grip man.
 
Last edited:
Oh dear.

What degree...2

What window size...51.

Get a grip man.

You need to be spoon-fed every single time?

What degree and window size did you use for 1 pass? And for each pass in the 2 pass version?
That's TWO questions.

You only replied to one.

Your reply to the other was a childish rant. See above. ETA: Reproduced below hoping it sinks:

What degree and window size did you use for 1 pass?
:)

2, 51

I suggest "others" don't make assumptions about what parameters they "think" may have been used on other graphs. Assumption is the mother of all...

(Though it can be entertaining)
 
Last edited:
That's TWO questions.
:dl:

You quote my response to WDC, which is about the 1-pass SG parameters, then whilst accusing me of not replying you quote me providing you with parameters for the 1-pass SG curve immediately after you ask, including me quoting the specific question I was answering so you wouldn't get confused.

LOL.

You only replied to one.
I know. You know full well I'm not going to post details for the old curve, no matter how many times you ask.

REALLY scraping the barrel Pedro.
 
Last edited:

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom