Merged Discussion of femr's video data analysis

This is all way way out of my patience limitations. My glace at this thread (and what I quote in public) is that femr has shown slightly greater than freefal acceleration of a single point on the north perimeter wall during a total of about one second of the 2.25 seconds NIST talks about. Does anyone disagree with that?

I also get from this that there is some rumbling and shaking a few seconds prior to collapse. Are some people saying that is caused by camera shaking or some other kind of error? My very nonscientific guess is that if there was building rumble, that could be evidence of internal structures falling apart prior to the external perimeter collapse.

I continue to say "possible greater than freefall acceleration" but I still believe femr's data is good enough to make that tentative statement plausible. Am I right? KISS
 
I also get from this that there is some rumbling and shaking a few seconds prior to collapse. Are some people saying that is caused by camera shaking or some other kind of error?
I personally have no doubt about the one a few seconds before because its magnitude is very clearly above the noise level. I only have problems about the one minutes before, and its origin if present. I think there's consensus that there was movement a few seconds prior to the collapse.


I continue to say "possible greater than freefall acceleration" but I still believe femr's data is good enough to make that tentative statement plausible. Am I right? KISS
In my case I still don't know. I'd keep the "possible" in there. Can't state it any more simply.
 
Last edited:
This is all way way out of my patience limitations. My glace at this thread (and what I quote in public) is that femr has shown slightly greater than freefal acceleration of a single point on the north perimeter wall during a total of about one second of the 2.25 seconds NIST talks about. Does anyone disagree with that?

I also get from this that there is some rumbling and shaking a few seconds prior to collapse. Are some people saying that is caused by camera shaking or some other kind of error? My very nonscientific guess is that if there was building rumble, that could be evidence of internal structures falling apart prior to the external perimeter collapse.

I continue to say "possible greater than freefall acceleration" but I still believe femr's data is good enough to make that tentative statement plausible. Am I right? KISS
What about the other millions of points which can't be seen? What was the overall system acceleration? Do you like the music like filter effect? What it the real acceleration? No big problem, but the work was done googling up the methods, and then using methods without reason.

What is the goal? Do you know? Do you see anything wrong with what NIST did? Did your read the appropriate NIST sections? What is the difference? How does this relate to debunking 911 truth?

WTC 7 fell due to fire, study of one point past what NIST did is a waste of time. The movement prior to collapse is not unusual. It is not surprising WTC 7 was totaled by fire, after 343 fire fighter and paramedic killed, they decided not to risk anyone in WTC 7, and it collapsed. Not the first high-rise to be destroyed by fire; Totaled.

What is the purpose of turning a set of data into silly curves? Go ahead explain which filter they should use, or if they should stick with a simple linear model. What would you do?

Did the building move, it moved all day long.
 
Third time lucky eh :rolleyes: (How many times did you actually abandon your original post ? :) )

Frankly, I don't see why you say I'm switching datasets, especially given you've been doing what you criticize all the time, including the posting of Dan Rather static point location when the static points being discussed were from Cam #3. That's called double standards according to my dictionary.
LOL. You switch between complaining about early motion "claims" and acceleration "claims" so often, the occasional (admitted) brain fart is inevitable :rolleyes:

I didn't have that much time. Had to cut somewhere. I'm forced to be selective. You have a tendency to post walls of text. (So does Ozeco, and amazingly he complains about other doing the same...)
How very convenient for you :rolleyes:

Not paying attention?
Always paying attention.

1) The error I took for that statement was tfk's 80 ft/s², as would be obvious if you looked at it in context. I was not serious using it, as would be obvious if you looked at it in context. It was an example of why error analysis matters, as would be... you know the rest.
Oh, so you don't have any issue with acceleration plot error bands ? :rolleyes:

2) You have claimed over-g. Several times during the last years.
I have indeed.

Like when discussing the number of times the corner went over g.
ROFL. No that was tfk, with his ludicrous claims of passing over the "over-g" threshold a specific number of times, based upon his wildly (and clearly method based) smoothed curve oscillations :rolleyes:

3) I assume nothing.
I beg to differ.

Again, unsupported ≠ false. Tattoo that in your forehead so you see it in the mirror as a reminder.
Read the thread(s). That I've posted information you haven't managed to spot yet is not my problem, it's yours.

I provided you with some information you claimed hadn't been posted "not long ago", yet you are still clinging to the notion that the details you "believe" have not been discussed (many times) are "absent". Get a grip man.

+/- 3 ft/s2 ring any bells ? :rolleyes:

All of these were responded in the post, if you looked at it in context instead of sentence by sentence. You cut the trees and then claim there's no forest.
Again... So you don't have any issue with acceleration plot margins of error ? :rolleyes:

I think you must be hitting reply before reading the post, and then reading and replying as you read each sentence, instead of reading, understanding, and then hitting reply. Which makes you miss the context completely. It's the best explanation for your MO.
ROFL. You can try and run away from your assertions but let me repeat your own words...

"Your claim of over-g is not supported by your analysis because the error is too big to claim a specific acceleration."

No "context" going on there. An assertion you've made many times for varying and shifting reasons....unless you're retracting it ? :rolleyes:

Exactly where does "context" come in ? :rolleyes:

*waves to your edit box* Hi, femr's edit box!
Hi pgimeno's third abandoned reply over the last six hours :rolleyes:

Done (even if the answers were already given in the context that you destroyed). My assertion is supported. It was from the beginning.
No, it's not. So are you personally happy with acceleration plots as provided ? :rolleyes:

My only problem with it is that currently it has the status of bare assertion unsupported by the evidence you've provided so far.
ROFL. What happened to your "context" based dismissals ? :rolleyes:

(Managed to spot me hinting that error estimation on acceleration has already been provided yet ? (perhaps not, early in the post I guess) :rolleyes:

They don't claim it's over g.
Quite correct. It is. WHY don't they say what it is ? Why do they arbitrarily say "AT" ? Why is it okay for NIST to perform zero error analysis, come out with results that indicate over-g then state "AT" g ? :rolleyes:

They clearly assume that any over g behavior is due to measuring error. Like the one at the 4s mark.
Assumption is the mother of all...

Defending NIST ? Nice :rolleyes:

Again, exactly why do you think over-g is an issue ?

Something along these lines might succeed in supporting the claim. As stated, that alone doesn't. As it stands, the claim remains unsupported.
No, those claims are utterly specific. You can quibble about magnitude error in the data, but the claims you are claiming are unsupported are "just not". :rolleyes:

At the time I posted that, you hadn't provided the static point location information.
ROFL. I provided that information in 2010 :rolleyes:

I still haven't evaluated the consequences because you keep diverting my scarce time into defending myself from your strawman attacks.
I'm not making strawman attacks. I'm showing your clear lack of understanding about the effect of perspective. I'm providing you with information you clearly don't understand. I suggested you should stop, but your ego has clearly kicked in. Bad idea. ROFL.

You might have a valid point here
No, I, unlike you, know what I'm talking about. You, unlike me, are talking utter faeces.

My "point" is absolute.

You can ask me what I'm talking about if you don't understand you know :rolleyes:

now that you've provided that information
The image I provided you with included two static trace regions. I already provided you with the relevant information. That your knowledge is lacking is not my problem :rolleyes:

I'll examine that when I have time.
If you don't have an instant understanding you REALLY shouldn't be making the inept assertions you have been diggin your own hole with.

Again, would you like me to explain ?

What good is you examining anything going to do ? You are showing repeatedly you "don't know your posterior from your elbow" in this arena :rolleyes:

Yes, including me, which doesn't mean it occurred. And before you jump the gun again, see above about what you should be tattooing in your forehead.
See later in my post where I provide you with resultant error analysis estimation :rolleyes:

I did. Your complaints were based on points having a huge difference in distance to the camera.
ROFL. No, they were not. They were based upon you dismissing perspective correction when tracing motion of the NW corner over a 200 pixel range, yet asserting that (hint) amplification of camera motion with differing camera-object distance, with a magnitude of 0.02 pixels was somehow more significant. :rolleyes:

ROFL. "DL". "Irony explode", ...

Let's go over this again
Again ? When did you dig this hole before ? :rolleyes:

huge difference in distance = huge consequences
Not necessarily. All manner of factors come in to play. Rewind to me informing you about the extra information you could obtain by tracing two static regions with differing distances from the camera ?

(A massive hint)

Exactly how large do you think the amplification of motion obtained from static region traces nearer to the camera is going to be further from the camera, in the specific case of the (surely) cam#3 footage ? [rolleyes]

(Another massive hint... if you trace two "distances" you can see the difference) [rolleyes]

negligible difference in distance = negligible consequences
Utterly viewpoint determined [rolleyes]

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.
ROFL. Fill that massively deep hole in quick son. I've already expressed (years ago) that the general profile shape is similar between cam#3 and Dan Rather datasets, however, YOUR "issue" is about magnitude...at which point the skewing caused by perspective effects becomes particularly significant [rolleyes]

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.
Say wha' ? :confused:

Maybe I should have said perspective distortion (which needs a corresponding correction) depends on distance to the camera. You usually have no problems with using the terms the way you want them to mean. Now having problems if someone else does the same? Dude, you're full of double standards.
ROFL. You clealry have no understanding of the effect of perspective upon trace data extracted from video. Can't resist (as it's been necessary for pretty much every other para...) [rolleyes]

Perspective correction methods use sophisticated techniques to find out the 3D location of the data
Deary me. Perspective correction from a known camera location is pretty trivial, and certainly doesn;t resolve position in full three dimensions [rolleyes]

in order to determine its distance to the camera from a given viewpointp
Distance ? Distance ? :) ROFL. Hows about "orientation" and "field of view" and ... [rolleyes]

and compensate for that by undoing the projection the camera has done in order to obtain a purely linear relationship, and undoing it depends exclusively on distance to the camera.
ROFL. Oh. My. Days. [rolleyes]

In the end it all boils down to that.
Just walk away [rolleyes]

The 3D to 2D transformation done by a camera (or our eye) is given by...
:) I'll wade through your "verbage" when I stop laughing. Can hardly see right now [rolleyes]

The buildings you chose as static points may well be about half the way towards WTC7
BuildingS ?! ROFL. No. One set close, one set further away [rolleyes]

that makes a 2x difference in perspective correction
ROFL. You're not even talking about perspective correction, and you don't even realise it. LOL.

Big deal, compared to the difference between distance to the top and distance to the bottom of the NW corner, which for the purpose of watching the shape of the acceleration plot, is negligible.
ROFL. #must remind myself to repost information about inter-window spacing [rolleyes]

You implicitly acknowledged that importance when you pointed to the Dan Rather static point location, saying that the building chosen was at a similar distance to the camera as WTC7 was.
The closer the better, absolutely ! What is hilarious is your inept extrapolation [rolleyes]

Therefore, I have not switched my standards whatsoever, quod erat demostrandum.
How many [rolleyes]'s does it take to sink in [rolleyes]

What? You posted some variance information, which is incomplete without the propagation of error in the measurements. And you said this yourself
:rolleyes: In the post you are responding to I provided you with error estimation for acceleration [rolleyes] (in other words propogation of error from displacement/time to acceleration/time [rolleyes])

You claimed that variance (and scaling metric) information didn't exist. Seems you were quite wrong. LOL. [rolleyes]

So yes, I'm sure about that, in context. NOW you have provided a bit of error information
ROFL. Information provided long, long ago son. [rolleyes]

taken from the symmetric difference method, the consequences of which I still have to analyze together with all the pending stuff.
Analyze ? [rolleyes] Given the "assuredness" of your "prose" you shouldn't have to blink twice [rolleyes]

For some reason I suspect it doesn't apply to any acceleration graph you've provided that you've used to claim over-g.
You can suspect what you please. You can flail about ALL OVER THE PLACE as much as you please. Given your recent "performance", forgive me if I point and laugh more than try and continue to help you [rolleyes]

What you have done so far has been to take my quotations out of context
Nope [rolleyes]

and building strawmen out of the decontextualized version of them.
Nope [rolleyes] Great nice longey wordy tha' [laughs big smiley]

I could only identify very few valid objections you've made
Would you like a list ? [rolleyes]

which I can't address because the amount of bullcrap, diversions, misinterpretations and strawmen in the rest takes all of my time.
What can I say, but...

:dl:

Dear. Oh dear. Oh dear. Oh dear.

I'm afraid the more discerning readers will see your sentence splitting and decontextualization and see through the way you manipulate it in order to make it seem you're right.
You just believe whatever it is that you want to believe. Great job sir.


Never had to place so many :rolleyes:'s in one post. [rolleyes]

(And had to change loads of them to text as the JREF forum doesn't allow 57 "images". LOL)
 
femr2 said:
he pertinent point being that it's data from 2010, in response to your silly assertion... "You are now focusing on the fact that I have not yet evaluated your newly provided data".
No, that's not the pertinent point
ROFL. Of course it's the pertinent point. It's not newly provided data. It's from 2010.

it's just wishful thinking
ROFL. No, it's fact.

you would like it to be the pertinent point in order to make you seem correct
ROFL. No, I am correct. The information you branded "newly provided" was provided in 2010.

It would be the pertinent point if you had provided it in a thread I was reading in 2010
ROFL. See above :rolleyes:

From my point of view it's newly provided data
ROFL. So ? Not my problem. Yours. Anyone told you personally that the world is not flat yet ? ROFL.

Again, don't expect me or anyone to pursue you to each forum and read each message you've posted.
I don't, however, when you make such inept statements as "You are now focusing on the fact that I have not yet evaluated your newly provided data"... when the data I have kindly (re)provided you with was provided to everyone, including you (whether you bothered to go find), in 2010.... I will point and laugh.
 
This is all way way out of my patience limitations.
Tell me about it :)

My glance at this thread (and what I quote in public) is that femr has shown slightly greater than freefal acceleration of a single point on the north perimeter wall during a total of about one second of the 2.25 seconds NIST talks about.
Pretty much, yes.

Does anyone disagree with that?
Uh oh... :)

I also get from this that there is some rumbling and shaking a few seconds prior to collapse.
There is traceable change in behaviour around 2 minutes prior to release.

Are some people saying that is caused by camera shaking or some other kind of error?
Yes, they are, despite clear data to the contrary.

My very nonscientific guess is that if there was building rumble, that could be evidence of internal structures falling apart prior to the external perimeter collapse.
Early motion is certainly a problem for all suggestion of "event->immediate descent" type scenarios.

I continue to say "possible greater than freefall acceleration" but I still believe femr's data is good enough to make that tentative statement plausible.
Thank you. I concur.

Am I right? KISS
I suggest you should be comfortable with that position, for the NW corner.
 
I continue to say "possible greater than freefall acceleration" but I still believe femr's data is good enough to make that tentative statement plausible. Am I right? KISS
Yes. The default hypothesis is "femr's data is good enough for some purposes." Which takes with it the implicit needs to:
1) Define your purpose;
2) Judge for yourself if the data is good enough for your purpose. (i.e. whoever "your" is.)

The hypothesis has not been falsified. Recent attempts at falsification have been countered by reasoned argument which the authors have chosen to not address.
 
Last edited:
This is all way way out of my patience limitations. My glace at this thread (and what I quote in public) is that femr has shown slightly greater than freefal acceleration of a single point on the north perimeter wall during a total of about one second of the 2.25 seconds NIST talks about. Does anyone disagree with that?
femr2 has suggested that. pgimeno rightly observes that femr2's deliberately obstructive video analysis has not provided sufficient evidence for a cautious skeptic to accept femr2's conclusion as proved, but my own independent analysis of femr2's data has led me to regard a brief period of over-g acceleration as likely for that particular point. On the basis of my own analysis, I think a half-second of over-g acceleration was more likely than "about one second."

I also get from this that there is some rumbling and shaking a few seconds prior to collapse. Are some people saying that is caused by camera shaking or some other kind of error? My very nonscientific guess is that if there was building rumble, that could be evidence of internal structures falling apart prior to the external perimeter collapse.
Rumbling and shaking before collapse seems likely on several grounds, but I don't think anyone has yet performed an analysis that would be adequate to rule out alternative explanations such as camera shake or other errors. If someone really cared about this, he/she would perform an analysis that takes both videos into account. (femr2 continues to argue against doing this, but femr2's argument is technobabble.)

I continue to say "possible greater than freefall acceleration" but I still believe femr's data is good enough to make that tentative statement plausible. Am I right? KISS
Yes. His video analysis hasn't been good enough, but I think his data are.
 
femr2's deliberately obstructive video analysis
Still frustrated I haven't given you exact details for replication of one out of many, many graphs eh ? :D

Ah, never mind eh. If it makes you feel better to make such incredibly thin proclamations (regardless of the great lengths I have gone to within this thread to provide endless detail, instruct those unfamiliar with the techniques in use, provide multiple datasets for others to use as they wish...etc...etc) then you just go ahead. Top job.

my own independent analysis of femr2's data has led me to regard a brief period of over-g acceleration as likely for that particular point
Where might pgimeno find your posts containing the error analysis he will require before accepting your assertion ? ;)

Could you kindly provide links please ?

I think a half-second of over-g acceleration was more likely than "about one second."
I have no great issue with that assertion, but would again request you provide links to where within this discussion you provided your "error analysis". Thanks.

Rumbling and shaking before collapse seems likely on several grounds, but I don't think anyone has yet performed an analysis that would be adequate to rule out alternative explanations such as camera shake or other errors.
We'll get to my data, I'm sure, however, have you actually read NCStar 1-9 Vol2 Appendix C ?

If someone really cared about this, he/she would perform an analysis that takes both videos into account.
Dan Rather viewpoint clip has, what, about 11s prior to release.

My "early motion" assertions relate to motion beginning 2 minutes prior to release.

Which "both videos" are you referring to ?

If you mean Cam#3 and Dan Rather...there's a wee bit of a problem... ;)

femr2 continues to argue against doing this
"This" ? What is "this" ?

I certainly continue to argue against using trace data from multiple viewpoints in order to construct fully three dimensional data coordinates for each sample, sure, for rather sensible and practical reasons given.

If you mean some other "this", as I suspect you do, then either stop massaging scope, or be particularly clear exactly what you mean by "this". Thanks :)
 
incorrect.

Actually, I think he's got this about right. The timing of "early motion". pretty much rules out anything outside of progressive failure (such as NIST describes). He would have easily detected any high energy charges. Although he doesn't like being clear, I think this is what he means here.

Early motion is certainly a problem for all suggestion of "event->immediate descent" type scenarios.
 
Actually, I think he's got this about right. The timing of "early motion". pretty much rules out anything outside of progressive failure (such as NIST describes). He would have easily detected any high energy charges. Although he doesn't like being clear, I think this is what he means here.

Oh I know... What I don't get is what the point is, beyond a childish 'i rule':cool:
 
femr2's deliberately obstructive video analysis
Still frustrated I haven't given you exact details for replication of one out of many, many graphs eh ? :D
Amused would be more accurate than frustrated, especially as you appear to think you were being clever.

my own independent analysis of femr2's data has led me to regard a brief period of over-g acceleration as likely for that particular point
Where might pgimeno find your posts containing the error analysis he will require before accepting your assertion ? ;)

Could you kindly provide links please ?
I'm not asking pgimeno to accept my assertion. I was trying to answer chrismohr's questions.

Parts 3 and 4 of my occasional series on Savitzky-Golay smoothing will explain my reasoning. I expect those posts will be of greater interest to pgimeno than to chrismohr.

Rumbling and shaking before collapse seems likely on several grounds, but I don't think anyone has yet performed an analysis that would be adequate to rule out alternative explanations such as camera shake or other errors.
We'll get to my data, I'm sure, however, have you actually read NCStar 1-9 Vol2 Appendix C ?
Yes, but I forgot about that. I stand corrected: NIST did perform that analysis, concluding motion did occur within the last few seconds before collapse.

That appendix also says slow changes occurring over intervals of 20 seconds or more were "likely due to movement of the camera." I take that to mean NIST was unable to rule out camera shake or other errors as possible explanations of the apparent earlier motion femr2 says he has been discussing:

My "early motion" assertions relate to motion beginning 2 minutes prior to release.
 
Thanks all. One other thing... in my YouTube video #18, freefall of Building 7, I use NIST data. This is more gross than measurements of femr and others but even this shows three places where greater than freefall occurs. Seems like pretty much everyone who has done this kind of careful analysis shows very slightly >g acceleration at some point or another.

OK thanks for simplifying for me. Now you can all go back to hashing out the minutiae!
 
Amused would be more accurate than frustrated
The not so gradual change in your posting behaviour would suggest otherwise, but you stick with that. Cool.

especially as you appear to think you were being clever
Clever ? The information I have gained about member intent within these "discussions" has been incredibly informative, by the simple act of withholding a couple of details about one graph, years ago.

Endless personal attacks, endless hilarious assertions about what the graph does and does not show, etc. tfk's defiant proclamations about inviolate laws were very amusing. Your multi-part derision laden lectures about savitzky golay similarly so.

If your focus was not "femr2" you'd have ignored my derivative plots years ago and discussed your own, as you have recently begun to do...now that (in order to try and score some kind of point) you have written your own SG implementation. It's gratifying, I suppose, that you have been driven by femr2 into personal exploration of Savitzky Golay, and may it stand you in good stead in the future (though if I were you I'd try and "wish away" much of the pblic discussion you've had with me surrounding such) but I digress...

I'm not asking pgimeno to accept my assertion.
Ah, I see, you've not actually "published" your analysis. Okay.

I was trying to answer chrismohr's questions.
But of course.

Parts 3 and 4 of my occasional series on Savitzky-Golay smoothing will explain my reasoning.
Splendid.

I expect those posts will be of greater interest to pgimeno than to chrismohr.
I would imagine you are correct.

Yes, but I forgot about that. I stand corrected: NIST did perform that analysis, concluding motion did occur within the last few seconds before collapse.
They did indeed, from around 13s prior to "start of main collapse".

That appendix also says slow changes occurring over intervals of 20 seconds or more were "likely due to movement of the camera."
They said "there are major changes in the location of the marker point that occurred over long time intervals of 20 s or greater. These changes were likely due to movement of the camera."...

(bolding mine)

However, NIST did not bother to extract camera motion from their data...and their extremely limited moire method would preclude doing so anyway.

I have extracted camera motion. Extended motion of the building is revealed.

Also, their early data can be shown to highlight issues with the moire method, which I shall discuss separately if necessary.

I take that to mean NIST was unable to rule out camera shake or other errors as possible explanations of the apparent earlier motion femr2 says he has been discussing
They didn't try to. They simply made the assertion. They were wrong.
 
You may have missed a couple of points in your previous response, accidentally of course...

I don't think anyone has yet performed an analysis that would be adequate to rule out alternative explanations such as camera shake or other errors.
As you have now discovered/remembered NCStar 1-9 Vol 2 Appendix C, which has been mentioned/cited within this thread ... numerous times in recent discussion , it strikes me that you have probably not discovered/remembered the details I have provided to highlight early motion.

Would you like another repost ? (I don't like the practice of linking to prior posts in isolation (like that above). It has the negative effect (that of course I'm sure you would not intend) of requiring those following the links to click again to see the full thread and responses to posts. Most of the posts you link to have rather damning responses immediately afterwards (which I'm sure you'll reject), yet I imagine numerous lazy folk don't see that, which leaves flawed assertions appearing to hold weight when they simply don't)

If someone really cared about this, he/she would perform an analysis that takes both videos into account.
Dan Rather viewpoint clip has, what, about 11s prior to release.

My "early motion" assertions relate to motion beginning 2 minutes prior to release.

Which "both videos" are you referring to ?

If you mean Cam#3 and Dan Rather...there's a wee bit of a problem... ;)

Again, which "both videos" are you referring to ?

(femr2 continues to argue against doing this, but femr2's argument is technobabble.)
"This" ? What is "this" ?

I certainly continue to argue against using trace data from multiple viewpoints in order to construct fully three dimensional data coordinates for each sample, sure, for rather sensible and practical reasons given.
Again, what "this" ?

ETA: Oh, and my response to W.D.Clinger's suggestion of "technobabble". Note how I have chosen NOT to link in such a way that surrounding posts/responses are hidden.
 
Last edited:
July, 2012 quote by Oystein belongs in this thread:

smoothing is manipulation, is data compression. Smoothing of derivative of smoothed first derivative is not analysis, it's nonsense.

With all six smoothing algorithms we still don't know what remains as signal once a real error analysis has been done.

Yes, chances are excellent that some features apparent in the smoothed stuff are real. Just don't put numbers to them.

Smoothing is not analysis. A smoothed curve without theory explains nothing.
 
July, 2012 quote by Oystein belongs in this thread:

smoothing is manipulation, is data compression. Smoothing of derivative of smoothed first derivative is not analysis, it's nonsense.

With all six smoothing algorithms we still don't know what remains as signal once a real error analysis has been done.

Yes, chances are excellent that some features apparent in the smoothed stuff are real. Just don't put numbers to them.

Smoothing is not analysis. A smoothed curve without theory explains nothing.
Yes, that's a nice summary. Tracing back through the sequence of links is also informative.

I have to disagree with the second sentence of Oystein's first paragraph. Smoothing a derivative of a smoothed first derivative is seldom best practice, but it could be part of an analysis. On the other hand, describing the result of that iterated smoothing as a Savitzky-Golay acceleration curve was dishonest and/or incompetent.

I had given femr2 the benefit of the doubt by assuming the latter. Now that femr2 is openly claiming credit for having made it hard for anyone to understand how he actually came up with his alleged Savitzky-Golay acceleration curve, I have revised my opinion.
 
Third time lucky eh :rolleyes: (How many times did you actually abandon your original post ? :) )
I don't know what you mean. If you are referring to my three editions, that's really silly. Here's the edit history:

First edition:

- Therefore, no switching of standards whatsoever, quod erat demostrandum.
+ Therefore, I have not switched my standards whatsoever, quod erat demostrandum.
​

Second edition:

- where x' and y' are the projected point's coordinates, k1 and k2 are constants that depend on the FOV and aspect ratio, x and y are the horizontal and vertical coordinates of the point to project, transformed to the camera's reference frame, and z is the depth, or in other words, distance from the camera. There's a slight difference between both concepts (distance and depth) in that the distance is sphere-shaped while the depth is plane-shaped, but at the usual distances, the radius of that sphere is so big and the region to consider so small in comparison, that the difference is negligible, therefore using the term "distance from the camera" as an approximation to refer to depth is valid.
+ where x' and y' are the projected point's coordinates (the pixels of the image), k1 and k2 are constants that depend on the FOV and aspect ratio, x and y are the horizontal and vertical coordinates of the point to project, transformed to the camera's reference frame, and z is the depth, or in other words, distance from the camera. There's a slight difference between both concepts (distance and depth) in that the distance is sphere-shaped while the depth is plane-shaped, but at the usual distances, the radius of that sphere is so big and the region to consider so small in comparison, that the difference is negligible, therefore using the term "distance from the camera" as an approximation to refer to depth is valid.
​

Third edition:

- where x' and y' are the projected point's coordinates (the pixels of the image), k1 and k2 are constants that depend on the FOV and aspect ratio, x and y are the horizontal and vertical coordinates of the point to project, transformed to the camera's reference frame, and z is the depth, or in other words, distance from the camera. There's a slight difference between both concepts (distance and depth) in that the distance is sphere-shaped while the depth is plane-shaped, but at the usual distances, the radius of that sphere is so big and the region to consider so small in comparison, that the difference is negligible, therefore using the term "distance from the camera" as an approximation to refer to depth is valid.
+ where x' and y' are the projected point's coordinates (the pixel coordinates of the image), k1 and k2 are constants that depend on the FOV and aspect ratio, x and y are the horizontal and vertical coordinates of the point to project, transformed to the camera's reference frame, and z is the depth, or in other words, distance from the camera. There's a slight difference between both concepts (distance and depth) in that the distance is sphere-shaped while the depth is plane-shaped, but at the usual distances, the radius of that sphere is so big and the region to consider so small in comparison, that the difference is negligible, therefore using the term "distance from the camera" as an approximation to refer to depth is valid.
​

The first edition was for clarity on who the subject of the corrected sentence was. The subsequent editions added or refined a minor explanatory detail. Nothing in these was worth any marking up or even adding an edit comment in my opinion. But if that counts as "abandon my original post" to you...


ROFL. You can try and run away from your assertions but let me repeat your own words...

"Your claim of over-g is not supported by your analysis because the error is too big to claim a specific acceleration."

No "context" going on there. An assertion you've made many times for varying and shifting reasons....unless you're retracting it ? :rolleyes:

Exactly where does "context" come in ? :rolleyes:
No, I have NOT made that assertion many times. That was the first time I made it and I was not serious, as I explained later.

Where does "context" come in? Are you serious? Have you read that sentence in its context?

Here it is:

tfk did a wonderful error analysis (yeah, right) and cam[e] up with (iirc) +/- 80 ft/s^2.

Just apply that and you'll be fine.
I'll do [apply that magnitude from the error analysis you suggest me to apply]:

Your claim of over-g is not supported by your analysis because the error is too big to claim a specific acceleration.

Thanks. It was useful.
In maroon, the references to the error analysis itself, which obviously refers to the one done by tfk. In olive, the references to the quantification of the error within that analysis. I've added an explanatory comment in [ ] to hold your hand walking you through the meaning of the sentence, seeming it's still unclear.

I hoped that it would be obvious that I wasn't serious, but I clarified later anyway:

1) The error I took for that statement was tfk's 80 ft/s², as would be obvious if you looked at it in context. I was not serious using it, as would be obvious if you looked at it in context. It was an example of why error analysis matters, as would be... you know the rest.

And you come now saying that that's my claim?

How many times do you fall in the same mistake?

Unsupported ≠ false

Here we go once again: I have not claimed in past that the error is too big. I have claimed that you haven't provided an error analysis that allows to judge whether the error is too big or not, in order to support your own claim of over g.

Once again, I don't say your claim is erroneous. I am saying it's unsupported.

I guess you didn't take my advice to tattoo that in your forehead. You really should take it.

And just in case you missed that too, that last thing was sarcasm. As should be obvious by the context, which you miss so often. Do you even know what "that" I am referring to in this same sentence? Probably by the time you read it it's already 5 quotes ago and you've forgotten. Never mind.


No, it's not. So are you personally happy with acceleration plots as provided ?
For what purpose? For looking at the overall shape, yes. For quantifying the actual acceleration of the building, no.


They don't claim it's over g.
Quite correct. It is.
Your burden of proof. You've failed.


Again, exactly why do you think over-g is an issue ?
Again, I don't think it's an issue. I just think it's not proved, that it's just a mere possibility that you've stated and nothing else. It seems W.D. has other thoughts. There's a difference with your claim, though: he qualifies that as likely, not as a fact. Still, I'll stay tuned in case he provides a justification.


ROFL. Fill that massively deep hole in quick son. I've already expressed (years ago) that the general profile shape is similar between cam#3 and Dan Rather datasets, however, YOUR "issue" is about magnitude...at which point the skewing caused by perspective effects becomes particularly significant [rolleyes]
Which means that you are in agreement with my assessment that cam #3 data is only good for looking at the shape. I don't see the problem nor do I know what hole you mean. I also agree that that skewing might be significant for looking at the magnitude as it introduces a small bit of error as the corner descends.


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.
Say wha' ? :confused:
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

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.


ROFL. You clealry have no understanding of the effect of perspective upon trace data extracted from video. Can't resist (as it's been necessary for pretty much every other para...) [rolleyes]


Deary me. Perspective correction from a known camera location is pretty trivial, and certainly doesn;t resolve position in full three dimensions [rolleyes]


Distance ? Distance ? :) ROFL. Hows about "orientation" and "field of view" and ... [rolleyes]


ROFL. Oh. My. Days. [rolleyes]


Just walk away [rolleyes]


:) I'll wade through your "verbage" when I stop laughing. Can hardly see right now [rolleyes]
I stand by my words, despite all the handwaving above. A bit more and you'll fly.

Orientation does have an effect over depth, as it changes the plane of view. FOV has an influence on how good the approximation of distance as depth is. For the case in point, it's a good approximation for extracting the conclusions I drew.

If being picky, I should have said "depth of the observed point with respect to the focal point". But the focal point is rarely more than 5cm away from the camera.

I will retract this part though:

There's a slight difference between both concepts (distance and depth) in that the distance is sphere-shaped while the depth is plane-shaped, but at the usual distances, the radius of that sphere is so big and the region to consider so small in comparison, that the difference is negligible, therefore using the term "distance from the camera" as an approximation to refer to depth is valid.
That's wrong for the general case; as FOV opens the difference becomes more significant. It applies to this case, though.


Here's a question for you: what kind of problems would you have with the cam#3 video to determine the metrics if the projection was orthographic instead of perspective?


BuildingS ?! ROFL. No. One set close, one set further away [rolleyes]
Silly pickiness.


that makes a 2x difference in perspective correction

ROFL. You're not even talking about perspective correction, and you don't even realise it. LOL.
Fair enough. I meant perspective distortion in this case, not correction. I have a tendency to say correction when I mean distortion.


The closer the better, absolutely !
Why? To amplify the error in compensating for camera shake (in the form of displacement)? Or are you performing perspective correction (correctly used this time) on the static point data to reduce that effect, perchance?


You claimed that variance (and scaling metric) information didn't exist. Seems you were quite wrong. LOL. [rolleyes]
Another strawman. I didn't make such claim. I know about your variance analysis. I claimed that you didn't do a decent error analysis for any acceleration plots that allows you to infer over g from them. You said so. Now you say that "a conservative estimation based upon wide window symmetric difference (rather than wide window savitzky-golay) would tot up to something around the +/- 3 ft/s2". Can you show the symmetric difference acceleration graph you're talking about?
 
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