Merged Discussion of femr's video data analysis

So, there are many hurdles cleared, but there are some pending. Thing is, I don't think I should be the one doing this, because I haven't made the claim. My view on that is similar to how proving works in mathematics: if I find a mistake in a theorem, a step that doesn't follow from the previous step, the theorem is not proved, and it doesn't correspond to me to correct the proof. Ozeco seems to think that "the error is small and the theorem is roughly proved". There is no such thing as a "roughly proved" theorem. It is proved or it isn't, yet he tries to throw the burden of proof on me to provide a proof showing that the magnitude of some error matters. femr2, on the other hand, has repeatedly asked me to show my proof that the theorem is false, when I haven't claimed it's false (if I did, I would indeed have the burden to present a proof). Now, is there a series of steps that can fix the broken link in the deduction chain and complete the proof? That's what we're working on.

I don't think the proper way to go about analyzing any claim involving measurements is from the point of view of a mathematical theorem. Real world measurements always carry some measure of uncertainty. There are sources of variation that will not be completely characterized in nearly any continuous measurement. The analysis should then proceed in a method similar to "roughly proved". In other words, there is a certain confidence interval, say 95%, that you operate from, and given that you can either accept or reject the null hypothesis on whether the data sufficiently differs from some set. That of course does not prevent you from searching for as many sources of variation as you can, which is as much as you can do to "fix the broken link".
 
I don't think the proper way to go about analyzing any claim involving measurements is from the point of view of a mathematical theorem. Real world measurements always carry some measure of uncertainty. There are sources of variation that will not be completely characterized in nearly any continuous measurement. The analysis should then proceed in a method similar to "roughly proved". In other words, there is a certain confidence interval, say 95%, that you operate from, and given that you can either accept or reject the null hypothesis on whether the data sufficiently differs from some set. That of course does not prevent you from searching for as many sources of variation as you can, which is as much as you can do to "fix the broken link".
Well, the problem is that we had no figures on "confidence interval" here. We were given the data and with no error estimation, the claim that the NW corner exceeded g acceleration. That's the "broken link". If you can't know whether the probability of it being wrong is under 5% or above 95%, then you can't claim that it's true.

ETA: The comparison with a theorem was a metaphor to explain my view on where the burden of proof lies. I am well aware of the differences.
 
Last edited:
Well, the problem is that we had no figures on "confidence interval" here. We were given the data and with no error estimation, the claim that the NW corner exceeded g acceleration. That's the "broken link". If you can't know whether the probability of it being wrong is under 5% or above 95%, then you can't claim that it's true.


As far as "confidence interval", I used the wrong term, I meant a confidence factor, that would be stated with the error figure or confidence interval. Typically 95% is the confidence factor used.

I agree on the error estimation point. The error estimation is needed in order to make any sort of statement as to whether the Null Hypothesis can be accepted or rejected.

In addition, what in essence is being done here is a "Measurement System Analysis". Femr's algorithm used in his acceleration determination is in essence part of his measurement system. The accuracy of the algorithm is thus another source of variation to be characterized.
 
I'll start here. Little time right now...

Good work.
Oh my many gods ! :jaw-dropp

Talk about double standards. Absolutely classic.

I provide a multi-page analysis, detailing multiple processes and methods by which accuracy is confirmed...

...which you instantly try to "reject"...

...yet you call a hotch-potch utterly useless "quick GIF" by one of your "preferred" members "good work".

ROFL.

:dl:n
femr2's estimation of Hezarkhani's camera position is definitely wrong.
As I've already told you, camera position is determined by:

a) Personal discussion with Carmen Taylor, who has confirmed Hezarkhani's approximate position.

b) Inclusion of foreground buildings to allow use of parallax effects to verify camera position....a factor your own (pointless) animated GIF shows quite effectively.

Your assertion simply shows that you are either:

a) utterly and woefully incompetent.

or

b) deciding to make false assertions based solely upon your desire to attempt to "prove femr2 wrong version the umpteenth".

Carmen Taylor's photograph has about the same width for the east wall as Hezarkhani's video, both notably different to ferm2's, proving that it's not a lens effect as femr2 said, but a perspective effect due to a different position of the camera
ROFL. Again, you are simply showing your incompetence I'm afraid.

Proves ?! My god man.

Greater levels of hypocricy and double standards I have not seen for some time, even here.
 
Last edited:
we can draw lines back to the camera position and see the relationship between the sizes of various objects in the frame
No, you can't.

Three dimensional world this, not two.

A quick and dirty check
I'm afraid it's quite wrong.

If I can be bothered, I may dig out the models and provide a true top-down view for your delectation...and the additional details you'd require to determine fuselage profile in glorious real-d...or have I already done so... ? :)
 
Last edited:
@femr2: Isn't it time to cut your losses with this rolling "Gish Gallop"?
That would undoubtedly have been a long time ago :)

Though I'd have to stress I've had little in the way of losses, other than wasting my time with folk who would argue with me about the sky ;)

And Mark's question starts to head in the direction of "quantification" which doesn't suit the tactics of "prove femr2 wrong version the umpteenth". (Neat phrase that one. ;))
But very apt more frequently than one would have thought possible. Ta.

pgimeno's tactic of limiting his acceptance of femr's analytical data to the requirement that femr must explain it to pgimeno's satisfaction. Plus a rolling "Gish Gallop" to make sure we all forget what the real issues are.
See "good job" assertion above. Funny as. And despicable at the same time of course.

A "funny as" moment that may arise "further down the line" is where you have to place the camera to force the NIST orientation to fit the real-world imagery...
 
Last edited:
My purpose is to make it usable.
For what purpose ? Acceleration ? If so, then the problem you have is that whilst the raw data is is certainly of higher accuracy than the Dan Rather data (for obvious reasons) after multi-camera lens diatortion and perspective correction (and the rest) the stacked-up errors are going to go through the roof...especially as you're ADDING Dan Rather data error into the mix in the process :rolleyes:

There's little point in having it there if it can't be used.
It has been used. Early motion.

I'm quite familiar with 3D graphics.
Clearly not, given your "good work" back-patting above.

It's not my own error. My interpretation of your words was valid. Within that interpretation I was right. Within your interpretation you were right.
My mistake, you're wrong ;)

My exact words
...whilst trimming context from mine. Good work :)
 
I have done it. Method and results will be published soon.
I do hope you're not going to try and foist a manual by hand lens correction upon anyone.

I do hope you're not going to make assumptions about the linearity of certain specific building features.

I'm working on the plane aircraft issue
Fine. A simpler problem.

You really need to retract/sort out your silly assertions above on that front rather quickly then.

Your end result will be that you too have identified seriour error with the NIST impact parameters.

I've already performed a lens correction analysis for Cam#3.
See above. Early motion. Hint.
 
Having issue with exact measurements used by NIST
Whilst comparisons are of course drawn, the true topic of this thread has little to do with NIST. But instead a near 3000 post "discussion" about the methods by which my data was produced.

One would have thought that if there were something seriously amiss it would have emerged within the last 3000 posts... :boggled:
 
I provide a multi-page analysis, detailing multiple processes and methods by which accuracy is confirmed...

...which you instantly try to "reject"...

...yet you call a hotch-potch utterly useless "quick GIF" by one of your "preferred" members "good work".
Yes, for adding a graphic that shows the elephant in the room in your analysis, namely the visible width of the east wall of WTC2.


As I've already told you, camera position is determined by:

a) Personal discussion with Carmen Taylor, who has confirmed Hezarkhani's approximate position.

b) Inclusion of foreground buildings to allow use of parallax effects to verify camera position....a factor your own (pointless) animated GIF shows quite effectively.
I know.


...whilst trimming context from mine. Good work :)
Can you show the context you claim I've trimmed, maybe, and explain? Or will you just handwave that as you often do, to just appear like you're throwing back at me the same criticism that I've previously thrown at you? It seems to me that the only context I've trimmed is where you claim that I have no idea of commonly used pixel aspect ratios, which is right (ETA: even though I did show that I factored that in), but that doesn't authorize you to throw a straw man on me claiming that an assertion that I didn't make was incorrect.


I do hope you're not going to try and foist a manual by hand lens correction upon anyone.
Wait until I provide it and judge then. But out of curiosity, what would be wrong with that?


You really need to retract/sort out your silly assertions above on that front rather quickly then.
Maybe my assertions aren't that silly. My analysis will tell. It will be open to public review in full.

Speaking of public review in full, can you provide the scene file you used, maybe?
 
Last edited:
Yes, for adding a graphic that shows the elephant in the room in your analysis, namely the visible width of the east wall of WTC2.
It appears you are so confused you can't tell that I'm talking about the erronious images from Reactor Drone :rolleyes:

You really can't understand why mapping an (incorrect) 2 dimensional measurement onto a photograph of a three dimensional environment is wrong ? :jaw-dropp (see "good work" references)

What were you saying about your understanding of perspective ?

In reference to your own image, I'm afraid you simply don't appear to have enough knowledge to accept what I've already told you. What you erroniously call a white elephant is in fact pretty trivial.

Yet you clearly don't understand why the conjunction of the two (or even the second alone) precludes significant camera position error, a basic and trivial fact. I suggest you investigoogle parallax.

Wait until I provide it and judge then.
Sounds like you have then. Ew.

But out of curiosity, what would be wrong with that?
I'll wait until you provide it and judge then ;)

Maybe my assertions aren't that silly. My analysis will tell. It will be open to public review in full.
Joy.

Speaking of public review in full, can you provide the scene file you used, maybe?
Perhaps, at a later date, after I critique your analysis.

As I said, if you do a good job of it you'll replicate my results, which will be splendid. Good luck.
 
Last edited:
Just a few quick points....

So, there are many hurdles cleared, but there are some pending. Thing is, I don't think I should be the one doing this, because I haven't made the claim. My view on that is similar to how proving works in mathematics: if I find a mistake in a theorem, a step that doesn't follow from the previous step, the theorem is not proved, and it doesn't correspond to me to correct the proof.

More or less echoing some others: I don't think proving a theorem is the right frame. Most normal science proceeds by people presenting their best empirical work so far, making more or less tentative inferences from that work, and also assessing the contributions and limitations of each other's empirical work.

I actually think it may still be possible to reduce the chest-thumping in this thread and have something more like normal science. It appears that the thread started over two years ago explicitly as an effort to discredit femr2's work. But it doesn't have to stay that way. It can look more like normal science. It has more potential for that than most of what happens in this subforum. I am still rooting for y'all.

I think it's OK to discuss multiple topics about femr2's video analysis in this thread, but it imposes special responsibilities in order to avoid seeming to leap (or "gallop") from criticism to criticism.
 
It appears you are so confused you can't tell that I'm talking about the erronious images from Reactor Drone :rolleyes:

You really can't understand why mapping an (incorrect) 2 dimensional measurement onto a photograph of a three dimensional environment is wrong ? :jaw-dropp (see "good work" references)

What were you saying about your understanding of perspective ?

While drawing sight lines onto an image certainly doesn't tell the whole story it does give useful information. For instance, I could never get the sightlines to quite work with your foreground buildings so I stitched together a map and checked their position and the camera position against yours. You may want to double check a few things. With the buildings in their correct location and the cameraman on land the sightlines worked out much better. I've placed the camera as far back as possible but he could have been closer along that line. With the in plane overlays all the buildings now line up much better with the apparent fuselage length of the NIST baseline orientation looking slightly large and Femr's looking slightly small. Since the camera is looking up at about 10° the apparent fuselage length will of course look a little smaller than the in plane view but since that's good for the NISTs and bad for your alignment I'm not sure why you're complaining about the lack of 3D.

Map overlay

WTC2impactNIST12_zpsd20520b8.jpg


WTC2impactFEMR3_zpsd5721295.jpg


Obviously there's a bit of wiggle room with the orientation, I'm just saying you might not want to be quite so adamant about your position on the aircraft orientation while you only have draft work on it and some fairly easily spotted errors.
 
I don't think the proper way to go about analyzing any claim involving measurements is from the point of view of a mathematical theorem....
That is the main point of misunderstanding revealed by these recent posts. I note that MarkLindeman has also identified that issue.

The technical issue under discussion - if we can remember what it is - measurement of noisy data using mathematical tools to make the data useful. (plus some issues of geometry if people want to separate them out as a distinct category.) Measurement in this setting is an analogue process. It is not binary "right/wrong" or "proven/unproven".

And that misunderstanding underpins other issues such as those in this portion of pgimeno's statement:
...Ozeco seems to think that "the error is small and the theorem is roughly proved". There is no such thing as a "roughly proved" theorem. It is proved or it isn't, yet he tries to throw the burden of proof on me to provide a proof showing that the magnitude of some error matters....
1) 'Ozeco seems to think that "the error is small..."' -- I have been explicitly clear that I regard femr's explanations as a prima facie statement of accuracy. Including all the evidence posted over many months which - summarised as the famous "Sixteen Points" all of which show an intrinsic high level of accuracy and greater accuracy than NIST. Those claims have been "on the table" for months if not years. pgimeno wants to start from zero base. He is entitled to ignore femr's claim and put forward his own. What he is not entitled to do is quote mine and nit pick details ad nauseam bereft of context - whether femr's context or his own. (And, BTW, such 'context denying nit picking' is one of the central features of a "Gish Gallop". :rolleyes:)

2) 'Ozeco seems to think that....the theorem is roughly proved.." -- No way would I make that claim - it is not a "proof of theorem" setting. Even if it was a "theorem" setting I wouldn't refer to a "theorem" as "roughly proved". Given that pgimeno seeks to invent those crazy ideas and impose them on me is it any wonder that I choose to withdraw from "debate".

3) 'There is no such thing as a "roughly proved" theorem.' -- A truism which I naturally accept. But the truth is not the deception. The deception is in how it is posted as a strawman criticism of me not related to what I have said but fabricated by pgimeno - see the previous two points.

4) [There are three or four more inaccuracies in the short bit I quoted. I won't bore people by explaining them. I'll take a rain check on the burden of proof confusion. :)]

Bottom line is that a lot of pgimeno's confusion is built on false foundation premises and an idiosyncratic style of discussion.

The misunderstanding of analogue measurement as a "theorem" is only one of those faulty foundations.

I suggest that the current topic under discussion is "femr's data and analyses are accurate enough to be useful". Call that "Version 1".

Naturally I would accept as "Version 2" the reversal of "femr's data and analyses are NOT accurate enough to be useful".

Provided we all discuss the same topic. Or, if we must switch topics in mid debate, we are clear as to which claim we are addressing and who is making that claim. (That should get us out of the "reversing the reversal of reversed burden of proof" nonsense. I'm not sure if I counted the levels of attempted reversal correctly :o - but the point is should be clear. :boggled:)

I think the status is that femr has made a claim. I accept it as a claim to "prima facie" standard - i.e. strong enough to warrant reasoned objective rebuttal. If anyone has a different understanding which they can express with equal or better clarity go for it. If the wish is to continue drifting through fog I will simple go back onto the sideline.

My thanks to LSSBB and MarkLindeman for identifying that one of the key issues
 
Last edited:
That is the main point of misunderstanding revealed by these recent posts. I note that MarkLindeman has also identified that issue.

The technical issue under discussion - if we can remember what it is - measurement of noisy data using mathematical tools to make the data useful. (plus some issues of geometry if people want to separate them out as a distinct category.) Measurement in this setting is an analogue process. It is not binary "right/wrong" or "proven/unproven".

And that misunderstanding underpins other issues such as those in this portion of pgimeno's statement:
1) 'Ozeco seems to think that "the error is small..."' -- I have been explicitly clear that I regard femr's explanations as a prima facie statement of accuracy. Including all the evidence posted over many months which - summarised as the famous "Sixteen Points" all of which show an intrinsic high level of accuracy and greater accuracy than NIST. Those claims have been "on the table" for months if not years. pgimeno wants to start from zero base. He is entitled to ignore femr's claim and put forward his own. What he is not entitled to do is quote mine and nit pick details ad nauseam bereft of context - whether femr's context or his own. (And, BTW, such 'context denying nit picking' is one of the central features of a "Gish Gallop". :rolleyes:)

2) 'Ozeco seems to think that....the theorem is roughly proved.." -- No way would I make that claim - it is not a "proof of theorem" setting. Even if it was a "theorem" setting I wouldn't refer to a "theorem" as "roughly proved". Given that pgimeno seeks to invent those crazy ideas and impose them on me is it any wonder that I choose to withdraw from "debate".

3) 'There is no such thing as a "roughly proved" theorem.' -- A truism which I naturally accept. But the truth is not the deception. The deception is in how it is posted as a strawman criticism of me not related to what I have said but fabricated by pgimeno - see the previous two points.

4) [There are three or four more inaccuracies in the short bit I quoted. I won't bore people by explaining them. I'll take a rain check on the burden of proof confusion. :)]

Bottom line is that a lot of pgimeno's confusion is built on false foundation premises and an idiosyncratic style of discussion.

The misunderstanding of analogue measurement as a "theorem" is only one of those faulty foundations.

I suggest that the current topic under discussion is "femr's data and analyses are accurate enough to be useful". Call that "Version 1".

Naturally I would accept as "Version 2" the reversal of "femr's data and analyses are NOT accurate enough to be useful".

Provided we all discuss the same topic. Or, if we must switch topics in mid debate, we are clear as to which claim we are addressing and who is making that claim. (That should get us out of the "reversing the reversal of reversed burden of proof" nonsense. I'm not sure if I counted the levels of attempted reversal correctly :o - but the point is should be clear. :boggled:)

I think the status is that femr has made a claim. I accept it as a claim to "prima facie" standard - i.e. strong enough to warrant reasoned objective rebuttal. If anyone has a different understanding which they can express with equal or better clarity go for it. If the wish is to continue drifting through fog I will simple go back onto the sideline.

My thanks to LSSBB and MarkLindeman for identifying that one of the key issues

Yet you won't say what that claim actually is.
 
You really can't understand why mapping an (incorrect) 2 dimensional measurement onto a photograph of a three dimensional environment is wrong ? :jaw-dropp (see "good work" references)

What were you saying about your understanding of perspective ?
The distance gives a good estimation of the error, and I didn't need to make any error evaluation to reach an informal conclusion that the error in the width of the east wall is bigger than that produced by perspective effects.

I hope to give a formal analysis if I can get past the equation problem I'm facing. I'm trying hard to find an analytical solution to an equation but I may end up needing to solve it numerically.


In reference to your own image, I'm afraid you simply don't appear to have enough knowledge to accept what I've already told you. What you erroniously call a white elephant is in fact pretty trivial.
What is your explanation for the differences in visual width of the east wall?


Yet you clearly don't understand why the conjunction of the two (or even the second alone) precludes significant camera position error, a basic and trivial fact. I suggest you investigoogle parallax.
I perfectly understand it and see your point. It can be the case that the parallax effect analysis is correct but the position of the camera is still wrong. Wait until you see my analysis and conclusions.


I'll wait until you provide it and judge then ;)
So you won't tell me what's wrong with that approach?

Can we at least agree that lens distortion is minimized around the center of the image?
 
The distance gives a good estimation of the error, and I didn't need to make any error evaluation to reach an informal conclusion that the error in the width of the east wall is bigger than that produced by perspective effects.
Whilst I've not finalised the study (as I'm very satisfied with the results I have) there's no reason for you to assume the early draft linked to was not progressed...

938786605.gif

(Image from 28-08-2009)

As I said, other viewpoints were also drafted as cross-checks, though they also are not included in the early draft.

I hope to give a formal analysis if I can get past the equation problem I'm facing. I'm trying hard to find an analytical solution to an equation but I may end up needing to solve it numerically.
Visual feedback via rotoscoping is rather handy.

What is your explanation for the differences in visual width of the east wall?
I already told you.

I perfectly understand it and see your point. It can be the case that the parallax effect analysis is correct but the position of the camera is still wrong.
Whilst, of course, numerous factors can introduce error, such as WTC placement, tertiary building placement, camera location (including elevation), aircraft placement, and even earth curvature...whether you personally "believe" my results or not is really not much of an issue to me. I've shown over a long period of time that accuracy of fine detail is way up there on my list of priorities and all elements in the study performed with significant attention to detail.

Unsurprisingly the time investment in producing the study (even the early draft) was not insignificant, and it is of course irritating to be confronted with basically unfounded criticism by folk who have not "been through the process" in any kind of detail, but instead "shoot from the hip" hoping something sticks.

Wait until you see my analysis and conclusions.
As I said, you should converge on my results as you increase accuracy.

So you won't tell me what's wrong with that approach?
Why would you ask ?

Parameters for lens correction for a wide range of hardware is available, for a reason, normally provided to 6dp.

Lens information for Cam#3 and Dan Rather footage cameras has not surfaced.

Can we at least agree that lens distortion is minimized around the center of the image?
No. Depends upon the lens. Not necessarily in the center.
 

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