Merged Discussion of femr's video data analysis

Oh. My. Days :eek::(

Seriously though, the location of the observer is placed on the map and sightlines are drawn between him and various vertical surfaces in the frame.
The aircraft is not at the same vertical level.

The blue line representing the fuselage length and orienta[t]ion are the[n] placed on the map
I very much doubt you are accounting for perspective correctly when orienting your "map".

Plan View Examples (though not from the "final" setup)...

Top:


Front:


Left:


White - NIST
Green - Draft femr2

the proportions of the baseline will remain consistent
Your "sightlines" do not all stem from the same vertical level.

Not in dispute
Good.

Not in dispute
Good.

although your layout differs from the map and satellite imagery
Oh. My. Days.

Do you really think I'd have the bleedin' towers in the wrong place, oriented incorrectly and scaled incorrectly ? :jaw-dropp

Can you not think of another reason why your overlay might not match what you expect it to ?

Ye gads man. Hint: There's a difference between clearly indicated draft non-final camera location and camera location for the last rotoscope

Not in dispute.
Good.

I thought your position might have been wrong because I was initially aligning the camera and WTC which threw off the position of the intermediate buildings.
You were wrong to suggest my secondary building placement was incorrect. Fine.

I'm not changing the camera position and if you looked at the images of my sightlines you'd have seen that we're within a few metres of other.
You've changed camera position several times.

You now accept near to this location...
126578805.png

Fine.

In your virtual word the aircraft doesn't line up, not necessarily the case in the real world.
:dl:

3DSMax is rather good at rendering three dimensional environments.

The towers are in the right place.
The other buildings are in the right place.
The camera is in the right place.

The render matches almost perfectly with video taken from the camera location, pointed at the same place, with the secondary buildings in the right place.

The ONLY thing that's out of place is the NIST oriented aircraft.

:eye-poppi

Figured out how the tail of the aircraft lies within the visual edge of the building from that other camera angle yet?
Yes. A problem between your keyboard and your seat.

Figured out that the Carmen Taylor viewpoint is slightly different to the Hezarkhani viewpoint yet, and that the aircraft in the Carmen Taylor photo has not yet made contact with the building ?

Maybe you could cross check how your 3d world looks from other camera locations.
Done a long time ago.

Enough of this nonsense.
 
Last edited:
...The ONLY thing that's out of place is the NIST oriented aircraft....
I remember the story of the proud grandmother at the Military Passing Out parade for her grandson's graduating course.

Says Grandma:"Why is everyone including the band out of step with our Johnnie?"

:boggled:


No, it wasn't me. Both my grandmothers had left this world long before my Graduating Parade.
 
As I am, of course, a thoroughly generous chap (and I really don't want to see further nonsense on this specific point), perhaps this annotated plan view will help you to realise where you've been going wrong...

730776256.png


Says Grandma:"Why is everyone including the band out of step with our Johnnie?"
:)
 
Last edited:
You've changed camera position several times.

As have you. Even my frst shot was within 20m of my later position.

Figured out that the Carmen Taylor viewpoint is slightly different to the Hezarkhani viewpoint yet, and that the aircraft in the Carmen Taylor photo has not yet made contact with the building ?


I didn't really need to do much figuring there, they are obviously different and I've always done separate sight lines for each image.

Done a long time ago.

You might want to try a location that's more than 1/3 of a degree different.

Nice to know your WTC buildings are aligned 27° off north though.
 
As have you.
Not in the last 3 years :rolleyes:

I didn't really need to do much figuring there, they are obviously different and I've always done separate sight lines for each image.
They have slightly different camera viewpoints, and the aircraft in the Carmen Taylor photograph has not reached the tower :rolleyes:

You might want to try a location that's more than 1/3 of a degree different.
Say wha' ? :eek:

The link I provided you with points to these images...

982554226.gif


284533061.gif


Get a grip man.

Admit your errors, move on.
 
As I am, of course, a thoroughly generous chap (and I really don't want to see further nonsense on this specific point), perhaps this annotated plan view will help you to realise where you've been going wrong...

[qimg]http://femr2.ucoz.com/_ph/2/2/730776256.png[/qimg]


:)

Let's see, the nose and the tail are in different positions and the larger the impact angle the larger the difference is. Yep, that's pretty much what everything I've done has shown, that's why there's a different sight line for the front and back of the aircraft.

The link I provided you with points to these images...

Didn't point me anywhere except to a title page.
 
Let's see, the nose and the tail are in different positions and the larger the impact angle the larger the difference is. Yep, that's pretty much what everything I've done has shown, that's why there's a different sight line for the front and back of the aircraft.
Deary me.

The reason I provided you with the image, that it would appear you are incapable of realising, is to highlight the width of the nose-tail profile width relative to the width of the tower in a true plan view.


True plan view...
femr2 - 7.8% of tower width
NIST - 19.7% of tower width

YOUR plan view...
femr2 - 20.0% of tower width
NIST - 36.3% of tower width

YOUR revised plan view...
femr2 - 28.0% of tower width
NIST - 40.4% of tower width


Your translations from photo to plan view are clearly screwed.
 
Apply them to the trace data.
Good.

It was looking very much like you were planning on modifying (aka destroying) the image pixel data.

Not the simplest of tasks once you include the time variable, but hey...good luck.
 
Deary me.

The reason I provided you with the image, that it would appear you are incapable of realising, is to highlight the width of the nose-tail profile width relative to the width of the tower in a true plan view.


True plan view...
femr2 - 7.8% of tower width
NIST - 19.7% of tower width

YOUR plan view...
femr2 - 20.0% of tower width
NIST - 36.3% of tower width

YOUR revised plan view...
femr2 - 28.0% of tower width
NIST - 40.4% of tower width


Your translations from photo to plan view are clearly screwed.

My relative widths are always the same from a viewpoint perpendicular to the wall, 15.5% for NISTs and 4.5% for yours. Different viewing angles and the perspective correction of the sightlines will give different projected values but lines drawn out from the walls at 90° will always be the same. Since none of the camera people were directly in line with the buildings "True" plan view ratios will not be seen.
 
My relative widths are always the same from a viewpoint perpendicular to the wall, 15.5% for NISTs and 4.5% for yours.
They should be approximately:

femr2 - 7.8% of tower width
NIST - 19.7% of tower width

Different viewing angles
Yes, my bad. Forgot about the rotation. (See how easy that is :))

My point remains though...

Your translations from photo to plan sightline view are clearly screwed.

As I've made a trivial oversight amongst the tide of denial coming from you, I guess I'll have to show you why :( (Should have time this evening)
 
Sooo...

13646968.png


Notes:
  • The aircraft models on the plan view are correctly oriented.
  • The "sight lines" eminate from the stated camera location.
  • The correct footage (Hezarkhani) is used (rather than your erronious use of the Carmen Taylor photograph)
  • The frame at the point where the nose of the aircraft contacts the building is used (rather than your erronious use of a photograph before that point)
  • There is no treatment of aircraft vertical orientation (so the nose-tail "width" is slightly larger than it should be on the image overlay)
  • It's an awful bleedin' method far inferior to the correct three dimensional modeling I've performed.

So what do we end up with...

  • The femr2 orientation is clearly a far better match than the NIST orientation.
  • The NIST tail location matches where my rendering says it would be.
  • Your translations from photo to sightline view are clearly screwed.

938786605.gif

Good job, done well, even if I do say so myself :D
 
Last edited:
I'll explain my argument very clearly.

1. Femr2 claims the NW corner went over g.
Unless femr2's position/displacement data are wrong, he's almost certainly right about his over-g claim.

2. As a proof of that claim, he shows a graph where the NW corner went over g.
Yes, and he has consistently refused to describe the methods and parameters he used to construct that graph in enough detail for others to reproduce his acceleration curve. He prefers to make others guess.

3. For the graph to constitute a proof, the error in it must be less than ~6.7 ft/s² (the difference between g and the peak his graph shows). If the error is bigger than that, the claim can't be sustained.
4. He doesn't show the error in his graph. ETA: Indeed he shows a different graph differing in over 10 ft/s² at some points from that one.
5. Without an error estimation, the proof in (2) is invalid.
It isn't hard to show that femr2's alleged Savitzky-Golay acceleration curve contains errors greater than 6.7 ft/s². I don't think anyone has yet done that, because explaining such calculations to technically disinclined readers is a lot harder than doing the calculations, but pgimeno has done a good job of explaining why we should suspect that curve contains errors greater than 6.7 ft/s².

6. Therefore we don't know if the claim in (1) holds, because femr2 has given no valid proof. It can be true or false; we just don't know.
You can't know without doing your own independent calculations.

As it happens, however, femr2's claim is almost certainly correct. To know that, you'd have to do your own calculations, because femr2's curves of unknown provenance and poor accuracy are inadequate to support his claim.
 
femr2's curves of unknown provenance
Curves ?

There's only one acceleration curve you are unaware of the generation procedure for.

pgimeno is welcome to use any of...

1) The "1-pass Savitzky-Golay" curve, for which, as you know, full replication details have been provided...
621711303.png


2) The linear regression of a 2s section of the descent, for which full replication details have been provided...
305517728.png

...or a shorter section focussed upon period of maximum acceleration as indicated by my plots...
2723862.png


3) His own acceleration plot...
855646088.png


4) Your acceleration curves...
32653665.png
 
Last edited:
Sooo...

[qimg]http://femr2.ucoz.com/_ph/2/2/13646968.png[/qimg]

Notes:
  • The aircraft models on the plan view are correctly oriented.
  • The "sight lines" eminate from the stated camera location.
  • The correct footage (Hezarkhani) is used (rather than your erronious use of the Carmen Taylor photograph)
  • The frame at the point where the nose of the aircraft contacts the building is used (rather than your erronious use of a photograph before that point)
  • There is no treatment of aircraft vertical orientation (so the nose-tail "width" is slightly larger than it should be on the image overlay)
  • It's an awful bleedin' method far inferior to the correct three dimensional modeling I've performed.

So what do we end up with...

  • The femr2 orientation is clearly a far better match than the NIST orientation.
  • The NIST tail location matches where my rendering says it would be.
  • Your translations from photo to sightline view are clearly screwed.

[qimg]http://femr2.ucoz.com/_ph/6/938786605.gif[/qimg]
Good job, done well, even if I do say so myself :D

I have no idea how you got that orientation but it should be more like this,

orientation_zpsc1c65eda.jpg


The image always has to be perpendicular to the line of sight. You can think of it as standing a photograph up on the map with the lines of radiating from the observers position to objects on the photo. As you can see the NE corner of WTC 2 has no relationship to the orientation you used.

Using the Carmen Taylor photograph is not erroneous. Different observer locations are used for each photograph. Since I'm not trying to locate the impact point the fact that the aircraft hasn't impacted when the photograph was taken is largely irrelevant to the question of orientation.

You might also want to check your angles. If that's a god's eye view they seem a little high.

angles_zpsd2d311ba.jpg
 
Jeez...

I have no idea how you got that orientation
How I GOT it ?!

I didn't "get" it. It's simply unmodified satellite imagery with unmodified Hezarkhani frame. The rendered aircraft is simply rotated to match WTC2 South face (as the render is from pure plan view).

but it should be more like this
Ye gads man.

All I've done is project points from the camera to the South face of WTC 2. All I'm interested in is the relative position in that plane.

By doing so much of the need to consider perspective and three dimensional space is removed. The remaining perspective issue is the vertical orientation of the aircraft, which affects the point at which the tail would project into the South face of WTC2.

The image always has to be perpendicular to the line of sight.
Rather method dependant. You are attempting to generate three dimensional information and have zero depth cueing. No accounting for perspective outside of a single plane.

That is why your method errs (though without detail of exactly what you're doing with your map reorientation you could also be causing further issues)

the NE corner of WTC 2 has no relationship to the orientation you used
Of course not. It's in a different plane.

Using the Carmen Taylor photograph is not erroneous.
It is if you use the Hezarkhani camera location, or compare to my renders.

Different observer locations are used for each photograph.
Wonderful.

I'm not trying to locate the impact point
Then you have an aicraft isolated in it's own 3D space, with no reference points on the WTC2 South face surface :eek:

the fact that the aircraft hasn't impacted when the photograph was taken is largely irrelevant to the question of orientation
You're not "undoing" any of the perspective effects involved. You have an object in it's own 3D space.

You can orient your "photo" to match the plane of WTC2 South face if you like but as soon as you include objects nearer or further than that plane you're walking into deep water.

You might also want to check your angles. If that's a god's eye view they seem a little high.
Jeez...

The aircraft is oriented in THREE dimensions ! :rolleyes:

You can't just ignore one axis.

NIST used 13 degrees lateralrotation about the aircraft axis, not 12.

(Oh, and I suggest you find out the relationship between the aircraft nose and tail)


This discussion is nonsense.

I've shown you that the elements in my scene are correctly placed.

I've shown you that the render is a superb match to real-world imagery.

The ONLY thing out of place is the NIST oriented aircraft.

Unending nonsense with your inappropriate methods is going to change nothing.

My rotoscoping is correct.

My suggested aircraft orientation parameters are correct (within a very small margin of error)

The NIST orientation is shown to be clearly wrong by a wide margin.
 
Last edited:
Using the Carmen Taylor photograph is not erroneous.

It is if you use the Hezarkhani camera location, or compare to my renders.

Let me know if I ever do that then.


NIST used 13 degrees lateralrotation about the aircraft axis, not 12.

(Oh, and I suggest you find out the relationship between the aircraft nose and tail).


NIST used 13° yaw with 1° sideslip giving an aircraft orientation in that plane of 12° relative to a line perpendicular to the building face (also 6° down with 1° nose up pitch giving 5° orientation nose down.) I'm also well aware of the height difference between the tip of the radome and the APU exhaust. You can of course use the straight part of the fuselage as a reference line.


The aircraft is oriented in THREE dimensions !

You can't just ignore one axis.


Changing the orientation in the Y plane shouldn't change your angle in the X plane when viewed from above. Nose and tail positions will move but a line parallel to the fuselage datum line won't change angle.


ETA- Oh, just for completeness I should ask what your WTC building orientation is relative to north in your 3d environment, just in case the photo reference you used did have any obliquity errors.
 
Last edited:
Let me know if I ever do that then.
This entire discussion relates to you disputing (for no justifiable reason) my rendered rotoscope views, which clearly show the extent of the NIST aircraft orientation error.

NIST used 13° yaw with 1° sideslip
Sideslip has no relevence in a static view. Your use of 12 degrees is still wrong.

giving an aircraft orientation in that plane of 12° relative to a line perpendicular to the building face
Nope. Again, sideslip has no relevance in a static view.

also 6° down with 1° nose up pitch giving 5° orientation nose down
However, NIST state that they aligned trajectory with orientation, and no detail within the report indicates that was not followed through into simulation. You can use 5 degrees if you like. It's the lateral angle that's the main problem.

Changing the orientation in the Y plane shouldn't change your angle in the X plane when viewed from above.
Rotations are not about the X and Y plane :rolleyes:

They are about the aircraft axes, of course :rolleyes:

Remember the aircraft is also "banked" with a roll of ~37 degrees.

Nose and tail positions will move but a line parallel to the fuselage datum line won't change angle.
See above :rolleyes:

Oh, just for completeness I should ask what your WTC building orientation is relative to north in your 3d environment, just in case the photo reference you used did have any obliquity errors.
I'd have to dig them out of the piling system from 3 YEARS ago.

Accurately placed over satellite imagery, with verification via vector maps.

There is no reason to doubt placement.

For completeness, do you know if the 2 pools on the current google satellite imagery are in exactly the right place, oriented exactly as the towers were, and how do you define the exact external footprint of the towers with the pool structures ?

This discussion is still pointless.

Your primitive method is flawed. There's no way to account for off-plane perspective. Even I can't easily resolve that issue for you.

My renders have been re-verified, again. Top-notch.

The ONLY thing out of place is the NIST aircraft orientation.
 
Last edited:
For completeness, do you know if the 2 pools on the current google satellite imagery are in exactly the right place, oriented exactly as the towers were, and how do you define the exact external footprint of the towers with the pool structures ?

The pools appear to be in the same place with the same orientation as the original buildings but the pool area appears slightly smaller at around 58.5m. That's why I've always added an overlay.

I only ask about the orientation because the layout in your draft only really matches if the WTC buildings are 31° off north and then it matches very well with all buildings and observer locations matching. Obviously, if that is the orientaion in the final 3d environment it would effectively add 4° to any aircraft orientation with respect to the buildings which were, of course, 27° off north. This would put your figure much closer to the lower bounds of the NIST baseline orientation.
 
Sideslip has no relevence in a static view. Your use of 12 degrees is still wrong.


Nope. Again, sideslip has no relevance in a static view.

Aircraft orientation with respect to the direction of travel is relevant. If the aircraft was approaching the WTC at 13° but was yawed 13° to the left(i.e sideslipping to the right) it would appear to be 90° to the surface at the point of impact.

Therefore if you're trying to visually match the aircraft it needs to be oriented where it's pointing rather than where it's going.
 

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