Merged Discussion of femr's video data analysis

Threw together a slightly higher resolution map and ran the sightlines method agian on the Carmen Taylor and Michael Hezarkhani images. Camera location was again fairly close to Femr2's although it's hard to be precise since his grid only contains WTC 1 and 2 and the foreground buildings and needed some scaling and rotation to get a reasonable match. I did notice Femr2's WTC buildings are rotated about 2.5° clockwise compared to my map. The results were similar to the lo res check and again, nothing to suggest the 12° figure is too far off. Since Femr2 argued that the method is flawed when analysing 3d imagery (or at least 2d imagery of 3d space) I thought I'd test it out against a shot from much closer with a higher look up angle. Again the method results in pretty accuate matching of sightlines across the plane of the image so unless the method gives a similar degree of error regardless of the viewing angle I'm not seeing anything that would suggest an 8-9° error.

carmentaylorsightlinesNIST12_zpsb4f699b5.jpg


closeviewhighangle_zpsb96ac229.jpg

This is a valid sanity check.
Representing 3D space on a 2D surface has been known since the Renaissance.




stock-photo-10128521-albrecht-durer-s-method-of-perspective-drawing-circa-1530.jpg


A more accurate way of drawing the perspective lines is as shown on the sketch below. In school we learned how to draw 2 point perspectives in the following manner. Drawings are rough and not to scale.
The picture plane (PP2-blue) and station point can be any place.
A to A1 along the sight line. A1 to perspective, perpendicular to picture plane. And so on.

picture.php




You can take femr's plan sight lines and NIST photo, put the picture plane, say where you had it on the first post, and take it from there.
tumblr_maqaoxG0YZ1rr4yh8o1_500.jpg



2 point perspective FLWright and subsequent photo.


images
 
Last edited:
Threw together a slightly higher resolution map
Your "camera" location is clearly incorrect.

Not that you actually appear to have a camera at all.

There are a multitude of free 3D modelling implementations available.

I suggest you abandon your 2.1D "sight-lines" methods and try glorious three dimensions.

Camera location was again fairly close to Femr2's although it's hard to be precise since his grid only contains WTC 1 and 2 and the foreground buildings and needed some scaling and rotation to get a reasonable match.
Deary me. How about you start here...
image004.jpg


ETA: Or, as you folk simply choose to dismiss carefully researched information in preference to your own guesswork... http://www.internationalskeptics.com/forums/showthread.php?p=3099536#post3099536

Yawn.

I did notice Femr2's WTC buildings are rotated about 2.5° clockwise compared to my map.
I've already stated that Google building placement is not to be trusted. Placement must be performed with reference to satellite imagery...a long and tedious job.

The results were similar to the lo res check and again, nothing to suggest the 12° figure is too far off.
938786605.gif

Ahem. Red aircraft "far off" from blue aircraft.

Since Femr2 argued that the method is flawed
It is. Clearly.
 
Representing 3D space on a 2D surface has been known since the Renaissance.
Of course, however, could you point out where I have presented a representation of 3D space which is NOT on a 2D surface ? :rolleyes:

I'm afraid Reactor Drone's results are flawed, clearly.
 
I'm afraid Reactor Drone's results are flawed, clearly.

Let's see your preferred 3.5° angle from that camera position,

femr35degree_zpsc1160b70.jpg


So, what 3d perspective effect gets the tail of the aircraft within the visual edge of the building?

Maybe, if you're the only person getting a particular result, it isn't because you're the only person who's got it right. In those cases it's far more likely that you're the one getting it wrong. :p
 
While I had a few minutes spare today I thought I'd have a quick look at Femr's aircraft impact angle. Using the camera and building position information Femr has provided we can draw lines back to the camera position and see the relationship between the sizes of various objects in the frame. By adding a line representing the fuselage length of the boeing 767 you can easily compare the visual length at the differnt impact angles, 3.5° from perpendicular to the face of the building for Femr and 12° for the NIST baseline. Since the downward angle of the aircraft only changes the length of the fuselage in the horizontal plane by about 6" I ignored that parameter for this quick check.

So here's the NIST baseline angle,
WTC2impactNist_zps9e46a14c.jpg


And Femr2's,

WTCimpactfemr_zpscefdd7c9.jpg


A quick and dirty check but I know which one looks closer to my eye.

You can project the white sight lines up to the picture plane (border between picture and drawing) and then drop perpendicular (red lines), enlarging or reducing as needed, matching points on both. You can then measure the plane angle on the drawing.
 
Last edited:
What algorithm does GIMP 2.6 use to apply lens distortion correction ?
From the code:

Code:
[i]Input parameters are:
  vals.centre_x
  vals.centre_y
  vals.square_a (barrel/pincushion)
  vals.quad_a (border)
  vals.scale_a (which I don't use)
  vals.brighten (which I don't use)
Precalculation goes like this[/i]

  calc_vals.normallise_radius_sq =
    4.0 / (width * width + height * height);
  calc_vals.centre_x = width * (100.0 + vals.centre_x) / 200.0;
  calc_vals.centre_y = height * (100.0 + vals.centre_y) / 200.0;
  calc_vals.mult_sq = vals.square_a / 200.0;
  calc_vals.mult_qd = vals.quad_a / 200.0;
  calc_vals.rescale = pow(2.0, - vals.scale_a / 100.0);
  calc_vals.brighten = - vals.brighten / 10.0;

[i]Actual calculation goes like this:[/i]

  off_x = i - calc_vals.centre_x;
  off_y = j - calc_vals.centre_y;
  radius_sq = (off_x * off_x) + (off_y * off_y);

  radius_sq *= calc_vals.normallise_radius_sq;

  radius_mult = radius_sq * calc_vals.mult_sq + radius_sq * radius_sq *
    calc_vals.mult_qd;
  *mag = radius_mult;
  radius_mult = calc_vals.rescale * (1.0 + radius_mult);

  *x = calc_vals.centre_x + radius_mult * off_x;
  *y = calc_vals.centre_y + radius_mult * off_y;

Therefore it uses the biquadratic model: r' = a·r4 + b·r2 + c. Saves the programmer from calculating a square root per pixel.

But why is that important? I already said I'm going to dismiss that one anyway. I want to try with a cubic one. First I'll try with LensFun but it doesn't seem to have center correction. Depending on the result I may try with Hugin, which doesn't have a Gimp plugin so it will be harder to do trial and error with it.
 
So, what 3d perspective effect gets the tail of the aircraft within the visual edge of the building?
Oh my days.

This is ridiculous.

By whatever means you're placing your blue lines upon your partial map fragment...it's nonsense.

Get a grip man.

a) The position where the docked ferry from which Hezarkhani filmed the footage is known. I've even provided you with a link within this forum from folk who went to the trouble of going to NY and taking photos to clarify.
b) The position of WTC 1 & 2 is rather well known.
c) Placement of Downtown Athletic Club and Whitehall Building was performed via rather tedious overlay onto 8192*8192 px satellite imagery, in conjunction with vector maps. As I've told you, don't trust Google 3D building placement.
d) Orientation of the NIST aircraft (red) is exactly as specified. It doesn't match. To make it match, regardless of intermediate buildings, you have to move the camera a vast distance. Approximate camera location is FIXED. It's not up to YOU.
e) Intermediate buildings allow refinement of camera location around the specific ferry dock (which was refined following my draft).

And would you believe it...
938786605.gif

...everything lines up, except the NIST aircraft.

Maybe, if you're the only person getting a particular result, it isn't because you're the only person who's got it right. In those cases it's far more likely that you're the one getting it wrong. :p
Your method is fatally flawed, either through you omitting additional required factors, using invalid map information or simply your own incompetence. Whichever of those it is, you're clearly way off.

I've been patient with you, but enough.
 
Last edited:
Mostly argue with you about some detail. Everyone needs a hobby, I figured you'd get this. :cool:
It's a bit more fundamental than that in this particular case. I know what pgimeno is trying to achieve. I'm checking that he does ;)
 
Maybe, if you're the only person getting a particular result, it isn't because you're the only person who's got it right. In those cases it's far more likely that you're the one getting it wrong. :p

To stem the flow of nonsense images, try this one...

http://femr2.ucoz.com/lineofsight1.jpg

2016px*5544px (2Mb)

The simplest of cross-checks.

From approximate camera location to SW corner of WTC1.

Lines up with both intermediate building features as it should.

I'd suggest you have some "explaining" to do but I'm really not interested. Whatever you're doing, the results are nonsense.

Er, :p

:(
 
It's great to see such optimism but I wouldn't count on it.
Hmm, well...

Reactor Drone surely could not continue to argue given such compelling evidence that he's doing something seriously wrong.

BasqueArch could surely not continue to support Reactor Drone from the side-lines.

And pgimeno has already (I think) accepted the validity of my study, though has decided not to comment upon or criticise Reactor Drone's "methods", "results" or "claims" in any way since the initial "good work" debacle. How strange :eek::rolleyes:


But you may be right. Folk are terrible at following links, so perhaps a few small sections of the image above will (in the long run) save me time responding to further nonsense (yeah, I know :))

So...

261746380.png

Red line extending from SW corner of WTC1.


126578805.png

Red line terminating at specified camera location.


And...

606458126.png


Do I need to annotate the relevant building features ? Ah, what the hell...

927663495.png


I still won't hold my breath though.
 
The above yawner has answered virtually every post on this thread for nearly three years. Boredom is not evident.

ETA - On the other hand, learning that the forum software allows a certain maximum number of eyerolls was enlightening.
I think considering the camera position was on a boat the following are key points to accuracy:

  • Actual current speed
  • Level of river (as in mean sea level)
  • Number and weight of passengers on ship
  • Distribution of passengers.
  • Number and distribution of over-weight passengers

I'll think of more but, surely everyone MUST recognize these conditions as key.

Until we do, we CAN'T know the truth about the planes position to an acceptable level of accuracy.




:boxedin:
 
...Until we do, we CAN'T know the truth about the planes position to an acceptable level of accuracy.




:boxedin:
Which naturally implies "acceptable" to who and for what "purpose" they find it "useful" :boxedin:

For those of us who don't lose sight of the overall picture here --- those of us whose interest goes beyond watching femr's patience and accuracy in "Mole whacking" --- there is a delicious bit of irony in the recent "goal post shift" to aircraft impact. And no-one has commented on it so far. ;)

...contrast with the discussion of accuracy relevant to WTC7 that is.

...and it still deals with the same key aspects of who finds the data "useful" and for what "purpose"

:)
 
Last edited:
What, exactly, is it that you are attempting to do ?

Again, I suggest you think carefully about your response.
I don't know what exactly you want to know.

At the "lowest" level, I am attempting to get a better estimate for the lens correction parameters of the collapse video than no correction at all.

At the "highest" level, I am attempting to obtain correct metrics with known error out of the video, and these corrections are not really that important since my goal is to know the error more than to minimize it.

But having found that way of reducing the error, I thought it'd be worth to spend some effort in it.
 
I don't know what exactly you want to know.
I'm trying to get you to think about what you are doing.

At the "lowest" level, I am attempting to get a better estimate for the lens correction parameters of the collapse video than no correction at all.
And what are you going to do with them ?
 
Oh my days.

This is ridiculous.

By whatever means you're placing your blue lines upon your partial map fragment...it's nonsense.


With photoshop, the same as I'm placing the red lines :p 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 blue line representing the fuselage length and orientaion are the placed on the map and the sightlines from the nose and tail are drawn on. Cutting the map perpendicular to the main sightline (centre of frame) provides relative spacing of the objects within the frame, including the aircraft. The cut can be made anywhere on the line since they are similar triangles and the proportions of the baseline will remain consistent.

Get a grip man.

a) The position where the docked ferry from which Hezarkhani filmed the footage is known. I've even provided you with a link within this forum from folk who went to the trouble of going to NY and taking photos to clarify.


Not in dispute

b) The position of WTC 1 & 2 is rather well known.


Not in dispute, although your layout differs from the map and satellite imagery.

c) Placement of Downtown Athletic Club and Whitehall Building was performed via rather tedious overlay onto 8192*8192 px satellite imagery, in conjunction with vector maps. As I've told you, don't trust Google 3D building placement.


Not in dispute. 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.


d) Orientation of the NIST aircraft (red) is exactly as specified. It doesn't match. To make it match, regardless of intermediate buildings, you have to move the camera a vast distance. Approximate camera location is FIXED. It's not up to YOU.


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.

e) Intermediate buildings allow refinement of camera location around the specific ferry dock (which was refined following my draft).

And would you believe it...
[qimg]http://femr2.ucoz.com/_ph/6/938786605.gif[/qimg]
...everything lines up, except the NIST aircraft..


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


Your method is fatally flawed, either through you omitting additional required factors, using invalid map information or simply your own incompetence. Whichever of those it is, you're clearly way off.

I've been patient with you, but enough.

Figured out how the tail of the aircraft lies within the visual edge of the building from that other camera angle yet? Maybe you could cross check how your 3d world looks from other camera locations.
 

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