Summary of all measurements
Earliest detectable movement leading up to collapse initiation
The earliest detectable movement is traced back to 9.5 seconds before the visible collapse initiation here:
Upper West Wall Pulls Inward 9.5s before Collapse
Antenna Base Shifts Eastward 9.5s before Collapse
Large antenna drop
for multiple points:
Over 200 frames
Same with frames stretched
Same over 700 frames
East tilt large antenna
, measured by tracking the horizontal difference between 2 points on the antenna.
Features of the graph:
a) The earliest eastward tilt drift is detectable in frame -100 (or frame 70??)
b) The tilt measures about 0.3 degrees in frame 220
c) By frame 320 the tilt has not yet reached one degree
Evolution of east tilting of large antenna visualized (angle exaggerated)
Features of the east tilting of large antenna :
a) The data of the east tilt shows a gradual drift leaning of the antenna eastward, beginning more than 4 seconds before the first visible movement. Downward movement of the antenna begins in frame 155 and falling begins in frame 215. Notice how drift leaning to the east is detectable as early as frame -200.
b) The whole antenna is drifting eastward while it leans.
The same highlighting initiation frames only
South tilt large antenna
, measured by tracking the vertical difference and convergence of 2 points on the antenna.
Features of the south tilt graph:
a) A south lean is first detected near frame 175
b) The antenna is leaning about 0.3 degrees to the south in frame 220 (it is later shown that the antenna starts to fall downward at a significant fraction of g in frame 220
c) The tilt reaches one degree in frame 295
d) By frame 320 the tilt has reached 2 degrees
For comparison, movement of stationary object in foreground
Large antenna drop compared to east antenna drop
NW corner drop
Multiple measurements of NW corner drop
a) The graphs clearly show there is no downward movement until frame 220
Comparison of north (Sauret) with NE (NBC) viewpoints: Large antenna angle
Features of the comparison:
Large antenna drop compared to NW corner drop
, Sauret viewpoint
The graph shows the drop data of the antenna compared to the NW corner. The measured downward velocities of both points are also shown.
a) Once again we see no movement of the NW corner until frame 220. Velocity measurements confirm this. The antenna moves downward as early as frame 155. The velocity graph of the antenna also seems to show a slow downward movement from frame 155 to frame 215, after which the antenna starts to fall at a significant fraction of g along with the NW corner.
b) Release points:
NE corner drop
NE corner, NW corner and large antenna drops compared
NW corner drop velocity reductions
red is drop in ft
orange is velocity
NW corner drop vs velocity
Large antenna drop compared to SW corner drop,
We only care about the lines marked "real fire" and "real antenna".
104 fl fire drop raw
a) Downward movement of the antenna is measured before the SW corner is observed to move.
Multiple measurements of large antenna, NW corner and SW corner drops taken separately and stretched, NBC NW viewpoint, show the same downward movement of these 3 points relative to each other:
NW corner drop
Large antenna drop
SW corner drop (104th fl fire)
Static point for comparison
Another measurement of the large antenna, SW corner fire and NW corner washer drops
from the NW NBC viewpoint shows once again that the antenna was moving downward before the SW and NW corners of the building.
(notes: Black - Antenna
Red - Washer
Purple - SW Fire
59.94 fps - Resolution Doubled. 1 pixel on the graph is 0.5 pixels on the original video.)
NE NBC viewpoint: Measurements of large antenna movement
Angular movement of large antenna from NE
, measured by tracking the horizontal difference between 2 points on the antenna
Evolution of angular movement from NE viewpoint visualized (angle exaggerated)
Features of the angular movement:
a) Between frame 165 and 217 the antenna tilted faster and reached about 0.75°. After frame 217 the tilting of the antenna stopped for the next 40 frames. During that time the antenna just shifted towards south while the north face (upper floors) apparently tilted more.
In the GIF above the x component of the movement is extremely stretched. Both axis are numbered each 10000 units. A thin blue line shows the position 10 frames ahead and a thin yellow line shows the position 10 frames before. The thin horizontal green lines show the measured antenna section. I extended the vertical lines to see at which elevation the "antenna" rotates (floor 98 is about at y=70000). It's obvious that the antenna started to descent at virtually 0° tilting. At about 0.6°tilting the north face gave way and stopped further tilting for the next 40 frames.)
Trace measurements and visual evidence of relative movement between the large antenna mast and the small antenna east of the mast show concave roofline deformity
The moving smoke created problems for the trackers used to trace fixed points. Therefore some frames were cut away (between -218 and 153) because the smoke would have distracted the trackers. The cut in the video distracted the trackers to but the trackers returned back to the best fitting position. The antennas show the positions relative to frame -307 (my start frame for the measurement.
All positions measured one time in a bright and low contrast version of the video and one time in a dark and high contrast version. The major difference is visible in the behavior of the yellow and the orange curves. It's both times the upper tip of the small East antenna but the the bright video (yellow) the area based measurement followed the sagging wire right behind the top.
The GIF covers frame 168-208:
This means that after frame 153 the antenna mast was already ahead of the east antenna. Until frame 207 the antenna mast sagged an additional 1500 units while the east antenna sagged about 700 units (orange curves) or less (yellow curves). That behavior is visible to the naked eye by the release of tension of the east guy wire supporting the antenna. and in the way the large antenna mast moved quite differently than the small antenna to it's east. This behavior cannot be explained by tilting in any direction. A part of the overall movement was caused by tilting but the difference in the way each antenna moved is a certain sign of roof deformation.
Visual evidence of roof deformation during the earliest motion
The antenna leans east and moves downward relative to the small antenna and the roofline. The guy wire looses tension. The small antenna remains stationary.
There is no doubt the antenna moves before the 104th floor fire along the SW corner, west face in this gif from the NW:
The antenna and perimeter do not rotate together.
The antenna falls towards the east and downwards and the window washer on the NW corner is pulled east. The small antenna and the east side of the north perimeter remain stationary. Deformation of the roof is the only explanation. The gif flashes between the images from around frame 123 and 208.
The east side of the north roofline remains stationary as the antenna falls downward toward the east.
The small antenna falls after the large one does
relation of small to large antenna prior to 9-11-01
gif of earliest detectable antenna movement
synchronization: frame number minus 347. This gif covers the 8 second interval before visible movement. The observed drift movement is first detected 9.5 seconds prior to visible movement. T
First movements in the antenna and NW corner become detectable on the level of individual pixels about 4 seconds before visible movement: From frames 800 and 900 (OWE numbers - release point of NW corner around 907), Sauret video. Shifts in pixels between frames 800 and 900 seen below:
The washer, and the NW corner with it, moves left (easterly component) a little:
The antenna moves to the left more, and down some:
The results of trace measurements visualized:
Pull-in of the upper NW corner over 500 frames (8 seconds) measured by a sub-pixel tracking technique, visualized - horizontally stretched:
Pull-in of upper NW corner compared to the eastward leaning of the antenna over 500 frames (8 seconds)
The pull-in of the NW corner and the sagging of the antenna are visible to the naked eye in these gifs:
Antenna visibly sags over frames 120 to 208:
The earliest visible antenna movement shown from 3 viewpoints synchronized:
In the center we see a narrow view that shows the movement of the 104th floor fire along the SW corner, west face. Notice how the antenna moves at least 1 seconds before the SW corner fire, meaning they could not have rotated together as a rigid object. Antenna movement clearly and visibly comes before perimeter movement.
Youtube version: http://www.youtube.com/watch?v=xGAofwkAOlo&feature=channel_page
The first 20 seconds are the original Sauret video.
1) The camera shakes about 9.5 seconds before collapse.
2) The antenna and the NW corner of the building begin to slowly move during or immediately after the camera shake.
achimspok Sauret version: starts some seconds prior to the shaking camera.
Audio: Dolby AC3 48000Hz 6ch 224Kbps [AC-3]
Video: MPEG2 Video 720x480 (16:9) 29.97fps 9608Kbps [Video]
The unfolded video fields into frames gives 7440 frames at 59.94fps. Just a segment of it is used. Frame 2005 of the unfolded version is the first frame "000".
Link to the original file:
Link to the upscaled segment used as numbered JPEGs and separated into the video fields.
The architecture in that area doesn't allow for many possibilities and those possibilities would affect the results in a very small way since the most important relation is the appearance of the antenna in the projection plane of the north face. For this projection plane we get the pixel-height relation from the video. It wouldn't change at all. So the results should be very close if the Sauret camera is placed 1342m north and 662m east of the center of the north tower at 32m above ground. The north tower is 29° turned towards the east.
NW NBC viewpoint
camera position relative to the center of the north tower:
28° tower turned towards east
NE NBC viewpoint
camera position in relation to the center of WTC1
on the roof of the reddish building here
NBC NW-corner and Sauret synchronized
Download available at: http://www.megaupload.com/?d=QIYRZNR3
It is a packed rar. It includes numbered JPGs for every frame at 59.94fps and the numbering Achimspok used. The videos are 29,97fps! Therefore you always have the same frame twice. The Sauret included is the blown up "field 0" of the original interlaced frames.
CNN west clips
FOX south clips
OWE Sauret version: framerate is 29.97 (recombined fields). The set of points are over-determined because there's some uncertainty in matching. Hopefully this assists. An entire thread devoted to this may provide additional info:
and a comment on why OWE frame numbers are off by 10 from the beginning of the segment are the first couple of paragraphs of this post:
There is a difference of 195 frames between OWE's and achimspok's versions.
Seen from the NE an antenna leaning leftward may have a southern and/or eastern component of tilt. Using the rotating model shown in the gif below, we can measure how a given angle of rotation is seen from the NE NBC perspective.
Some known rotations are mapped vs the angle the antenna appears from the NE perspective here:
Over the first few degrees rotation the angle is perceived as 80% of the actual angle.
The same model tilting eastward shows the NE viewpoint sees 70% of an eastward tilt.
gif of earliest antenna movement
synchronization: frame number minus 347.
For better understanding: That movement happened during 8 seconds prior to the collapse.
Angular movement of antenna from NE, measured by tracking 2 points on the antenna.
(vertical scale is wrong. Shape of plot is correct. Scale should be stretched 25%)
The image should be replaced with the following:
This one shows the raw curves unchanged and the fat curve 25% stretched for the maximal possible south tilt for east tilt = zero.
I gave the measurement of the NE corner drop a second tricky try. I chroma-keyed the moving smoke and replaced it with vertical lines as guides for the area trackers.
I guess the result is pretty good. Here are the raw curves (horizontal grid = 10ft):
The NE corner started to descent in frame 211. The velocity curve kinks down in 209 because of symmetrical averaging. The roof line reaches a tilt angle of about 0.25° in frame 235.
The faster east tilt of the antenna started in frame 185. Once again, if the core was pulled east by a perimeter collapse of the 98 SE corner we may expect between frame 185 and 211 twice the east angle of the antenna at the south face and both south corners. The fire at 104 SW corner should move 0.2 meters towards east but...
...the x-component of the 104 fire shows no motion at all (eastward motion = downwards slope of the curve). May be the top tilted towards south at the very same time and nullified the x-component. If so then we should see at least a downwards motion of the fire between 185 and 210 but...
...the y-component of the 104 fire apparently moved downwards after frame 210 and may be between 145 and 185. Between 185 and 210 all 19 curves show an upwards tendency (flame up due to core collapse?).
If the perimeter forced the core to tilt then it appears to be impossible that the perimeter stopped to tilt right when the core started to tilt and vice versa. Hence, there is no detectable perimeter driven cause for the east tilt of the antenna.
Created on 06/28/2012 11:44 PM by admin
Updated on 06/28/2012 11:47 PM by admin