this is a great scheme!
Hi guys. I will present my final study of the geolocations of all points of interest. Since my last update I develop a method to convert pixels in a specific footage and covert to line of sights of objects that is possible to identify in the Google earth satellite picture. The method result is a grid-map off medium and maximum deviation of the horizontal position in pixels. The location with minimum maximum or minimum medium tends to be the best geolocation of footage.
The method has its limits because it is necessary to know where the horizon is in the footage so vertical lines can be located on the object to measure the horizontal distance in pixels.
In total I used 13 footages and 43 screen shots.
This study was based on minimizing the error of geolocations. To minimize the error the best reference points are light poles, visible edges of buildings, visible tree trunk, projected edges of buildings, projected trunk of trees.
Geolocation of light poles:
The best way to find the geolocation of light poles is to draw lines of shadow and the light pole, the point where it cross is the geolocation of base of light pole. here is an example of that:
Elevated position correction:
Reference points that are in elevated position must be corrected since the satellite picture is not perfectly vertical, but at angle of around 25 degrees in the case of 11/05/2021. This angle results in altering the geolocation of elevated position to the West, so the correction must be in the opposite direction to the East. For example, the frontal wall building 6 has 3m of elevation that in the satellite picture results in a dislocation of that position 1.4m to West. So, when defining any reference point on that wall the geolocation must be corrected 1.4m to East.
Using a drone footage of that wall I projected that image on Google Earth on the front of the building. So, when I need a geolocation of any point on that wall, I use this image to geolocate.
Observe that the West end of the building does not coincide with the roof image, also that difference will occur on any point on that wall.
Now the order of the footages will be important since some of them are more precise than others. The most precise one is RSBN, because the angle of the camera is wide, the footage is close to horizontal, and is Almost a still position. I used 2 methods to geolocated RSBN.
method 1: footage pixel model
For RSBN I used 5 different print screens of different times that covers different angles. Then I estimate the geolocation of each footage. Here is the map with all the reference points for RSBN.
The resulted geolcation of each image differ in around 2m. So, what I did I find the minimum error between all footages and the best geolocation found was here:
This grid is 3x3m pointed to the North and the middle is a reference point (586684.00; 4523384.00)
Also, I was able to find the angle of the camera without zoom of 70 degrees +/- 0.2 degree.
Result RSBN: 586684.48; 4523383.68
method 2: drone image.
This drone image has a lot of error, to minimize that I found points of references that are close to RSBN.
Drone photo:
Drone photo orientation: North: 40.857441°, East: -79.970249°, South: 40.856761°, West: -79.971883°, Rotation: -23.2355°
Here are the 3 points of references used on the zoomed image close to RSBN:
Here are the same points of references used on the satellite picture:
Here is RSBN location (method 1) on the RSBN podium of the drone image.
The error from method 1 and 2 is around 0.3m. Since the method 2 depends on defining where the camera is in the RSBN podium, method 1 is more reliable.
NTD and C3P:
With the line of sight of RSBN on NTD fottage, it reduces the points of tests for NTD locations. The same happens to C3P footage.
NTD sees RSBN at 11:56:
C3P sees RSBN and possible sees NTD at 1:27.
C3P lines of sight of RSBN and NTD in yellow and NTD lines of sight of RSBN in purple.
Also, after having RSBN NTD and C3P I was able to project lines of sight from Trump, James, David, the Bleacher top positions and the Teleprompter positions. This is all done without the drone image from the Trump podium.
The Trump podium top drone photo:
This photo has some distortion that are difficult to overcome. But a manage to use the perspective correction from PowerPoint and achieved some pretty good results.
Here is the perspective correction used on PowerPoint:
Rotation X: 358°; Rotation Y: 19°; Rotation Z: 359° perspective: 20°
And here is the orientation: North: 40.857287°, East: -79.970452°, South: 40.856745°, West: -79.971403°, Rotation: -23.5482°
Here are some the points I was able to identify.
The error was reduced from 1.5m to 0.3m. Now even the Barns are parallel to the satellite image and the level of accuracy is very high been possible to identify small variations on the roads.
Elevated positions corrections:
Now comes the correction of elevated positions on that image. Crossing all possible vertical lines, I found the most probable geolocation of the drone (586719.64; 4523375.42) and its elevation of 119m. Since the white tents have the same height, I was able to estimate corrections of elevated positions. The final result came pretty close the geolocation of bleachers using the footage method 1 with errors of less then 0.3m.
Here is the Bleachers corrected in blue.
The most distorted is the Middle bleacher.
Source 5:
This source is far away from the podium and since it is zoom in (only 5 degrees of angle of view) the geolocation from the pixel method is imprecise and the range can differ 1 meter to side and 5 meters to front and back. Still here is the geolocation (586658.66; 4523405.68) and the grid of error map projected on Google Earth.
James and David positions was checked by RSBN NTD, source 5 and TMX.
With the defined position of James I used James footage to validate his location and Trump location. It differs around 0.2m, but James could have moved to the side on the footage.
Middle bleacher footage that sees Dstew was synchronized with Dstew footage. This way I was able to identify Dstew line of sight in two times: at 1st shot and at the time Dstew sees Crooks seconds before shooting.
Synchronized footages when Dstew sees Crooks seconds before shooting:
Dstew footage was located during his movement during the shooting using the Middle bleacher footage, his on footage and some line of sights. Also his movement was considered linear during 1 and 4 shots.
The same method was applied to source 4.
Ross location was determined by his on footage and by source 4 seen him near shot 10.
Using Ross footage and Dstew Crooks position was determined by the line of sights from the footages. This was compared to the geolocation of Crooks using post images of the scene and the correction of the roof elevated position the locations differ around 0.2m.
The BWC footages was used to define the South Barn Snipper position by 2 print screens and the 9th shot snipper I used 2 prints also.
Here is the final result for all points of interest:
| ref | E (m) | N (m) |
|---|---|---|
| RSBN_V3 | 586684.48 | 4523383.68 |
| NTD | 586691.16 | 4523403.54 |
| c3p | 586733.45 | 4523409.60 |
| DRONE_STAGE1_height_119m | 586719.64 | 4523375.42 |
| Corey | 586736.21 | 4523417.70 |
| S5 | 586658.66 | 4523405.68 |
| James_v3 | 586722.04 | 4523372.12 |
| David_v3 | 586719.97 | 4523371.22 |
| Trump_v2 | 586726.82 | 4523394.08 |
| Trump_MIC | 586726.64 | 4523394.28 |
| SBlower | 586722.47 | 4523376.15 |
| TMX_v2 | 586740.37 | 4523498.02 |
| MB_N | 586744.48 | 4523394.30 |
| Dstew@1shoot | 586774.15 | 4523486.16 |
| Dstew@2shoot | 586775.89 | 4523486.19 |
| Dstew@3shoot | 586779.50 | 4523486.33 |
| Dstew@4to9shoot_V2 | 586781.91 | 4523486.56 |
| Dstew@10shoot | 586775.11 | 4523488.07 |
| source4@1shoot | 586693.00 | 4523528.92 |
| source4@2shoot | 586691.84 | 4523529.60 |
| source4@3shoot | 586690.60 | 4523530.21 |
| source4@4shoot | 586683.23 | 4523537.98 |
| source4@9shoot | 586682.84 | 4523539.08 |
| source4@10shoot | 586683.83 | 4523544.57 |
| Ross | 586683.91 | 4523542.97 |
| Crooks | 586767.73 | 4523529.40 |
| SB_Snipper 2 | 586763.45 | 4523360.11 |
| SB_Snipper 1 | 586762.39 | 4523360.54 |
| Snipper Shot 9 | 586761.86 | 4523426.72 |
| Cruizer | 586793.39 | 4523525.12 |
11 Likes
It was not subsonic ammunition. Which would be an odd choice for snipers anyway who would want a longer, flatter shooting round one would imagine?
But, regardless, weâve got the 10th round with a fat sonic crack over Trumpâs mic. Itâs definitely supersonic.
5 Likes
Wow, @vt1 , well done! And thank you for providing the background on how you arrived at the results. Iâd appreciate it if you would add another column to your data table that shows elevation.
5 Likes
@rough_country_gypsy do you have a photo or video that shows AGR6 from the perspective of a person standing near the entrance to the North Barn? Iâd like to know whether the tree blocks the shooterâs position.
Edit: A drone shot from the shooterâs perspective toward the barns would also be helpful. I looked back in this topic to see whether you had posted that. I found 6 YouTube links for your drone video clips, but none of these are from the roof of AGR 6.
Edit 2: Spa Guy drone footage @ 3:34 does a good job of addressing my question
very cool!
a couple of days ago, I made a conversion tool for UTM â Google Maps/Earth coordinates which results in the following:
ps: I took the liberty of correcting a few typos
3 Likes
Hi Greg.
I will work on elevations now.
Also, I would like to validate positions with other 3D models. I could be wrong in some cases. If @roger-knight @howdoiknowthisinfo @kincses-zsolt @pk2019 @offtheback @rough_country_gypsy @bigtim could validate or find some position that is not coherent I will reevaluate.
5 Likes
will do.
could you share the image you use as a ground overlay with the coordinates to have it imported the same way you do?
thanks!
1 Like
nice work with the mock-up 
what just came to my mind here is the Higgins Preliminary Report. There he states:
âShot 9 hit Crooksâ rifle stock and fragged his face/neck/right shoulder area from the stock breaking up.â
and further
âI believe the shot damaged the buffer tube on CrooksâAR. I wonât be certain of this until I can examine Crooksâ rifle, but Iâm 99% sure, based upon reliable eye-witness ESU tactical officers who observed Crooksâ rifle before the FBI harvested it as evidence.â
Assuming his observations are correct, that somehow would rather support he first shot 3 rounds from a rearward/down position and then moved slightly to a forward/up position to release shot 4-8. There was a gap of about 2.8sec between shot 3 and 4, that would be enough to move something like 1.5 feet forward/up. He would then also be more exposed so the SWAT guy could see him better and take his shot 9.
Also Trump ducked down quite quick after shot 3 (about 1.5sec). Maybe Crooks couldnât see his target anymore and therefore moved forward and a bit up to see better where to aim. And because he couldnât make out Trumps red Cap anymore he went into kind of burst mode and released shot 4-8 pretty fast.
Iâm not sure but would the spread of the ejected casings likely be bigger in case of fast fire? So maybe casings of shots 1-3 and 2 of shots 4-8 landed north roof side, and 3 of shots 4-8 landed south roof side.
1 Like
I just dropped a file on @brian60221âs server:
https://barf.bz/files/howdoiknowthisinfo/kml/vt.20240918.kml
it contains the points of interest @vt1 enumerated.
as soon as I get the unskewed ground overlay picture, I will add it to this file such that it resembles the rally site betterâŠ
save this file to your local computer and open it with google earth proâŠ
ps: public/public is what you need to access this fileâŠ
1 Like
Gary also took some good drone footage from the barn roof sniper positions. Watch around 20:45-25:30 and 57:50-1:05:00 (with specific distances and gps elevation data)
Dangerous Liberty Ep88 - Brand New Butler Drone Footage and Mapping Data - YOU Get To See First! (rumble.com)
2 Likes
I think this is correct and correlates with my findings. I think heâs back to being prone and on the stock when the 10th shot comes in as well.
1 Like
And here is the orientation: North: 40.857287°, East: -79.970452°, South: 40.856745°, West: -79.971403°, Rotation: -23.5482°
1 Like
cool & thanks!
a few weeks ago, @rough_country_gypsy flew his drone over the rally area and I stitched the images together with (the trial version of) ptgui (hence the watermarkâŠ), which results in the following base layout:
with the following orientation (north, east, south, west, rotation):
40.857935°
-79.969881°
40.856281°
-79.972149°
-115.4646
you will see that your positioning is a little off to the left and a little too high:
without changing the aspect ratio of your image, I have moved your image a bit to the right and a bit lower, and now it matches RoughCountryGypsyâs drone layout almost perfectly, updated orientation for your image (N, E, S, W, rot):
40.857267°
-79.970436°
40.856725°
-79.971387°
-23.5482
this results in, where the road across the seated area across the stage/podium aligns very well with the base layout:
I have updated the kml file on @brian60221âs server and added the ground overlay (download File Browser and the images directory File Browser (make sure the images are in a directory called âimagesâ in the same folder/directory as where you saved the kml file); use public/public if necessary).
I think something went wrong with the positioning of Drone Stage1, height 119m (I corrected the name to say cm instead of m, but that is not the point): this point is now located in the right bleachers instead of near Trumpâs stageâŠ
your position for crooks is also a bit too far away from the parking sideâŠ
the âDRONE STAGE1 height 119cmâ point maps on the right bleachers and should move more towards Trump:
crooks and the witnesses near the trees:
overview with all your points of interest:
I used the following renaming scheme for the labels:
.replaceAll(Pattern.quote(âSnipperâ), âSniperâ)//
.replaceAll(Pattern.quote(â@â), ", bullet â)//
.replaceAll(Pattern.quote(âshootâ), " â)//
.replaceAll(Pattern.quote(â_â), " ")//
bottom line: the positioning of your labels makes much sense, but I believe they are a bit off for
- the stage and bleachers area when we look at the imho very accurate footage of RoughCountryGypsyâs droneâŠ
- the position of crooks is a bit too far from the parking side when we compare with the location of his body:
so, if you agree with the shift of your layout image as suggested above, I think it will be quite easy to finetune the coordinates of the points of interest of the bleachers, stage (and roof of AGR building 6, but this is unrelated to the shift)âŠ
Here is the drone image. I think itâs in the best location for me. I used all the vertical lines from the white tents, the barns doors and light poles to estimate that position. All the vertical lines converge to the point of perspective where the drone took the picture.
This is how I know the elevation, because considering the height of the side of white tent (3m) I can trace back to drone elevation.
This dimension in red is 0.92m and the yellow dimension is 34m. Considering that the tent height is 3m it possible by triangle similarity to calculate the elevation of the drone.
This estimation resulted in 111m, in other cases I achieved higher results. I considered the medium of them all that resulted in 119m.
This way now I can do the reverse and with a known elevation correct the bleachers and the teleprompter positions. Note that all the Bleachers in blue rectangles also trace back to the point of perspective. Also, the resulted bleacher in blue is a perfect rectangle with right angles and the sizes of the North (13.6x5.97m) and South (13.6x5.92m) bleacher are almost same.
For Crooks position I used the correction for elevated positions, which dislocate his position on the roof 1.5m to East because he is around 3.8m in elevation to ground.
This will happen to any roof in the satellite image. As you can see the east side of the building is due to the satellite not been exactly vertical to that location, but at an angle close to 24 degrees.
So, any geolocation in an elevated position to ground must be corrected. Also, you can see the projected image of the south wall dislocated to East for the same reason, in this case the dislocation is 1.4m to East. This makes a lot of difference when trying to make lots of images from the same footage result in the same location.
Also, I always use the 05/11/2021 satellite image, because the most sharp and clear image.
And I do not use the 3D buildings activated.
3 Likes
Hi-res 5000x3334px
https://s183.convertio.me/p/yXRl0UaYFTdb258jYBonWA/833e652d266c57551ca9faa2bbf95cf1/arial-view-large.jpg
Hi-res 3885x2895px
1 Like
it looks to me that the bullet scratched the top surface of the railing and moved along towards the JCB hydraulic liftâŠ
the grey puffy cloud clearly originates from the point indicated by the green mouse pointer in the following pictures from x.com
I think the striking point is on top of the railing, not below the top of the bleachersâŠ
1 Like
Looking more in to this, 3D building may compensate the satellite distortion, but the images used in 3D is probably the 14/06/2014. That is not a good image. So I prefer to use the 2021 image more clear and sharp lines.
Here is your image on the 2021 satellite picture. The location of crooks is closer to real spot.
1 Like