Can you say approximately how high the muzzle velocity and the bullet velocity at the stage microphone would have to be if a second shooter had fired from the roof of the rear two-story building in order to achieve a crack/thump value of 0.22s?
if you tell me how to compute it, I can immediately add it to the pins of each point/person of interest
On ShotCutās timeline, I kept going back and forth using the arrow keys until l heard the initial sound of the shot for the first time. I thought of recording the audio via Audacity and importing it back to ShotCut (shotcut really is bad at visualizing audio waveform), but I chose to keep it simple and use the AV alignment I get from the downloaded video, since weāre using frames per second as our unit or measurement, and audacity could make me go to partial-frame territory (maybe), so I kept it simple.
I know what you meanā¦ I used the words to locate myself, but eventually I just decided to try and find manually the shot sound. If I remember correctly, I place the shot right after she finishes saying āRossā. Itās my frame 38, and it happens one frame after the casing has the view obstructed (37). I believe the sound reached the cellphone 1 frame after the casing is no longer visible.
It was my first time using that stabilize filter (and shotcut, for that matter), so I chose some parameters, clicked Analyze and let it do itās thing. It kind of rotates the whole image mid-playback, but I left as such without tinkering too much with the parameters. One thing I noticed is that the watermark changes position. I was running short on time to look up how to better use those filter parameters. I do remember that I kept thinking how your version looks sharper than mine, because you were able to get the shell to appear better on the early frames. Your use of filters was very good IMO.
Seems Interesting. What would that mean for us? Audacity would work better with, say, videos with full video frames, instead of traditional audio units of time?
Well, are you saying that TMZ is in on the conspiracy, making fake videos? Do remember Iāve been saying to you for weeks now that I saw the casing and the recoil movement. That clip I made for X is from August 6thā¦
I played itā¦, I see the casing fly, the hair move and the head moveā¦ same as Iāve been saying beforeā¦ recoil is there but not monstruous.
I donāt know exactly how it is calculated.
The AI thinks that a projectile velocity of about 954 m/s would result in a crack/thump value of about 0.22s.
This would require a muzzle velocity of about 1200 m/s.
I wanted to know from you whether this is realistic.
i did have a look at that video at Scooper. In the file properties and itās details, the Length and Media Created date fields are empty. The data rate is 3154 kbps.
With the video at TMZ, The Length is 00:01:33 and Media created date is 07/14/24 12:32 PM, almost 1 day after the event. The data rate is 2350 kbps.
The size of these two files are different and leaning towards transcoding variations as for why. i do get what you are looking for and been on my radar as well.
The first version of this video i saw was on Twitter. The media creation date of this file was 7/14/24 1:45 PM, a bit over an hour after TMZ finished their version. It was also posted on Twitter at the same time, so it looks like the media creation date got set at the time of upload.
There are also other differences, no start logo, no enhanced frames at the end, has a Butler Pennsylvania mark. The frame rate of this file was set at 60fps causing some frame problems. The image resolution is also poor making it impossible to identify any thrown casings. I have been able to spot the slight shoulder recoil in this version when knowing what to look for.
1 Like
Audacity does not handle any video, just audio. it can pull out the audio of a mp4 video file if you install the ffmpeg library. Otherwise will have to try and export a wav sound file from Shot Cut to use in Audacity.
With the way Shot Cut does work with audio, it is limited in how it steps through one frame at a time. it can be misleading as it steps in and out of each frame, especially when trying to determine the exact moment the report is detected.
Audacity does excel with being able to zoom in, select parts of the sound file and alter the play back speed without the interference of steeping in and out of the frame chunks.
I do remember that I kept thinking how your version looks sharper than mine
I exported all the frames to individual image files, then used Gimp to align each image and add in the text and other markings.
1 Like
Hi Kwaka,
So, I guess we can only trust videos from first hand, downloaded directly from the cell phone?
The title of the version I got on Friday from the TMZ homepage was named:
(Audio) Trump Rally Shooter Seen Opening Fire Getting Killed Seconds Lat-01
By looking at that name it does not seem to be the original either. I agree, you see a slight sliver in one frame in that version. In the first video you refer to on X you donāt see anything.
The newest title from the TMZ video is:
Trump Rally Shooter Seen Opening Fire Getting Killed Seconds Later in New Video
and this version seems to be edited to add more dynamics and the casing has been enhanced as per explanation in my post above. And you can see the result: One day after uploading 5 or 6 new gifs come out claiming to see the dynamics!
This does not pass the smell test, could it be that the mainstream media is changing the goal posts to fit their narrative by editing this video?
1 Like
Iād also like to see that answered. 
Because to achieve a 0.22 second crack-thump, it would be easier from your hidden nest at around 145 meters than from that HighRoof 240 meters away. (distances aprox., donāt have actual numbers here where Iām at). Your hidden nest distance to podium is practically the same Crooks-Podium distance, afterallā¦
From HighRoof, the crack-thump of 5.56x45 caliber heard at the podium mic would be around 0.4-0.43 second. For caliber 7.62x39 (thatās the AK-47 round), the crack-thump from HighRoof to Podium would be around 0.3-0.33. @sgt-raven Check me, please, if you get the chance.
yes, all these sound analyses have been hyped and used to decide authoritatively to consider this or that hypothesis, but very few people really know how these analyses work and they are highly dependent on guestimatesā¦
the 220ms seems to have been carved in stone, but in this line of work, there are no certaintiesā¦
the microphone does some processing, the registration software does some processing, audio format conversion tools do some processing, and when playing the recorded audio, there is some processing in the reverse order too.
most of do these steps not skew a lot of the signal, but there is no lossless image or sound processingā¦
some time ago, I used this type of hearing protection (https://multimedia.3m.com/mws/media/1087488O/hearing-protection-for-impulse-noise-technical-bulletin.pdf), similar to
this thing introduces a small delay such that it can analyse the incoming sound and cap it when too loud.
the small delay is not zero and not too large such that normal conversations do not require lip reading 
the bottom line is that the audio that is registered is not the exact audio that people could hear live.
Mike Bell reported that both RealDjStew and the witnesses under the trees recorded the first 3 shots TOO EARLY compared to the subsequent 5 shots.
this means that the location from which these two sets could have been fired may have been different and RealDjStew and the witnesses near/under the trees had a recording advantage.
2 Likes
Sounds good.
Please see my third video @ 8:17 where I go into detail on shots 9 & 10.
Okay, but what evidence is leading you in that direction? Donāt we hear a distinct difference in sound between the two shots, indicating that they are at different distances and/or different weapons?
You cannot hear a separate report for shot 10 on the podium mic, but my model is using a crack-boom time of 13 ms for that, as computed by the formula.
it is standard practice to work with the original data where you can, each step that degrades something ends up with lower quality at the end. As for what TMZ has been doing to get these more enchanced images, I donāt know. The results do look impressive with how the rest of the scene does have more clarity.
I have come across one image enhancement techniques in the past that combine multiple frames to achieve better resolution in a single frame. It does get deep and too much for me to get into it at the time. Something like this maybe?
I came across one report 10-15 years ago that Hollywood was working on technology to quickly edit and update its movie library. Is this kind of thing coming to social media. Powerful way to gaslight the population if it can be done. I am not aware or have detected this kind of capability at this time. The discrepancy around Niel Armstrongās famous quote has been one example of this kind of action at work.
I am finding it unlikely that TMZ has been paid off by the deep state to cover things up. Michael Difrischia was filmed by others taking the video at the time. If the deep state wanted to cover things up, easier to delete it than go into some elaborate deep fake. If it should come out that TMZ would fake something like this, the blow back would be bad. Psychological operations are usually the job for CNN.
The more I have been digging into this, the more I do have to consider that is was a single shooter.
Roger, are you saying that the original and copy frames 105 are from two different uploads? I donāt see the pixel information in the one marked āoriginalā that would allow traditional image enhancement techniques to pop out the cartridge I see in the copy. Do you have other comparison frame pairs from the nine or so frames when the cartridge is in flight?
I was already noticing some artifacts in the version everyone is looking at here that donāt look like normal video coding would produce them (when bitrate constrained, video codecs throw away high frequency edges first). It would be helpful to be able to compare other frames.
BTW, if the cartridge IS really there, I donāt believe what we are seeing is it being ejected from the gun. At that angle of rise, it would have been found far, far away from Crooks. The most likely explanation would be that the AR ejected it sideways, it hit one of the standing seams on the roof, and popped up in the air. That, however, adds even more delay that we would have to account for, maybe a half-frame or more for the cartridge to hit the seam, and then a whole frame (at a velocity consistent with the parabolic arc we see) to rise to where we first see it. It doesnāt sound like much, but in 1 1/2 frames the gun report has traveled 57 more feet that we have to rationalize between the audio and video.
Hi guys. I was looking to geolocate Dstew at the shooting time with precision and discovered something worth sharing. I synchronized the Dstew footage and the middle bleacher footage. I used the officer walking near Dstew and his relative position to the cop car on the back. Conclusions:
1 - The supposed suppressed shots are actually the sounds of Dstew walking. One less conspiracy theory to debunk.
2 - At the time Dstew capture Crooks on camera, using the real location of Dstew on the middle bleacher I confirmed my original geolocation of Crooks (586767.73 ; 4523529.4)
3 - At the exact time of the second shoot and at the exact location of crooks on the middle bleacher footage is possible to see a white dot that disappears 13 frames after. Still donāt know how to explain that.
4 - Here are the locations I manage to find with this analysis and other previews location.
| Shoot | x | y |
|---|---|---|
| Dstew @ Shoot 1 | 586774.15 | 4523486.16 |
| Dstew @ Shoot 2 | 586775.89 | 4523486.19 |
| Dstew @ Shoot 3 | 586779.58 | 4523486.82 |
| Dstew @ Shoot 4 to 9 | 586782.7 | 4523487.79 |
| Dstew @ Shoot 10 | 586775.14 | 4523488.1 |
| MBfootage | 586744.99 | 4523394.47 |
| Crooks | 586767.73 | 4523529.4 |
5- If the audios are synchronized with the video the delay of the report is 0.237sec of the two footages. The margins error is within one frame (0.033sec). That is consistent with Crooks position with 0.023sec error on my calculations. Since I canāt upload the video here and Iām not an audio expert and if anyone wants to improve this calculation the middle bleacher footage starts at the frame 8460 @ 4:42sec on the Dstew footage. @greg_n, do you think this can help your geolocation of shoots?
Edit: here is one of the frames used to sincronize the videos:
3 Likes
I find it very dangerous to use AI for brainstormingā¦
the AI system is extremely biased in favour of the views of its programmers (the big companies that have been mining our personal information for years now and their configuration (the clients and investors of these big companies) such that their views are pushedā¦
if you ask for such scenarios, it will give you only the things that are in line with what they want you to consider and it will not give you alternative scenarios that are not to be considered from their viewā¦
I have tested different AI systems and asked them to produce things that are in my field and should be straightforward, but the AI system refused to provide me with any useful response because the answer would infringe the policy it has been programmed to enforce and comply withā¦
1 Like
speed of sound also depends on atmospheric pressure, altitude and humidity
but no worries. I will use the formula it specifies for the different points of interest
thanks!
First, it is a very poor video quality so Iāll base my re-iteration on the fact it is really hard to see. So difficult my initial casual views did not see a casing eject or movement. Upon close frame-by-frame review, I do see what could be a casing.
It is ejected in a pattern consistent with an AR. The observed direction is consistent with where cases found to Crooks right and forward.
@vt1, I apprecate the amount of time you are taking to pinpoint locations like this. Youāve gone to a deeper level of video analysis than what I did, so I will take your Podium and DJStew locations, incorporate them into my model and see what happens. But not for a couple of days, since Iām trying to focus on work that I get paid for this week vs. the Rally stuff.
I looked at your Shot 1 lat/lon and see that it is 3 m SE of my estimate:
I would also appreciate it if you would derive a precise location for the police cruiser. I have estimated 586793 m E, 4523524 m N, but if you have a more precise lat/lon, Iāll use it. I know that some people question the authenticity of the cruiser audio, but since it agrees with the other, older sources, I currently accept it as authentic, and I really like the fact that it is stationary!
And, while youāre at it, your location for the TMZ āRoss!ā video, as well.
4 Likes
Something for you:
1 Like
this is all relativeā¦
if Trump shifts 5m north-east, I would guess the position of crooks on building 6 will shift with more or less the sameā¦
I notice that there are different things to consider:
- if you project a ground overlay in google earth, you have a couple of options, and each of them have disadvantages:
- if you clamp the overlay to the ground, the image gets scrambled up due to the image processing that tries to stretch or shrink certain parts of the image to match the ground level āas good as possibleā. you can easily see this with the rows and rectangle sized bleachers: the āclamp to groundā makes the benches/rows/borders of these bleachers look very zigzaggy
2.you can use an absolute height such that the image is not screwed up by the stretching/shrinking crippling, but this has the advantage that if you rotate along the vertical axis, you will see that the a pin-placemark will pivot around the overlayed image, even though the longitude-latitude coordinates match where that pin connects to the ground. so, this is a visually pleasing solution, but makes people believe that the coordinates are wrong due to the pivotingā¦
- the date of the tiles the google earth server. each date has its own coordinates. the distances and angles between objects in one set of tiles match the distances and angles between other objects
all in all, as long you stay within one set of tiles, the coordinates of different points of interest will have comparable distances and angles compared to another set of tiles. the exact values of the coordinates will change, but the internal distances and angles do notā¦
2 Likes
thanks for including the high roof shooter position!
1 Like