Dear Roger
Could you spare a few minutes of your precious time to redo this analysis with the corrected time stamps to account for the error caused by the beginning of the video not containing 30 frames in the first second?
What do you get?
under oath
Under oath, before I began my studies, we all had to swear to uphold two important commitments. However, the oath was not made to the government or the leading party (majority). The two pledges were:
- We would never speak or distribute foolish theories that do not align with our studies.
- We would never use our knowledge to deceive less-educated individuals.
I can still vividly remember these pledges, even though the exact wording of the oath has faded over time.
I cannot find it, but there are similar texts:
“I will respect and preserve the scientific integrity of the work I perform, never falsifying, misrepresenting, or plagiarizing any data. I will not deceive or mislead others in order to gain any advantage. I will uphold the principle of honest and transparent communication in my research, ensuring that my work is always conducted with a sense of responsibility toward society and the environment. I will collaborate in a spirit of mutual respect, always striving for the collective good of scientific progress, while treating my colleagues and the public with dignity.”
“I solemnly swear to uphold the ethical standards of scientific research. I pledge to maintain objectivity and rigor in all aspects of my research, to promote transparency in my methodology, and to ensure that my findings are accurately presented without distortion. I will not allow personal, financial, or political interests to compromise the integrity of my work. Above all, I will use my knowledge and skills for the benefit of humanity, ensuring that my work remains accessible, responsible, and ethically sound.”
“I solemnly swear to conduct my research with integrity, upholding the highest standards of honesty, objectivity, and accountability. I will never spread misleading or unfounded theories that could harm the pursuit of knowledge. I will not misuse my research for personal gain or to deceive others. I will seek the truth, even when it challenges my own beliefs, and I will share my findings openly, for the benefit of society and future generations of researchers.”
“I pledge to seek knowledge with integrity, to ensure that my research and its conclusions are grounded in truth, and to never knowingly contribute to misinformation or deceit. I will promote science for the greater good of humanity and for the advancement of understanding, always striving to uphold the honor of my field and never to use my knowledge for exploitation or harm. I will educate and share knowledge in a way that respects all people, without intention to manipulate or mislead.”
I will never spread misleading or unfounded theories
never knowingly contribute to misinformation or deceit
maintain objectivity and rigor in all aspects
I will not misuse my knowledge to deceive others.
never to use my knowledge for exploitation or harm
I will not deceive or mislead others in order to gain any advantage
You might add to the end: “So help me God.”
Basic MP4 File Structure:
An MP4 file consists of a sequence of atoms, which are essentially binary blocks of data. For video, the important atoms include:
ftyp(file type)moov(metadata box)
trak(track box)
mdia(media box)
minf(media information box)stbl(sample table box), which contains thestts(time-to-sample) atom. This is where the timestamps and frame durations are stored.
Approach:
- Reading and parsing MP4 atoms manually: You’ll need to traverse the MP4 file’s structure, locate the
moovatom, and then find thesttsatom that contains frame durations (in terms of time).- Timestamp Extraction: The
sttsatom contains pairs of sample count and sample delta. The delta is the time interval between frames.- Rate Change Detection: By inspecting the deltas, you can detect when the frame rate changes (i.e., when there’s a sudden difference in the delta values).
Unfortunatelly I have no time to do it today.
Explanation of the Code:
ExtractTimestampsMethod: This method reads the MP4 file, searching for themoovatom, which contains metadata. It iterates over the atoms in the file, seeking out thetrakatom and then thesttsatom.ParseMoovAtomMethod: This method parses themoovatom, which contains metadata for all tracks (video, audio, etc.). It looks for thetrakatom, which describes the properties of individual tracks.ParseTrakAtomMethod: This method parses eachtrakatom, which contains information about the track (e.g., video or audio). It looks for thesttsatom within thetrakatom.ParseSttsAtomMethod: This method parses thesttsatom, which contains the frame timing information (sample count and delta). Thedeltais the time interval between each frame or sample, and this is what you need to calculate the timestamps.ReadUInt32Method: This helper function reads a 4-byte unsigned integer from the stream.
When you said 16 video frames were missing from the video’s first second, I guessed that the audio had been started ‘early’, by the time equivalent of 16 video frames.
7786044 bytes of ‘moshe.mp4’, unchanged by me, are here:
https://superfly.co.nz/mp4/moshe.mp4
Could you see if QuickTime finds 30 frames in its first second of time?
QuickTime finds 16 frames in its first second of time.
I re-analyzed a massive video (over 11 hours of recordings) from Real America’s Voice (RAV) in which the shots against Trump are replayed four times. For each replay, I extracted the video sequence and used Audacity to watch the exact time of the cracks of shots 1 and 2, and used QuickTime to watch the moment when Trump begins to raise his hand to his ear. The results are as follows:
Excerpt 1
Crack 1: 3.628 s
Crack 2: 4.486 s
Raising hand: 5 s + 5 frames or 5.167 s
Delay compared to crack 1: 1.539 s
Delay compared to crack 2: 0.681 s
Excerpt 2
Crack 1: 16.440 s
Crack 2: 17.298 s
Raising hand: 17 s + 26 frames or 17.867 s
Delay compared to crack 1: 1.427 s
Delay compared to crack 2: 0.569 s
Excerpt 3
Crack 1: 18.881 s
Crack 2: 19.739 s
Raising hand: 20 s + 2 frames or 20.067 s
Delay from crack 1: 1.186 s
Delay from crack 2: 0.328 s
Excerpt 4
Crack 1: 41.860 s
Crack 2: 42.718 s
Raising hand: 43s + 1 frame or 43.033 s
Delay from crack 1: 1.173 s
Delay from crack 2: 0.315 s
Therefore, the time between crack 1 and the start of the hand raise varies between 1.173 s and 1.539 s, with an average of 1.33 s.
And the time between crack 2 and the start of the hand raise varies between 0.315 s and 0.681 s, with an average of 0.47 s.
The timings vary from one broadcast to the next, which shows that the audio-video synchronization exhibits slight variations, but Trump begins raising his hand about 0.5 seconds AFTER crack 2 and about 1.3 seconds after crack 1.
I’ll let you draw your own conclusions…
The initial frames of a video are not presented accurately in VLC or QuickTime, as explained by ChatGPT:
Why VLC shows “2s” when the real PTS is ~1.5s
- MP4/H.264 streams often have edit lists or B-frame reordering. This means that the first “presented” frame doesn’t start at exactly 0, but maybe at 0.5 s.
- ffmpeg’s %{pts} overlay shows the true presentation timestamps embedded in the stream.
- VLC (and many other players) will normalize those timestamps for convenience: they shift the displayed clock so the first frame starts at 00:00:00.000. That’s why VLC shows a “nice” 2.0 s when the internal PTS might actually be 1.5 s.
So when your quicktime is showing 2 seconds it is actually 1.5 seconds. I verified it by using a simple stopwatch started at 0 and stopped at 4 seconds. So Quicktime shows 4 seconds into the video but the stop watch shows 3.5 seconds.
You method with Quicktime is therefore incorrect. Why not simply use the command with ffmpeg. It puts each frame number and the exact time stamp.
Here is the simple command you need to enter and everything is clear for everybody:
.\ffmpeg -i "Moshe Schwartz - New video from the moment shots rang out at the Trump rally.mp4" ^
-vf "drawtext=text='Frame %{n}':x=400:y=270:fontsize=28:fontcolor=yellow:box=1:boxcolor=black@0.5,
drawtext=text='Time %{pts\:hms}':x=400:y=300:fontsize=28:fontcolor=yellow:box=1:boxcolor=black@0.5" ^
-c:a copy "Moshe Schwartz - New video from the moment shots rang out at the Trump rally with frames and time stamps.mp4"
Please verify this with a stopwatch: you will see that the timestamps in the first frames reported by QuickTime are incorrect, while those indicated by ffmpeg are accurate, QuickTime then reports a time with a consistent absolute offset of 0.5 seconds.
However, since our comparison focuses on dynamic actions occurring later in the video, this initial discrepancy in the first frames does not affect the analysis.
As mentioned in the previous post, QuickTime does not display the first frames correctly. For example, when QuickTime shows 1 second, the actual time is only 0.5 seconds.
Simply by listening to the video’s audio, you can pinpoint this with high accuracy: the “crack” of the first shot occurs exactly as Trump pronounces the “e” in the word “happen.”
Reading Trump’s lips, however, provides an even more reliable method than analyzing the audio track, since live broadcasts often introduce slight audio–video offsets, which can lead to additional errors and misinterpretation.
After the first shot grazes his ear, Trump reaches up and touches his ear approximately 0.968 seconds later . Can we all agree on this and move to the last part of this analysis?
1 Like
As I explained to @phiphi-the-frenchie VLC does not indicate the correct time in the first 2 frames. Therefore if you count 15 frames the actual time is 0.5. I verified it with a stop watch. Starting at 0 and stopping at 4 seconds in VLC, the stop watch indicates 3.5s.
Since you are a frequent user of ffmpeg, why not add frames and timestamps with this command:
.\ffmpeg -i "Moshe Schwartz - New video from the moment shots rang out at the Trump rally.mp4" ^
-vf "drawtext=text='Frame %{n}':x=400:y=270:fontsize=28:fontcolor=yellow:box=1:boxcolor=black@0.5,
drawtext=text='Time %{pts\:hms}':x=400:y=300:fontsize=28:fontcolor=yellow:box=1:boxcolor=black@0.5" ^
-c:a copy "Moshe Schwartz - New video from the moment shots rang out at the Trump rally with frames and time stamps.mp4"
It will make things much easier and allow us to conduct the analysis in a far more professional way.
1 Like
Since we will soon finalize the analysis of “Shot 1 hitting Trump’s ear and then the bleachers,” I will begin examining other topics as promised.
Please start by providing the source video of bald man — that is, where this video was obtained—and include a brief explanation of your claim.
Keeping my promise to analyze the Hercules 2 sniper testimony, I must frankly admit that I have not spent much time on this topic. What I recall comes only from the “Who Shot Trump” video: one sniper claims the shot came from his left, while the other claims it came from the right—implying it must have been somewhere in between them.
Is this correct? If possible, could you provide me with more details about this testimony? Specifically, do they mention which shot they are referring to—was it Shot 1? If this did occur, what would be the theoretical impact point: the bleachers or the hydraulic pipe?
Any additional information would help me form a more informed opinion.
Thank you for your support.
1, or 1.1 second to react to a shock surprise is good, and I wouldn’t have reacted in 1/2 second.
Also, the audio has failed you in many ways, because it’s so easy to remove it, or move it, relative to the video.
I’ve seen Trump’s reaction from West and from North(Corey’s) to conclude that the 1st shot hit him.
Trump was already hit with Shot 1, heard by David Dutch, before Dutch was hit with Shot 2.
As you hear the first crack and pop at the podium mic, try to hear a 3rd sound, as I did ~8 pm on 13 July…the 3rd sound is Trump saying ‘ohhhh!’
Source video:
https://www.youtube.com/watch?v=pe1U9O3Fm18
contains hundreds of frames showing people on the roof.
Ten frames contain images of a bald man, kneeling, probably on the head and/or body of another man.
Start with
010.png
011.png
012.png
at my website:
At what time in the video does the bald man appear?
bald man appears close to the 3 minute mark, of that 7 minute 34 second video.
But the top half of a bald head/face appears ~1 minute earlier…117 consecutive frames(3.9 seconds of real time) show a person with a raised fist, as HorseLady rides East.
Hello. I come back here for a visit a year later and see that some of you are still going strong with the theories. Your dedication to the cause is impressive!
I plan to watch the “Who Shot Trump” documentary since I see that it was mentioned here. I hope that our collective research is shown in it after all of the effort we put into analyzing the data!
3 Likes
I doubt the cloud of dust follows a Brownian motion model. I think the particles got kicked up with a certain amount of velocity, and they will decelerate from air resistance. At the same time, gravity and wind are acting on them. Once the original velocity has dissipated, the wind and gravity will “have their way” with the particles.
1 Like
You’re right, Roger, you’re right, once is not always the case, don’t get into bad habits!
This whole discussion wasn’t pointless; it allowed us to better understand how QuickTime and VLC work, which can be useful…
So, all I have to do is subtract the duration of 14 frames, or 0.467 seconds, from the QuickTime timings. And I get that Trump starts raising his hand about 0.86 s after crack 1 and at the same time as crack 2.
And he touches his ear about 8 frames later, or about 0.27 s later, so 1.13 s after crack 1 and 0.27 s after crack 2.
So Donald is rather slow, taking a little less than a second to react, but perhaps he’s reacting more to the blood flowing down his ear than to the pain!
I wish I could be more specific, but when I look at my spreadsheet where I’ve copied interesting bits of testimony from the House and Senate reports, I’ve got one lonely sentence plus the links to the House and Senate transcripts.
Perhaps others can give the more useful quotes.
The first and longer one is the Senate testimony, given Aug 16:
The other is the house testimony, given Oct 7
If no one steps up I can try to put together some of the most important quotes in a day or two. Meanwhile I promised my substack people I’d put out a post on something else in “a day or two” two days ago.