Tranche 4 has just been released
https://www.war.gov/UFO/
https://www.dvidshub.net/search/?q=uap&view=grid&sort=publishdate
https://www.war.gov/UFO/
https://www.dvidshub.net/search/?q=uap&view=grid&sort=publishdate
Department of War - UAP Release 4
- Thread starter flarkey
- Start date
Looks like a bunch of balloons...
https://www.dvidshub.net/video/1014124/dow-uap-pr116-unresolved-uap-report-atlantic-ocean-2020
https://www.dvidshub.net/video/1014124/dow-uap-pr116-unresolved-uap-report-atlantic-ocean-2020
The familar 6 pointed camera artefact is there again...
https://www.dvidshub.net/video/1014101/dow-uap-pr104-unresolved-uap-report-yellow-sea-2025
https://www.dvidshub.net/video/1014101/dow-uap-pr104-unresolved-uap-report-yellow-sea-2025
RobertBillyBob
Senior Member.
Four of the videos are marked as "featured" on the war.gov website.
The "featured" column is a new addition to the .csv for this release.
RobertBillyBob
Senior Member.
PR024 and PR030 are in this release, they are filling gaps we had in the numbering from previous releases. PR020 and PR025 are still missing, they might come out one day.
RobertBillyBob
Senior Member.
I missed this because I was only looking at videos, but that's a pdf containing one frame, I don't consider this as "released".
John J.
Senior Member.
I'm guessing that whoever submits a UAP image/ video to AARO isn't obliged to consider the pattern of diffraction spikes that the imaging system might show. But I'm a bit disappointed that AARO doesn't comment on what are obviously diffraction spikes.
DOW-UAP-PR104, Unresolved UAP Report, Yellow Sea, 2025 https://www.dvidshub.net/video/1014101/dow-uap-pr104-unresolved-uap-report-yellow-sea-2025External Quote:
The six-pointed star seems to be the focus of interest. It's a diffraction spike pattern.
If the original observer/ image analysts aren't aware of this, it poses questions about their training. If imagery like this is being submitted by other people who have access to the imagery but don't understand it, this also raises questions.
The description states that the report consisted of the image in its present state. I don't think there is more to it than what's been released.
Chief Gleeby
Member
A lot of hoopla's being made over these images of what appears to be a slender triangular-ish object Columbia captured.
But here's the thing, the mission Columbia was on (STS-80) was carrying payloads: 2 Satellites that they released, let float in orbit for a while, and then re-captured!
Photos taken from within the shuttle of a distant object when one of their objectives is to recover some satellites that they themselves deployed... hmmmm....
The fact that none of this is mentioned on any of the image's pages is great too, really keeping this gravy train going lads, good work.
Also, as relayed by one of the crew for this mission, they saw diddly squat.
A few minutes of googling was all it took! Hey DOW, hire me!
But here's the thing, the mission Columbia was on (STS-80) was carrying payloads: 2 Satellites that they released, let float in orbit for a while, and then re-captured!
Photos taken from within the shuttle of a distant object when one of their objectives is to recover some satellites that they themselves deployed... hmmmm....
The fact that none of this is mentioned on any of the image's pages is great too, really keeping this gravy train going lads, good work.
Also, as relayed by one of the crew for this mission, they saw diddly squat.
A few minutes of googling was all it took! Hey DOW, hire me!
RobertBillyBob
Senior Member.
PR102 has a few frames you can almost read :
MyMatesBrainwashed
Senior Member
Starting to wonder what doesn't look anamolous in infrared.
Ravi
Senior Member.
It all is insanely laughable. How can I "have an open mind" when we are fed with complete and utter nonsense videos? Ah, I suppose "the good ones are still being hidden from us!" Sigh (^3)
Fritzkquzerk
Senior Member
That's a good point.
It's the perfect camera setting for seeing weirdness where there isn't.
They will claim this the next 79 years, it's a never ending story...
NoParty
Senior Member.
Apologies if this was explained somewhere already, but was there some criteria given, for
the order in which these duds would be released?
I ask because I'm thinking that if the "best" ones had been prioritized in the first batch
(I gotta google a better synonym for that noun), it seems it would be hard to get many folks' attention
at this point, with a 4th batch, when loads 1, 2 & 3 failed to provide much sizzle...
the order in which these duds would be released?
I ask because I'm thinking that if the "best" ones had been prioritized in the first batch
(I gotta google a better synonym for that noun), it seems it would be hard to get many folks' attention
at this point, with a 4th batch, when loads 1, 2 & 3 failed to provide much sizzle...
ExclusionZon
New Member
Following up on the PR104 mention earlier in this thread - agreed with the diffraction spike call, and since AARO's own label on it is "Unresolved" figured it was worth actually running the numbers instead of just eyeballing it.Pulled the actual file from DVIDS (two independent copies - official high-res master plus a separate remux of the public HLS stream, both hash the same structurally) and ran the same kind of frame analysis I did on the Roswell tape a while back.
Some of what came out of it:
Averaged 16 frame crops centered on the core and the six spikes come out sharper, not blurrier - only makes sense if they're locked to the screen rather than the object. Did a more granular version too, tracking rotation across 63 frames at 4fps, and the pattern doesn't drift at all over the full clip (0.55° average, no progressive trend). A camera panning and drifting for 15 seconds would show some rotation in anything actually attached to the object it's filming. Not seeing that here.
Also ran the same pipeline against PR46 from the first release (the "football-shaped body" one) as a control, since it's got literally the same HUD design - same corner brackets, same north indicator position, same censorship layout, so almost certainly same sensor family. PR46's object moves WITH the frame between shots and has no long radial spikes. Same system, different behavior depending on whether it's looking at an extended object or an unresolved point source. That's exactly the discriminator you'd expect if this is diffraction.
Built a quick optical sim too - point source through a 3-strut aperture, standard diffraction PSF, saturation/blooming added for the black-hot look. Spits out six spikes in three collinear pairs within half a degree of the real angles. Not just "looks like diffraction," a 3-parameter model actually reproduces it.
One genuinely odd thing I couldn't resolve: the official description says "18 seconds of footage" but the published file is 15.67s on both copies. Checked five other PURSUE videos across two releases and they all match their stated duration exactly - this one's the only outlier. No way to tell from public material if that's a redaction or just a drafting error on the DVIDS page.
Full writeup with all the numbers/methodology: https://www.theexclusionzone.com/yellow-sea-star-pr104-forensic-analysis-six-pointed-uap/
Happy to share the frame stacks or the sim script if anyone wants to poke at it.
Some of what came out of it:
Averaged 16 frame crops centered on the core and the six spikes come out sharper, not blurrier - only makes sense if they're locked to the screen rather than the object. Did a more granular version too, tracking rotation across 63 frames at 4fps, and the pattern doesn't drift at all over the full clip (0.55° average, no progressive trend). A camera panning and drifting for 15 seconds would show some rotation in anything actually attached to the object it's filming. Not seeing that here.
Also ran the same pipeline against PR46 from the first release (the "football-shaped body" one) as a control, since it's got literally the same HUD design - same corner brackets, same north indicator position, same censorship layout, so almost certainly same sensor family. PR46's object moves WITH the frame between shots and has no long radial spikes. Same system, different behavior depending on whether it's looking at an extended object or an unresolved point source. That's exactly the discriminator you'd expect if this is diffraction.
Built a quick optical sim too - point source through a 3-strut aperture, standard diffraction PSF, saturation/blooming added for the black-hot look. Spits out six spikes in three collinear pairs within half a degree of the real angles. Not just "looks like diffraction," a 3-parameter model actually reproduces it.
One genuinely odd thing I couldn't resolve: the official description says "18 seconds of footage" but the published file is 15.67s on both copies. Checked five other PURSUE videos across two releases and they all match their stated duration exactly - this one's the only outlier. No way to tell from public material if that's a redaction or just a drafting error on the DVIDS page.
Full writeup with all the numbers/methodology: https://www.theexclusionzone.com/yellow-sea-star-pr104-forensic-analysis-six-pointed-uap/
Happy to share the frame stacks or the sim script if anyone wants to poke at it.
FatPhil
Senior Member.
Looks like your struts are at 0, 120, -120, and the angles aren't quite a perfect match - can you do one at 0, 130, -130, say? That's closer to the actual mechanics of some of the pods I've seen photos of. But you're right, these are classic diffraction artefacts.
jarlrmai
Senior Member.
Useful work, some members also did some simulation in the chandelier case
https://www.metabunk.org/threads/pr038-the-chandelier-ufo.13307/post-309157
ExclusionZon
New Member
Went and actually measured it instead of eyeballing - ran a 16-frame stack of the high-res master with a 0.125°/bin angular profile and parabolic peak interpolation. Real spike-line angles come out at 29.5 / 94.0 / 148.9°.
Checked total angular error against a few models: symmetric 3-strut (0/±120) comes out at 5.7°, our original 0/±125 was actually worse at 13.4°, and your proposed 0/±130 is worse still at 23.4°. So if anything the data pulls toward ~0/±120, not ±130 - and it turns out our own initial sim was off too (I'd misremembered it as 0/±120 when it was actually run at 0/±125, my mistake). Updated the sim and the writeup to match.
With about +1° of global rotation allowed, the symmetric model lands within ±1-3° of the measured stack, which is inside the noise floor (per-frame angle estimates jump around by ~5° from sea clutter, the stacked average is the number to trust).
One thing worth flagging: the measured spike SEPARATIONS aren't quite even either (64.5/54.9/60.6°) - hints at maybe ±5° asymmetry but it's right at the edge of what's measurable from this compressed public file. Would need the uncompressed sensor original to actually resolve that, which isn't out there.
Script's the same one from before if anyone wants to rerun it.
Checked total angular error against a few models: symmetric 3-strut (0/±120) comes out at 5.7°, our original 0/±125 was actually worse at 13.4°, and your proposed 0/±130 is worse still at 23.4°. So if anything the data pulls toward ~0/±120, not ±130 - and it turns out our own initial sim was off too (I'd misremembered it as 0/±120 when it was actually run at 0/±125, my mistake). Updated the sim and the writeup to match.
With about +1° of global rotation allowed, the symmetric model lands within ±1-3° of the measured stack, which is inside the noise floor (per-frame angle estimates jump around by ~5° from sea clutter, the stacked average is the number to trust).
One thing worth flagging: the measured spike SEPARATIONS aren't quite even either (64.5/54.9/60.6°) - hints at maybe ±5° asymmetry but it's right at the edge of what's measurable from this compressed public file. Would need the uncompressed sensor original to actually resolve that, which isn't out there.
Script's the same one from before if anyone wants to rerun it.
Fritzkquzerk
Senior Member
Can you explain this to a simple mind like mine?
What are you implying?
ExclusionZon
New Member
Sure, plain version: the "six-pointed star" isn't the shape of anything flying. It's made inside the camera itself.
When a camera looks at a really intense point of light (or heat, for an IR sensor), the light bends slightly around anything sitting in the optical path, like the thin supports holding the lens elements in place. That bending smears the point into thin rays. Same reason stars in James Webb telescope photos have those spikes, the spikes belong to the telescope, not the star.
What we did was measure the angles of the six rays really precisely across the whole clip. Three things came out of it: the rays sit about 60° apart from each other, which is what you'd get from three support struts spaced evenly around the lens. The star never rotates relative to the screen for the full 15 seconds even though the camera's panning the whole time, a real object would basically have to stay magically aligned with someone else's camera to pull that off, while a lens artifact just does it automatically. And a simple simulation of "bright point plus lens with three struts" reproduces the star almost exactly.
So what we're saying: the video shows an intense heat source, could be a flare, an engine, something hot and far away, no idea really. And the dramatic star shape is basically the sensor's own optics doing what optics do. What the actual heat source was is still unknown since the telemetry's redacted, so "unresolved" is a fair label. The six-pointed star part just has a pretty boring, well-understood explanation.
When a camera looks at a really intense point of light (or heat, for an IR sensor), the light bends slightly around anything sitting in the optical path, like the thin supports holding the lens elements in place. That bending smears the point into thin rays. Same reason stars in James Webb telescope photos have those spikes, the spikes belong to the telescope, not the star.
What we did was measure the angles of the six rays really precisely across the whole clip. Three things came out of it: the rays sit about 60° apart from each other, which is what you'd get from three support struts spaced evenly around the lens. The star never rotates relative to the screen for the full 15 seconds even though the camera's panning the whole time, a real object would basically have to stay magically aligned with someone else's camera to pull that off, while a lens artifact just does it automatically. And a simple simulation of "bright point plus lens with three struts" reproduces the star almost exactly.
So what we're saying: the video shows an intense heat source, could be a flare, an engine, something hot and far away, no idea really. And the dramatic star shape is basically the sensor's own optics doing what optics do. What the actual heat source was is still unknown since the telemetry's redacted, so "unresolved" is a fair label. The six-pointed star part just has a pretty boring, well-understood explanation.
Fritzkquzerk
Senior Member
Yeah, thanks, I figured it was a camera artifact because it's basically identical to the Chandelier Video, but I got
So basically it's just the same as the previous "Chandelier star UFO" video, but with different sensors?
Also, in the chandelier video smoke is visible, that's why many thought of a flare, could this new one be the same thing? Or a jet engine? Missile exhaust?
Edit:
I didn't mean "lens flare", I mean "flare flare", or whatever you refer those ones to distinguish them from the "camera artifact" flare.
I mean those things that burn in the sky, deployed by planes.
FatPhil
Senior Member.
I'm sorry, but if you think those pairs of lines are parallel, you need better glasses.
If whatever tool you are using is telling you that those lines are better alinged than any alternative, then you need a better tool.
ExclusionZon
New Member
Yeah, same underlying camera artifact, different sensor/platform. The star shape itself doesn't tell you what's making the heat, that part's the same regardless of what's actually burning down there.
Not familiar with the Chandelier video specifically, got a link? Curious if it's the same six-strut pattern or something else.
On flares (the burning kind, not lens flare), that's honestly a solid guess for something like this. Countermeasure flares are basically a small intense point source of heat, which is exactly the kind of thing that produces this artifact on a sensor. Missile exhaust or a jet engine plume would probably look more elongated/diffuse rather than a single tight point, though I haven't tested that specifically. Can't rule any of it out from this file, the telemetry that would help narrow it down is redacted.
If the Chandelier video has visible smoke and this one doesn't, that alone doesn't tell you much either way, could just be viewing angle, wind, or the smoke just isn't hot enough to show up clearly on IR.
Not familiar with the Chandelier video specifically, got a link? Curious if it's the same six-strut pattern or something else.
On flares (the burning kind, not lens flare), that's honestly a solid guess for something like this. Countermeasure flares are basically a small intense point source of heat, which is exactly the kind of thing that produces this artifact on a sensor. Missile exhaust or a jet engine plume would probably look more elongated/diffuse rather than a single tight point, though I haven't tested that specifically. Can't rule any of it out from this file, the telemetry that would help narrow it down is redacted.
If the Chandelier video has visible smoke and this one doesn't, that alone doesn't tell you much either way, could just be viewing angle, wind, or the smoke just isn't hot enough to show up clearly on IR.
Fritzkquzerk
Senior Member
https://www.metabunk.org/threads/pr038-the-chandelier-ufo.13307/
There you go!
Thanks again for the answers about flares!
ExclusionZon
New Member
Fair, they're not perfectly collinear, already flagged that in the numbers actually. Measured separations were 64.5/54.9/60.6°, not an even 60/60/60, so there's a real asymmetry of a few degrees between the pairs. Noted it as being right at the edge of what's measurable from this compressed public file, could be genuine or could be noise from the sea clutter, can't fully separate the two with what's public.
Which pair are you looking at specifically, the sim overlay or the real footage stack? Can pull the exact per-line residuals if that's useful, might be a few degrees off on one specific pair rather than all three.
Similar threads
- Locked
Latest posts
-
Department of War - UAP Release 4
- Latest: ExclusionZon
-
-
Trending content
-
Thread '341.ROSWELL.41 - actually looked at the viral "crater" clip properly'
- ExclusionZon
Replies: 16 -
-
-
