Hacker News Reader

Comment by squarefoot | Hacker News Reader

That video was taken from far distance because the drone had to operate unnoticed in enemy territory; pretty sure that they had much better optics and transmission capabilities even back then, but it had to be very far from action. The distance can be roughly measured, if the bullets speed is known, by counting the time between the shots audio (assuming it's properly synced) and when they're seen hitting the area.

Today police drones could look at your backyard from a few dozen meters, and the same qualified immunity that let them execute people in the streets would prevent you to do anything against them; those drones wouldn't even need to hide themselves.

ps. Upvoted because I noticed you were downvoted, probably by a flesh drone programmed to quench any reference to Wikileaks and the Assange cause, without whom that video would never become public.

View on HN

One thing which bothers me is satellite imagery.

In the 80s, 30+ years ago, license plates were discernible from orbit. I often wonder, what if you had two distinct lenses, each with separate transmit?

EG, both operating at once. Or multiple lenses?

Now you could do things, potentially, such as use a high-fidelity lens to tag individuals. EG, license plates. Facial recognition. That sort of thing.

Once tagged, another lens, one which records the entire surface of the continental US, would track objects the size of vehicles, or even people. Determining who or what each was, would not be vital. Only high probability tracking, and those other lenses coming into play for closer viewing, to tag.

Now you're just tracking dots as they move. And once someone is tagged (phone, or license plate), they don't need to be retagged.

If not already being done, it won't be long until it is done. And once tagged, you'd never shake that tag easily.

We're all familiar with facial recognition tech used by SV giants, and we know of many things, such as gait recognition used in China. Yet what amount of tagging, facial recognition could be done, when budget was of absolutely zero concern. If computing power was of zero concern.

All you need at the images from the satellite. Ground based computing power can do the rest.

What if you had trillions in computing power to throw at this?

Well. Regardless. If it isn't a problem now, it will be a problem eventually.

In one of the local traditions, the voters in the state all assemble in a town square to vote in person. (this year they exceptionally used ballots)

https://www.ai.ch/politik/landsgemeinde

In the old days, every voter was given a special sword, and admission to the gathering was conditioned upon possession of the sword (my former landlord still has his, it might still be a requirement).

Would it be possible to develop some sort of equivalent, but as an antenna rather than a sword, reflecting on wavelengths illuminatable from the ground and observable from the air, that would allow popular votes to be taken via spy imagery? Presumably confirmation would involve multiple independent drone captures...

Do we even have capabilities to prevent another nation from just doing this on American citizens? They are probably doing everything you've described right now in China on people in China, but I wonder how long it will be until the Chinese point there cameras at Americans as well, if they haven't already.

> if they haven't already

Gary Powers was shot down in 1960.

> In the 80s, 30+ years ago, license plates were discernible from orbit.

This is absolutely false.

https://www.theatlantic.com/technology/archive/2011/12/top-s...

It was also known by the likes of William Safire that our satellites could "read the license plates on the cars of Kremlin officials."

If untrue, please provide a counter source to many of the like statements as cited above. (You may be right, but it seems that many believe you are wrong)

https://news.google.com/newspapers?id=nugbAAAAIBAJ&sjid=kFEE...

Under 'twenty questions', second column, end of column, under highlights "Big Bird" satellite.

Again, you may be right. However there are many claims that this capability existed in the 80s.

Suggest you research some of the satellites of the time, including ones which worked on film, and jettisoned rolls of it for terrestrial pickup.

> Again, you may be right. However there are many claims that this capability existed in the 80s.

I wouldn't be surprised if they could do that but there was no proof around to demonstrate it. The military tend to be very secretive about what they can do because by revealing it they would implicitly give hints on what they cannot do.

> Suggest you research some of the satellites of the time

Suggest you do the same.

Here's the "Big Bird" satellite you referred to: https://en.m.wikipedia.org/wiki/KH-9_Hexagon

> The cameras could scan contiguous areas up to 120 degrees wide, and achieved a ground resolution better than 2 ft (0.61 m) during the later phase of the project.

The sources you cite are, to put it nicely, propaganda. The physics simply doesn't agree that this was even remotely possible at the time.

Another commenter correctly pointed out that these claims would violate the diffraction limit by more than an order of magnitude, and that's assuming zero atmospheric interference, which is absolutely not the case.

Why do you think film is relevant here? These satellites are not noise-limited like your phone vs. an SLR, they're diffraction limited. This is an optical resolution limit that has nothing to do with the photon capture medium or the size of the photographic plate. The only way to do better is to increase the effective area (of the mirror, not the plate) or to decrease the effective wavelength.

They were not doing interferometry with these satellites, and certainly not the multi-axis interferometry you'd need to be able to resolve centimeter-thick lines in text.

> Another commenter correctly pointed out that these claims would violate the diffraction limit by more than an order of magnitude, and that's assuming zero atmospheric interference, which is absolutely not the case.

Yet this comment was refuted, and you have not addressed that fully.

> Why do you think film is relevant here?

As per the timeline, my comment with the film reference was written prior to any discussion about noise, or diffraction.

It was simply there to demonstrate the lengths gone, 2 decades prior to the 80s, in the 60s, to get imagery back to Earth.

I was curious how they did it in the early days, as video transmission technology was nascent. It quite certainly would have been a potential massive blocker on quality, from satellite to surface. Film seemed to solve that aspect of things in those early days.

I found this highly interesting. Film, from multiple lenses, the ultimate sneaker net (yes, it's analog, but still...)

Anyhow, over the years I've seen so many technical issues overcome, with sideways attacks on the problem. When people start saying "But... impossible!", I tend to have a high degree of skepticism.

I will say this though. Yes, it certainly could have been propaganda.

Doesn't mean it was, but I certainly do agree the potential is there.

LargoLasskhyfv added a link with an awesomely detailed PDF.

I'll say two things here.

First, I'll still say "post film manipulation and processing may have enhanced imagery, and they were constantly modifying this gear from what I've read in that PDF."

Second, "Big Bird" was decommissioned in the mid 80s.

https://www.thespacereview.com/article/3662/1

There's no way it wasn't replaced with something else. What was that?

An example, is the https://en.wikipedia.org/wiki/KH-11_Kennen system, which wikipedia says was launched in the 70s, has multiple iterations, and is still being used.

https://en.wikipedia.org/wiki/KH-11_Kennen#Imaging_sensors_a...

Shows some of these versions, and blockIII seems to have very nice resolution. Launched in 92, supposedly.

And yet, as these all of the programs? I doubt it. For example "Big Bird" (KH-9) and KH-11 were used in parallel. I bet others existed too.

I'm going to say that throwing out "No, it couldn't be done" in the 80s is hard for me to believe. Was it done?

Well if not, it was very close to the 80s, as in the very early 90s.

The Rayleigh limit in spatial resolution of an image forming device is given by:

l = 1.22 * λ * L / D

Where λ is wavelength, L is distance and D is optical element diameter.

For satellite at an altitude of 150 km with 2.4 m diameter mirror and 0.4 μm blue light, this gives a spatial resolution of 0.3 m. To read a number plate requires an order of magnitude better than this.

Thus an interferometric imaging system can theoretically do it ( two input scopes separated by 24m). The hard part is the atmosphere.

In one axis, and I don't think they were capable of doing it at the time suggested.

Weren't U2, blackbird and other spy planes highly classified? It could possibly be that airplane footage (by planes that at least in theory could be legally intercepted / shot down) was passed off as "magical" sattelite intelligence?

Maybe better links:

[1] https://web.archive.org/web/20120429200224/http://www.canadi...

[2] https://web.archive.org/web/20120108034640/http://www.nro.go...

Thanks for this. Especially the PDF link.

To anyone reading this:

Pages 100 onward, you get to see diagrams of how it was built, and pics of the retrieval system and info on methods.

Page 120 onward talks about that return unit in more detail.

Page 122 I found interesting, details about measures to prevent the return unit from falling into enemy hands.

Page 128 is a good summary image, with some details on the lower right.

Page 165 onward, especially 168 have actual shots. 168 looks better than the first Google Earth images, and you can easily make out and identify how many 'things' the enemy has.

Sadly I have no more time currently to read, and more sadly some details are still redacted. Understandable, when I ponder the nature of some of those redacted parts. EG, privacy, current contractors still employed, tech methods still used on newer gear, at time of declassification.

The future, imo, is sensor fusion. Cameras everywhere, in the sky and on the ground. Those cameras feed back to a large tracking system which is able to detect and track objects as they move between cameras.

This information can be crosschecked against cell phone location data and could flag anyone travelling without a phone, or with the wrong phone, for further scrutiny.

It sounds like paranoid delusions now, but I think it's what we're moving towards in 50 years or so after society has normalized mass surveillance.

Where I live we already have cameras on larger street crossings for traffic management, and along every bus line, not to mention the ones in busses, metros and stations.

What is lacking is the fusion you mention. But with ubiquitous 5G it's only a matter of when, not if.

The smaller ones clipped onto the top of traffic lights look rather eyeborgian ( https://www.imdb.com/title/tt1043844/ )

Here is a BAE camera system that can track all cars in a large area:

https://www.baesystems.com/en-us/product/airborne-widearea-p...

This tech is available to private corporations, and almost to individuals. If you've called a solar company to get an estimate recently, they have full 3-D LIDAR maps of your house from orbit. They can map out the trees down to the branch level and then model how much sunlight will hit a given spot on your roof if you trim them. I was on the phone with a sales rep from LG yesterday and it was a bit surreal: I knew she was looking at a satellite image of my house, so I was like "We're planning to cut that whole branch off the middle tree that's shading the south roof of our house, and we're going to thin the branches of the corner tree" and she could follow along.

Much of this is also available to private citizens on Google Maps too - my home is not visible from the street, so when we were buying it, we used the satellite imagery to determine how big a back yard it had, how much parking was available, roughly how big the decks were, and we could even see the previous owner's patio furniture.

In the 80s, 30+ years ago, license plates were discernible from orbit. Are you sure about that? I'm not even sure about that now. Any credible information on the net, somewhere?

I thought that was movie stuff, like in https://www.imdb.com/title/tt0120660/

I'm aware that this is difficult to prove, because secrecy, national security and so on, but license plates from orbit? In the 80ies?

>That video was taken from far distance because the drone had to operate unnoticed in enemy territory

That video is not taken from a drone at all. It's from an AH-64 Apache, a manned attack helicopter with a flight crew of two.

Reply on news.ycombinator.com