Satellites Taking Pictures of Rockets Carrying More Satellites | Hacker News Reader

Satellites Taking Pictures of Rockets Carrying More Satellites

planet.com

214 pointsprivong46 comments

Each time I'm sad we're not exploring galaxy star trek style yet and the space conquest made a flop, I remember how many commercial satellites are around here.

It really has this regular taste of things you think just died after being hyped (the moon landing) and which you suddenly realize has got slowly, steadily, discretely omnipresent since then without anybody noticing much (granted, satellites are not as cool as visiting other celestial bodies, but that's still growth, and going the right direction).

I would love to see a graph of the amount of satellites in activity over time.

You may be excited about SpaceX plans.

It has basically fallen into a classic hype cycle - https://en.wikipedia.org/wiki/Hype_cycle

This is partly why I'm skeptical of technologies when the media is talking about them. It probably means we're somewhere on the peak of inflated expectations.

Indeed - Elliott waves come in mind too. As un-scientific as those ideas can be, not being expressed in any kind of mathematics, they seem to point at some possible reality in growth patterns. Notably, the fact that there's always a plunge after a first surge - where everybody think it's dead and where it starts a new uptrend - is something that makes me really curious.

Note that it's not just about people expectation either : humans actually went on the moon, they didn't just talk about it, and they didn't go any further in decades. So beside expectation, it doesn't seem far fetched to say there indeed was a plunge in achievements, at least if we consider distance traveled by humans (which should grow again when/if we go on Mars).

The peak of inflated expectations was the Saturn V. Custom built, massive, cost around $10,000/lb in present day dollars to low earth orbit.

The trough of disillusionment was the Shuttle. Much smaller payload capacity, trapped in LEO so it ended manned deep space exploration, and cost around $50,000/lb to LEO.

The Falcon Heavy is the highest capacity launch system since the Saturn V, and will cost $1,000/lb to LEO. I think the plateau of productivity approaches.

Would love to see an actual speed version of this... what's the hurry?

They stiched still images together to make the video, perhaps that created the speed up?

Yes but that's configurable with a good tool.

I presume their satellites aren't equipped to take more than one image per second. (For most of their applications, it's not necessary, and adding that ability to a microsat is would definitely increase the weight/cost/power.)

You can watch it on 25% speed on YouTube. It's pretty jittery, so it would be nice if someone made an interpolated version.

I tried using butterflow and it seems ok at first, but I guess the algo is not made for this weird shot.

https://www.youtube.com/watch?v=8a6QoKQNGWA&feature=youtu.be

This one might be better.

FPS is set to 60 and speed is the same:

https://youtu.be/b7Ua4mXO23E

Their satellites have a one dimensional sensor that scans continuously.

They are using 2D TDI CCDs, just like everyone else.

This timelapse was likely shot with frame by frame exposures though. I'm not sure how it's possible with TDI.

Still inherently 1D. TDI sensors are not like area arrays.

Think about the system[1], you've got to take a picture, store it, and then downlink it. If you take too many pictures your downlink is overloaded and you run out of memory. Assuming that an HD frame is about 2 megabytes once it is compressed, that is a 2 - 5 mbs downlink speed (depending on error correction). The satellites they launch are tiny and low power so they will have a limited budget for down link bandwidth.

That said, it makes me wonder about things like NROL76 which was a classified DoD payload that SpaceX launched and had no upper stage video from. At what point are there so many of these tiny satellites in orbit that they can pick up the fairing deploy and satellite separation independently? Or have them watch the X37 as it maneuvers around in orbit?

[1] https://www.planet.com/company/approach/

If bandwidth were the main limiting factor, they would just take video intermittently and downlink it over a long period of time. This would be useful for documenting real-time changes on the ground during important isolated events (e.g., rocket launches). But they apparently don't have this capability, and I think it's because the hardware for taking many frames per second just isn't worth the cost and complexity.

As mentioned in parent, the rate you can capture depends on the resolution (i.e. image size in bytes), downlink speed, available memory, and length of the event.

As the length of the event approaches infinity, the framerate you can capture is basically bandwidth / image size. `(mb/s)/mb = 1/s` (aka hz). Beyond that limit, the memory size / image size gives you a hard limit for the number of additional frames you can buffer. To get a constant frame rate, you want to spread that quantity over the duration of the event.

I don't think memory is limiting design factor either.

The playback speed can also be adjusted by changing the frame rate of playback. Each frame just stays on the screen for a longer time. Similar to the -delay <n> parameter when creating a GIF in ImageMagick.

That's not in question. Yes, playback framerate is independent from recording framerate.

The question is the availability of resources (memory and bandwith) to the satellite and how many frames it can store without needing to transmit them over its slow downlink.

The frames are already back on earth now, as evidenced by the video being on Youtube. Adjusting the playback rate of those frames is what I'm talking about...

Whoosh. Nobody is arguing that they don't have the frames that they have. As we can plainly see, they have plenty of frames from which to make a video that runs in closer to real time.

All they need to do is adjust the frame playback rate to make the video last longer. Yes it will have little jerky movements but it would give time for the viewer to enjoy and ponder what they are seeing.

You could play the video at a slower speed if that is what you want, the youtube player goes down to 0.25x speed.

Indeed. That gets us part of the way there.

There is a live HD zoomed video stream from the space station. Maybe someday they will catch one of these.

Any of you have any idea of how much the images from planet.com cost?

Satellite image prices through a broker, including Planet's RapidEye constellation

http://www.landinfo.com/satellite-imagery-pricing.html

Prices are per square kilometer, but there's a minimum order of 500 square kilometers, and you're paying for a licence for a certain number of people to use them in a certain way.

TLDR: best have at least a few hundred dollars handy even if you're only wanting to look at one city.

A lot of lower resolution satellite data is free and open though

Has somebody recombined low res data from many satellites to a high res free version? Sort of like many synced cheap oscilloscopes can form a expensive one?

There is pansharpening (effectively improving the spatial resolution of colour images by combining them with sharper, higher spatial-resolution monochrome images) and using processing techniques to combine low-res images can certainly improve the effectiveness of detecting features which are smaller than the image pixel size, but I don't think anyone's likely to get comparable detail to a sub-1m or even a 5m pixel image out of combining 20m and 30m pixel images taken at different times.

Why not? You can treat it as a statistical problem where you are given low Res samples of the actual data. I've done it myself on a school assignment. A matrix inversion was required. There are things like sensor shift cameras commercially available too. And stacking software for astronomers etc...

Any idea why the minimum area requirements exist? It completely kills localized applications like a single farm. It seems like a minimum price, which you could also reach by committing to buy regular photos for a period of time, would be a lot more flexible.

I suspect it's related to the need to possibly re-orient the satellite to have a particular place visible by the camera. It must be relatively slow, and possibly even cost reaction mass, thus the commitment to shoot more of the same area.

Rapideye and SkySat aside, Planet primarily operates a monitoring constellation, rather than a tasking constellation. The doves (satellites used to take this and the satellites being launched) are focused on imaging the land surface of the Earth with an approximately daily cadence. They don't target a particular area, try to reorient, or even have active propulsion (just reaction wheels).

This is an exception -- the satellite was tasked to try to capture this and did reorient itself to do so.

For the most part, though, the doves just point straight down and capture images whenever they're over land and it's not forecast to be completely cloudy. There are over a hundred of them imaging the earth like a line scanner as it rotates beneath their orbit.

The Rapideye satellites operate completely differently, though. They are tasked, and do have to reorient as you mentioned, thus the minimum area requirement for acquisition of Rapideye imagery.

Thanks very much for this excellent comment.

Their strategy is particularly awesome: a hundred of small short-lived satellites (really small, they're 3-unit cubesats, 30x10x10 cm) operating in a swarm. No big deal if a single one is lost: operation can continue with other ones once they are in position. Also, costs are spread due the possibility of servicing several customers simultaneously in separate locations around the globe.

Oh jeez, I didn't even consider that they'd actually be orienting satellites in response to orders. (Although, does that make sense for "archival imagery"?)

I guess maybe this restriction might be lifted when Planet deploys enough coverage to be imaging the entire Earth daily?

https://www.planet.com/pulse/planet-launches-satellite-const...

Kind of offtopic but I figured I'd ask: how close are we to enemy of the state style satellites that can track movement like a overhead CCTV camera?

Beyond close. We've been touching for years.

In 2001 I was in a lecture with a world authority on remote sensing, and he was unwilling to talk about the highest resolutions..

There are probably satellites capable of very high resolution pictures, but tracking a person with a continuous video feed is probably not practical - you'd need a ridiculous number of satellites due to the orbit.

What we do have, and have had for a couple of decades, is drones, and they can definitely do that.

20 years ago they pretended it was good enough to differentiate between Saddam Hussein and his decoy lookalikes.

It is possible however that they made that up to not reveal an internal source of intel.

I, for one, think we have been there for a while. The tech is not hard: camera with huuuuge optics will bring you there.

The military have secret satellites with undisclosed specs, that's what they have been doing since Sputnik. You bet they are good at it by now.

The government never made those claims. The media has fanciful ideas of what our imagery capabilities are and it isn't in the government's interest to correct them.

The tech might not be hard, but stationkeeping an orbit on top of a target is, unless you have optics that can handle 36,000 kilometers of distance (geosynchronous). All other, lower, orbits involve the satellite hauling ass across the ground, usually at 5+ kilometers per second. The video would be great for about five seconds.

Real spy satellites have time windows and image like hell within the window, then adjust for the next orbit and wait 90 minutes. They’re often in polar orbit, too, because then you get different parts of Earth as it rotates under you. Satellite surveillance is quite tedious, and the enemy knows where and when they are.

Huge optics is one option

https://en.m.wikipedia.org/wiki/Synthetic-aperture_radar

https://en.m.wikipedia.org/wiki/USA-215

Things in space can often be seen with a good from the ground.

https://www.youtube.com/watch?v=fCrB1t8MncY

With this kind of video, it would be interesting to see how the North Korean rockets actually do.

Something tells me the CIA/NSA have already used this to get an answer. And at military resolution too.

Vampires pretending to be humans pretending to be vampires

Reply on news.ycombinator.com