World's Most Powerful Digital Camera Sees Construction Green Light 89
An anonymous reader writes: The Department of Energy has approved the construction of the Large Synoptic Survey Telecscope's 3.2-gigapixel digital camera, which will be the most advanced in the world. When complete the camera will weigh more than three tons and take such high resolution pictures that it would take 1,500 high-definition televisions to display one of them. According to SLAC: "Starting in 2022, LSST will take digital images of the entire visible southern sky every few nights from atop a mountain called Cerro Pachón in Chile. It will produce a wide, deep and fast survey of the night sky, cataloging by far the largest number of stars and galaxies ever observed. During a 10-year time frame, LSST will detect tens of billions of objects—the first time a telescope will observe more galaxies than there are people on Earth – and will create movies of the sky with unprecedented details. Funding for the camera comes from the DOE, while financial support for the telescope and site facilities, the data management system, and the education and public outreach infrastructure of LSST comes primarily from the National Science Foundation (NSF)."
What does the DoE get out of this? (Score:2, Informative)
Strange.
Re: (Score:1)
Re: (Score:2)
In a consumer product based price comparison start from this
https://www.cinema5d.com/canon... [cinema5d.com]
The Canon ME20F-SH – A Lowlight Camera with 4 Million ISO is closer to the design
needs of this telescope.
This telescope will have low temperature sensors (heavy) to increase the IR side and
reduce over all signal to noise problems.
As for the Defense Department ... I recall a discussion of a program to detect and track rocks in space
that might impact the earth. Then there was DARPA and TCP/IP wit
Re: (Score:2)
But that still doesn't answer the question of what the Department of Energy gets out of such a camera.
Re: (Score:2)
Re: (Score:1)
Remember Department of Energy is the successor of the Manhattan project and is responsible for funding much of the United State's fundamental research in physics. Most particle physics research (Higgs boson), nuclear research (including fundamental research into how neutron stars explode as supernovae), and dark matter research is funded by the DoE. The DoE is building this camera because it will make precision measurements of Dark Energy by seeing how our universe has grown with time.
Re: (Score:2)
I can see a major use being patrol searches for possible Earth-colliding objects. Think of it as a follow-on to LONEOS.
Re: (Score:3)
It's not so much what the DoE "gets" out of it as the LSST is being ran by Stanford Linear Accelerator Center (SLAC) as part of the Stanford National Accelerator Laboratory, which is ran by the DoE. All of the national laboratories are ran under the DoE and do a lot more basic science research than just figuring out how to make nuclear reactions and seeing how fast they can smash particles together.
From the bottom of the article:
Distance? (Score:5, Funny)
World's Most Powerful Digital Camera Sees Construction Green Light
Yes, but how far away is the green light? If it's only a few feet away then the fact that the camera can see it really isn't such a big deal.
Re: (Score:3)
Seeing its own construction light would make that the world's biggest selfie. XXIst century priorities, you see...
Re:Distance? (Score:4, Interesting)
Seeing its own construction light would make that the world's biggest selfie. XXIst century priorities, you see...
Speaking of selfies, by the time this is completed in 2027 (planned time + overruns), you'll be able to get the same resolution on the iPhone 23. It's like using computers for code-breaking, the best way to break crypto that takes ten years to attack is to wait 9 1/2 years and then do it in six months on the computer you can get then. The best way to get this camera is to wait until a year before it's due to be comissioned, then buy the sensors that'll be available then. Oh, and in the meantime you can be collecting interest on the money you're not spending.
Re: (Score:2)
I don't know about that. A couple of back-of-the-envelope computations make me think that 10 years is not a long enough timeframe to make such a camera anywhere near common. Consider, for instance, the 3 ton weight. Suppose that technology develops such that an equivalent sensor halves in weight every year. Ten years then represents halving the weight 10 times, giving a weight of approximately 6 lbs. That definitely isn't iPhone weight, and comes from a pretty optimistic assumption about how quickly th
Re: (Score:3)
"he stretched out his arms toward the dark water in a curious way, and, far as I was from him, I could have sworn he was trembling. Involuntarily I glanced seaward – and distinguished nothing except a single green light, minute and far away"
Re: (Score:2)
But it can see its own light even before it's built.
That's pretty impressive.
Re: (Score:2)
Re: Distance? (Score:1)
Re: (Score:2)
Since it hasn't been built yet, seeing the light that triggers its own construction means that the camera is indeed far away - a few lightyears from Earth, at least.
Re: (Score:2)
Re: (Score:2)
And in 50 years (Score:2)
We'll all be walking around with one in our smartphones.
Re: (Score:2)
We'll all be walking around with one in our smartphones.
A 3 ton camera? I doubt it...
Lots more information (Score:5, Informative)
here [energy.gov]. (Warning: 50 page graphics intensive PDF.)
Optical path on page 26. 6Gb of raw data every 17 seconds (page 32).
Re: (Score:1)
hmm, how will they store all the data? imagine even having to catalog it especially since only a tiny amount of the total data will be useful.
Re: (Score:2)
Should be quite simple compared to the LHC's data output.
Re:Lots more information (Score:4, Interesting)
1) It's not in the middle of Europe, it's on a remote mountain in Chile. A bit harder to get super high speed internet up there
2) The data off the LHC can mostly be analyzed by computer. While some of the LSST data can be (transient stuff), discovery of interesting new things is going to be a lot harder to automate, so trying to figure out how to get people to actually look at the torrent of info coming off of it will be a challenge.
That said, they aren't very worried about the actual data itself- they are starting with a 150TFLOP computer to do the initial analysis and figure they will need about 950TFLOP after a decade of use, which is fast but not world record setting. ~60PB of info over a decade is doable with a variety of tech
And in addition (Score:3)
3) The LHC does not run as continuously as a telescope. Optical telescopes run 12x7x365.
Re: (Score:1)
Re: (Score:2)
Read again: 6Gb every 17 seconds (reading a big sensor is slow work)
That's 353Mb/s, which neatly fits in a Gig-E pipe and can be recorded by the puniest 5" HDD out there.
To sustain that rate for the whole night (10 hours or 36ks), you need about 12.7Tb of storage... a 3TB drive will do with room to spare.
Build a good size RAID, put tape backup (you have the whole day), not exactly a challenge.
(FYI, the LHC is many orders of magnitude higher: http://home.web.cern.ch/about/... [web.cern.ch])
Re: (Score:2)
Just went by the OP's numbers.
Even 15TB per night is a laughable amount, when you remember it's at best 6 years away. That will probably be one drive per night by then...
Re: (Score:2)
I may not get "it", but I know how to read the original thread question (AC @05:08PM): "hmm, how will they store all the data?"
Oh, and I do get your point too. Massive image processing acceleration via FPGAs might be somewhat related to my day job.
Re: (Score:2)
You see, you take your truck full of tapes down your mountain, past the desert and into the city, then you drive into a weird place called a datacenter, and upload the whole thing into a system designed for high availability, duplication, indexing/processing, and resilience. You can't have the datacenter near the telescope for many reasons, but you can save the truck trip by laying a bunch of optical fibers alongside your electrical mains, if the distance and the budget allows. If you don't know much about
Not the biggest, nor the best resolution. (Score:2)
Is the phrase "World's Most Powerful Digital Camera" carefully chosen to exclude the much larger cameras just above this world?
(All, except one, looking down at us)
Gigapixels? (Score:2, Funny)
How big is that in football fields?
Re: (Score:2)
I was going to say, HD TV's aren't a high enough resolution unless you're just trying for a big number. Why didn't they just go for 10,100 SD TV's?
Honest question: why? (Score:1)
part of the image at a time. One pic b/c free $bil (Score:1)
I imagine they'll look at, or process, a small part of the image at a time. If you eyes can only see a few megapixels at a time, and your GPU can only analyze a few megapixels at a time, "why not just take a thousand images of 10 megapixels each?" you might ask. Because the tax payers won't give you a half billion dollars if you do that. When the government is ready to hand you half a billion dollars, why not go ahead and have bragging rights to the highest resolution in a single image? At least u
Re: (Score:3)
So you can crop it. A lot.
Re: (Score:3)
You can zoom and crop like this [htwins.net] and your prints will still be fine.
Re: (Score:2)
It's not about displaying. It's about analysing the results.
half a billion (Score:2)
The budget is $483 million.
I do wonder if the iPhone 9 won't have similar resolution, and be completed at around the same time.
Pr0n? (Score:1)