Google Brain Creates Technology That Can Zoom In, Enhance Pixelated Images (softpedia.com)

← Back to Stories (view on slashdot.org)

Google Brain Creates Technology That Can Zoom In, Enhance Pixelated Images (softpedia.com)

Posted by BeauHD on Tuesday February 7, 2017 @10:45AM from the zooming-progression dept.

Google Brain has created new software that can create detailed images from tiny, pixelated images. If you've ever tried zooming in on an image, you know that it generally becomes more blurry. You'd just get larger pixels and not a clear image. Google's new software effectively extracts details from a few source pixels to enhance pixelated images. Softpedia reports: For instance, Google Brain presented some 8x8 pixel images which it then turned into some pretty clear photos where you can actually tell facial features apart. What is this sorcery, you ask? Well, it's Google combining two neural networks. The first one, the conditioning network, works to map the 8x8 pixel source image against other high-resolution images. Basically, it downsizes other high-res images to the same 8x8 size and tries to make a match on the features. Then, the second network comes into play, called the prior network. This one uses an implementation of PixelCNN to add realistic, high-res details to that 8x8 source image. If the networks know that one particular pixel could be an eye, when you zoom in, you'll see the shape of an eye there. Or an eyebrow, or a mouth, for instance. The technology was put to the test and it was quite successful against humans. Human observers were shown a high-resolution celebrity face vs. the upscaled image resulted from Google Brain. Ten percent of the time, they were fooled. When it comes to the bedroom images used by Google for the testing, 28 percent of humans were fooled by the computed image.

146 comments

Min score:

Reason:

Sort:

Google fags create horse shit by Anonymous Coward · 2017-02-07 10:47 · Score: 0

Turns pixelated images into new ways to track people
1. Re:Google fags create horse shit by epyT-R · 2017-02-07 10:58 · Score: 1
  
  No. Google hipsters just want to make handing over your privacy as slick and painless as possible, hoping it'll make you less likely to kneejerk. That's why that creepy-friendly corporate newspeak in big anti-aliased fonts is their favorite way to communicate with users.
2. Re:Google fags create horse shit by gl4ss · 2017-02-07 16:48 · Score: 1
  
  nonono.
  ways to track people they made UP on the spot!
  you see, they're already tracking everyone so they'll just make these computer generated faces to track them.
  (you can't track people effectivey with a 8x8 image. you could however generate fake evidence that shows that it _could_ match this and this dude, or even 99% of population.).
  
  --
  world was created 5 seconds before this post as it is.
Google can put together images by hackwrench · 2017-02-07 10:50 · Score: 2

Google can put together images based on smaller images that look like faces.
No CSI by Thelasko · 2017-02-07 10:51 · Score: 5, Informative

I don't care how fancy the algorithm is, the original data was lost. This is still just a guess about the original content. It's just a better guess than was possible before.

I just hope law enforcement doesn't think they can use this to solve any crimes.

--
One of our competitors trademarked the term "hypothesis". From now on, we will call them "boneheaded ideas".
1. Re: No CSI by Anonymous Coward · 2017-02-07 10:58 · Score: 0
  
  On Monday, February 6, at 4:03pm, an 8x8 pixel we have identified as Thelasko was seen entering the designated transgender bathroom and then exiting five minutes later.
2. Re:No CSI by rmdingler · 2017-02-07 10:58 · Score: 5, Insightful
  
  Law enforcement would never rely upon unproven methods to improve conviction rates.
  
  --
  Happiness in intelligent people is the rarest thing I know.
  Ernest Hemingway
3. Re:No CSI by K.+S.+Kyosuke · 2017-02-07 11:04 · Score: 1
  
  The interesting thing would be if a system like this could actually recognize details from previously-seen better-resolution pictures of the same thing to actually fill in details in a justified way.
  
  --
  Ezekiel 23:20
4. Re:No CSI by im_thatoneguy · 2017-02-07 11:15 · Score: 0
  
  Why couldn't they solve crimes with this? It might not be fool proof but I see no reason why it couldn't narrow the search.
5. Re:No CSI by Anonymous Coward · 2017-02-07 11:18 · Score: 0
  
  Oh they *will*. The fact that it's been altered is going to be the least of their worries, too, thanks to a good two decades of everyone being told by those TV shows that "zooming and enhancing" shows you the real thing. LE cannot wait for the planting and parallel construction opportunities this will grant them.
  "This bodycam shot may at first look like the man was unarmed, but when we zoom in and enhance, we can now clearly see exactly what officer jones himself was looking at. A derringer!"
6. Re:No CSI by Anonymous Coward · 2017-02-07 11:22 · Score: 0
  
  That may not be the case here. It's possible that there's more information in the image than what the human eye+brain can see, and this is simply altering the image to put that information into a format that humans can more easily handle.
  I mean, we've been under no evolutionary pressure to be good at extrapolating pixelated images clearly, the same way we have at extrapolating very small (far away) images. If anything, the ability to see through pixelation may be detrimental to fitness in modern society (c.f. japanese porn).
7. Re:No CSI by Thelasko · 2017-02-07 11:23 · Score: 2
  
  Sure, it could narrow a search. However, this shouldn't be used as evidence in a trial, or even to obtain a warrant.
  
  --
  One of our competitors trademarked the term "hypothesis". From now on, we will call them "boneheaded ideas".
8. Re:No CSI by Anonymous Coward · 2017-02-07 11:28 · Score: 0
  
  Well considering there's a very large, but finite, amount of data in the picture, I'd say ... yes, but you would have to quantify the algorithm's uncertainty somehow. It would probably work in theory much better than a human, but it would rely on the human's judgement in exactly the same way (training it on the right set of images, for example, and using realistic quantifications of uncertainty)
9. Re: No CSI by Anonymous Coward · 2017-02-07 12:00 · Score: 0
  
  It works on CSI because they live in a different future than we do. Their FBI has already planted microscopic HD cameras every few inches all over the entire country including inside homes and businesses.
  They use the crappy original image to feed into the surveillance network and it grabs the HD feeds from that time and "enhances" the image.
10. Re:No CSI by Hognoxious · 2017-02-07 12:12 · Score: 2
  
  Absolutely not. If that was allowed to happen then before you know it they'd be acting as an army (except with different coloured uniforms) and we all know that'll never happen because FREEDOM and NUMBER ONE!
  
  --
  Confucius say, "Find worm in apple - bad. Find half a worm - worse."
11. Re:No CSI by PPH · 2017-02-07 12:15 · Score: 2
  
  I just hope law enforcement doesn't think they can sell this to a jury.
  FTFY.
  CSI has already proved problematic in that jurors have developed unreasonable expectations of what is possible.
  
  --
  Have gnu, will travel.
12. Re:No CSI by wbr1 · 2017-02-07 12:43 · Score: 2
  
  This needs so many more points. As someone who has been to prison - guilty as charged and pled to it, there are many who were not. There are plenty who were guilty in prison, but one that is not is too many.
  One particular case stands out to me. Navy soldier, on leave, drunk. Seen in an altercation in a bar with someone. that someone was known for instigating fights. Person winds up stabbed to death later.
  The suspect is arrested, and drunk, and with a huge lack of sleep, under duress from trained psychological tactics confesses. He later recants. Blood at the scene does not match hos or the victims blood type. None matches his. No physical evidence shows he was there. Convicted of 2nd degree murder.
  He has a family member (through an attorney) years later make an inquiry if the evidence is still in storage. Murder cases are supposed to have evidence kept for a very long time (if not forever) in my state. They are told yes. Attorney gets innocence project involved. A few months later the innocence project requests evidence, and is told sorry, it was 'lost' in a move between labs, never to be found again.
  This man did more than two decades before mandatory release. For a crime he possibly did not commit. He is not the only one with stories like this.
  Find this video from 2006 and ask has it gotten any better? http://www.cnn.com/CNN/Program...
  "The degree of civilisation in a society is revealed by entering its prisons." -Fyodor Dostoyevsky
  "If you want to see the scum of the earth, go to any prison - at shift change" - Paul Harvey
  
  --
  Silence is a state of mime.
13. Re:No CSI by Anonymous Coward · 2017-02-07 12:55 · Score: 0
  
  Even using it to narrow the search might be counter productive. Look at the example photos in the article. It turned a white lady into an asian guy in the first image and a guy who looks like Nic Cage into Rush Limbaugh in the third image.
14. Re: No CSI by Anonymous Coward · 2017-02-07 12:58 · Score: 0
  
  Google already has a copy of every image ever posted online. Of course Google can "enhance" a pixelated image by matching it against the original which Google already has.
15. Re:No CSI by Anonymous Coward · 2017-02-07 13:01 · Score: 0
  
  Perhaps the purpose of prisons is to give sadists and sociopaths a place to have a job, so they don't become DMV clerks and middle school gym teachers.
16. Re:No CSI by aiht · 2017-02-07 14:10 · Score: 1
  
  Yeah, no. If you want to zoom in on a tiny blur that you know is a face, this tech will give you just that: A face. A different face that it found in a different (higher-res) photo. That can't possibly ever do anything other than mislead.
17. Re:No CSI by rmdingler · 2017-02-07 14:17 · Score: 1
  
  So, the Navy goes to Great Lakes Naval Training Center for incarceration... were you also Navy or did you get your mail at the USDB?
  
  --
  Happiness in intelligent people is the rarest thing I know.
  Ernest Hemingway
18. Re:No CSI by Anonymous Coward · 2017-02-07 14:28 · Score: 0
  
  I couldn't agree more. While one might be able to guess with a fair degree of accuracy, it's still a guess.
  Much like DNA.
19. Re:No CSI by mmell · 2017-02-07 14:45 · Score: 2
  
  That's part of the problem though, isn't it? We all root for the protagonists on CSI, Criminal Minds, Law & Order, and at least half a dozen other police dramas when they routinely use some mystical all-knowing government database or "hack" into government and private (corporate) databases and nail the bad guy beyond all shadow of doubt. Many of us would groan when the protagonists would "hack" into live bank ATM security cameras and enhance the medallion number of a cab half a block down from a dozen greyscale pixels (my own wife constantly reminds me "it's only television" when I point out how ridiculous it is that they can create data out of the ether).
  So - will this algorithm always come up with medallion #1313, because that's what it "learned" a medallion number is? When the system "guesses" at the missing data based on what it "learned" during "training", will it always identify Christy Brinkley and Chuck Norris as the perpetrators? Some of this is valid - I can often read information which is blurry or even pixelated, but not always. Intuitively differentiating an 8 from a 0 based on how dark that few pixels are is different from proving that it was an 8 or a 0.
  One other concern - in the past, enhanced video has been useful in court, when there was someone able to explain exactly what transformations were done, someone prepared to prove why the enhanced video is still representative of the portrayed reality. Learning systems may often produce surprisingly accurate results from seemingly inadequate data, but it's considerably harder to prove that the enhanced information accurately portrays reality - perhaps impossible, given the complexity of such systems.
20. Re:No CSI by toastjam · 2017-02-07 15:09 · Score: 1
  
  You're misinterpreting the result there. NN stands for "nearest neighbor". It's just the closest match in the DB (I guess the one they used to train?), not what their algorithm produced.
21. Re:No CSI by msauve · 2017-02-07 15:22 · Score: 1
  
  But, Google.
  
  Who are you to say that something from nothing isn't possible? :)
  
  --
  "National Security is the chief cause of national insecurity." - Celine's First Law
22. Re:No CSI by RogueWarrior65 · 2017-02-07 15:29 · Score: 1
  
  It doesn't matter if law enforcement thinks it can solve crimes. What matters is if lawyers think juries will believe the results. Juries think that CSI is real and forensics can all be accomplished in less than an hour.
23. Re:No CSI by Anonymous Coward · 2017-02-07 15:42 · Score: 0
  
  Of course it's possible!
24. Re:No CSI by skids · 2017-02-07 16:32 · Score: 1
  
  Because it is not image enhancement, it's image embellishment. It would quite often "narrow" the search onto the wrong track whenever it encountered an atypical situation, and in court it could be claimed that its results were only a convenient pretense, not a probable cause for suspicion (an authority could shop multiple such competing services until one hit on someone they wanted to harass.) The shit would really hit the fan when the algorithm was inevitably shown to have some significant accidental bias towards a categorizable subpopulation.
  
  --
  Someone had to do it.
25. Re:No CSI by Rei · 2017-02-07 22:05 · Score: 1
  
  It's fine just as it is for a lot of non-legally-questionable uses, such as image compression (smaller number of real pixels to achieve the same aesthetic effect), upscaling old images (picture a high quality upscaler in an emulator for old games, for example), etc.
  By the way, this appears to be basically the same thing that Magic Pony Technology already did. Another thing that MPT demonstrated with their software was infinite texture generation - given a fixed size texture, it can create an infinite amount more that "looks like it". Also, a neat thing was its use on images that contained text. If the neural net recognizes a letter, then it can obviously draw the letter in full, clear detail. I've had quite a few PDFs that were exported at too low of a resolution for their images that I would have liked that on; it's no fun having to zoom in and tease out the letters one-by-one manually.
  
  --
  Next to my desk we have an Ire Extinguisher. Our boss is really assertive, so we like the idea of having it.
26. Re:No CSI by Rei · 2017-02-07 22:07 · Score: 1
  
  Indeed. It's equivalent to asking an artist paint a face onto a blur.
  
  --
  Next to my desk we have an Ire Extinguisher. Our boss is really assertive, so we like the idea of having it.
27. Re:No CSI by aod7br · 2017-02-08 00:04 · Score: 1
  
  I could be used for law enforcement, not to confirm a candidate, but to discard suspects that don't match.
28. Re:No CSI by AmiMoJo · 2017-02-08 01:01 · Score: 1
  
  We all root for the protagonists on CSI, Criminal Minds, Law & Order
  I don't. Especially those bastards on CSI. "Why don't you take a DNA test, rule yourself out of the investigation?" Fuck off.
  
  --
  const int one = 65536; (Silvermoon, Texture.cs)
  SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
29. Re:No CSI by parkinglot777 · 2017-02-08 03:17 · Score: 1
  
  Law enforcement would never rely upon unproven methods to improve conviction rates.
  That are good points. Law enforcement are supposed to use each of those evidence as leads. However, they tend to fixate on the what fits most to the case instead of thoroughly check out those leads. Once they have a suspect, they will try to fit as much evidence to the suspect as they can. They also eliminate the parts that don't fit (or not admit them to the court). As a result, it causes an innocent to be convicted.
  I don't know why they want to close the case ASAP. Possibly, it is like a trophy for how many cases they can solve, so that they would get promoted? I have no idea...
30. Re:No CSI by Script+Cat · 2017-02-08 04:13 · Score: 1
  
  The algorithm can paint anything it wants into those squares. Want a video of Donald Trump getting peed on? Just film a low res video and have the algorithm paint his face into the place of the actors.
31. Re:No CSI by Anonymous Coward · 2017-02-08 05:15 · Score: 0
  
  Finally, the tech to bring to life the stereotypical CSI scene:
  Hard-bitten Detective: We need more evidence! Is there video?
  Geeky Tech: Yes, here it is! I instantaneously found the relevant 5 seconds of footage from among a network of 50 cameras and over 2000 hours of recordings!
  Hard-bitten Detective: I can't see the perp's face. Enhance!
  Geeky Tech: (pushes one button). Done!
  Hard-bitten Detective: Now we've got the SOB! You're going down you piece of filth!!
32. Re:No CSI by Anonymous Coward · 2017-02-08 06:28 · Score: 0
  
  Ah, but they will, because, of course, they are idiots. Fortunately, since most of the training is on images that already appear on the internet, which is heavily weighted toward celebrities, once the police zoom in on the hallucinated suspect, they will end up arresting George Clooney or Johnny Depp.
33. Re:No CSI by PPH · 2017-02-08 10:09 · Score: 1
  
  Now we've got the SOB!
  Just hope that the training data set used to tune the enhancement software doesn't include your face.
  
  --
  Have gnu, will travel.
34. Re:No CSI by wbr1 · 2017-02-08 23:23 · Score: 1
  
  Not all crimes are prosecuted by the military. Sometimes the local or state authorities do, often with a discharge or additional military justice. I was never in the militar, but knew several in state prison that were at the time of their arrest.
  
  --
  Silence is a state of mime.
Bedroom images? by Anonymous Coward · 2017-02-07 10:53 · Score: 0

Improving on low-res porn?
1. Re: Bedroom images? by Anonymous Coward · 2017-02-07 11:09 · Score: 0
  
  You really need to see the actresses' zits?
2. Re: Bedroom images? by taustin · 2017-02-07 11:19 · Score: 1
  
  I'd setting for even low res porn that didn't have all those long, loving close ups of the guy's hairy ass and shaved balls. Is all porn made by amateurs these days, and they have their gay buddy run the camera cuz he's not interested in the hot chick they hired?
3. Re: Bedroom images? by Anonymous Coward · 2017-02-07 12:05 · Score: 0
  
  Amateur porn is hot for a reason. We want to see people who have fun doing it. Not professionals who are basically emotionless. Or, actually acting. Nothing kills a boner like a whore hamming it up because she thinks she is an actress now because lots of porn was made with her.
4. Re: Bedroom images? by taustin · 2017-02-07 14:12 · Score: 1
  
  Yeah, I'm still not interested in close-ups of some guy's hairy ass and shaved balls.
5. Re: Bedroom images? by Anonymous Coward · 2017-02-08 05:08 · Score: 0
  
  Would you be interested in a shaved ass and hairy balls? It's porn, there's no combination of fetishes someone hasn't put on film, sometime, somewhere!
6. Re: Bedroom images? by virtuosonic · 2017-02-08 15:08 · Score: 0
  
  I think they call it female friendly porn... Trump will make porn great again!
  
  --
  http://agender.sourceforge.net/ get a free schedule tool
Blade Runner-esque? by threadsafe · 2017-02-07 10:55 · Score: 2

Is this Blade Runner-esque? Decker summoned some wicked camera technology. Don't bother me with those pesky limits to the physical laws.
1. Re:Blade Runner-esque? by threadsafe · 2017-02-07 10:58 · Score: 4, Funny
  
  I'm actually a dog on the internet because I didn't proof or spell check my previous reply.
2. Re:Blade Runner-esque? by FatdogHaiku · 2017-02-07 12:52 · Score: 4, Funny
  
  I'm actually a dog on the internet because I didn't proof or spell check my previous reply.
  Just paws before you hit Submit...
  
  --
  You have the right to remain sentient. If you give up the right to remain sentient, you will be elected to public office
3. Re:Blade Runner-esque? by Anonymous Coward · 2017-02-07 13:00 · Score: 0
  
  should have used Pee-view
4. Re:Blade Runner-esque? by uvajed_ekil · 2017-02-07 18:21 · Score: 1
  
  *Deckard
  But yes, my thoughts immediately jumped to Bladerunner upon reading the title.
  
  --
  This is a hacked account, for which the owner can not be held responsible.
The future is here by Anonymous Coward · 2017-02-07 10:56 · Score: 0

The future is here, and it's lame as fuck! Now we can use voice commands and advanced AI to yell "Enhance!!" at our screens.
Stupid by Anonymous Coward · 2017-02-07 10:56 · Score: 1

This is just pure, 100% guesswork done with a computer. You cannot 'enhance' information that simply does not exist. The 10% of "fooled" people just mean those people were not familiar enough with what that celebrity actually looks like to tell the difference.
CSI "enhance it!" remains fiction, sorry.
1. Re:Stupid by ceoyoyo · 2017-02-07 11:24 · Score: 3, Informative
  
  I see you still don't actually know what a Fourier transform is.
2. Re:Stupid by Anonymous Coward · 2017-02-07 11:29 · Score: 0
  
  But it IS guesswork. Every time a round of blurriness is added to to a photo, you lose a round of detail. The more rounds of blurriness that are added, the less likely what is restored will match the original.
3. Re:Stupid by Anonymous Coward · 2017-02-07 12:07 · Score: 1
  
  You'd think the name would be enough for you, but apparently it isn't. It's called the Fourier TRANSFORM. It transforms data. It does not create it. Specifically it's used to take what is commonly called "time domain" representations, and TRANSFORMS them in to "frequency domain" representations. If you want to get in this argument I will, and you will be proven a fool.
4. Re:Stupid by Anonymous Coward · 2017-02-07 12:33 · Score: 0
  
  Come on man, you just destroyed your street cred. Now all you are is a old liberal cus. FFTs do not work like that at all.
5. Re:Stupid by Anonymous Coward · 2017-02-07 13:15 · Score: 0
  
  100% guesswork done with a computer. You cannot 'enhance' information that simply does not exist.
  Of course you can. Scientists use the Fourier transform equation to do this every day. It's not "guesswork".
  <YODA>The stupid is strong in this one</YODA>
6. Re:Stupid by PopeRatzo · 2017-02-07 15:28 · Score: 1
  
  I see you still don't actually know what a Fourier transform is.
  What is anti-aliasing if not providing data that's not actually there?
  http://www.sci.utah.edu/~csche...
  When used with the BBM equation, it can do predictive error-correcting. I know this because I am friends with both of the "B"s in BBM.
  
  --
  You are welcome on my lawn.
7. Re:Stupid by Anonymous Coward · 2017-02-07 15:37 · Score: 0
  
  It's one thing to shoot the shit on slashdot, speculate about supercool tech stuff and live the life of that nerdy guy who's so smart he works with computers.
  But it's totally another when you enter the realm of math and engineering. You're not tall enough for that ride.
  You will get smacked down if you try to feign knowledge in those domains because those of us who know our shit have zero tolerance for bullshit
  Answer: No, once information has been lost it's gone. No amount of time or frequency domain analysis will restore it.
8. Re:Stupid by PopeRatzo · 2017-02-07 15:45 · Score: 3, Interesting
  
  You will get smacked down if you try to feign knowledge in those domains because those of us who know our shit have zero tolerance for bullshit
  
  No, once information has been lost it's gone. No amount of time or frequency domain analysis will restore it.
  https://arxiv.org/pdf/1101.007...
  
  --
  You are welcome on my lawn.
9. Re:Stupid by Oligonicella · 2017-02-07 16:02 · Score: 1
  
  What is anti-aliasing if not providing data that's not actually there?
  It's providing data not recovering data.
10. Re:Stupid by PopeRatzo · 2017-02-07 16:36 · Score: 1
  
  It's providing data not recovering data.
  That's my point. Data "provided" by probability is still data. Otherwise, noise reduction wouldn't work.
  Nothing in the article said anything about data recovery. The whole, "once data is lost it's lost" argument is a red herring in this context. It's based on a concept of "data" from computer science rather than from actual science. ; )
  
  --
  You are welcome on my lawn.
11. Re:Stupid by Anonymous Coward · 2017-02-07 21:22 · Score: 1
  
  Please point out EXACTLY where you see the claim that the method described "recovers data" or "enhances information".
  This approach simply provides a prior probability distribution over a specific image domain, which IS useful. Even though it may attain the claim (made up by you) of being capable of "recovering data", it lets you sample from the space of "plausible high-resolution images in a specific domain, given a low-resolution image of it".
  Going from a low-resolution to a high-resolution image is always going to take some measure of "guessing" (since decimation is obviously not bijective). The difference is whether you use a naive method (e.g. bilinear interpolation) that will give you shit results OR a method that actually gives you sharp, plausible results.
12. Re:Stupid by elgatozorbas · 2017-02-07 22:55 · Score: 2
  
  Anti-aliasing is not "providing data that is not actually there", anti-aliasing is removing data that is actually not there, more specifically high-frequency aliases of the low-frequency information. Hence the name anti-aliasing.
  In practice, this means that the best any algorithm can do to improve a (general) blocky image, is smooth the edges of the blocks. What this algorithm apparently does, as far as I understand, is just correlate the blocky image to a library of known facial features. This has nothing to do with FFT or anti-aliasing whatsoever.
  Apart from that, unless your name is "M", I am afraid
  
  I know this because I am friends with both of the "B"s in BBM.
  doesn't really mean anything.
13. Re:Stupid by Anonymous Coward · 2017-02-08 00:40 · Score: 0
  
  No. Pigeonhole principle. There are more high resolution pictures than there are low resolution pictures, thus multiple high resolution pictures must map to the same low resolution pictures. There is no way to tell which was the original source of the low resolution picture or which scene resulted in both one of the high resolution pictures and the low resolution picture. It's guesswork. You can make a high resolution images that depict something which would result in the low resolution picture, but you can make an astronomical number of those and which one you choose is guesswork.
14. Re:Stupid by AmiMoJo · 2017-02-08 01:19 · Score: 2
  
  A Fourier transform takes a signal and converts it into a bunch of sine waves that when combined will reproduce the original signal.
  A signal is just varying amplitude over time. A sine wave is a signal where the amplitude is sin(t), where t is time. It turns out that all signals can be constructed by combining sine waves. The more sine waves you have the close to the original signal you can get.
  Why it is useful to convert a signal to sine waves? Say you decide you are going to use 20,000 sine waves, the first one a 1Hz wave, the second one a 2Hz wave, 3Hz, 4Hz all the way up to 20,000Hz. You use a Fourier transform to convert the signal to those sine waves. Now you increase the amplitude of the low frequency sine waves, say 200Hz and below. Now do another Fourier transform to convert from sine waves back into a signal. Congratulations, you just pumped up the bass on your music.
  In technical terms the signal is in the time domain, and the sine waves are in the frequency domain.
  Disclaimer: This is a simplification, it's more complex than this but without getting into the heavy maths of it, this is basically what's going on.
  
  --
  const int one = 65536; (Silvermoon, Texture.cs)
  SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
15. Re:Stupid by Script+Cat · 2017-02-08 04:25 · Score: 0
  
  FFT's so instead of a bunch of squares it's a bunch of blurry squares. What the technology does is paint in arbitrary face data into the squares. Or dogs there are dogs everywhere! Bbbbooooowwww OWOwwwwwwwwwww......~~~*8,.9()090...
  http://i.vimeocdn.com/video/52...
16. Re: Stupid by Anonymous Coward · 2017-02-08 05:57 · Score: 0
  
  I don't get it. Why not just use the higher resolution photos in the first place. If all high res photos have a low res copy, as you suggest.
17. Re:Stupid by david_thornley · 2017-02-08 07:22 · Score: 1
  
  When I have my glasses off, I can recognize things my eyes can't resolve. Human stereo vision in relative depth perception depends on measurements that are more precise than the layout of cone cells would suggest. I can reliably get information that I can't see. Why can't a computer?
  
  --
  "When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
18. Re:Stupid by Anonymous Coward · 2017-02-08 07:39 · Score: 0
  
  that is a super cool paper that doesn't really dispute the claim "Once information has been lost it's gone". In this case, the information is still present in the image that is processed to find a statistically probable original image.
  it's super cool because it's got ghosts and Lena in it.
19. Re: Stupid by ChrisMaple · 2017-02-08 18:46 · Score: 1
  
  We don't use the higher resolution photo in the first place because we don't know which high resolution photo is the proper one. All that we're sure of is that we've got a crappy low resolution picture.
  Consider the extreme case: we've got a picture in which a person's head fills one pixel. It's a black and white photo, and the value of that pixel is 128 (on a scale of 0-255). Now, from that, determine which photo, from all photos taken over the last 200 years, best matches that one pixel. Have fun!
  
  --
  Contribute to civilization: ari.aynrand.org/donate
20. Re:Stupid by ChrisMaple · 2017-02-08 19:08 · Score: 1
  
  There are several things that help humans.One is that you can take several perceptions over the course of a fraction of a second and unconsciously merge them in a manner that improves resolution. Another is that humans can do pattern recognition of things buried in noise that a computer can't do unless specifically programmed to do that specific thing. (Think of resolving a head of hair in in a dim corner in a dim room.) Another is that the brain just makes up stuff (read about the blind spot.) You may be picking up clues that you're not conscious of, and fitting them into patterns that you're not conscious of, either.
  You've got a lifetime of experience using a highly developed adaptive imaging system. Computers have to be told explicitly how to accommodate all sorts of corrupting factors that you do automatically, like shadows, tilt, lens distortion, blur, lighting levels, lighting colors, makeup, bruises, 5-o'clock shadow, etc.. It's a lot of stuff to compensate for, it all has to be done and combined properly.
  
  --
  Contribute to civilization: ari.aynrand.org/donate
21. Re:Stupid by david_thornley · 2017-02-09 04:46 · Score: 1
  
  Sure. In many ways, I'm a far better computer than has ever been built in silicon. That doesn't mean that what I do can't be partially replicated in a powerful computer. I'm using myself as a proof of concept, that it is possible to beat the resolution, not claiming that any given system is or is not able to do the same.
  
  --
  "When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
Ideal test case by fibonacci8 · 2017-02-07 10:57 · Score: 3, Interesting

Feed it minecraft screenshots and japanese porn, and see what the result is.

--
Inheritance is the sincerest form of nepotism.
1. Re:Ideal test case by Anonymous Coward · 2017-02-07 11:04 · Score: 0
  
  Pixellated tentacles?
2. Re:Ideal test case by Somebody+Is+Using+My · 2017-02-07 11:09 · Score: 1
  
  Feed it minecraft screenshots and japanese porn, and see what the result is.
  I was thinking Super Mario Brothers sprites myself. Those were 8x8 sprites, if I remember correctly.
  As the article points out, this is less, "Zoom, Enhance" and more "Best Guess". Unfortunately, years of bad computer science on TV is just going to confuse people into believing this algorithm can do far more than it actually does. I wish Google would open it up so we could test it with our own images and show others how untrustworthy these programs really are.
  More useful are the programs which enhance an image from a video based on data from surrounding frames. I wouldn't be surprised if we start seeing fast-shooting CCDs coupled with similar "enhancement" algorithms built into cameras eventually; you take one picture, the CCD takes five or ten, and then in the background builds a higher-resolution image than the sensor itself would normally allow.
  On the other hand, while Google's software isn't itself that useful, it does show how quickly computers are advancing in recognizing images, which is probably the whole point.
3. Re:Ideal test case by Anonymous Coward · 2017-02-07 20:16 · Score: 0
  
  I was thinking Super Mario Brothers sprites myself. Those were 8x8 sprites, if I remember correctly.
  They were stored as 8x8 pixel sprites, but small Mario was made up of four such sprites, and big Mario eight. See image here: https://nesdoug.com/2015/11/22/6-sprites/
4. Re:Ideal test case by Djoulihen · 2017-02-07 21:02 · Score: 2
  
  Well, if you are into that kind of stuff (neural network generated kind-of porn but not really), I suggest you check this out: https://open_nsfw.gitlab.io/ (Warning: somehow NSFW, or just weird, I don't know, but it will give you nightmares).
It's about time! by npslider · 2017-02-07 10:59 · Score: 4, Funny

So in other words... from a small picture of the earth viewed from orbit, Google can now show me my house AND the address on the UPS package sitting at my doorstep?
Amazing!
1. Re:It's about time! by mmell · 2017-02-07 14:54 · Score: 2
  
  Yes, they can! And the flip-side of this technology is that you can zip a 500GB database down to 200GB, zip that down to 700MB, zip that down to a few hundred KB . . . when it's 1B, you can embed it on a carbon atom on my DNA and I can carry it back to Kronos, for the glory of the Empire!
2. Re:It's about time! by Baron_Yam · 2017-02-08 02:22 · Score: 1
  
  There was once a company trying to sell infinite compression technology.
  Information theorists chuckled knowingly, and of course the 'inventors' failed miserably to decompress their compressed data... but at some time somebody actually thought it was possible and tried to bring it to market.
3. Re:It's about time! by StormReaver · 2017-02-08 02:24 · Score: 1
  
  So in other words... from a small picture of the earth viewed from orbit....
  Only if Google already has a high resolution picture of your house with the UPS package already sitting on your doorstep.
  The algorithm does NOT enhance pixelated images in any meaningful way. It is only able to match a pixelated image with an already existing high resolution image of the same thing, and only by scaling the high resolution image down to a pixelated form suitable for comparison with the existing pixelated image.
  The only thing that makes this at all interesting is that pattern matching algorithms are a bit better than they were in the 80's when every teenage programmer with a home computer thought of the same thing. It has been an introductory programming exercise in computer graphics for decades.
4. Re:It's about time! by npslider · 2017-02-08 05:01 · Score: 1
  
  Black Hole Inc.?
5. Re:It's about time! by npslider · 2017-02-08 05:08 · Score: 1
  
  "I can see my house from here!" ;)
  (Simulation of Kronos, the Klingon capital city. Star Trek Enterprise Season 1 Episode 05 - Unexpected)
6. Re:It's about time! by DEVALEV · 2017-02-08 05:28 · Score: 1
  
  Kill Joy...
7. Re:It's about time! by david_thornley · 2017-02-08 07:26 · Score: 1
  
  A friend of mine was involved in a vendor demo for compression to reduce the number of CD-ROMs needed to contain software. He insisted that the information was already compressed, and everybody was ignoring him until the vendor's software increased the number of CD-ROMs.
  
  --
  "When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
8. Re:It's about time! by david_thornley · 2017-02-08 07:30 · Score: 1
  
  That's easy. Here's a compressed version of the Library of Congerss: 1. I'm still working on the decompression algorithm. That will take a little more time.
  Back when Bruce Schneier facts were a meme, one claimed that, when Bruce Schneier wanted to write a book, he generated a random string of the correct size and decrypted it.
  
  --
  "When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
9. Re:It's about time! by Anonymous Coward · 2017-02-08 12:01 · Score: 0
  
  I've figured out the decompression algorithm and I was able to encrypt the data. I can compress and encrypt anything down to a single bit. For security, you need to have your password be the same as the data as that way we can verify that it's your data before decrypting and uncompressing it.
10. Re:It's about time! by ChrisMaple · 2017-02-08 19:11 · Score: 1
  
  I use an indexed table. The first entry is the Library of Congress.
  
  --
  Contribute to civilization: ari.aynrand.org/donate
11. Re:It's about time! by ChrisMaple · 2017-02-08 19:24 · Score: 1
  
  "I am Chronos, God of Time" (Johnny Bravo - Bearly Enough Time - 1997)
  
  --
  Contribute to civilization: ari.aynrand.org/donate
What about zoom out ? by alexhs · 2017-02-07 11:01 · Score: 3, Funny

If it doesn't do uncrop, it's lame.

--
I have discovered a truly marvelous proof of killer sig, which this margin is too narrow to contain.
1. Re:What about zoom out ? by K.+S.+Kyosuke · 2017-02-07 11:06 · Score: 5, Interesting
  
  Already done. ;)
  
  --
  Ezekiel 23:20
Fake Hot Chicks by Anonymous Coward · 2017-02-07 11:02 · Score: 0

So if I go to a bar fully of ugly chicks, I'll no longer have to become drunk to think I'm talking to/screwing $FAMOUS_HOT_CHICK? This can be even better than beer. Shut up and take my money!
Obligatory by edx93 · 2017-02-07 11:04 · Score: 1

Sorry, but the VT state troopers already have this technology: https://www.youtube.com/watch?...
"...it was quite successful against humans..." by turkeydance · 2017-02-07 11:06 · Score: 3, Funny

now we know the perception and purpose,
Biased selection? by Anonymous Coward · 2017-02-07 11:08 · Score: 1

So Google fed its algorithm millions of high-resolution images. It then feeds a pixellated image of a celebrity, and AMAZING! The algorithm can create an image of that celebrity so detailed that it fools people.
Of course, chances are a good portion of the source material is pictures of that celebrity. This would be far more impressive if it could construct a police artist profile image from a pixellated source. One that could be reliably matched to an individual that's not even in the massive collection of images fed into the source.
1. Re:Biased selection? by omnichad · 2017-02-07 12:10 · Score: 1
  
  When you think about what they'll probably do with it, this makes perfect sense. If you've ever used Google Images' "Visually Similar" feature, it doesn't work all that well - especially when you're trying to find a higher resolution image.
  If they can do a pseudo-reconstruction, they're partway there.
PixelCNN? by Anonymous Coward · 2017-02-07 11:10 · Score: 0

PixelCNN?
Those are Fake Pixels!
no need to photoshop by zlives · 2017-02-07 11:12 · Score: 2

this is perfect for your "real" profile pic on dating sites, just upload a google enhanced image of your self created from your 8x8 pixel image. yes this celebrity is really me.
They've been watching too much TV by taustin · 2017-02-07 11:14 · Score: 4, Insightful

TV Detective: "We have this security video showing the murder."
TV Lab Rat: "It's too grainy to tell how is it?"
TV Detective: "Can't you enhance it?"
TV Lab Rat: "Sure. Who do you want it to look like?"
Nothing new here by Anonymous Coward · 2017-02-07 11:23 · Score: 0

This technology has been available to the FBI in the movies for years.
Nothing new by ripperapid · 2017-02-07 11:23 · Score: 1

This is nothing new. They have been doing this in movies for ages.
Really impressive progress by PhrostyMcByte · 2017-02-07 11:23 · Score: 3, Informative

A while ago, someone made the nnedi upsampler that uses neural networks to upsample. It's still one of the best image upsamplers available.
Google's approach is quite a bit different. Where nnedi worked to better extract detail out of what was already in the image, Google seems to literally fill in detail that was probably in the source but maybe not. Much, I guess, like how our own memories work. It's an interesting approach and the results look quite fantastic. My only question is how well it will work on a random sampling.
move over, Mount Rushmore by epine · 2017-02-07 11:31 · Score: 1

Move over, Mount Rushmore, Google now has an algorithm that can wallpaper the Ceres asteroid with the face of every American who has ever been photographed—all the way back to a pinhole camera exposing an onion skin soaked in lemon juice and potato starch.
1. Re:move over, Mount Rushmore by Tablizer · 2017-02-07 11:52 · Score: 1
  
  Google now has an algorithm that can wallpaper the Ceres asteroid with the face of every American who has ever been photographed
  Was that intended as a joke? Google already has such technology
  
  --
  Table-ized A.I.
Shannon turning in his grave by dv82 · 2017-02-07 11:34 · Score: 1

The only reason this works as well as it does (poorly - 10% to 28% according to the post) is that humans are hard-wired to find faces among images. We'll accept anything with two small ovals on the same line with another shape roughly in the middle, below. Nose optional. Try the equivalent with random images or music, and I'll bet the success rate would drop to nil.
FIF was a completely different approach. by safetyinnumbers · 2017-02-07 11:39 · Score: 1

Fractal Image Format was compression that used self-similar parts of the same image to decopress to higher resolution https://en.wikipedia.org/wiki/...
Fractals by thinkwaitfast · 2017-02-07 11:43 · Score: 1

In the 1980's people were trying to this with fractals. I remember reading about something very similar and wish I could have afforded the book. I wonder what google is doing. Probably brute force.
I was really interested n fractals and fractal compression for a time. And while you could get some insanely high compression ratios, the technique was lossy and decompression took 50+ hours (and 2 hours to compress). A potential use of the technology was picking out interesting artifacts from low rez space ohotos.
1. Re:Fractals by thinkwaitfast · 2017-02-07 12:06 · Score: 1
  
  FRACTAL COMPRESSION AND ANALYSIS ON REMOTELY SENSED IMAGERY: An advanced Compression and Analysis Method for Remote Sensing Images
2. Re:Fractals by Anonymous Coward · 2017-02-08 05:49 · Score: 0
  
  AlienSkin Blowup
Bad summary by swillden · 2017-02-07 11:51 · Score: 5, Informative

The summary's explanation of what this does isn't correct. It says:

Google's new software effectively extracts details from a few source pixels to enhance pixelated images.
It doesn't extract details from a few source pixels. It invents details to add to those source pixels, based on the knowledge that the pixelated image is of a face, and of what faces look like. It produces something that plausibly fits the input data. How close this is to the original image, pre-pixelation, depends on what images were in its training set.
This is an interesting piece of work, but it doesn't mean that you can recover data that has been discarded.

--
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
1. Re:Bad summary by mrclevesque · 2017-02-07 12:20 · Score: 1
  
  ^^ this
2. Re: Bad summary by Anonymous Coward · 2017-02-07 13:32 · Score: 0
  
  But... can't you? At least in the eye of a regular Joe? Sure I understand you're technically correct but for the average Joe, recovering an average image (by definition what this tool should excel at) it will seem like it's recovering missing data. Which is how we'll get into misapplication problems by law enforcement, as other commenters have noted.
3. Re: Bad summary by belthize · 2017-02-07 13:36 · Score: 1
  
  It's trained on plausible images, so it's going to produce plausible images but there's nothing that says that the resulting plausible image is a reasonable reproduction of the original, only that it's good enough to make a human think 'yep, that's another human'.
4. Re:Bad summary by Anonymous Coward · 2017-02-08 00:21 · Score: 0
  
  This is an interesting piece of work, but it doesn't mean that you can recover data that has been discarded.
  Very much this.
  The only extra piece of information you can ever get from a camera is the sensor model and lens arrangement.
  If you can get those, you can make a pseudo multi-focus lens from that and the data, and extract a tiny bit more detail from the overlap of EM across several nearby pixels.
  This technique is being used to make lensless camera sensors better, which will allow for paper-thin cameras. (phones to come after! except Samsung!)
  It has a slight extra cost in making the sensor since you need to know it has a pretty perfect yield for all the pixels, or have a map of all the dead pixels at least, then you can patch it over with firmware per-sensor.
  But the lack of a complicated lens system more than makes up for that extra cost.
  They've been currently pushing the resolution of the system up higher.
  The system is basically similar to how eyes evolved in the first place.
  Photosensitive cells over bodies, figuring out how to create pictures out of the noise which led to those ones surviving over the ones that failed to, then some evolved mutations with trapped bubbles, making the image clearer, and iterating onwards to more complex forms. (we can even do this with the human tongue and an array of electrodes connected to a similar simple retina, it's being actively researched for blind people, military and divers, only takes a little while of training for it to work)
  But there are still plenty of animals that use the lensless sight system. Due to evolution tending towards lowest energy, they've not massively improved over the last billion odd years seemingly.
  We don't have as restrictive a limit as living creatures do since we are talking cameras with high-energy batteries, so we can do far more advanced operations to the data.
  On a related note, it is quite similar to how people say that the brain will never be beaten, they are wrong.
  The brain is SHIT, it is hugely redundant and "low"-energy. It only has us beat on connectivity, the energy requirements are NOT an advantage! In fact, our huge bodies are also a huge disadvantage to us. When will we evolve in to Hobbits already?!
  Once we overtake that connectivity, evolution is going to go crying home to its momma. Fricken nerd.
  Plasticity can be emulated in software at faster speeds than the brain can do it with synapses, given at least triple the connectivity a typical brain has. This is based on the most recent and huge change to how many synapses are actually in the brain, from those deep scans we did a couple years back that shows how highly organized the brain is, instead of a mess of synapses like previously thought.
  All of this depends on us getting to graphene-based computing, when we do we'll be more than capable of reaching that.
  Optical computing will be even more dense since we can make a cube with all the computing on the outsides and the middle space as the interconnects and virtual logic using a great many frequencies to connect said cube. The Borg were right all along! In fact, optical will likely come sooner than expected since researchers have created vastly superior photodetectors in the last decade, which is needed for the switch from lightelectrical to ever be feasible.
Think video by Doub · 2017-02-07 12:08 · Score: 1

On a single image the additional detail is guess work, the information isn't there. But think multiple slightly different but related images of the same subject, like any video filmed in the past 100+ years. Now you can cross-check and improve guesses with time correlation, and really find things that are there in terms of information, but invisible to the eye because each bit is spread over multiple frames.
They do that in movies all the time by OneHundredAndTen · 2017-02-07 12:26 · Score: 2

Does this imply that movies have been lying to us all along? :-)
1. Re:They do that in movies all the time by Anonymous Coward · 2017-02-07 12:36 · Score: 0
  
  yes. Next question.
2. Re:They do that in movies all the time by iggymanz · 2017-02-07 18:09 · Score: 1
  
  nope it's legit. really good systems can even uncrop
  https://www.youtube.com/watch?...
Google Invents by Anonymous Coward · 2017-02-07 12:37 · Score: 0

technology that already existed, news at 11!
pixelate photos by p51d007 · 2017-02-07 12:40 · Score: 1

When I post photos that have personal information, license plates on cars, house numbers, names etc, I will do a two fold thing. Use the lens blur to blur them out, then go back over them with a masking color. Pretty simple in photoshop.
1. Re:pixelate photos by Anonymous Coward · 2017-02-07 13:06 · Score: 0
  
  Just black rectangle, the way censorship was done last century. Nazis were smarter than you, idiot.
What do you mean, ages? by Cyberpunk+Reality · 2017-02-07 12:54 · Score: 1

Movies won't be able to do this for two more years!

--
Rule 35 of the internet: "If it can be hacked, it will be". - Charles Stross
Blade Runner by Anonymous Coward · 2017-02-07 13:13 · Score: 0

Once again, Scifi foretells the future.
1. Re:Blade Runner by uvajed_ekil · 2017-02-07 18:28 · Score: 1
  
  Once again, Scifi foretells the future.
  Good thing Bladerunner has gotten almost everything else wrong so far. At the very least it had the timing all wrong, since we're pretty unlikely to have replicants, off-world colonies, lots of flying cars, an inability to buy real snakes, or that level of smog or whatever choking LA by 2019.
  
  --
  This is a hacked account, for which the owner can not be held responsible.
2. Re:Blade Runner by Anonymous Coward · 2017-02-08 03:05 · Score: 0
  
  Are you sure? We got two more years and Trump is in the Whitehouse, I think there is a good change that LA could look like that two years into the Trump presidency.
3. Re:Blade Runner by Anonymous Coward · 2017-02-08 08:13 · Score: 0
  
  Give it time. AI and replicants are coming.. all the pieces are starting to fall into place.
  Flying cars, in cities, may be here soon too.
4. Re:Blade Runner by Lotus456 · 2017-02-10 10:58 · Score: 1
  
  Thin 80s ties, constant rain in L.A., video payphones, "Cityspeak" lingua franca, advertising blimps...
  Some other stuff BR got right: Atari still in business, guns with LEDs, giant animated billboards, and revival of analog synthesizers. ^_^
  
  --
  "It's a good computer... for I to BM on!" - apologies to Triumph, the insult comic dog
Want to try one of these algorithms out? by Anonymous Coward · 2017-02-07 13:21 · Score: 0

http://waifu2x.udp.jp/ has a working one that you can try. It works *very* well for upscaling things within its particular artistic style.
Finally! by Anonymous Coward · 2017-02-07 13:36 · Score: 0

...time to depixelate all those Japanese porn flicks
1. Re:Finally! by Anonymous Coward · 2017-02-12 18:42 · Score: 0
  
  I fucked a Japanese girl once and her pussy was actually pixilated.
Come on Google by CODiNE · 2017-02-07 13:58 · Score: 3, Insightful

At least make the world a better place...
by making this a MAME scaler.

--
Cwm, fjord-bank glyphs vext quiz
Pixelated Images by b783719 · 2017-02-07 13:58 · Score: 1

For some reason, the first thing that comes to mind is a 2D pixelated Luigi now transformed into a six packed ultra good looking guy with the most macho mustache ever.
Maybe I need more coffee.
1. Re:Pixelated Images by Anonymous Coward · 2017-02-08 04:25 · Score: 0
  
  Or a cold shower...
Nothing to see here, move along.... by Anonymous Coward · 2017-02-07 18:39 · Score: 0

Attention grabbing headline masks underlying simple image comparison solution...
They needed a complicated neural network setup to do image comparison on an 8x8 grid ? Seriously ?
Image comparison has been used for years to do photo matching, I'm betting they are also making use of a high res source image and a low res pixelated image that had the same ratio and dimensions to start with. If you don't do that then it becomes damned near impossible to achieve because the resampled images end up with almost zero similarity when reduced down to a significantly smaller grid size.
If they were provided with some real world test images I imagine they would not be able to duplicate their results. What they are demonstrating is a Uri Gellar mind trick.
JPEG Rot by Anonymous Coward · 2017-02-07 19:04 · Score: 0

Cool. Hopefully this can be used to combat image decay. Not to mention the hordes of idiots who screenshot images on their phones, apply a filter, and post the screenshot.
8x8 by dschiptsov · 2017-02-07 20:08 · Score: 1

> 8x8 pixel images which it then turned into some pretty clear photos where you can actually tell facial features apart. This is bullshit. It is mathematically impossible to compensate such loss of information. They might use other images to reconstruct original form a pattern, but this, obviously, has nothing to do with reality if the other images used as sources of details have not been made at the same moment.
ffmpeg by Anonymous Coward · 2017-02-07 20:10 · Score: 0

so when is this feature going to be in FFMPEG ?
I don't see why not by bugs2squash · 2017-02-08 04:30 · Score: 1

If they have, say, a minute of pixelated video presumably they could estimate the orientation and position of key features of the face and then make progressively improving estimates of a higher resolution image.
I know that was not the focus of this research (to match a pixelated image to one of a number of high resolution alternatives). But much of the crappy blurry images I see are in the form of video and it seems to me that there would be multiple independent images of a face that could be assembled into a single better image using related information in other frames.

--
Nullius in verba
1. Re:I don't see why not by CaseCrash · 2017-02-08 12:59 · Score: 1
  
  That's a very interesting idea/application of the tech. Nice work. All those crappy convenience store robbery videos might actually be of help after that.
  
  --
  No, that link you posted to a web comic we've all seen a hundred times is not "obligatory."
Mulitple low resolution images. by PeterJFraser · 2017-02-08 05:18 · Score: 1

If there are multiple low level images such as would occur in a surveillance camera. It should be possible to combine them a get a much better guess at a true image.
Ha ha When Trump will ask identity of protesters by Anonymous Coward · 2017-02-08 06:31 · Score: 0

I can even imagine the scene.
DonalTrump: Who dare to protest against my wisdom ?
StevewBannon: All we have is a drone photo. But thankfully the guys from NSA siphon the identification algorithm from Google.
Let's use it.
DT: I can't believe how all this woman's where only number 9's and 10's !!! Why they're all hate me ?
SB: Seems like the whole crowd was composed only of clones of Taylor S'es, Beyonce's, young Claudia Shiffer's , and some Brad Pitt's ,Justin Biber's.
DT: Oh no. I will send right away a twwet ....
Not Completely New by Anonymous Coward · 2017-02-08 12:09 · Score: 0

The first papers describing deep neural nets noted you could fix some of the output and run the network in reverse to generate potential inputs, such as images of digits (it was a paper on OCR). So since the dawn of deep neural nets we've know you could train a network on something, give it a partial end result, and come up with an input projection. Google is doing it a little differently, but the basic idea was already known.