Hacker News Reader

Comment by fredley | Hacker News Reader

...or just use steganography. Steganography in videos can be made to be invisible to the naked eye, and extremely robust - even to video compression. It's also difficult to detect, and even more difficult to remove/tamper with.

What the author of the article is proposing is actually a kind of steganography, albeit a rather crude one, and probably rather difficult to implement (since scene timings will be different for each episode, and you'll need to keep a separate log of exactly which timings correspond to each person you've distributed the file to), compared to running a video through a process that embeds a steganographically hidden id pattern at each keyframe.

View on HN

Came in to say this. It's the obvious answer, and presumably the IP owners know about it even if the author of the article does not. It's also obvious that the author doesn't understand video encoding if he is proposing measuring millisecond differences in the length of specific scenes.

This.

The minimum difference in scene length allowable would be a single frame.

Do you actually know that or are you speculating?

I wouldn't be surprised if MP4 (or other video formats) provided a mechanism for tweaking the timing of a given frame. Of course, if you're using such a primitive watermark you might as well just add a note in the metadata...

But wouldn't it be trivial to notice that in the reencoding step and throw it away?

"Huh, this frame is supposed to be delayed by 3ms. Ah, screw it, [checks 'force to 23.97 fps' button]."

Yes, we know that. Video playback has physical constraints; a file format where image elements are sampled at irregular time intervals is obviously possible and useful for very "scientific" purposes like measuring when images are acquired and interpolating the raw data, but not reasonable for playing back on a screen with good compression and easy decompression.

Huh? VC-1 in WMV and H.264 in MKV both support VFR. It's not uncommon for handheld video cameras or some anime sources.

Yes, I know that - I work in film and post-production for a living and I can explain the origin and workflow aspects of video frame rates to you in excruciating detail.

You know how annoying it is when you watch a movie and there's some sort of 2-dimensional 'hacker' character and s/he just says some gobbledegook about 'hacking the password' or whatever, you see something a progress bar briefly on screen, and then the actor says 'we're in!' In short, how stupid movies can be about computer topics? Well that's how it feels when someone who knows nothing about video formats or editing comes up with suggestions about adjusting the length between scenes by a few milliseconds and having it slightly different for every viewer.

thank you

This is a good time to bring up the Gell-Mann Amnesia effect(xhe).

It's where you read a page full of comments about something you really and truly deep know something about and see how very, every wrong nearly everyone is...

...then you go on to take every other page pretty seriously without remembering just how far off people were on the topic that you know about. I experienced this the other day when people were talking about beep.js on HN.

(xhe) http://www.goodreads.com/quotes/65213-briefly-stated-the-gel...

> This is a good time to bring up the Gell-Mann Amnesia effect

Thanks. Now I know the name for this.

VFR exists, but would be very obvious in the metadata and would make the file unplayable on many devices.

its very straightforward. video comes in frames, what happens in between frames is that the same image is being displayed for that whole time. there is no way to end a scene half way through the frame - there is no visual cue or anything.

not to mention how you would go about defining the ends of scenes...

no standard format that i am aware of has variable framerate. it makes no sense.

Yeah, it's not really possible and as others have said it's by far the silliest possible way to accomplish unique file fingerprints.

Of course we have no idea if they used steganography or not, right? He bases the article on the idea that they have no tracking info except the watermark, but I don't see any evidence that they don't know more.

To be honest, when I heard that the files were protected with a watermark, I assumed that they actually meant a steganographic watermark. I thought that this tech had been adopted into the industry ages ago, I'm rather surprised if it isn't the case, given the arms-race between the media and illegal distribution.

Same here. I was really surprised. That's why I thought that my very simple idea warranted writing a post.

Your initial assumption was correct, it is widely deployed in the industry, there are multiple vendors that provide fingerprinting technology that is orders of magnitude more advanced then what the article suggests.

The visible fingerprint might just have been there to deter the clueless. In any case even with a fingerprint its hard to proof without a doubt, the individual who actually stole it.

The visible fingerprint might also be there to prove intent or add circumvention charges in court.

I don't think they need to prove that a certain individual stole it. The party which received the advanced copies likely had to take on liability, so HBO can go after them. Of course, that doesn't mean they won't turn around and lean on the actual culprit, but whether or not they catch someone HBO will have their due.

It's there to advertise HBO. It's that simple.

Arms race? The entertainment industry has cantered out onto the jousting pitch with their horse and lance. Meanwhile illegal distribution is calibrating the laser cannons in their orbital battle station and dispatching their aircraft carrier battle groups.

I don't think "arms race" really applies when the two sides are building entirely different weapons. The entertainment industry is interested in pursuing the source of leaks - as the other comments here have mentioned, they've gotten pretty good at it.

I think you're wrong about the two sides. The guys with the orbital battle stations are the distribution channel. Watermarking isn't there to deter them, so their lasers don't matter.

Watermarking is a deterrent to the people who upload the content in the first place. The ones who're careless about who they'll share things with. A unique watermark (steganographic or visible) is about finding the source of the leak. In that arms-race the entertainment industry is facing people who can remove a watermark - but we have no idea whether or not they're winning. The fact that torrents exist with watermarks intact makes me think that perhaps they are.

The entertainment industry operates some 40 actual satellites that were put up into space using actual rockets, engages in petabyte-scale asset management as a matter of course, and constantly reworks its content acquisition pipelines as new digital and analog technologies come to market. We're a little bit more technologically sophisticated than you seem to imagine.

>reworks its content acquisition pipelines as new digital and analog technologies come to market.

I would have gladly paid them to watch a stream at 8pm last night, yet somehow that still isn't an option in 2015.

Instead I ended up with 4 episodes from a torrent at 8:15. Tell me more about embracing new technology as it comes to market.

I thought HBO Now (I'm not a subscriber) allowed one to stream 5 minutes after air?

I remember hearing lots of noise about them offering a stand alone streaming service, I was even excited to try it out and see how it handled the traffic of a season premiere (HBOgo has been notably terribly on this front)

I googled around for about 15 minutes and was only able to find news articles announcing it. I'd be happy to pay them for what i've already watched if anyone could direct me to a sign-up for this elusive streaming service.

Here's the site: https://order.hbonow.com/

I don't have an iOS device or their specific supported ISP, so I haven't signed up.

In this case there's someone else who paid for you not to be able to stream it: Apple.

HBO Now is Apple TV exclusive for 3 months.

No, you can get it on Sling TV, too.

I kind of assumed the laser cannons were on flying ostriches.

no arrests, i am certain that if one of these early release approved people was caught they would make a huge song and dance about it to deter others. at oscar time the screeners make it onto torrent sites with ease. if they were able to find the leakers using current methods we would know about it.

They have tracked down leakers before. The issue is that even if you get a clear cut high profile case (like howard stern[1]) then what? It is pretty hard to prove intentionality or where exactly in the chain the data went astray. These things just get left around. If a screener disappears during a party, who even knows who took it (A guest, a caterer, or did it just end up under a couch)?

I've heard of minor reviewers and third party firms getting cut off after a leak is traced to them, but for the most part the back office people simply don't have the power to enforce against anyone of note. Also the evidence just isn't enough for an actual law suit against someone who can hire a lawyer.

1: http://torrentfreak.com/super-8-screener-leaks-with-howard-s...

I will say that steganography is definitely in use already, at least on feature films. (It's particularly easy in animation, where literally anything can be changed seamlessly). A film pirated from a theater (even by a camera pointing at a screen) can easily be traced to the exact theater and showing of the movie, and IR cameras in the theater permit authorities to identify the exact individual with the camera.

In short, don't try to film a movie in a theater; they will find you.

Yes. The author of the article is barging through an open door: It's very likely that those Game of Thrones episodes had individual hidden watermarking, and that using those they can tell exactly who leaked it.

The author jumped to the conclusion that those don't exist, apparently because they aren't visible (but that's the point!) and that there is a visible watermark which the author apparently assumed was the only one (but it almost certainly isn't).

While stego might be a route, blurring the watermark may not be enough to obscure the source.

Given a known frame, they could blur the same portion of the same frame on all watermarked copies, and the one with a matching output is your culprit.

Blurring loses information, but is more akin to a hash than a deletion, as it's a deterministic process.

Edit: this sort of thing. http://dheera.net/projects/blur

It's a deterministic process, but it still has inputs. You'd have to know those inputs or test for every possible set of inputs to whatever blurring method was used.

I imagine HBO knows every reviewer they gave a watermarked copy to.

It depends on the blurring method. If you do a gaussian blur, for example, you might be able to do a pretty good inverse blur by finding black frames and performing blind deconvolution. If I were at risk I'd use a nonlinear injective filter instead, to make sure information is lost (linear filtering is not only deterministic but perfectly without noise and no frequency response nulls).

This might be a very naive suggestion (I know nothing about the subject) but couldn't you just take two copies, XOR them and blank out the difference to defeat such signatures?

We're talking about images, so if you take the pixels for example they're just {0,...,255}^n, integers.

You can if you have the original file, but then what's the point?

Otherwise, take image a=original+f1, image b=original+f2. Then if you e.g. average them you still have c=original+f1/2+f2/2 -- so both signatures are kept (theoretically if either you have a large number of sources you could average the signals out of existence, or perhaps the system isn't robust and can't detect multiple sigs). This assumes the introduced signatures are indistinguishable from the source (quite reasonable).

Even if that worked (maybe it does, I don't know) you'd still have the very major obstacle of getting two potential leakers (from presumably separate publications) to find each other and collude without getting caught.

I think that could be avoided by making sure that large parts of the signatures overlap. There's probably some scheme to be able to recover which copies were used to make the leaked copy (up to some finite number of copies, but it quickly becomes impractical to get enough copies).

Over 4 separate files, depending on how they encode the IDs, you might still find out who the leakers are. They might share some bits and not others. The studio can take its time if it has a serious leaking issue.

What the author of the article is proposing is actually a kind of steganography, albeit a rather crude one, and probably rather difficult to implement

...but easy to explain (and sell?) to the entertainment industry, who have shown themselves to be mostly technically illiterate. Some cynical, unscrupulous hacker could probably extract a lot of money from them for such a service...

Author here. That's my goal, but I don't have their email address so I simply wrote it down there.

Wouldn't a simple re-encode break any sort of stenography?

No. There are algorithms out there that can survive compression[1] by encoding statistically over an image instead of just bit manipulation.

You can also embed data in the audio channel by alternating between two imperceptible echo kernels for 1/0[2]. This is extremely compression resistant.

1: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumb...

2: http://www.slidefinder.net/a/audio_steganography_echo_data_h...

Surely it would be possible to create encoding resistant stenography. Will the stenography be affected, sure, but it should be possible to create error-resistant stenography up to a certain point.

Stenography: encode information such that it is below the human detection threshold.

Lossy compression: discard all sub-threshold information.

Thus, in the general case it's fundamentally impossible. You can make a scheme that will survive certain known encoders and bitrates, but you can't make one that will survive all lossy compression.

How will that survive transcoding? What if you bump it down to 477p resolution?

There will be a limit to how much compression it can take, but some methods e.g. Echo Hiding are extremely compression resistant (I've implemented this over a phoneline - it works).

> and extremely robust - even to video compression. It's also difficult to detect, and even more difficult to remove/tamper with.

Split & merge frame by frame, in random, from multiple video sources. Visually they will be the same, but steganography has a high chance of been broken.

Any steganography solution on the market today that is worth its salt would actually be robust against this. The amount of different sources you would have to combine to protect against this is higher than you'd realistically be able to find.

Disclaimer: I used to work at one of the leading companies in this field.

the algorithm could tell exactly which sources are used in the final video?

impressive!

Are you being sardonic? There are extremely obvious ways to do this particularly since they've got both space, color, sound and particularly time to play with. Unmixing signals isn't rocket science, especially if you've provided the signals that are getting mixed.

That's very interesting. I always assumed averaging (maybe with varying weights across frames and time), or applying a few rounds of other steganography, would degrade the encoded message into noise.

Could you link to some papers? If you try to encode lots and lots of messages with different keys into the video, does it ruin the quality before ruining previous messages?

I'm not sure in how much details I can go since a lot of this is behind patents and NDAs and whatnot, but just googling for "video watermarking" should already yield quite a few interesting details for you.

it's amusing to me, how much effort people are willing to put in to get free entertainment.

Even better? Those of us who do it all after paying for the service.

Sometimes, it's not about the money.

It it were a full-screen steganograph it would need to be subtle as to not interfere with viewing. So could that be combatted by re-compressing with another codec?

> it would need to be subtle as to not interfere with viewing

Pre-release screeners often have big ugly on-screen watermarks. Almost anything is better than a scrolling mid-screen line of text "PROPERTY OF STUDIO NAME DO NOT PIRATE THIS FILM".

Steganography (which may or may not have been used) wouldn't have any effect whatsoever on the number of downloads

I don't see anyone anywhere claim that it would. It would only aid in tracing the source of the leak.

True - and then what? The leaked episodes may have been stolen or can be claimed to have been stolen.

Anyway, the horses have bolted and the damage is done.

Yes, but the leaker would be blacklisted from all future HBO (and possibly partner) pre-release materials.

And probably sued into oblivion.

Do you know for a fact that the leaker is the same person (organisation?) that has received the official HBO material? Maybe it is more like this: http://www.ibtimes.co.uk/expendables-3-online-leak-two-arres...

Maybe so, but HBO isn't obligated to prove guilt beyond reasonable doubt. Arguably if someone can't keep the screener secured from internet thieves they're still not a trustworthy guardian of the content.

Reply on news.ycombinator.com