News for hackers. Stuff that matters.
Re:Tell W3C: We don't want the Hollyweb by Daiz on Friday March 22, 2013 @12:12PM (#5423842)
It wouldn't really be a feasible solution for video, though. You'd have to re-encode the whole thing for every individual user, which would be incredibly inefficient. Even if you're showing the watermark in just a few places, it'd still require a damn lot of resources. There's also quite a few things out there for removing video watermarks[1], so unless the watermarks were incredibly obtrusive, they'd probably be quite easy to get rid of. And of course you couldn't really put the watermark outside the actual video stream either, since then the user could just strip it out.

[1] Like various Avisynth plugins for example: http://avisynth.org/mediawiki/External_filters#Logo_Removal

Re:Tell W3C: We don't want the Hollyweb by im3w1l on Friday March 22, 2013 @01:50PM (#5424433)
You could split the stream into N chunks. Every time you watch the movie, M of them are selected and fudged just a little bit. Since M << N, this can be done in real time. The next user sees the same chunks you saw, except that some new chunks have been fudged also. Thus the movie slowly drifts away from the original. To counteract this, fudges are more likely to carry the chunk towards the original then away from it. This is what I came up with in 5 min, so there might be problems with the approach.
Re:Tell W3C: We don't want the Hollyweb by derefr on Friday March 22, 2013 @08:05PM (#5426610)
Sounds like the steganographic equivalent of Y-chromosome genotyping.
Re:Tell W3C: We don't want the Hollyweb by Silhouette on Friday March 22, 2013 @10:37PM (#5427041)
You'd have to re-encode the whole thing for every individual user, which would be incredibly inefficient.

It would be inefficient, but not necessarily prohibitively so if you're talking about the kind of business where piracy is a serious problem so there is real money to spend.

There's also quite a few things out there for removing video watermarks

If you're operating on the kind of level we're talking about, you don't just put in a visible watermark, you convert the entire file/stream to embed a unique forensic footprint tied to the account of whoever downloaded it. Such a footprint will stand up to just about any manipulation you can think of that doesn't reduce the quality of the data so much that the value is destroyed anyway, and you're probably not removing it unless you have a PhD or two in the field and you know someone's secret key.

The limitation with any watermarking approach is that fundamentally it's still based on the honour system and obviously does nothing to prevent private copying between people who don't care about ripping someone off. It's not a technical limitation.

Re:Tell W3C: We don't want the Hollyweb by benologist on Friday March 22, 2013 @12:23PM (#5423910)
Is it really true that they can't do it on the fly? Live streaming is creating the video on the fly completely isn't it?
Re:Tell W3C: We don't want the Hollyweb by trotsky on Friday March 22, 2013 @01:13PM (#5424201)
no, they can do it. There is overlay support in current generation iptv support that can target scrolls and texture replacements at a resolution of less than a zipcode. Take that gear and apply a few static watermarks with timecodes and then swap the different streams at unique times for each viewer and you can end up with a stream thats uniquely identifiable.

Motorola's (Google) new gear can do this, though the purpose is targeted advertising not watermarks. The truth is streaming DRM is about preventing your mother from giving a copy to the neighbors, almost all the content is already on P2P by the time it streams.

Re:Tell W3C: We don't want the Hollyweb by Daiz on Friday March 22, 2013 @12:36PM (#5423982)
In case of live streaming, the video is only encoded once and then distributed to a whole bunch of users. If you were doing unique watermarks, you'd have to encode an individual stream for every single user. Encoding is much more taxing than decoding. The end result would be that you'd need a metric crapton of additional processing power, and the quality would be worse since you'd have to use sufficiently fast encoding settings to be able to do it in realtime (as opposed to being able to use a lot more expensive options when the encoding is done just once and separately from the actual streaming process).
Re:Tell W3C: We don't want the Hollyweb by benologist on Friday March 22, 2013 @02:36PM (#5424724)
What about onlive, they stream(ed) cpu/gpu intensive games in real time on a one-to-one basis?
Re:Tell W3C: We don't want the Hollyweb by coldtea on Friday March 22, 2013 @06:05PM (#5426081)
No, they don't. You have the game assets downloaded already on your PC. They just stream control messages (move the camera there, that player hits the other player, etc).

Real time gaming is not streaming video...

Re:Tell W3C: We don't want the Hollyweb by jbri on Saturday March 23, 2013 @12:06AM (#5427197)
OnLive[1], which is what was being talked about, is literally streaming video to you of a game that's running in a datacenter somewhere.

[1]: http://en.wikipedia.org/wiki/OnLive

Re:Tell W3C: We don't want the Hollyweb by wmf on Friday March 22, 2013 @04:36PM (#5425533)
There's been some interesting research on compressed-domain forensic watermarking; check it out.