"Cinavia"/Verance BluRay audio watermark protection

[asarian]

I've tested all the commercial players for Cinavia. *So far* none have it. But you can expect them all to get it eventually.

Well, they can't have their cookie and eat it. If indeed, as kurkosdr pointed out, they make the mistake of putting it into PowerDVD, TMT, etc., then the Cinavia detection algorithm is up for grabs by anyone who can handle a debugger. So, the choice is theirs: either leave it out of software players -- allowing us to build HTPC's without Cinavia (for which we don't even need the commercial software, of course) -- or put it in, and watch their precious Cinavia be cracked. :)

The only manner in which I can envision us to be really screwed, is when Cinavia becomes integrated in audio chipsets on the motherboard (like the popular Integrated Realtek ac97 audio chipset). As to whether that will ever really happen remains to be seen. Intel's plans to integrate DRM into their CPU's, for instance, has never materialized either.

[SamuriHL]

I'm sure there's ways to make it difficult to defeat Cinavia by debugging a commercial player. TMT, for example, won't even run with a debugger open. So there's a lot of protection and obfuscation to get through. Could someone determined enough break through it? Probably. Eventually. Then people would have the algorithm needed to detect Cinavia. Then, HOPEFULLY, a way to break that detection can be found. So, even with Cinavia implemented in software players, it still won't be a cakewalk. Doable, most likely, by the best and brightest. But not easy.

[ChiDragon]

Now this is just out of interest:
I dont know if you can answer this safely, but have you heard a watermaked portion of audio and the same portion without mark.

I posted FLACs on page 16 and everyone ignored it...shrug.

[asarian]

I posted FLACs on page 16 and everyone ignored it...shrug.

Guessing here; but I reckon because maybe people realize that, of course, you really don't have an FLAC of the same track without Cinavia (for the simple reason that if you did, Cinavia would already have been broken). What you do have, instead, is a completely different track (though to the ear it may sound similar, or the same even), which just so happens to be also DTS-MA.

[Klendathu]

I posted FLACs on page 16 and everyone ignored it...shrug.

You did not use a spectral frequency display. There is not a lot to say about your standard waveform display of entire tracks. Sorry, but those screens are useless.

[jdobbs]

With all sophisticated signal processing one can imagine: The watermark must be darn robust as it seems to survive re-encoding into different formats, re-compression (non-linear), poor quality microphones etc. Of course on the other hand all of those CODECs etc are designed to reproduce the original signal without noticable differences...Yes, but how are we going to get the firmware out of the chip it's stored in inside the standalone? Have Conan the Barbarian squeeze the chip really hard till the firmware leaks out? People do this every day... firmware patches that make slight mods are all over the internet, and it isn't really that hard if you know the chipset -- the problem is volume, who wants to spend all their time reverse engineering and patching hundreds of them?

[Krawhitham]

You did not use a spectral frequency display. There is not a lot to say about your standard waveform display of entire tracks. Sorry, but those screens are useless.

He posted the audio files not just pictures. If you click one of the images it allows you to download the audio files

[xenex]

So, I downloaded the FLAC files posted by ChiDragon earlier. Very useful, I think. I loaded both into Audacity and only kept the left channel of the Cinavia and Non-Cinavia files.

With the two files "linked," ran Audacity's ->Analyze->Beat Finder (with default value 65) and came up with this:

http://stashbox.org/1034089/cin.png

That sure does look like a series of tiny "echoes" there between around the 32 and 41 second marks. I also tried with other channels, and found similar results, but in different places.

This is really easy to do, I'd appreciate if someone else gives this a try as well, just to verify this result.

EDIT - Here is a "Zoom In" on the right channel. This REALLY shows the echoes:

http://stashbox.org/1034097/cin2.png

[Ghitulescu]

One cannot remove cinavia, it's embedded in the sound.

The results would be similar to any de-logo or de-subtitler plug-in for video. Except that the sight and the hear work somehow different and probably an extremely narrow notch-filter would lead to "passable" artifacts. I also think that cinavia won't be implemented into AC-3 tracks on the same BD, only in the lossless ones, as it won't survive well the compression and might trigger false alarms.

[rotty]

So, I downloaded the FLAC files posted by ChiDragon earlier. Very useful, I think. I loaded both into Audacity and only kept the left channel of the Cinavia and Non-Cinavia files.

With the two files "linked," ran Audacity's ->Analyze->Beat Finder (with default value 65) and came up with this:

http://stashbox.org/1034089/cin.png

That sure does look like a series of tiny "echoes" there between around the 32 and 41 second marks. I also tried with other channels, and found similar results, but in different places.

This is really easy to do, I'd appreciate if someone else gives this a try as well, just to verify this result.

EDIT - Here is a "Zoom In" on the right channel. This REALLY shows the echoes:

http://stashbox.org/1034097/cin2.png

Hi Xenex
Nice one, that would confirm what a lot of us thought in the first place, short term echo's. Very interesting.

Even poor audio recoding will still be able to reproduce this, as would even microphone recording. Very robust.

[asarian]

Hi Xenex
Nice one, that would confirm what a lot of us thought in the first place, short term echo's. Very interesting.

Even poor audio recoding will still be able to reproduce this, as would even microphone recording. Very robust.

So, now all we need is some good oldfashioned echo-cancellation software. :)

[rotty]

So, now all we need is some good oldfashioned echo-cancellation software. :)

hi Asarian

LOL, Well of course, at best our comments are pure speculation, could be barking up the wrong tree. Doesn’t really matter our thoughts are just for interest and to bounce ideas off each other, and just curiosity, we are not the boys doing the fix.

Hopefully the people doing the work know the inner workings of this for sure.

[Ghitulescu]

Nobody said that one cannot remove cinavia.

Generic/blind echo cancellation would bring nothing but sensible alterations of the source. Verance patents encompass a large palette of permutations that includes: 2 or more carriers (one has been identified, 15kHz, but it could change), variable phase-shift, delays, offsets and data-embedding (which may tell the host which algorithm to use for triggering "positive alarm")....

Since the cinavied signal must be "invisible" to the ears, so its inventors, it cannot be placed inside the audible spectrum, any lossy and aggressive (low bitrates) algorithm would remove it, or at least its simpler implementations.

Anyway, either the playback would stop or one will end with a low quality copy of the audio. Win:win situation.

[xenex]

One cannot remove cinavia, it's embedded in the sound.

Nobody said that one cannot remove cinavia.

Ghitulescu, which one is it? Perhaps you should decide one way or another before your next post.

[asarian]

Generic/blind echo cancellation would bring nothing but sensible alterations of the source.

Hence the smiley behind my comment. :) It's still interesting to see Peer was apparently on the right track with his ideas (now why doesn't that surprise me?). As to whether Cinavia can be removed, so as to keep the HD quality, I'm still not a huge believer in that; but 'From the minds that gave us AnyDVDHD' is good for a lot of credit over here.

Indeed, you can bet data is being embedded as well to instruct the player. Which might be an Achilles heel, if you could manipulate said data in telling the decoder that this track has the lowest 'alarm' setting (or none at all). Hmm...

At any rate, we'll just see what the big boys come up with.

[Ghitulescu]

Ghitulescu, which one is it? Perhaps you should decide one way or another before your next post.

One cannot remove cinavia without degrading the quality. That's the whole idea (which is by way not new, it dates back in the 60ies, when the analog recorders hit the market).

It can be removed/inactivated, but with quality issues and the sound won't be HD-lossless-like quality, or it can't be removed, if HD-lossless-like quality is to be maintained.

Read the posts in their context.

PS: I'm not a purist, however I dislike to get "distorted" A/V items, even more when they are marketed as "listen XYZ in the studio-master-quality" but you know this is not true. Because this is not copy protection, it's putting into circulation of fakes, and therefore it should be outlawed.

EDIT
Here is a recaptioned video frame from Alien (the original frame had embedded subtitles).
http://joeclark.org/appearances/AEA/2007/AEAimages/160Alien.jpg
Were the old subtitles removed or not?
I could answer YES, because they are no longer visible, I could also say NO, because should they have been removed in the first place, there were no longer need for a black frame on top of which the new subtitles were written. Both answers are correct, depending on how the things are considered (what is more important). Sorry for the bad image, it's all what I could find in this short time.

[rotty]

One cannot remove cinavia without degrading the quality. That's the whole idea (which is by way not new, it dates back in the 60ies, when the analog recorders hit the market).

It can be removed/inactivated, but with quality issues and the sound won't be HD-lossless-like quality, or it can't be removed, if HD-lossless-like quality is to be maintained.

Read the posts in their context.

PS: I'm not a purist, however I dislike to get "distorted" A/V items, even more when they are marketed as "listen XYZ in the studio-master-quality" but you know this is not true. Because this is not copy protection, it's putting into circulation of fakes, and therefore it should be outlawed.

EDIT
Here is a recaptioned video frame from Alien (the original frame had embedded subtitles).
http://joeclark.org/appearances/AEA/2007/AEAimages/160Alien.jpg
Were the old subtitles removed or not?
I could answer YES, because they are no longer visible, I could also say NO, because should they have been removed in the first place, there were no longer need for a black frame on top of which the new subtitles were written. Both answers are correct, depending on how the things are considered (what is more important). Sorry for the bad image, it's all what I could find in this short time.

Hi Ghitulescu

I have to disagree with you on this one.

Its worth remembering that Watermarking is not the same as subtitling.

Firstly, the subtitle is designed to be seen, and seen well at that. Certainly NOT transparent to the viewer, quite the opposite.

A watermark is designed to to be stealthy and un-obtrusive to the viewer/listener

Putting aside Audio watermarking for a second, lets take Video watermarking. As you already know, video watermarking is embedded on CCTV to stop frames being removed/tampered with for evidential reasons. These frames have a few pixels per frame altered according to an algorithm.

They are for the most part unseen by the viewer and can be easily destroyed and removed/blended. The result is totally unseen and the video frames look fine and un-damaged.

Not the same as subtitling.

[Ghitulescu]

The point was the removal of a watermark. I had to use an image here, not a sound.
I couldn't find a better image, and a personal one would take some weeks to be approved. I was looking for a de-logo image, actually.

Yes, indeed, the watermarking is supposed to be transparent, but is not.
The idea was how to remove it, and the answer to anyone that has read the cinavia patents, is that it cannot be done without damaging the audio (it might be noticeable or not, but this is not the point, the point is that the AC-3/DTS track might be of a better quality than the lossless one). It was not designed to be removed, au contraire, it was designed to be detected only by a device not by a human, and in the same time difficult to remove.

[rotty]

The point was the removal of a watermark. I had to use an image here, not a sound.
I couldn't find a better image, and a personal one would take some weeks to be approved. I was looking for a de-logo image, actually.

Yes, indeed, the watermarking is supposed to be transparent, but is not.
The idea was how to remove it, and the answer to anyone that has read the cinavia patents, is that it cannot be done without damaging the audio (it might be noticeable or not, but this is not the point, the point is that the AC-3/DTS track might be of a better quality than the lossless one). It was not designed to be removed, au contraire, it was designed to be detected only by a device not by a human, and in the same time difficult to remove.

Hi Ghitulescu

I find myself un-usually poles apart from you on this one.

Removing Oliver Hardy from a picture of Laurel and Hardy would not be seamless but removing/changing a few pixels which is the watermark would be.

Same with audio, removing the piano from a syphany would be very noticable, removing sounds. that were not niticed before would not be noticed.

Im not saying that Cinavia marking is simple to remove, just that the comparison of a subtitle/logo being the same problem as a watermark in this case is completley wrong.

[spotter]

Nobody said that one cannot remove cinavia.

Generic/blind echo cancellation would bring nothing but sensible alterations of the source. Verance patents encompass a large palette of permutations that includes: 2 or more carriers (one has been identified, 15kHz, but it could change), variable phase-shift, delays, offsets and data-embedding (which may tell the host which algorithm to use for triggering "positive alarm")....

Since the cinavied signal must be "invisible" to the ears, so its inventors, it cannot be placed inside the audible spectrum, any lossy and aggressive (low bitrates) algorithm would remove it, or at least its simpler implementations.

Anyway, either the playback would stop or one will end with a low quality copy of the audio. Win:win situation.

If it's placed in inaudible spectrum area, then one should be able to make a relatively perfect copy of the uncompress PCM stream (or whatever its called) with a simple filter to filter out those frequencies. If it's inaudible for other reasons (i.e. in an audible area, but mixed in for a way that's inaudible) that's a different story.

[rotty]

If it's placed in inaudible spectrum area, then one should be able to make a relatively perfect copy of the uncompress PCM stream (or whatever its called) with a simple filter to filter out those frequencies. If it's inaudible for other reasons (i.e. in an audible area, but mixed in for a way that's inaudible) that's a different story.

Yes I would agree. I beleive its well within the audible spectrum. Its just un-noticable, it wouldnt be able to survive a bad microphone recording otherwise.

There could of course be more than one signal type, audible and outside our spectrum, either of which if received are valid.

[whurlston]

(one has been identified, 15kHz, but it could change) I wouldn't be so sure of that. If you look at ChiDragon's non-Cinavia sample, you will see that it has the same 15kHz characteristics.

[Ghitulescu]

Some people hear 15kHz some not. The easiest way of checking this for yourself is to have an old CRT TV and listen carefully if you hear the buzzing of the line transformer (which is driven at 15625 Hz).
Some people pretend to hear the FM stereo signal (19kHz), too.

And there's a net separation between what is claimed (theory) and what is technically and economically possible. This is why cinavia will be implemented in lossless audio only. It is not feasible for the manufacturers to produce gear that would not yield false positives because of the inability of cinavia to withstand some compression schemes 100% reliable. Such a device would not sell.

[rotty]

Some people hear 15kHz some not. The easiest way of checking this for yourself is to have an old CRT TV and listen carefully if you hear the buzzing of the line transformer (which is driven at 15625 Hz).
Some people pretend to hear the FM stereo signal (19kHz), too.

And there's a net separation between what is claimed (theory) and what is technically and economically possible. This is why cinavia will be implemented in lossless audio only. It is not feasible for the manufacturers to produce gear that would not yield false positives because of the inability of cinavia to withstand some compression schemes 100% reliable. Such a device would not sell.

Hi Ghitulescu

Do we know if the 15k pilot is Cinavia for sure.

I can hear the 15.625 Khz from a LopT still but this is not resolved by a poor mic, and we know that mic recording of Cinavia is retained. Also, kids would be driven mad by this on a good sound system.

If this pilot is Cinavia then this suggests that there are at least two systems involved, one possibly using echo effects that would survive microphone/poor quality recordings and the 15k approach that can instantly be detected for digital to digital recording and non notched line recording.

What is strange though is that if there are two systems then why have the pilot at 15k on HD tracks, why not higher.

Adding a 15khz signal is not good for keeping sound fidelity, in fact it is far too close to normal hearing range and will give strange intermodulation effects with sounds well within the range.

Even the 19K pilot for analogue stereo FM can cause problems even though its totally notched, although, because further removed from normal spectrum is less troublesome.

In this day and age of demands for perfect audio, what’s the point in having an HD audio track that is being compromised by a 15k tone.

The reason that HD audio extends well above normal human range is because sounds we cannot hear do interact with sounds we can hear, and by keeping these sounds (even though we cant hear them) is to reproduce the effects we hear in the real world.

Tampering with that, well we may as well have average quality MP3.

[Ghitulescu]

Hi Ghitulescu

Do we know if the 15k pilot is Cinavia for sure.

I can hear the 15.625 Khz from a LopT still but this is not resolved by a poor mic, and we know that mic recording of Cinavia is retained. Also, kids would be driven mad by this on a good sound system.

Adding a 15khz signal is not good for keeping sound fidelity, in fact it is far too close to normal hearing range and will give strange intermodulation effects with sounds well within the range.

Even the 19K pilot for analogue stereo FM can cause problems even though its totally notched, although, because further removed from normal spectrum is less troublesome.

In this day and age of demands for perfect audio, what’s the point in having an HD audio track that is being compromised by a 15k tone.

The reason that HD audio extends well above normal human range is because sounds we cannot hear do interact with sounds we can hear, and by keeping these sounds (even though we cant hear them) is to reproduce the effects we hear in the real world.

Tampering with that, well we may as well have average quality MP3.

Exactly my point - with cinavia or "decinavied", the lossless track is doomed, that's exactly why the Philips system (and later the CDS one) failed for CD-DA, and SACD and DVD-A are stillborn.

[rotty]

Exactly my point - with cinavia or "decinavied", the lossless track is doomed, that's exactly why the Philips system (and later the CDS one) failed for CD-DA, and SACD and DVD-A are stillborn.

Hi Ghitulescu

And there lies its Achilles heel, in fact worse, its self inflicted destruction.

[asarian]

Hi Ghitulescu

Do we know if the 15k pilot is Cinavia for sure.

Good question. Although I'm by no means an audio expert, I must say I still highly doubt Cinavia is so frequency-range bound.

Embedding a signal in a certain (narrow) frequency range makes it primarily 'vertically' present, to put it that way; whereas a subtle echo, or other type of shift, possibly across the entire spectrum, makes it 'horizontally' applied. The latter is much, much harder to remove (if you don't know exactly what to look for), and would explain why Cinavia has been so incredibly resillient against re-encodes (even those of very poor quality): because compressing to MP3, or some such, typically sacrifices areas of the frequency spectrum believed to be inaudible, but hardly mangles the 'horizontal' flow of things, so to speak.

Besides, as has been pointed out, putting it at 15Khz would be odd, as I could still hear my TV at that frequency until I was way in my twenties. It's simply too close to what we can normally hear.

[Ghitulescu]

Verance patents encompass a large palette of permutations that includes: 2 or more carriers (one has been identified, 15kHz, but it could change), variable phase-shift, delays, offsets and data-embedding (which may tell the host which algorithm to use for triggering "positive alarm")....

Embedding a signal in a certain (narrow) frequency range makes it primarily 'vertically' present, to put it that way; whereas a subtle echo, or other type of shift, possibly across the entire spectrum, makes it 'horizontally' applied. The latter is much, much harder to remove (if you don't know exactly what to look for), and would explain why Cinavia has been so incredibly resillient against re-encodes (even those of very poor quality): because compressing to MP3, or some such, typically sacrifices areas of the frequency spectrum believed to be inaudible, but hardly mangles the 'horizontal' flow of things, so to speak.

That's in the patent too :)
Identifying such patterns also needs computing power, so in the end the CPU would spend more time in decrypting keys and copy protections than in decoding the HD video (properly or not) ;)

[asarian]

That's in the patent too :)
Identifying such patterns also needs computing power, so in the end the CPU would spend more time in decrypting keys and copy protections than in decoding the HD video (properly or not) ;)

So, the 15Khz is in the patent? Gotta say, that only adds to my annoyance about the whole thing. :(

[Ghitulescu]

No, phase shifting and stuff.
And there are several patents, not only one.

[setarip_old]

@Ghitulescu

Hi!

Your posts to this thread would be far more meaningful if you provide a link each time you make/made reference to specific portions of specific patents...

[Krawhitham]

@Ghitulescu

Hi!

Your posts to this thread would be far more meaningful if you provide a link each time you make/made reference to specific portions of specific patents...

http://www.avsforum.com/avs-vb/showpost.php?p=19396029&postcount=65

[setarip_old]

@Krawhitham

Hi!

Certainly, when making a post about a SPECIFIC passage of a SPECIFIC patent, a poster can, as I asked, provide a DIRECT link to that SPECIFIC passage...

[coolalibaba]

Hi I'm not very familiar with video/audio rip/processing. Can we confirm that when an audio is ripped/extracted from the BD, it contains all the original data? I ask this question because I saw someone posted this in other forum:

The way Cinavia is working is they are embedding a high and a low frequency into the dvd and blu rays now the problem isnt trying to remove these frequencies from ripped materials it is adding them. i have pulled the audio apart from a bluray disc and a copy i made of the bluray the actual purchased bluray has the frequencies in the files, the ripped version of it does not no matter if it was a exact copy. the high and low frequency goes on for the entire film meaning any breaks in it is when you get your error code 3 from cinavia.... now this is a very good system to have for the media now the downfall is every new bluray that i rip to my server can no longer stream to my ps3. I actually need to put the stupid disc in. (I am Lazy... What can i say) this is going to be on all the new bluray players and the firmware updates. Now the xbox 360 (for now) does not have this issue. It will stream everything still but that will probably change here soon. The only way i am THINKING to get around this is a new version of ripping software that will not throw these frequencies away when compressing the information. Its crappy i know and i am
feeling it also.. but i have found a use for my xbox 360 for now.... thank you microsoft.

[SamuriHL]

Hi I'm not very familiar with video/audio rip/processing. Can we confirm that when an audio is ripped/extracted from the BD, it contains all the original data? I ask this question because I saw someone posted this in other forum:

Sounds like complete nonsense to me. The only thing that happens in a lot of my copies is the AACS encryption is removed. Nothing more, nothing less. The streams aren't altered in any other way.

[Ghitulescu]

Hi I'm not very familiar with video/audio rip/processing. Can we confirm that when an audio is ripped/extracted from the BD, it contains all the original data? I ask this question because I saw someone posted this in other forum:

Cinavia is like a stain of paint on your T-shirt. When you undress yourself (ie rip the BD), you'll get all the original "data" (ie the T-shirt with the coloured spot), because cinavia is part of the original. It was designed to be so. Someone from the Amazonian rainforest or from Papua-New Guinea can't discern the difference between this stain and the "Polk High" (:)) logo, because he can't read, for him it's all alike, coloured spots on something white (he knows no T-shirts either).

How to remove the stain without removing "Polk High" in the same time, that's the problem.

[RugRat]

Cinavia, ok been playing around with this problem, here are my findings.
Salt DC, Stream form pc to Ps3 NO CINAVIA.
M2ts. On a memory stick NO CINAVIA.
AVCHD disk 20 chapters CINAVIA
AVCHD disk 1 chapter CINAVIA
Since it’s the same m2ts that plays on the memory stick and not a disk, could it be that the chapters have to be the same as the original
If someone can tell me how many chapters are on Salt DC blue-ray disk, or the losers blue-ray disk, I will try my theory out

[asarian]

Cinavia, ok been playing around with this problem, here are my findings.
Salt DC, Stream form pc to Ps3 NO CINAVIA.
M2ts. On a memory stick NO CINAVIA.
AVCHD disk 20 chapters CINAVIA
AVCHD disk 1 chapter CINAVIA
Since it’s the same m2ts that plays on the memory stick and not a disk, could it be that the chapters have to be the same as the original
If someone can tell me how many chapters are on Salt DC blue-ray disk, or the losers blue-ray disk, I will try my theory out
That's a pretty big memory stick then, as the SALT m2ts file is a good 15 Gig (Director's Cut; playlist 0003).

Also, have you tried streaming the m2ts file? If your theory about chapters is right (which I find pretty incredulous), then streaming just the single (re-assembled) m2ts file should not trigger Cinavia either. I highly doubt it, though.

EDIT: Playlist 0003.mpls has 16 Chapters

[SamuriHL]

So then how do you explain my PS3 detecting Cinavia on a memory stick with an MP4 with stereo AAC made from The Losers blu-ray? I didn't add any chapters at all.

[wswartzendruber]

Pardon me for skimming this thread so fast, but I took the center channel from those two FLACs and...

1. Cut about a millisecond of time from the Cinavia one to compensate for a slight delay.
2. Inverted the phase of the non-Cinavia one.
3. Played them mixed together.

The above procedure produces some very odd phasing results...almost as if I applied Audacity's Phaser effect, except I didn't. Is this normal?

[RugRat]

Can’t explain that, but it does not show up with a m2ts file with ac3 and it dose stream ok with ps3 media server, If you stream it but copy it to the ps3 which makes a mpg2 file and then copy and burn it to a disk cinavia dose not show up ether

[SamuriHL]

That's pretty interesting. Maybe a bug in the PS3's Cinavia detection routine?

[asarian]

That's pretty interesting. Maybe a bug in the PS3's Cinavia detection routine?

Wait... I think he's using an existing AC3 track from the Blu-Ray, which will likely not have Cinavia in it to begin with (as it's not HD audio). Which is, of course, not the same as down-sampling an existing HD track (with Cinavia in it) to a custom-rolled AC3 track.

[SamuriHL]

Wait... I think he's using an existing AC3 track from the Blu-Ray, which will likely not have Cinavia in it to begin with (as it's not HD audio). Which is, of course, not the same as down-sampling an existing HD track (with Cinavia in it) to a custom-rolled AC3 track.

Oh, yea, could be. Would be nice if it was clarified that the tracks were the original HD audio or not....even if re-encoded. In my case, they were.

[RugRat]

It is the original hd audio, streams with no problem, but when i put it on to disk i get cinavia,

[SamuriHL]

Thanks for the clarification. That's definitely interesting for sure.

[RugRat]

I thought it might be the chapters i.e. so many chapters have to have passed or a chapter has to start exactly where cinavia is

[SamuriHL]

Can't be. Cause my MP4 has no chapters and Cinavia's detected. So, I would guess a bug in the PS3 implementation.

[RugRat]

Ok it was just an idea

[SamuriHL]

Not a bad one, just that it doesn't fully explain the situation. I honestly don't have any real idea what's going on there. I'm also guessing. :)