I think we should consider this a topic of open conversation, where everyone’s input adds to the discussion and broadens our understanding of it.

I would love to hear more about how the FBI sees it. I’ve talked to a few people from the FBI and some who work with them, but I’ve never spoken to the actual decision makers within the FBI.

The explanations you gave are consistent with what I’ve heard them say. I thought I explained that in my blog post, but I can see you you’ve added the bit about predictive frames.

I agree that, from a theoretical and logical point of view, predictive frames do seem to offer an additional element that could make the video challengeable in court. However, this is where non-technical people get caught up in the details, seeing a nit on the bark of a tree, while missing the forest.

This, by the way, is quite common. In fact, a writer recently wrote about one of the big problems we have in government is that so few of our leaders really understand technology. What they end up relying on are simplified understandings that are often misleading if not used in the proper way.

This is exactly what we have here. The idea of predictive frames is causing the problem. Not the predictive frames themselves.

The concern is that lawyers could argue that what the video compression has done is altered what really happened. And indeed, lawyers will argue anything they can, if it could throw out evidence against their clients. But the point I’ve tried to make is that the FBI is needlessly concerned about this technicality.

The idea that the video compression is altering what actually happened is technically true, but is unimportant. Why? Because the imperfections created are designed to be unnoticeable.

We watch compressed Hi-Def BluRay movies, and the detail is incredible. It seems perfectly sharp, as if you were there. We don’t notice all the fuzziness created during the compression, because it is designed to be fuzzy in areas that our perception doesn’t notice.

If you wanted to take your argument about modification of what really happened, we could apply this to any form of reproduction. Take faxes, for example. They reduce the image down into dots. A great deal is lost in the process. Some copiers actually use character recognition to compress documents. Copying always alters what was really there.

Film based photographs rely on the grain of the film. If it is black and white, you lose all the color. If it is dark, the image can be quite grainy. Digital cameras have similar limitations. Moving objects can become blurred.

No representation is ever the same as the original. Something is always changed. The question is whether what is lost or modified is significant enough to distort what we can see.

MPEG and H.264 compression is designed specifically to minimize visual distortion, while maximizing compression. It is by far the best solution.

If you take the FBI’s approach, which is based on a misunderstanding of technology, then you end up with a much worse solution. Why? Because you will have a much more compromised version of reality using their approach.

We have limited storage space to record video. Lossless compression techniques take up so much more space and bandwidth that, in the real world, will force people to use far lower frame rates and far lower resolution images. Both are compromising the details that matter most. You will lose far more valuable and important visual information using that approach, than using forward and backward predictive frames.

If bandwidth were unlimited and free, and if storage was infinite and free, then, sure, why not use lossless compression? I agree. But the whole point of using MPEG and H.264 compression, is to maximize the visual information (including all the most important details) with minimal file sizes. Millions of hours of engineering effort have gone into optimizing the accuracy of the perceptible details.

This is why, when lawyers have tried to argue on the technicality that video compression can modify the image, it has been rejected in court. Judges have ruled on this. They aren’t going to throw out real and valuable evidence if it is visually clear enough to recognize what happened, unless the defense can prove that important distortion or modification has actually occurred. The theory of modification is a technicality and not compelling enough to throw out valuable evidence.

But poor quality video is often thrown out, if it can’t give you enough visual information to accurately determine the scene or the people involved.

Courts are much more concerned about tampering of video, than the compression artifacts introduced, and have urged manufacturers to use watermarks or other means to make it possible to verify if the video has been intentionally altered. However, even without watermark protection, video is still allowed as evidence, unless it can be shown that there is specific evidence for believing the video was tampered with.

The bottom line: The FBI’s approach will lead to worse video evidence, not better. They are better off with higher resolution images and higher frame rates (up to 15 fps, more than that doesn’t buy much), because it will give you far more visual information. They have every right to insist on watermarking of the video, to assure it has not been intentionally tampered with. But insisting on lossless compression is misguided.

Their argument might seem logical, but it is based on a technicality, which comes from having a too simplified understanding of technology. Without real understanding, logic can lead to bad conclusions.

[...]The Myth of MPEG in Court « Spot On Security[...]…

kalagate now part of LGC Forensics

I’m just getting a chance to respond.

Is what you are suggesting a part of the MPEG standard? We aren’t familiar with this, but we’d like to know more.

I agree that making the video encoding traceable to the encoder would be a nice addition to the watermarking that is commonly done.

I have seen some discussion that legal experts and law enforcement would prefer that the video not be altered and would rather see metadata that authenticated the video. Is what you are suggesting actually altering the video?

I’m interested in hearing more.

Thanks.

Doug.

By the way, here is a good link to Sam’s blog at Security Systems News. He has some interesting added comments:

http://www.securitysystemsnews.com/blogs/?p=2218

Doug.