All iPads Kyle Rittenhouse Trial iPad

[Forensic_Dude]

I'm a Certified Digital Forensic Examiner and Media Collector. I also do lighter weight data recovery. Here's my take.

EXPERT BLUF: There are several reasons why the Prosecution is wrong and the Defense is correct. The pinch to zoom function does alter the data it doesn't interpret the data. Here's why. During the course of a digital forensic investigation or digital forensics portion of an investigation among many things collected files are hashed, and all the identifiers are logged (type, size, etc). This is because when presented to the court, jury, etc; the forensic examiner / collector needs to prove 100% that data hasn't been altered.

1. FACT - Data is being manipulated on the fly and presented, not presented in its original form at a scale. In this instance the original video format would be scaled or aspect ratio changed as needed to fit the display. The frames stay the same and the screen they are rendered on stay the same. But more importantly the original file stays the same. But with this the portion that is zoomed in for viewing alters the data not only by adding but more importantly subtracting the pixels of the original file and enhancing the new rendering. The old pixels aren't magically rendered somewhere else because its not efficient computing (consumes ram, cpu, gpu resources, and battery). Now when you zoom back out they are recalled. The artifacts someone mentioned earlier is pretty close technically. A marker would be left each zooming motion to see which pixels of the original file were subtracted from rendering so they can be recalled instantly without the user ever realizing. So in short, the data is being modified from its original forensic form. Its different with zooming a picture though but that is a whole other discussion. The difference here is the medium that is being used to display information (Video).

2. FACT - I was unable to locate Apple's video software approved for use by DHS/NIST Forensic Validation Program (this software would be likely be under "Analysis" but a general search yielded nothing). This means it hasn't been forensically vetted or failed its vetting.

3. OPINION - On the adding pixel arguement. Without physically examing the ipad it is impossible to know for sure this portion, it is possible that the prosecution has ProRes on or used it to increase the quality of the video which would alter the original file and add more pixels and may create pixel bleed which may give an appearance of an object thats not there or in the original position of the original video. Seeing both an original and rendered video it would be easy to dissect in a pixel analyzer.

4. OPINION - It is highly likely that the pinch to zoom function utilizes some portion of the ProRes Software without even having it enabled because it would make the video look better and we all know how Apple likes showing off cool stuff (oooh.........ahhh.....).

As for me I obtained my certification (and several others) through the Department of Defense, with over 10 years experience in the IC with among many other things NAT SEC Investigations, and I have over 30 years computer experience. I hope this will help dispel some of the false information or at least clarify things for people who might have questions. I'm not a Democrat or Republican but I have to say The Engadget article on this topic is fake news, it was written by a Weekend Editor not a Digital Forensic Examiner.

 

[Forensic_Dude]

Bread Analogy

Essentially a video in this instance is just a sliced loaf of bread, you pull it apart and put it back together again it fits. But you make a hole in a slice or a series of holes and only view the circle you just cut out, you've altered your bread. And no matter what you do, that slice isn't going to be in its original form ever again.

For the nerds, forgot to mention also, another thing you could do is play the regular video while capturing the RAM and then separately do a 2nd RAM capture in which you use pinch to zoom. Then carve the player and played videos out of the RAM image, hash the files, and compare their HEX values, sizes, etc while also looking at what processes conducted function calls to where and what result. As a baseline having a decompiled version of the player with the original function calls as a separate copy. Would be a fun project.

 

[Sanlitun]

When you distort an image like this then it becomes a little bit like looking at clouds and you see what your mind wants you to see.

In this case some are seeing guilt when he is actually innocent.

 

[cybercheck]

I would guess this also calls the recorded imagery in doubt.

As anyone can see - all phone cameras and indeed any camera has AI or at the very least image compression and enhancement. The phone manufacturers use their AI as a selling point!

Check any review about smartphone cameras and you can see they ALL distort the original.

My Samsung photos are very different from my Huawai - yet they are the same image.

My wife's Samsung shows different images than either of mine.

Also - how else do you get 'Beauty' filters and so on?
I see many peoples faces and ladies instantly gain a non sharp chin - its all over their online personas.
Surely they alter the original!

Again - how do you get 'night vision' buttons? - they alter the original.

As to adding artefects - plainly they do - you may check the NSI (UK) standards for images that would be acceptable in court.

Check you tube and you will see countless examples of 'weird' stuff (ghosts and 'unexplained' things) dating right back to the origin of photography.

It appears deeper than just a zoom being fought over.

Maybe we shouldn't just trust our eyes either! How many optical illusions are there? Countless..

Food for thought..

 

[BR3W]

Maybe we shouldn't just trust our eyes either!

We definitely should not trust our eyes (brains). It's a robust finding that our brains deceive us, which is unfortunate given eye-witness testimony is held as a gold standard.

I was going to mention this earlier but I didn't want to generate a tangent that would distract from my other points--but, yes, our eyes will fill in gaps for our brain to make sense of our world...kinda like interpolation

 

[cybercheck]

We definitely should not trust our eyes (brains). It's a robust finding that our brains deceive us, which is unfortunate given eye-witness testimony is held as a gold standard.

I was going to mention this earlier but I didn't want to generate a tangent that would distract from my other points--but, yes, our eyes will fill in gaps for our brain to make sense of our world...kinda like interpolation

Brains... Much better analogy - thank you. 😀

I personally agree your points for sure - compression & interpolation are there.

Apple themselves say their AI adds pixels when zooming - specifically interpolation and anti-aliasing...

"If you change the interpolation value to Image.Interpolation.medium, SwiftUI smoothes out the pixel data to produce an image that isn’t as pixelated:"

The manufacturers sell those fiendishly complex topics openly.

And again from published research touting new methods of interpolation mentioning 'artefacts' or things not there in the original...

"Image-zooming is a technique of producing a high-resolution image from its low-resolution counterpart. It is also called image interpolation because it is usually implemented by interpolation. Keys' cubic convolution (CC) interpolation method has become a standard in the image interpolation field, but CC interpolates indiscriminately the missing pixels in the horizontal or vertical direction and typically incurs blurring, blocking, ringing or other artefacts. In this study, the authors propose a novel edge-directed CC interpolation scheme which can adapt to the varying edge structures of images. The authors also give an estimation method of the strong edge for a missing pixel location, which guides the interpolation for the missing pixel. The authors' method can preserve the sharp edges and details of images with notable suppression of the artefacts that usually occur with CC interpolation. The experiment results demonstrate that the authors'method outperforms significantly CC interpolation in terms of both subjective and objective measures."

Interesting is - both sides have use zoomed images in their cases...

I attached 2 simple pareidolia even though they may be off topic as 'pinch & zoom'. I am sure the famous 'Solway Spaceman' is familiar - took years to resolve by 'experts'

Kudos this site generated the first examination of the trial case on google search on my PC out in Cambodia.

 

[Forensic_Dude]

Brains... Much better analogy - thank you. 😀

I personally agree your points for sure - compression & interpolation are there.

Apple themselves say their AI adds pixels when zooming - specifically interpolation and anti-aliasing...

"If you change the interpolation value to Image.Interpolation.medium, SwiftUI smoothes out the pixel data to produce an image that isn’t as pixelated:"

The manufacturers sell those fiendishly complex topics openly.

And again from published research touting new methods of interpolation mentioning 'artefacts' or things not there in the original...

"Image-zooming is a technique of producing a high-resolution image from its low-resolution counterpart. It is also called image interpolation because it is usually implemented by interpolation. Keys' cubic convolution (CC) interpolation method has become a standard in the image interpolation field, but CC interpolates indiscriminately the missing pixels in the horizontal or vertical direction and typically incurs blurring, blocking, ringing or other artefacts. In this study, the authors propose a novel edge-directed CC interpolation scheme which can adapt to the varying edge structures of images. The authors also give an estimation method of the strong edge for a missing pixel location, which guides the interpolation for the missing pixel. The authors' method can preserve the sharp edges and details of images with notable suppression of the artefacts that usually occur with CC interpolation. The experiment results demonstrate that the authors'method outperforms significantly CC interpolation in terms of both subjective and objective measures."

Interesting is - both sides have use zoomed images in their cases...

I attached 2 simple pareidolia even though they may be off topic as 'pinch & zoom'. I am sure the famous 'Solway Spaceman' is familiar - took years to resolve by 'experts'

Kudos this site generated the first examination of the trial case on google search on my PC out in Cambodia.

I really appreciate you gathering that information. It further demonstrates why process is not forensically sound. Modified data is modified data and wouldn't be admissible in court if it is required to be displayed in this manner. They'd hire an expert to look at every detail. This would have been worked out prior to going to trial. One thing I'd like to clarify its about video not pictures though. Pictures are another discussion. If you cut up the video into pictures by frame with the zoom on, the hash values would be different from the original and that's why it would get thrown out or be inadmissible. A VERY VERY VERY deep dive needs to be conducted on ProRes and if attorneys are using it while displaying evidence, because that could potentially amount to evidence tampering, not just in this case but writ large.

 

[cybercheck]

I really appreciate you gathering that information. It further demonstrates why process is not forensically sound. Modified data is modified data and wouldn't be admissible in court if it is required to be displayed in this manner. They'd hire an expert to look at every detail. This would have been worked out prior to going to trial. One thing I'd like to clarify its about video not pictures though. Pictures are another discussion. If you cut up the video into pictures by frame with the zoom on, the hash values would be different from the original and that's why it would get thrown out or be inadmissible. A VERY VERY VERY deep dive needs to be conducted on ProRes and if attorneys are using it while displaying evidence, because that could potentially amount to evidence tampering, not just in this case but writ large.

Totally agree... I apologise for my simple examples of a deep subject on a world-wide trial. It IS a Landmark case in my opinion and without a doubt your points vis "Hash Values" have to now be used. It was certainly a serious introduction to the court and the prosecution are legally obligated as far as I can tell to either 'prove it' or 'exclude it'. Extremely good defence team! Good call by the judge! Stopped everyone in their tracks. Almost along the lines of "Rising Sun" regarding tampering!

 

[agamoto]

Thank you for providing that screenshot. It certainly appears to be the standard Windows Media Player. That said, I'm not familiar with WMP anymore although I'm aware there are enhanced versions of that app that provide more granular control of how the video rendering is managed.

It's not Windows Media Player, that's a completely different application, which is optionally installed in Windows 10 and up. It's Windows 10 "Movies and TV". It's extremely basic, and it doesn't perform alteration of the media, nor can you alter the renderer used to play it.

Stepping back from this point/counterpoint discussion over what is/isn't used in courts, let's re-visit the core of what I was explaining. I hope we can both agree that the issue for consideration is mainly about software rather than hardware (with the caveat that displays, like the 4K TV, can and do "manipulate" images in their own ways).

Well, then we'd be ignoring the fact that nearly every phone and tablet out there today, including all iPads and iPhones, use a processor-based embedded GPU for dedicated hardware video processing. PC's with certain types of video cards in them (I can assure you, experts like Black most definitely would be using them on his PC) can be set to perform all manner of hardware-based antialiasing and interpolation tricks in the process of decoding and rendering video for the screen. Are they certifying those machines and checking their settings before they're allowed in court? I doubt it.

None of that really matters. The reality here is that the defense pulled this argument out of their ass with the intent of denying the prosecution proof to show Rittenhouse lied. They are 100% incorrect in their assertion, and the judge bought it because he's "touched.

There isn't anything inherently bad about an iPad or a Surface or really any display device. The issue is whether the device, and it's software, has been authenticated for use in a courtroom and, more specifically, about each and every piece of evidence entered into a trial. That can be done on an iPad, a Windows tablet, or any other device...but it must be cleared beforehand. Usually it's not a problem for a prosecutor to present something like this to me but when it comes down to a couple pixels I would expect some pushback and expertise to resolve any concerns. These would be presented to the judge beforehand, outside the presence of the jury, and not dropped into evidence in the way that they were in the trial yesterday.

It's not a matter of a couple of pixels. If you've got a decent monitor, you can clearly see he raises his rifle at Rosenbaum, w/o having to zoom. Zooming into the video puts it into the realm of beyond reasonable doubt. That's the problem for the defense.

The burden of demonstrating that authenticity falls on whomever is attempting to enter it into evidence. The foundation for entering evidence like that is not typically laid on cross. I don't know if you are aware of the source of that video, but it should have never been entered into evidence in the way that it was.

The source of the video? It's the FBI's drone footage. I would suspect that if the source of the video was in question, the defense would have objected about that. The video, played previously from that same iPad, was ALREADY admitted into evidence without objection earlier. Maybe this is part of the problem with the feed you're watching that's not showing you the video that the prosecution is playing.

Pinch/Zoom is essentially no different than using a magnifying glass on the iPad. If you don't believe me, go try it. It's the defense who is making sh*t up here. The onus should be on the DEFENSE to prove their false claim that Apple inserts pixels that shouldn't be there. Forcing the prosecution to do so makes about as much sense as asking atheists to prove God doesn't exist. What the defense did is a knee-jerk hail mary to take advantage of the judge's ignorance, which wasn't too difficult to do.

Today, the judge got up out of his seat and watched the "unzoomed" video in question on the 4K TV. By getting up close, and honestly he was probably TOO close, it essentially had the same effect as what the iPad does when you pinch zoom. He allowed the unzoomed video to be presented like that to the jury. Hopefully, they've got decent eyesight. I did not need to zoom in to see Rittenhouse raise his rifle at Rosenbaum before he followed Rittenhouse. Of course the defense absolutely insisted that the prosecution not be able to tell the jurors where they should be looking on the screen... I can't recall if the Judge agreed to that or not.

Again, I was making a larger point that we don't bring devices into the courtroom and just pass them around for people to (mis)handle. For those following from the beginning, remember the state had an expert witness who was trying to enhance these pictures but they couldn't make heads or tails of what they were looking at. It was after that the prosecutor tried to argue for the veracity of using pinch/zoom to enlarge the image to sufficient acuity to understand which way the barrel was pointed.

The video, once again, already accepted into evidence, could have been set to repeat the sequence on the iPad without any zoom, and a clerk of the court could have held it out for each juror to see with a large magnifying glass. That should have been what the prosecutor suggested because it instantly would have shut down the ignorant claims by the defense. It would have been no different than pinch-zooming or standing 5 feet in front of the 4K TV.

One can argue over the minutiae of the objection the defense raised, but it was genuine and legitimate. Furthermore, no one should be advocating for the state's ability to enter into evidence unauthenticated digital evidence...or any evidence, for that matter.

My ass is also genuine and legitimate. Come on, man. Again, the video was already admitted. If pinch/zooming to focus on a particular area can't be allowed because it may introduce pixels that aren't there, then what's stopping defense attorneys from gumming up the wheels of justice by bedazzling judges who can't set the time on a microwave with similar bogus technological claims?

You know, if you jump on the slippery slope toboggan, it could be argued that using features like zoom, image enhancement, frame-by-frame, or even using a laser pointer to draw attention to parts of the screen... All of those things alter our perception of what people are looking at, whether the original images are altered themselves or not. The entire reason 24 frames per second works for film is because the brain is capable of some marvelous analogue interpolation. Our minds insert "pixels" which do not exist in between every single frame of video we watch. Should I be allowed to deny a prosecutor the ability to show a video in evidence in a certain way because it might cause the viewers to perceive the video differently?

Are you watching the trial right now? It appears they're going through the issues with the video now.

Not now of course, but I was earlier, sitting there shaking my head at what I was listening to. It's too bad they couldn't get themselves experts who were actually capable of describing what's going on, instead of letting some defense attorney dictate his false interpretation of how scaling works or how Apple zooms in on videos. Even Dr. Black, with his 23 years of video analysis experience fumbled through explaining simple things like NTSC and frame rates.

 

[agamoto]

Totally agree... I apologise for my simple examples of a deep subject on a world-wide trial. It IS a Landmark case in my opinion and without a doubt your points vis "Hash Values" have to now be used. It was certainly a serious introduction to the court and the prosecution are legally obligated as far as I can tell to either 'prove it' or 'exclude it'. Extremely good defence team! Good call by the judge! Stopped everyone in their tracks. Almost along the lines of "Rising Sun" regarding tampering!

The linked site describes a method for app developers to resample images in their iOS applications. That is unrelated to pinch/zoom, which does not resample a video. The onus should have been on the DEFENSE to prove what it was asserting, not the other way around.

 

[cybercheck]

The linked site describes a method for app developers to resample images in their iOS applications. That is unrelated to pinch/zoom, which does not resample a video. The onus should have been on the DEFENSE to prove what it was asserting, not the other way around.

Interesting perspectives. Good reading for sure.. I am not a lawyer so I couldn't comment about admissibility of evidence. I am fairly clear that Apple clearly state the image is manipulated on zoom as shown in their official quote from their website above. Not having seen the said evidence - as it has not been allowed - and neither has anyone else outside the defense and prosecution teams I also can't comment about what it shows or does not show or even what video they are referring to - sorry about that.

My personal work is only providing Security CCTV systems capable of producing images and footage that may be used by the Police and Courts. PSDB publication 17/94 CCTV Operational Requirements Manual.

In the UK it is clearly stated that if you cannot see the image clearly it will usually be classed as inadmissible - ZOOM will not be expected (except on the original sytem software) and not relied upon...

Here is the exact wording:

It is for the court to decide whether the pictures are accepted, and this is done on the grounds of relevance to the case, reliability of the evidence, etc. The appropriate resolution, level of compression and number of pictures per second will be determined by what you wish to see in the recording. If you can’t see it then it’s not fit for purpose.

It should not be expected that enhancement features, such as zoom controls, will provide extra detail.

My work involves making sure the CTTV is correctly set up, positioned and fit for purpose.

So on the other hand I would say I DO use Zoom on the ACTUAL system hardware and the level of detail shown is certainly good enough to show serial numbers on money counted at the bank, fraud at casinos when slowed down and of course can show a nasal hair on people faces who enter the door.. Export of the data has a custody chain and no alteration is allowed of the exported files.

Due to the Chain of Evidence and also Who Exactly Can Access the Videos etc we have a Rigorous Standard BS 10008 which is our Code of Practice for Legal Admissibility and we can say although personal phone footage may possibly be allowed it is unlikely... Neither would Post-Manipulation of Images or Videos. Oh and the export MUST include the software used to capture/view/replay the original!

As I say I am not a legal expert - my work is only high quality footage only on secure systems where access is controlled and strictly logged.. saves the headache shall we say?

But The UK and US laws may be different....

 

[cybercheck]

For those wondering about Zoom etc - and may have read my comments in this thread - here is some clarification about the use of CCTV in UK courts.. This is what is expected from images and videos. This is how I set up a system for banks, casinos, embassies etc... There are 4 levels - There may be more than one camera on an area of activity or a zoom/pan/tilt or a combination. However the recorded images and videos follow guidelines. In short the recorded image is expected to be like THIS without Zooming the recorded images/videos. The CCTV operator must make sure the recording follows these guideline... Pays to follow!

To simplify the situation and provide guidance to a system specifier, five general observation categories have been defined, which are based on the relative size that a person appears on screen (figure 4). As part of the OR
development, the user will be asked to decide which of these four categories best reflects the type of activity being observed. The CCTV installer will then be able to fit a suitable camera to meet the requirement.

Monitor and Control: A figure occupies at least 5% of the screen height and the scene portrayed is not unduly cluttered. From this level of detail an observer should be able to monitor the number, direction and speed of
movement of people across a wide area, providing their presence is known to him; i.e. they do not have to be searched for.

Detect: The figure now occupies at least 10% of the available screen height. After an alert an observer would be able to search the display screens and ascertain with a high degree of certainty whether or not a person is present.

Observe: A figure should occupy between 25% and 30% of the screen height. At this scale, some characteristic details of the individual, such as distinctive clothing, can be seen, whilst the view remains sufficiently wide to allow some activity surrounding an incident to be monitored.

Recognise: When the figure occupies at least 50% of screen height viewers can say with a high degree of certainty whether or not an individual shown is the same as someone they have seen before.

Identify: With the figure now occupying at least 100% of the screen height,picture quality and detail should be sufficient to enable the identity of an individual to be established beyond reasonable doubt.

To help I have attached the guidance for the above for images recorded WITHOUT ZOOM that are usually admissable in court without question...

The Monitor or Detect categories may be suitable for covering a wide area such as a car park.

The Observe category is useful in areas where it is necessary to monitor a group of individuals such as outside nightclubs and pubs, or in town centres, as it provides reasonable detail of the subject whilst simultaneously providing some context to their activity by monitoring the area around them.

The Recognise or Identify categories would be used for the cameras providing close-up images at the entry and exit points. In scenarios where the purpose of the camera is primarily access control and identity verification, a figure much greater than the 100% Identify setting may be required.

The purpose of these categories is to suggest appropriate image sizes to aim towards when specifying a system to meet a particular requirement, rather than to define a minimum standard. It does not follow that it will be impossible to recognise or identify an individual if the image size is smaller than the 50% or 100% figures suggested. Equally, there is no guarantee that individuals will be identifiable just because they occupy >100% of the screen.

Trust this helps

 

[agamoto]

Just wanted to remind everyone that this is what the defense had to say to convince the judge to put the onus on the prosecution to prove pinch to zoom doesn't create magic pixels:

"iPads, which are made by Apple, have artificial intelligence in them that allow things to be viewed through three dimensions and logarithms."

Following the precedent here, I've come up with some other things a defense team can say, should they feel the need to stump the prosecution and befuddle the judge the next time a prosecutor wants to pinch in on video that's already been admitted into evidence.

Objection!

"iPads, which are made by Apple, utilize nanoscale graphics microcontrollers to de-clutter spatial/dimensional audio/visual displacement markers!"

"iPads, which are made by Apple, have embedded acceleration algorithms which compensate for suppositional imageblock discrepancies!"

"iPads, which are made by Apple, use third-order, phased intermodulation to invoke tri-linear interpolated jitter!"

"iPads, which are made by Apple, convert capacitive input into temporal chromatic and photonic quasi-particles!"

"iPads, which are made by Apple, offset a digitally-oscillated matrix resulting in full to partial optical wave genesis!"

Have fun! Make up your own! Just make sure the jury isn't in the room to hear it.

 

[cybercheck]

Just wanted to remind everyone that this is what the defense had to say to convince the judge to put the onus on the prosecution to prove pinch to zoom doesn't create magic pixels:

I am sorry to slightly disagree but appreciate your points.

Pinch and Zoom in court is not actually allowed - at least in the UK.

It DOES create pixels as Apple themselves say it does on their official website. Original Recordings are allowed on the Original Software used to create the image/video.

The defense team is a lawyer NOT an Image expert and says he is. So it would be great if the image expert does testify.

The UK is quite clear on this and the fact of Innocent until Proven Guilty and the onus is on the Prosecution to Prove their case.

I do admit the lawyer bungled the phrasing however the point still stands that Zoom on equipment in court is inadmissible in the UK - any defence lawyer would throw it out if tried.

IN SHORT The Image must be Zoomed by the original equipment at the moment of recording.

Maybe US laws are different...

I have included a quick picture in case you wish to look up the basis of the argument and why UK courts adopt the view of what can and cannot be admitted. From the Scientific Development Branch of our Government. Please also not we are governed by other laws. The Police and other Experts may Zoom Before trial and present this evidence but in Court itself shown to the jury It would not nbe allowed to interactively Pinch and ZZoom or not use the original recording software to view.

I hope this helps sir.

 

[tankapotamus]

Did you watch the trial?

At issue is whether an image of someone holding a firearm from across the street can be pinch/zoom'ed enough to differentiate which way the firearm's barrel was pointed--not whether there was a firearm present.

Yeah, sure, an image "manipulated" in this way won't create a firearm out of a broomstick or vice versa, but it certainly does matter when trying to differentiate whether someone's firearm is pointed at a 60 degree angle vs. a 90 degree angle from 100 yards away.

The question anyone still stumbling over this is why the prosecutor didn't use any trial technology. Believe it or not, we don't typically use iPads to show jurors videos...for good reason. We certainly don't resort to the Windows default video viewer when whatever we wanted to use isn't available, either. Does everyone arguing for this genuinely believe witnesses (or the state) just wave their iPads and iPhones around for evidence?

I have to disagree. With that small of an image, a person absolutely could think a broomstick is a gun. It happens all the time to people at night, and in person. It's insane the number of people that say, "It doesn't add or change anything!", and a sentence later say, "It's just an algorithm that guesses what should be the and interpolates the pixes to smooth it out". It's like they know damn well it does, but want him to be guilty so bad they just can't admit the truth they already know.

 

[agamoto]

No, it doesn’t help.

Where does Apple say pinch to zoom adds pixels to zoomed video? Got a link?

As I’ve mentioned before, there are so many different pieces of hardware and software at play here in the courtroom touching this video, there is no guarantee even the forensic software used to play back video is displaying things correctly.

Also, obviously, neither the defense nor the prosecution are playing back video using the same system used to record that video.

Anecdotally, I’m typing this on an 12" iPad Pro right now and I have all the requisite materials to duplicate what they are attempting in the trial… in my tests, there is functionally no meaningful difference between pinch zooming into an iPad video at 4x, using a 4x magnifying glass to watch a 1:1 video, or standing 4 to 8 feet away from a screen capable of playing the video at its native resolution without any post processing enabled.

Negligible differences, but if I had to say which was best, the greatest clarity seemed to come playing the video at 1:1 and using a magnifying glass, the worst standing in front of the 4K screen where I am casting the video to.

I should add that I’m not just some armchair nitwit. I’m a former TV anchor/reporter with a journalism degree and television engineering experience. I started as a studio grunt back before everything went digital. I also have a degree in IT security with 35 years of PC experience, 20+ years experience using Adobe photoshop, illustrator, AE, premiere. I’m a also a part-time professional hummingbird photographer! My wife is an executive engineer for terrestrial, satellite and streaming for a major global media company.

 

[agamoto]

I have to disagree. With that small of an image, a person absolutely could think a broomstick is a gun. It happens all the time to people at night, and in person. It's insane the number of people that say, "It doesn't add or change anything!", and a sentence later say, "It's just an algorithm that guesses what should be the and interpolates the pixes to smooth it out". It's like they know damn well it does, but want him to be guilty so bad they just can't admit the truth they already know.

There is absolutely zero confusion as to whether Kyle is holding a broom or a gun. The goal here is to determine whether Kyle is holding his broom up when he told the court he wasn’t. Pinch zoom would have been more than sufficient to clarify this. In fact, if you have good enough eyes and know where to look (the prosecutors were Not allowed to tell jurors where to look), you can see it without any magnification.

 

[clueless22]

I don't understand how digital zoom interpolation is even being questioned or debated on a tech forum.

This is a known issue with digital media and has been known for decades. One cannot "add" information to a picture that doesn't already include it. In order to make a "larger" picture, digital zoom uses interpolation to estimate what it "thinks" the pixels should be; it doesn't simply make pixels "bigger."

Hi, I have recently found myself interested in this trial and I found this forum while doing some research. I have very little knowledge of any of this stuff but I've been trying to wrap my head around it. In the method known as "Nearest Neighbor Interpolation" (as described here : https://theailearner.com/2018/12/29/image-processing-nearest-neighbour-interpolation/ ), can it not be said that there is indeed no new information added? If so, is this the technique that is used in so called "court technology"?

 

[BR3W]

I have to disagree. With that small of an image, a person absolutely could think a broomstick is a gun. It happens all the time to people at night, and in person. It's insane the number of people that say, "It doesn't add or change anything!", and a sentence later say, "It's just an algorithm that guesses what should be the and interpolates the pixes to smooth it out". It's like they know damn well it does, but want him to be guilty so bad they just can't admit the truth they already know.

I agree with everything you wrote here.

I was conceding the fact Rittenhouse was carrying a firearm to advance the conversation. That Rittenhouse was carrying a firearm was not in dispute. If I was defending this case, objecting on the grounds that it was somehow disputable he was carrying one would be a bad faith argument.

That also leads into the answer as to why it's of no consequence which video players were being utilized to playback videos whose accuracy is not in dispute--neither side was trying to show something in order to resolve a dispute of fact.

But, yes, you're correct in how you've framed and described part of the problem with this kind of evidence.

 

[agamoto]

Just in case anyone was wondering what specifically the prosecutors hoped to show enlarged on the ipad, here it is.

https://twitter.com/i/web/status/1458644126572990466

Is bicubic interpolation or AI making it appear as if Rittenhouse raised his rifle at Joshua and Kelly Ziminski in these frames? Pretty clear to me that's what's happening.

It's a stretch.

It's this precise moment in time that's critical to the outcome of this case. If Rittenbaum raised his rifle at the Ziminskis, then the defense has a very serious problem because it presents Rittenhouse as being the initial aggressor.

 

[BR3W]

It's this precise moment in time that's critical to the outcome of this case. If Rittenbaum raised his rifle at the Ziminskis, then the defense has a very serious problem because it presents Rittenhouse as being the initial aggressor.

It's not a critical element of the case--it's only critical to self-defense. If the jury decides he pointed his firearm at Ziminksis it changes self-defense to imperfect self-defense, which still isn't murder.

Furthermore, anyone claiming that video you posted is "clear" in any way, shape, or form has already formed a conclusion before watching it.

 

[agamoto]

You don't think there's a lot of value to demonstrating him lying under oath?

 

[BR3W]

You don't think there's a lot of value to demonstrating him lying under oath?

There is value in impeaching a witness when one wants the jury to disregard/question their testimony.

That's not why this was introduced as evidence, though, and it's not why you posted it in this thread. You claimed it was critical to the case for the purposes of demonstrating the fact that he raised his weapon at a particular person at a particular place and time. It was not introduced for the purposes of impeaching the witness.

Furthermore, I'd argue that if one's case critically depends on evidence this questionable then that's a larger problem beyond the technological issues this thread has been discussing. If it is critical to the case then the prosecutor should have followed all of the rules of evidence and best practices in order to make sure he could present it to the jury in the way that he wanted.

I asked if you knew the source of the video. You responded it came from an FBI drone.
Here's how a federal prosecutor would have been expected to handle the evidence in a federal trial:

• If any enhancement of the video and/or still images is required, this should be completed only by a qualified expert witness, not the prosecutor.
• For video files to be introduced into evidence, they must be authenticated. This can occur when an eyewitness with knowledge testifies that the video file is a fair and accurate representation of what transpired, or when no eyewitness is available to testify, the “silent witness theory” can apply. If introducing video evidence under this theory and absent a stipulation, it is a good practice for a prosecutor to prepare to call, if necessary, a witness who can testify to the operation of the device that recorded the video, the witness who recovered the video and placed it on evidence, any witness with knowledge, and/or any other relevant chain-of- custody witnesses.

(https://bja.ojp.gov/sites/g/files/x...nal-video-evidence-primer-for-prosecutors.pdf, Pg 7).

That drone footage should have been submitted to the defense for evaluation, both sides have their experts evaluate the evidence, the judge weighs it for its evidentiary value, and then (and only then) can it be submitted as evidence. During the trial, the prosecutor doesn't just pass an iPad around the courtroom. They introduce it as evidence via a witness (the prosecutor is not, and cannot be, a witness to the trial) who then explains the evidence roughly along the same guidelines the DOJ has posted for its federal prosecutors to follow.

The reason the FBI wasn't there to explain all of this is because they didn't want to have to expose their technology to the defense. They don't have to...unless it's exculpatory evidence (FYI, this often overlooked rule is another possible reason the prosecutor didn't follow the normal rules of evidence). I've had a case where my client didn't keep his mouth shut when the police raided a party. They found some "unknown substance" and a gun in the building. When asked about them, he said they weren't his and to, "go ahead and test 'em. My prints won't be on that gun because I've never seen it or touched or nothing" and "that's not even drugs--it's vitamins. Test it!"

So now I have to defend this mess because they never tested any of it. If they had, they would have found it was vitamins and that he had, in fact, never handled the gun and they would have been required to disclose that to me and then explain it to the jury when I asked them under oath. Instead, since they didn't test any of it, they were still legally able to testify, "our training in narcotics interdiction leads us to believe this substance is methamphetamines." and that the gun was the defendant's (the prosecutor argued it was his via "constructive possession").

The rules of evidence are in place for many reasons. One of the most important reasons is to ensure, to the best of our abilities, that we don't convict innocent people of crimes they did not commit. They are also important to ensure that guilty people aren't sent to prison based on the inappropriate actions of LE or the prosecutor, like what almost happened the other day in this case.

 

[cybercheck]

I would say that the fact of so many really good experts on here disagreeing would mean the judge was right to try to eliminate a point which would lead to 'reasonable doubt'.

I certainly have reasonable doubt about my own expertise.

In a case of this magnitude and consequence it would only be right to ask the burden of proof to rest with the prosecution.

The 'point' was passed and the trial goes on.

I haven't seen whether an expert witness has been asked for.

As for the 3d - well sure AI does that although I have only seen google announce that particular algorithm.

For the video itself which people are referring to I personally am in doubt as to this really awful grainy video being acceptable - as the defence says - people can see in it what they want to see. I for one cannot make my mind up exactly what I am seeing and that is the truth.

I would have doubt to send anyone to jail for life based on just this, so the case would have to be built on other and supporting evidence.

Back to the point - unfortunately google search algorithms has buried the Apple Official Site beneath hundreds of 'experts' and 'news' articles - however anyone asking can look up the technology which shows the AI adds pixels based on 'best guess' 'nearest pixels' on zoom.

 

[cybercheck]

No, it doesn’t help.

Where does Apple say pinch to zoom adds pixels to zoomed video? Got a link?

From Apple:

Image sizes vary widely, from single-pixel PNG files to digital photography images with millions of pixels. Because device sizes also vary, apps commonly need to make runtime adjustments to image sizes so they fit within the visible user interface. SwiftUI provides modifiers to scale, clip, and transform images to fit your interface perfectly.

Rendering an image at anything other than its original size requires interpolation: using the existing image data to approximate a representation at a different size. Different approaches to performing interpolation have different trade-offs between computational complexity and visual quality of the rendered image.

It’s easier to see the effect of interpolation when scaling a smaller image into a larger space, because the rendered image requires more image data than is available.

SwiftUI smoothes out the pixel data to produce an image that isn’t as pixelated:

In short it adds pixels based on "best guess" "nearest pixels" - end of story

Here is 2 shots Apple use to describe the "effect"

Fitting images into available space | Apple Developer Documentation

Adjust the size and shape of images in your app’s user interface by applying view modifiers.

developer.apple.com