Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

“Apple’s iPad creates what it thinks is there, not what necessarily is there”

theverge.com

Post reply on HN
Print view View original

11–20 of 20 posts

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#11

Post by gillesjacobs »

True: For image CAPTURE on portable devices a lot of ML-based image processing is done.

Unknown but unlikely: video PLAYBACK on iPad to use AI upscaling. A Google search does not provide a definitive answer to this, but in all likelihood linear upscaling is used. AI-based upscaling does exist though, look at the many plugins for MPV for instance. Surely, if Apple has implemented this in Quicktime video player it would be all over their marketing.

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#13

Post by nradov »

This is a shit article by a lazy reporter who doesn't understand the rules of evidence in criminal trials. The judge is simply applying precedent here. Digital zoom does involve some manipulation of the original image. That manipulation is probably fine, but if the prosecution wants to introduce that evidence then they need to also provide an expert witness who can testify to that effect. If the prosecution failed to do so then they're incompetent and failed to do the necessary trial preparation.

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#14

Post by Rebelgecko »

nameless912 wrote: post #3

How in the world did a judge actually allow this argument to go forward? This is malpractice.

Perhaps the judge had read articles in the past about upsampling gone wrong (here's one that's also from the verge- https://www.theverge.com/21298762/face-depixelizer-ai-machin...). If the prosecution wanted to zoom in to show something that wasn't visible on the big TV, I think it's reasonable to be worried that whatever they want to show would be thrown off by sharpening algorithms or the rounding that happens when you blow up 97 pixels of video onto a 1920px screen

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#16

Post by nameless912 »

freejazz wrote: post #12
nameless912 wrote: post #3

How in the world did a judge actually allow this argument to go forward? This is malpractice.

Judges can't commit malpractice, they do not represent anyone.

I meant on the lawyer's part for making the argument in the first place. The judge is clearly not doing a good job here either though.

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#17

Post by nameless912 »

Rebelgecko wrote: post #14
nameless912 wrote: post #3

How in the world did a judge actually allow this argument to go forward? This is malpractice.

Perhaps the judge had read articles in the past about upsampling gone wrong (here's one that's also from the verge- https://www.theverge.com/21298762/face-depixelizer-ai-machin... ). If the prosecution wanted to zoom in to show something that wasn't visible on the big TV, I think it's reasonable to be worried that whatever they want to show would be thrown off by sharpening algorithms or the rounding that happens whe…

Sure, but the lawyer confused algorithm for logarithm. Apparently (eyeroll) an expert needs to be brought in to explain to them that normal video zooming isn't going to cause AI-enabled upsampling.

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#18

Post by nradov »

nameless912 wrote: post #16
freejazz wrote: post #12

Earlier quoted context omitted.

Judges can't commit malpractice, they do not represent anyone.

I meant on the lawyer's part for making the argument in the first place. The judge is clearly not doing a good job here either though.

How is the judge clearly not doing a job? Instead of vague insinuations please provide a specific citation on how the judge has violated the rules of evidence for this exhibit.

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#19

Post by Rebelgecko »

nameless912 wrote: post #17
Rebelgecko wrote: post #14

Earlier quoted context omitted.

Perhaps the judge had read articles in the past about upsampling gone wrong (here's one that's also from the verge- https://www.theverge.com/21298762/face-depixelizer-ai-machin... ). If the prosecution wanted to zoom in to show something that wasn't visible on the big TV, I think it's reasonable to be worried that whatever they want to show would be thrown off by sharpening algorithms or the rounding that happens whe…

Sure, but the lawyer confused algorithm for logarithm. Apparently (eyeroll) an expert needs to be brought in to explain to them that normal video zooming isn't going to cause AI-enabled upsampling.

Do you know what algorithm they use? Even non-ai algos like nearest neighbor vs bicubic can have drastically different outputs

Re: “Apple’s iPad creates what it thinks is there, not what necessarily is there”

#20

Post by freejazz »

nradov wrote: post #18
nameless912 wrote: post #16

Earlier quoted context omitted.

I meant on the lawyer's part for making the argument in the first place. The judge is clearly not doing a good job here either though.

How is the judge clearly not doing a job? Instead of vague insinuations please provide a specific citation on how the judge has violated the rules of evidence for this exhibit.

It'd be admissible because it impeaches Rittenhouse's testimony from that day. Also, hard to see how you could watch the video of the judge and think he's being reasonable.
Post reply on HN