Can reflections in eyeglasses actually leak info from Zoom calls? Here's a study into it Boffins at the University of Michigan in the US and Zhejiang University in China want to highlight how bespectacled video conferencing participants are inadvertently revealing sensitive on-screen information via reflections in their eyeglasses. With the COVID-19 pandemic and the rise in remote work, video conferencing has …
Those "multiframe super resolution techniques" don't get talked about much, though it's been vaguely referenced in news articles and cop dramas where fuzzy CCTV images are mysteriously enhanced to make faces recognisable.
Watching an old episode of 'Cracker' on Britbox recently, I was thinking how great it would be if there was such a thing as a technique to enhance the resolution of old TV recordings. Probably using some sort of multiframe tech with AI. Even getting it from what looks like 360p to a natural looking 1080 would be a major boost.
There seems to have been a lot of recent development so I'm wondering why the likes of Britbox aren't adopting this tech.
Without thinking about it too much.. static scenes wouldn't benefit from multi-frame interpolation (in a not dis-similar way to how detailed photographs don't benefit from resolution enhancement as much as cartoon images do - the required information just isn't present), and there a lot of static scenes in old TV shows so the results would be uneven, glitchy and probably feel more unreal than simply watching the low-res version and letting the more worldy-experienced AI in our heads figure out the detail.
Zoom meeting eyeglass reflection is probably a best case scenario because there'll always be small amounts of head movement to provide differing information over multiple frames.
An current webcams are painfully slow, they don’t enough onboard processing power, and have shitty lenses, not luminous at all. They capture very few frames per second, at low shutter speeds due to the lack of luminous lenses, and spit a fast lossy compressed video image that is then further processed and compressed by the videoconferencing software. Not the best source to interpolate a higher perceived resolution from multiple frames
Old telly could end up looking almost exactly *unlike* the modern stuff: the static backgrounds would remain blurry whilst the actors would be super-enhanced to sharp clarity.
Who knows, maybe we could get so used to actually seeing the actors clearly on Legend, Dave and all the other repeats channels that the compression on the new stuff gets improved.
It's been used for a lot of archive TV restorations primarily Doctor Who, which IIRC VidFIRE was developed for by the restoration team.
Simple explanation - https://tardis.fandom.com/wiki/VidFIRE
More detailed* - https://en.wikipedia.org/wiki/VidFIRE
* Shamelessly hijacked footnote from above "Beyond this, the technique (although critically applauded)[citation needed] has seen relatively little exposure, perhaps because of a belief within the broadcasting industry that public interest in the kind of archive television that would benefit from VidFIRE is insufficient to justify the cost of processing.".
Interesting, however VidFIRE is quite old technology now, completely pre-AI is seems.
Since then AI has vastly improved and cost of computing has come down. So I'm still surprised how little research there seems to be. The products I've seen demos of on YouTube don't seem to be ready for full movies. The issue I saw was not so much still images, but moving human forms and textured surfaces. The results with inanimate objects were remarkable though.
But if you take off your glasses you'll not be able to clearly see your colleagues in all their glamour.
Like Fred, who still feels like he is alone in his home office and continues his deep nasal exploration. Or Jill, who sits with her back to the window so she can't see the neighbours doing - what is that? - but gives you the perfect view.
Maybe, if wearing specs on video calls becomes a well-known security risk, we can turn around the old habits of hiring people just because they are good-looking.
"Plug-ugly: because Security by Obscurity works on video"
It’s absolutely true. And very valuable to me when I’m interviewing someone remotely - because I can see if they’re googling the answers. Vocal tics, unnatural pauses, reflections - all valuable evidence that something might be amiss.
One indicator on its own won’t condemn an applicant to a failed interview - but put them all together…
Ah, the number of "nah, it works" software bugs I've fixed over the years...
Closer to the topic, I'm pretty dubious about the value of interviewing software developers, and particularly of asking them to solve technical problems during interviews. I don't have any thing better to propose, though. Hiring developers who are competent, reliable, trustworthy, and good team members seems to be a Hard Problem.
Er, well don't turn the camera on then. Simples. Why do you need to see your face when what you say is the only thing that matters? Oh, and stick some tape over the camera just to make sure.
And if you have to have the camera on, 'Sorry boss, my bandwidth is very poor, and can't cope with the extra data.'
Why do they need to see your face when what you say is the only thing that matters?
Exactly the right question! It is a fact that the out-of-voice-band information is playing SOME highly weighted role(s), otherwise "face" conferencing wouldn't have the popularity that it does. Although there are some reasonable answers: e.g., it helps to remember who said what because faces are easier to remember than printed names of faceless voices (yes, but an icon would suffice), there would endless reams of tediously illogical, unreadable, or unreasonable answers as well.
I have been working from for decades, and find the whole COVID and beyond rush to video calls hilarious. When I was on a call, it was a real call on a real telephone. No bandwidth on modem dialup even for VOIP. Even as technology improved and video from home became a possibility, I never used it. Still don't now.
I worked from home for most of the past 30 years, and even when I was in an office on a regular basis, we worked with remote teams and had daily conference calls. Those were POTS (typically with speakerphones, so even worse than with headsets) until maybe 2005, when we went through a series of videoconferencing options.
A few of the teams want video for their calls, and I'll do it, because it's not worth the effort or social capital of objecting. Most don't.
I find video doesn't seem to add anything for me, except that it's mildly pleasant to see those of my remote coworkers whom I've met in the flesh and have a long history with. Some others report that seeing faces makes meetings significantly easier for them to follow and/or easier to sit through.
My tentative hypothesis is – and I know this sound crazy, but bear with me – different people are different, and some like video, and some don't, and some don't particularly care either way.
Many years ago (around 25) I was working with one of the oil companies with platforms in the North Sea. Communications had improved greatly since the days when tropospheric scatter was the channel of choice; on one of their relatively new rigs the daily call (a call each morning between onshore engineers and offshore supervision, to discuss the previous days work and plan from the upcoming stuff) - the call was via a satellite video link. On one of their onshore fields, the call was made over the standard phone line with a speaker phone at each end. The contrast in the quality and depth of communications was stark. The only reason the standard phone worked was because any real issues resulted in the engineering team being able to get in a car and take the hall hour drive to the site.
Communication is two way and being able to see each other adds a lot more than we often realise. When details really matter, the standard of communication matters - of course, not relevant when discussing cat videos (nor indeed for what may be >90% of meetings).
Enhance 224 to 176.
Enhance.
Stop.
Move in.
Stop.
Move in.
Stop.
Pull out, track right.
Stop.
Center and pull back.
Stop.
Track 45 right.
Stop.
Center and stop.
Enhance 34 to 36.
Pan right and pull back.
Stop.
Enhance 34 to 46.
Pull back.
Wait a minute.
Go right.
Stop.
Enhance 57-19.
Stop.
Track 45 left
Stop.
Enhance 15-23
Gimme a hard copy right there…
“Eyeglasses” is still a common term in Leftpondia. (“Spectacles” is still occasionally used too, but it is far older, going back to the late 14th century.) Like the season “fall” and the past participle “gotten”, older vocabulary sometimes remains in current use in various dialects.
It's quite an old film, but at one point in Twelve Angry Men, a bunch of the characters have a discussion about eyeglasses. Never thought of it till now, but it is an odd word to use. Over here in the UK, it's normally "glasses" or sometimes "specs". Also, "sunglasses" or "sunspecs".
Never thought of it till now, but it is an odd word to use.
The OED distinguishes “eyeglasses” from “spectacles”. If it has parts that make use of ears, then they’re technically “spectacles”; otherwise, they’re “eyeglasses” (e.g. a lorgnette or a pince-nez).
Over here in the UK, it’s normally “glasses” or sometimes “specs”. Also, “sunglasses” or “sunspecs”.
The first three are used here as well, but I haven’t heard “sunspecs” here — “shades” are used instead.
Sunshades - Some of use simply buy (Transitions or Reactolights) photochromic lenses to cope when we venture into the outside*.
*Very handy in winter to, until you walk into a shop from the snow covered Canadian exterior & are rendered instantly blind.
It's the new official El Reg House Style. North American.
Some PHB at El Reg has decided that having authors from different parts of the world writing in their own local version of English is confusing for their poor dumb readers and have chosen to do everything in American English instead of English. Maybe most of their writers are from the USA these days. Or maybe their "analytics" show most of their readers are in the USA. Or maybe USA readers are less likely than others to have script blockers blocking the "analytics" and so skewing the results towards the USA.
But I agree, of so many "Americanisms", eyeglasses feels more jarring to this Brit than most others. Worse even than necktie :-)
This post has been deleted by its author
At this writing, it’s been changed to “shone”.
As usual, the OED shines some light on the matter:
The regular str. pa. pple. is rare in Eng., being unrecorded in OE. and appearing only once in ME. sinen ; it was superseded by the weak form shined, which was in common use c 1300–1800; this was supplanted by the form of the str. pa. t., which first appears as pa. pple. in the second half of the 16th c. (Weak forms are found in some of the continental langs., e.g. (pa. t.) late WFris. schynd, MLG. schynede, OHG. scînta, early mod. G. schein(e)te.)
If Thomas Claburn is a native speaker of a dialect of North American English, then his dialect might still prefer “shined” to “shone” for some meanings of “shine”, perhaps due to influence from German.
“Shined” is always used here for the causative meaning, e.g. “He shined his shoes until he could see himself in them”; do you use “shone” for this meaning too?
There have been — e.g. the OED states that a knot of hair in a pigtail, and a “kind of low shoe fastened with a tie or lace”, were called “ties”. If you include non-human bodies, “the locking together of dog and bitch during copulation” is also called a “tie”.
“Tie” used to be a categorical term, including neckties, bowties, and cravats. The “knot of hair” meaning is the oldest of the meanings listed here.
bollocks to me. I'm far enough away from the screen when on a Teams call this would never happen. Not only that, the inbuilt camera isn't even HD so its never going to happen. Its not like in the Alan Partridge movie when she wonders into his office and you can see in his glasses he's watching porn. Maybe thats where they got the idea.
E.g. let it film you from a 30°, or bigger, angle to the side.
If you put the camera directly above the screen you are facing the issue occurs. But if you put the camera, for example, on your second monitor on the right, but look into your main monitor there is not much of a problem. Same for the Laptop: If you use that camera fine, but look mainly at your external screen - which should not be right above the camera, of course.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
