Comment by com2kid | Hacker News Reader

Comment by com2kid | Hacker News Reader

Parent Full thread

As someone with only 1 eye, this is quite suited to me! I am actually sad that the previous crop of "low cost, slim, not earth shattering" AR glasses all died out. A camera built into one stem and some simple text displayed for one eye would suffice for a ton of use cases.

A killer app for me would be floating name tags above people's heads, but I imagine that would require better HW than what is available here.

I'd love to see what this looks like actually installed on glasses, in different configurations.

Some simple AR use cases that I'd like to see: ]

* Turn by turn directions

* Incoming high priority notifications

* Weather alerts

* Front door notifications

* Upcoming meetings

* Interior building navigation

* Shopping Lists (hard to do the interactions on this)

The issue here is none of these are a killer feature, so I wouldn't be willing to pay more than a small premium to have them added on top of what I normally pay for frames, but since I already am paying $300-$500 for frames (before insurance), if someone offers the above for another $100 or $150, sure why not.

> A killer app for me would be floating name tags above people's heads, but I imagine that would require better HW than what is available here.

This is a societal issue, more than a HW/SW issue. Google Glass explicitly bans this type of feature:

https://developers.google.com/glass/policies#c_what_you_cant...

> Don't use the camera or microphone to cross-reference and immediately present personal information identifying anyone other than the user, including use cases such as facial recognition and voice print. Glassware that do this will not be approved at this time.

This is in response to Congress raising Privacy concerns: https://slate.com/technology/2013/05/google-glass-facial-rec...

I wouldn't expect any "official" facial recognition on any of these upcoming AR devices any time soon.

> > A killer app for me would be floating name tags above people's heads, but I imagine that would require better HW than what is available here.

Which is too bad, since I see AR as potentially being a massively useful device for those with memory issues, dementia, or who generally need some help with things like this.

floating name tags over heads

I'd have given anything for that 35 years ago. First thing on any contract job, I'd make a map of desks & names - but it always tripped me up when people slipped over to work at a different desk!

As someone who has struggled with names and dates and times my whole life, that is the single use that I was interested in.

As someone with prosopagnosia, that was my first thought too.

The dystopian use case of this is: See the name of someone + how they voted last time. Then you can „engage“ them to convince them of your „truth“.

The Google Glass backlash (which may well have happened to any AR glasses) has set back practical use of AR in private life so much. It's frustrating, especially because of the splash damage it's had.

What's particularly odd to me is that many people seem to be sensitive about, specifically, cameras in glasses. I sometimes literally wear a bodycam, not subtle at all, and it's mostly ignored, while with no-camera HUD glasses people get nervous that there's a camera there.

I hope that we can eventually get away with stuff like real-time facial ID as an adaptive technology feature - though the only way that will work is if we can get it pretty much invisible. Or we finally get used to all being recorded all the time anyway and stop freaking out about it.

the difference is if the camera is obvious then you at least know you're being recorded. playing a guessing game 24/7 is exhausting. being recorded once in awhile not such a big deal. having to watch every word and every action you do all the time? not nice.

This is a case where people's expectations really need to catch up. We're being recorded in so many ways in so many situations that "helping someone remember your name" is really the least of your concerns.

Unfortunately, that's also why "auto-name-tagging" is never going to happen. It doesn't help the gigantocorps who are already tracking us through wifi and IR and seismic patterns in public places. They've had data that could make the individual experience immeasurably better for nearly 15 years now, but they conveniently use "privacy concerns" to avoid having to acknowledge it in public. As long as there is a modicum of suspension of disbelief, they can continue to collect all the data they want for their own purposes and to hell with anyone who wants to use their own data for their own purposes.

It will happen eventually, like the end of Columbus day.

user will upload they pictures to cloud, now company have picture of everyone live in their village.

Not even an exception for user-entered data.

Like, I could conceivably take photos of people I meet, store those photos along with notes in the iOS notes app (or even a contact app) then create a panel of those photos so that I can tap the right person and bring up those notes when I see them again. AR would simply make this easier and faster. Unless the AR camera was hidden it would not make it different from a privacy perspective.

To make things more confusing, both the iOS photos app and Google Photos automatically find faces in your photos and the other photos containing them and let you easily tag each face with the correct name. The technology to do this already exists.

It’s ok if they are the ones to do it. Safe old big tech. It’s not ok if these nefarious nerds with bad memories do it.

Exactly what I was thinking.. Have the feature offline but that would require storage I suppose. Also, the tagging in photo apps were another aspect I thought of. It is ok to tag in one instance but not another.

> easier and faster

That's the thing. The privacy implications are different at scale.

A world of FOSS tools, where centralized ID databases are heavily restricted or banned, and people are using facial recognition technology on people they encounter in their lives and tag, that would be fairly harmless, perhaps even doing much more good than harm.

But it's not going to play out like that. How many of us run speech transcription software on our own machines, and how many of us use a remote service?

> Google Glass explicitly bans this type of feature:

But that's, like, 90% of what I'd want something like that for. Not for everyone in my field of view, of course, but when I'm out in the field I meet a lot of people in a short time and I can never remember everyone's names. Work shirts with embroidered names are an absolute godsend.

I'd love something like that too, but google glass wasn't designed for you or your needs. It was designed to plaster ads over everything you see while continuously collecting data on everything you point your face at. That's the problem with technology today. However useful it could be for you, it isn't really created for your benefit. Instead every new technology today is used against you, even when it does allow you to do something you want with it.

Self driving cars will track everywhere you go and when and ultimately decide where you'll be allowed to go and when, augmented reality will make it so you can't see anything without an ad playing, robots in your home will record you and catalog everything you own, the phone in your pocket already tracks you and everyone you care about, etc. It's hard to think of a single popular tech product that isn't collecting data and working for someone other than the "owner" who paid to bring it into their home.

Even this open source AR project is guilty of it. Their privacy policy says they'll collect your personal data to use for marketing. Any tool or tech that requires an account and/or requires connections to someone else's servers is likely going to be used against you, but I hope that someday we get a fully self-hosted, open source product that runs on trustworthy hardware so we can do things like keep track of people's names, record our experiences, and shape the way we see the world without putting ourselves and those around us at risk.

>This is a societal issue, more than a HW/SW issue. Google Glass explicitly bans this type of feature:

Seems like one of those things that can't be stopped. It can be slowed, as Google has done, but eventually OSS will allow for it. If it is massively outlawed then it can be reduced, but even then it'll still be on the level of hidden cameras.

> This is a societal issue, more than a HW/SW issue. Google Glass explicitly bans this type of feature:

Which is ridiculous, it is giving an advantage to people who were born with a good memory for faces.

Do not allow for lookups in any sort of online database, all tagging has to be done locally by the user.

> I wouldn't expect any "official" facial recognition on any of these upcoming AR devices any time soon.

I would, just not for the general public. This would be very helpful for police to immediately do a check on people around them and spot people with active warrants.

By that logic, why not just put cameras everywhere, use image recognition tech to do all the identification and enforcement, from traffic violations to active warrants, China-style?

That’s a great idea!!

Should they also ban people who have good memories? Or someone who has handlers who tells them who's who?

I get there's a balance, but at some point, it's a problem that holds individuals back and supports the elite.

That’s why open source is so nice right?

Pretty wild for congress to raise privacy concerns, considering the ungodly amount of data they harvest every day on both their own citizens and those of other countries across a variety of mediums, agencies, and subsidiaries...

gotta point out that explicitly banning putting names over people is explictly excluding a very useful accessibility accommodation for folks with any degree of faceblindness.

It can 100% be done in a non-sketchy way by simply limiting the names to the names you've manually entered for people you've seen and tagged before.

folks need to stop ignoring disabled people and / or going "yeah well who cares about disabled people's needs".

It'd be useful for teachers who are trying to remember the class' names! Geofenced, of course.

So Google basically banned Ironman! I am sure the government uses such devices responsibly ;)

maybe open source will help with that. maybe some nobody will make those features and you can sideload and it'll be quiet enough that no one will know for awhile.

As a father of a kid with 1,5 eye -- my daughter's left eye stopped developing at young age so we missed the window of opportunity to have it corrected on time, so one eye does most of the heavy lifting for her, and the other eye can only see a blur -- your comment made me think of possible use cases for the monocle for her. Maybe notifications using colors and shapes that could be perceived by the underdeveloped eye. I believe the signals from the monocle would not interfere in any way with the good eye, so from a brain's perspective, both things could coexist independently. And, as you suggested, it could well be a "clip-on" addition to the frame itself. That would be nice.

I think these are actually kind of lame use cases. They can almost entirely be replaced with a smart watch.

Where this device has a unique advantage is overlay AR. Turn by turn is somewhat there but I think a better way to look at this is passive information. You could walk by a business and see hours on a virtual sign. You could make an app to leave virtual art you could stumble on. I'd like to see more ideas like that. Just a new way to show notifications doesn't cut it for me.

> You could walk by a business and see hours on a virtual sign.

Most stores have those posted already. :-D

> They can almost entirely be replaced with a smart watch.

Agreed, unfortunately current smart watches are tasked with doing a lot, and as a result they are not very good at blending into the background of life.

I'd also love a life logger, being able to see snapshots of my day and quickly create journal entries. An automatic record of everyone I talked to, summaries of important conversations.

Human potential could be greatly expanded with current day technology but sadly we are limiting ourselves.

>Most stores have those posted already. :-D

Ok but it could be 3rd party data instead. It could be live open seating data, etc etc.

Or, hear me out.. it could be _targeted advertisements._ /s

Seriously though, I'd love to have a wearable adblock. Detect anything that resembles an ad, blank it out. Obviously not while you're doing anything safety critical, but most of the time - good trade.

Fix practically anything that instructions have been made for, unscrew here, unclip that, pull out that, replace this part; do it all backwards to put back together ... with visual overlays you're acting as a robot almost (that's good and bad).

Also, anything the AR can see you put down/drop can be remembered; anything you're looking for can be highlighted (in theory).

Would be fun to work on some other ideas.

Yes DIY is such a killer use case for this stuff, and what an awesome way to learn, imagine working on your car or anything really with an expert mechanic who can see exactly what your seeing highlighting where to turn your wrench or tell you that your technique is incorrect, or diagnose that weird sound you are hearing.

That would all be great but alas very battery intensive and things like dropped item tracking would require a high speed high fidelity camera.

>Shopping Lists (hard to do the interactions on this)

with a companion app, you could scan the barcodes of the items as you place them in the cart, and then remove the item from the visible list in the AR HUD. shit, tie that into the actual store's check out system so you don't have to use their stupid POS self-checkout lines. just bag it up and be on your way. other than the products that are sold by weight, this seems like something we're missing out on as a species.

This already exists. One example would be Ikea Scan & Go.

Hell, there are like two dozen fully autonomous Żabka convenience stores in Poland that do this. Unlocks the door for entry, identifies the items you pick up by CV, pays for them automatically without interaction, and unlocks the door for exit. No app or phone necessary, no cashier or staff, no checkout. You just pick the thing up and leave.

https://www.youtube.com/watch?v=5t_-Na63kUY, https://youtu.be/ZMVbwVpBkKc?t=115 for the tech

Even better, the device can have a camera that watches you put the item in your cart and marks it off the list.

Or it checks prices/inventory at other stores and lets you defer purchases to other stores to reduce how much you spend.

A killer app for me would be to make all the windows in my apartment seem larger. Scenery can stay the same just make them 1.3x the size.

Im blind in one eye also, and was thinking the same. But would certainly preffer the lenss to be a litle smaller. But for a first version, seens to have some potential.

> Turn by turn directions > Interior building navigation

Disclosure: I worked at one of the companies spending big on AR and worked on these exact use cases. Sadly, the company was not excited by these cases in the short term, so the projects were effectively cancelled. I fully expect them to be revived at some point, but not any time soon.

>* Turn by turn directions

I wouldn't touch any kind of AR use case while driving. Seems ripe for lawsuits when you're obscuring a driver's view (especially with an eyepiece compared to putting a HUD on the windshield)

> I wouldn't touch any kind of AR use case while driving. Seems ripe for lawsuits when you're obscuring a driver's view (especially with an eyepiece compared to putting a HUD on the windshield)

Cities, walking. :)

Walking around a city pulling out my phone to check directions is actively stupid, a HUD with arrows telling me what intersection to go down would improve safety on multiple fronts. (Theft and not situational awareness).

Would be nice for public transport, too, if indoors navigation works in some of the more complicated train/subway stations!

Looking at you, Japanese railway stations!

Though honestly on vacation there, I enjoyed getting lost in Shinjuku station. :D

Office complexes, finding someone's desk.

Yes, teenagers would obviously use it responsibly and not overload the view with distractions. /s

Also, there is an ongoing trend to ban mobiles for pedestrians crossing the streets (Honolulu, followed by Jiaxing, Lithuania, Poland, ...). It would be great if AR was regulated from the start to disappear in such semi-risky situations.

You might be right that it is ripe for lawsuits, but a properly designed low distraction overlay seems much better than what most people currently do, which is alternate between looking at their phone and the road.

Once head tracking is working, couldn't you show the AR view in the exact same place that a windshield HUD would appear?

The easiest way for that is to have the user set that position to account for the car model, driver height, and seat position, but you're not likely to find a way to pick a good safe spot to show it without asking the user.

But if you say this is safe to use while driving, you need to handle situations where "this AR app had a bug which drew a solid color over the entire view, obstructing my view, and caused me to miss the pedestrian", while a projected HUD has a limited area it can draw on.

If you can't/won't say it's safe to use while driving, then why develop the feature for driving?

This being an eye piece which has software control over your entire field of view has potential for danger.

Then you will have to make sure all of that doesn't replicate these results from the out set: https://arstechnica.com/cars/2015/07/heads-up-displays-in-ca...

Yes, yes you could.

It could work for walking though.

To add to this, I’d love real life real-time ad blocking. All billboards either blacked out or replaced with kitten pictures or master paintings.

Good list. I'd also find an HUD display like this useful for playing music:

- lyrics - chord changes - a visual metronome

Fun; I would just want to plug it to various device though MQTT and do trippy display.

Reply on news.ycombinator.com