Marvin
About
- Username
- Marvin
- Joined
- Visits
- 131
- Last Active
- Roles
- moderator
- Points
- 7,008
- Badges
- 2
- Posts
- 15,586
Reactions
-
Apple Vision Pro review one year later: time to exit the preview
There are a lot of components inside:crazyexcalibur said:You’re lucky that you don’t have any issues with the weight. I’ve tried at least ten different iterations of the face cushions and none have been able to make it where I can wear the AVP for longer than a couple hours. It’s not so simple to do this either. I’ve had to buy them, try them on for a week or so and return them. Then repeat the process. I’ve even bought third party head gear hoping it would help. I’ve found a few but they just transfer the weight from my eyes to my forehead. I only use it for watching media because of this, as media usually is about 2 to 3 hours, the max before my face hurts too much.
The weight is actually similar to most headsets:
https://www.statista.com/statistics/1337114/vr-headset-comparison-by-weight/
AVP weighs 600-650g for the headset and 350g for the battery pack.
The Big Screen Beyond VR headset (display-only) is lighter and weighs 170-185g:
https://www.uploadvr.com/bigscreen-beyond-weight-specs/
An iPhone 15 weighs 171g, including battery.
An iPhone 15 compute unit (minus display) + Big Screen Beyond headset would weigh 356g, almost the same as the AVP battery pack. It may need more battery power but they can still build a much lighter unit.
Hopefully the 2nd revision will see a big reduction in size and weight. It would be good if they could sit it closer to the eyes to increase field of view.
In September this year, they will have the 3nm A19 Pro chip, which could go in AVP 2 or a smaller model. This will run much cooler than M2 and drain less battery.
If Apple had made AVP1 like an iPhone strapped to a headset, even with 2K displays around $1499-1999, they'd have sold a few million units. It would still have been the best VR headset on the market and people would mostly use it for media consumption. Text would have looked more blurry but people used SD monitors for decades and this problem will get resolved in time as display manufacturing improves.
A more mainstream model would let them see what the demand is like and would drive more software and content having a larger userbase.
-
Development of Apple's smart glasses continues despite massive hurdles
There are companies that have been shipping AR sunglasses for a couple of years now so the first part about getting AR displays into a glasses form factor is available:AppleInsider said:Like the creation of its headset, producing smart glasses with AR features is a technical nightmare. It's a hard set of problems to solve, ranging from creating the image in front of the user's eyes to producing something light enough to be like regular spectacles.
All while dealing with other long-term issues like handling processing and communications, and somehow hiding a battery on the frame.
However, Apple is far from the only company to be working on smart glasses.Meta's Orion AR glasses prototype is expected to arrive as a product in 2027, making it a very early release in the field.
Then there is Google, which is producing the Android XR operating system intended for next-gen headsets and smart glasses. Gurman writes that Google demonstrated the operating system to him in December using various glasses, including some with displays.
Those prototypes were considered to be quite polished, but unlikely to reach the open market until the harder challenges like battery life are solved.
Batteries are a massive problem for VR headsets now, let alone lightweight spectacles. It's a weight that must be minimized and carefully placed so the glasses don't feel heavy to wear, which is extremely hard for a purposefully lightweight item like spectacles.
Apple certainly has to come up with a better answer than the current tethered battery on the Apple Vision Pro. But, short of magically making batteries as light as air, it's a difficult problem to solve.
That said, tethering to an iPhone or another device could help further, by handling processing for the glasses. This offloads another set of components and reduces the power draw, but it still means it'll be connected to another piece of kit.
https://us.shop.xreal.com/products/xreal-one-pro
This product offloads processing to a tethered device and only handles the display + motion tracking so no battery needed in the frames.
The power usage is said to be around 2W. Airpods Max has a 5Wh battery, iPhone 15 has a 13Wh battery.
Assuming a new product isn't offloading compute, it would use an iPhone chip for half the power of M-series chips around 5W (max, not average). The displays would be around 2W. The battery would have to be around double the Airpods Max at around 10Wh, increase the cushion on top to distribute the extra weight more, possibly have a thin strap at the back to stop it tipping forward and have lightweight displays that sit in front.
I don't think glasses are necessarily the end goal because even normal glasses are uncomfortable to wear for a while and they can still look unattractive to wear:
One thing that sets Apple products apart is they all look good. Almost every product they design is the best design in its class. Best looking laptops, phones, tablets, displays, headphones.
The end product can look something like this:
The display would wrap around with optional light blockers, mainly for above and below the eyes. The blockers are only needed for immersive content, AR content shows the surrounding environment so it doesn't need to block the light unless the environment light is too bright.
The connection point for the display would have to be the headband if the cups still have to swivel and they should be easy to push up to the headband out of the way. There would be no weight on the face or ears like glasses, the displays would be suspended in front of the eyes.
Retail cost should aim to be under $2000. If they can only manage 2K resolution at this price point, so be it. HDR and black levels are more important than sharpness for media content. It's not going to be great for text but usable and they can sell 4K ones at a higher price in a Pro model.
People are already walking around in public with this form factor and it looks perfectly normal.
No matter how much the current AVP form factor iterates, the bulk of the device is in the wrong place and needs too many cushions. The product itself is just the black/silver part at the front, which would be even smaller without EyeSight components and the compute parts:
I don't think it needs to take as long as 2027 to deliver something like this. They have all the parts this year to make a revision 2 with a more compact form factor.
This form factor would sell fairly well for 5 years while they iterate on getting things more compact and power efficient. It may end up that having compute on the wearable never needs to be the end goal if visuals can stream fast enough wirelessly to a nearby device but at a minimum they need to power the displays so a battery needs to go somewhere to get rid of the wire. If it goes on the front or back of the head, it's back to a helmet form factor again with weight on the face.
-
Thinner, smarter, more connected: What to expect from a 2025 Apple TV
GPUs have raw performance measures that come from core counts and clock speeds. These measures aren't always directly comparable but they tend to be fairly reliable. M4 is 4.6TFLOPs (around the same as XBox Series S), the PS5 is around 10 TFLOPs.oberpongo said:
Where does „half the performance“ come from.I am sure Apple can tweak some cores (more gpu etc) to easily double the PS5 performance
It's better to compare actual GPUs, the consoles use AMD hardware and the PS5 is close to an AMD 6600XT:
https://www.notebookcheck.net/Radeon-RX-6600-XT-vs-M4-10-Core-GPU_10939_12502.247598.0.html
There aren't many games on the Mac to compare but Shadow of the Tomb Raider is a well-optimized title. It shows M4 around 30% of the PS5-equivalent. With upscaling, it can get close to half.
M4 Pro is 9.2TFLOPs and roughly equivalent to PS5.
M4 Max is 18.4TFLOPs and roughly double the PS5, probably close to PS5 Pro.
A18 Pro in the iPhone is 2.5TFLOPs, around 1/2 the M4 or 1/4 the PS5.
Apple's cheapest M4 Max product would be the $2000 M4 Max Studio when it's released.
Nvidia's latest AI frame-gen can 4x the FPS. Apple will have at least 2x frame-gen in an upcoming MetalFX for the Cyberpunk 2077 port this year so this will help vs the PS5 performance, although PS5 has it too:
https://www.tweaktown.com/news/98570/amd-fsr-3-frame-generation-is-now-available-on-ps5-and-xbox-series-consoles-to-deliver-120-fps/index.html
The Nintendo Switch 2 is rumored to be around 4TFLOPs (10x Switch 1) and will support DLSS3 and frame-gen, this should perform closely to a PS5 without upscaling and frame-gen and similar to the base M4 iPad Pro. Apple would be competitive with Switch 2 in the iPad Pros and Mac mini and with upscaling and frame-gen can give PS5-like performance.
A TV dock would allow people to use their iPad Pros this way too and the dock could open a TV UI like Steam's Big Picture mode.
https://store.steampowered.com/bigpicture
Apple used to have something called Front Row, which was a media center software:
https://en.wikipedia.org/wiki/Front_Row_(software)
This is used for photos, music, movies. A dock could help get more Apple TV+ (and Arcade) users because way more people have iPhones than Apple TV boxes.
High quality games for this setup are still lacking but Ubisoft is having some financial trouble just now and seem to be putting themselves up for sale. Tencent owns a portion of the company but the owners have some contention about leadership. If Apple was more willing to let them lead the company, perhaps they'd sell to them instead (should cost $5-10b). Then they'd get a few decent franchises like Splinter Cell, Assassin's Creed, Far Cry, Rayman, Rabbids, Watchdogs, The Crew (like Forza), Trackmania, Just Dance. They'd get 2 game engines: Snowdrop and Dunia (fork of CryEngine). Snowdrop was used in the Avatar game:
They can port their existing library of games:
https://en.wikipedia.org/wiki/List_of_Ubisoft_games:_2010%E2%80%932019
https://en.wikipedia.org/wiki/List_of_Ubisoft_games:_2020%E2%80%93present
-
Meta CEO mocks Apple for 'sitting on' iPhone 20 years later despite doing the same with Fa...
According to the news, Meta used multiple 3rd party fact-checking services:Stabitha_Christie said:Snopes is not a Meta fact checker. Bringing them up is a complete red herring as they are in no way relevant to the conversation.
https://www.seattletimes.com/business/meta-moves-to-end-fact-checking-program/
"After enormous public pressure, Zuckerberg turned to outside organizations like The Associated Press, ABC News and the fact-checking site Snopes, along with other global organizations vetted by the International Fact-Checking Network, to comb over potentially false or misleading posts on Facebook and Instagram and rule whether they needed to be annotated or removed."
The inclusion of Snopes was just to highlight the vagueness in some kinds of facts that are being checked, it wasn't a critique of Snopes, they are a well-regarded fact-checker. Most fact-checkers are, they just tend to have a left-center bias, which affects how middle-ground issues are handled.I can't say Meta is trying to increase hurtful content but clearing the way for people to claim women are property and LGBTQ+ people are all mentally ill does increase the amount of harmful content and misinformation. It also doesn't make the platforms less oppressive. It opens the door for harassment.There are issues that are not so heavily agreed on. One Zuckerberg touched on is blasphemy where people in Pakistan wanted him sentenced to death for allowing images of their religion on the platform. The issue with censorship is where it's appropriate to draw the line. Just saying it opens the door for harassment covers a lot of things. Should they not allow people to say someone is fat or ugly or that they aren't good at their job. One of the top comments on a Youtube video about censorship was that people often mistake hate speech for speech they hate and if censorship is going to work reliably, it needs a consistent standard.
As you alluded to earlier, people all start setting limits of free speech from a common ground but fairly quickly draw the line at things they are personally sensitive about. Things one person is more sensitive to, is less sensitive for someone else. That doesn't mean they agree with it, they just don't think it reaches a level of offensiveness that requires removal.
He definitely doesn't live up to being a free speech absolutist and this was apparent before he took over because it's inevitable that people do this. People all have this same flaw. Inwardly people see themselves as the arbiters of truth because people have no reason to lie to themselves but they only have their own perspective.Elon Musk has said many things that are outright lies; it is mind-bending that anyone would defer to him. Also, he regularly surpasses speech on X that he disagrees with. Not that it is factually incorrect but he simply doesn't like it. He has in no way lived up to his claims to be a free speech absolutist, rather the opposite.
The BBC interview shows his mindset and intent, he hasn't lived up to this stated intent and I don't think anyone can run a large scale platform only using the law as a base guideline and then tacking on subjective rules. I think he is overly optimistic about how humanity operates at scale.
It's easy to lose sight of the scale of these operations, they are getting hundreds of millions of posts per day. They don't have millions of employees so they rely on software to automate the process.
That's why I think the solution lies in a better convention for conversation. They have to setup a social contract that rewards constructive input instead of destructive. This can be seen on forum sites, the comments on forum sites are usually way above the standard that you get on social media because there's a different convention for conversation. Neutering political conversation helps of course. Politics more often divides people instead of finding common ground.It's important to clarify the word 'fact' here because it's being taken at face value to mean something that is known to be true. Fact-checkers should really be called something else because what they are checking are not yet considered to be facts. The accounts they find to be false are not facts.No one decides what true or isn't. Facts are self evident.
Facts are immutable. Community notes are in control of the platform and its users.
Zuckerberg clarified that their existing policies surrounding facts will remain in place. If something is factually/indisputably terrorist activity, trafficking etc, this won't be left to community notes and will be removed.
Things that are left to notes are unverifiable accounts. If both sides in a war zone try to spread information of war crimes, they don't always have a way of verifying these accounts and their bias would tend to fall on which side of the war they hope will be successful. In the war in Ukraine, Western media is more likely to believe accounts from Ukraine than from Russia and justifiably so but this justified bias doesn't make an account factual.
The point about injecting bleach is valid and caused real harm and I'm sure these kind of statements will fall under their normal guidelines that will remain in place. This isn't something that's vague or debatable.
Thanks for the kind reply. For the AI, this would be some kind of machine learning. Generative AI has reliability issues but machine learning is good at consistent labelling. The models used could be biased but there can be open-source models. These would be primarily trained to detect civility in conversation but can also detect bias in different topics. As long as the same models are used on each platform, they will have consistent output. They can be used to test mainstream media sites for bias. With open source models, individuals won't be able to corrupt them for their own gain.First of all, thank you for composing such a well formed and articulate post. It’s always a pleasure to read a person’s opinion when it is presented so clearly. Whether or not I agree or disagree with what you’re saying isn’t the point. You’ve provided food for thought and added another perspective to consider along with all of the other perspectives stated here.
One response I have to your post is concerning using AI to moderate social media content. I’m not so sure that would work because we’d have to all agree on which AI we would all agree to use and there would always be claims that the AI is itself biased. In the end I doubt that humans will ever defer to AI in policing their personal behavior, which includes a lack of self discipline, empathy, and self moderation.
There will be other ways to accomplish the same goal of encouraging constructive conversation but on a large scale, there needs to be a consistent, automated solution. These sites will use AI in their tools already but they clearly aren't being used to encourage constructive input. They can even have monetary rewards for the most constructive users.
-
Thinner, smarter, more connected: What to expect from a 2025 Apple TV
It would be difficult to get a competitive price and maintain profit.oberpongo said:Make it an Apple TV Pro. Including support for 8k and with an M4 to rival any PlayStation 5 or XBOX in terms of gaming.
The Apple TV is $129, 64GB SSD, A15 (iPhone 13), 4GB RAM.
The Mac mini M4 is $599, 16GB RAM, 256GB SSD. This is half the performance of XBox/PS5, which sell for $450.
They'd have to use an iPhone chip to get the price down and they'd need 8GB RAM as some games use over 5GB:
It might be easier to allow people to dock their iPhone to a TV as there are tens of millions of them in use already, this video shows it works well (7:00):
Then it's more like a Nintendo Switch that docks to a TV.
They might be able to do A18 Pro, 8GB, 64GB for $299 but people would then be weighing up vs an XBox Series S at $300.
In terms of unit volume, getting people to dock iPhones to their TV with a first party dock and supported controller (Xbox One, PS5) would be higher. People can even dedicate an old iPhone to the TV dock.
