Valve Steam Frame Defeats Quest 3 on Privacy but Fails on Value

Valve Steam Frame在隐私上胜过Quest 3但性价比不足 Valve的Steam Frame相比Meta的Quest 3具有更出色的隐私保护和更低的面部发热温度，但其1,060美元的起售价和较差的原生透视功能阻碍了主流市场的普及。 Gamers Nexus对Valve Steam Frame与Meta Quest 3进行了全方位的技术基准测试，涵盖延迟、散热、电池续航及软件稳定性。Steam Frame在低延迟无线PC串流、舒适的面部发热控制以及避免生物识别数据采集的纯本地端侧隐私模式上表现优异。然而，其出人意料的高昂发售价格、不稳定的充电速度、较差的原生单色透视摄像头，加之更广泛的VR软件生态匮乏，使其难以获得极力推荐；而Quest 3则因Meta激进的家庭及生物识别监控政策被正式排除在推荐之外。 在行业内存和闪存成本暴涨高达600%后，Valve推出的Steam Frame VR头显售价为1,060美元（256 GB）和1,300美元（1 TB），突破了此前定下的1,000美元以内目标价。 (0:00) Steam Frame展现出相对于Meta Quest 3显著的无线串流延迟优势，通过其专用的6 GHz PC适配器在LDAT基准测试中实现了46.1毫秒的延迟，而Quest 3为72.5毫秒。 (20:21) 隐私评估揭示了显著差异：Valve在设备本地处理所有摄像头和眼球追踪遥测数据而不进行云端收集，而Meta则会提取大量的生物识别、房间几何结构及语音遥测数据。 (8:28) 测试发现Steam Frame存在严重的充电问题，其在低于12 W的功率下耗时215分钟才充满电，未能复现Valve声称的40分钟内从0充至65%的快充表现。 (34:50)

Gamers Nexus concludes that while the Valve Steam Frame delivers superior hardware privacy, lower latency, and better ergonomics than the Meta Quest 3, its $1,060 starting price and defective stock passthrough severely undermine its value. Conversely, Gamers Nexus treats the $600 Quest 3 as disqualified from recommendation due to Meta's invasive domestic data-harvesting policies. Contrasting Data Privacy and Sensor Telemetry (8:28) Meta requires Quest 3 users to accept roughly 80,000 words of legal terms authorizing extensive data collection within the home. Meta's telemetry captures room geometry, continuous ambient audio, accidental voice assistant activations, biometrics abstracted from gaze and facial tracking, and behavioral data from children aged 10 to 12. Citing congressional testimony from former Meta researchers regarding leadership deprioritizing child safety to protect ad engagement, Gamers Nexus rejects the Quest 3 purely on privacy grounds. In contrast, Valve relies on standard Steam agreements totaling 12,616 words and confirmed that camera and eye-tracking telemetry operate strictly on-device without cloud ingestion or third-party broker sales. Diagnostic logs leave the device only if a user manually submits a support ticket. Valve also allows users to air-gap the headset entirely from the internet by pairing solely to the included wireless PC adapter, and the company removed mandatory arbitration clauses from its subscriber agreement in late 2024. Broken Pricing Promises and Software Bottlenecks (0:00) Valve reversed prior pre-launch guidance that targeted an early 2026 release below $1,000 and undercutting the Valve Index. Instead, the Steam Frame launched in September at $1,060 for 256 GB and $1,300 for 1 TB. Gamers Nexus argues that this premium stalls mainstream adoption, particularly because essential comfort and utility items—such as the top head strap and upgraded passthrough cameras—are sold as separate accessories. This pricing barrier collides with an enduring software drought. Outside established titles like Half-Life: Alyx, niche flight or racing simulators, and flat 2D games streamed to virtual screens, standalone VR content remains limited. Although the Frame's ARM-based SteamOS translation layer (via FEX, DXVK, and VKD3D) runs transparently across roughly 130 launch-verified titles, Gamers Nexus found controller remapping across headset and host-PC menus deeply convoluted. Ergonomic and Thermal Gains Versus Passthrough Flaws (14:31) The Steam Frame's physical chassis proved exceptionally comfortable in testing. Its open nose cut-out and breathable gasket prevented the intense facial clamping and heat retention observed on the Quest 3. Controlled radiometric benchmarking using a Fluke RSE600 infrared camera after 30-minute gaming sessions confirmed the Frame ran cooler against the skin, recording 33.0°C at the brow line versus 35.0°C on the Quest 3, alongside lower temperatures across the cheeks and nose. However, Gamers Nexus labeled the Frame's stock monochrome passthrough cameras intolerable and nausea-inducing. The feed operates at a choppy 30 FPS with heavy optical distortion and spatial warping that renders text and phone screens unreadable. Resolving this requires Valve's optional $150 Arcturus dual-RGB add-on module, which outputs full color at up to 2464×2464 per eye over a 4-lane MIPI interface, though at the expense of higher battery consumption. Latency Advantages, Battery Limits, and Charging Quirks (20:21) Using an acoustic-trigger LDAT capture methodology, Gamers Nexus measured lower motion-to-photon latency on the Steam Frame than on the Quest 3. In native testing of Thumper at 120 Hz, the Frame clocked 56.5 ms, while streaming via its dual-band Multi-Link setup added only 9.2 ms (65.7 ms total), noticeably outpacing the Quest 3's wireless streaming (83.5 ms). Valve also implemented an aggressive 8% duty cycle on its displays, illuminating pixels for just 0.67 ms per frame at 120 Hz to minimize persistence blur. Conversely, the Quest 3 demonstrated superior battery longevity. In native Moss sessions, the Quest 3 lasted 120.2 minutes compared to just 47.0 minutes on the Frame under maximum local performance, though the Frame reached 124.6 minutes when restricted to its 72 Hz Playtime profile. During real-world Half-Life: Alyx gameplay, the Quest 3 outlasted the Frame by 20% (139.2 versus 115.8 minutes). Furthermore, the Frame exhibited anomalous charging: connected to an official 45 W supply, it drew only 11–12 W peak and required 215 minutes to recharge, falling far short of Valve's 90-to-120-minute specification. Gamers Nexus presents consumers with an unresolved market dilemma: the $600 Meta Quest 3 offers solid standalone performance but is disqualified by invasive corporate data harvesting, while the Valve Steam Frame provides commendable local data privacy and low latency but asks an uncompetitive $1,060-plus price burdened by flawed stock passthrough and erratic battery charging.

[9:04] Valve: Camera and eye tracking data is processed on device, and we don't collect it, nor do we collect byproducts of it. [12:21] Former Meta Researcher: Children drive profits. If Meta invests more in safety to get kids off of there, engagement goes down, monetization goes down, ad revenue goes down. They need them, they need them young... But Meta's smart, they know that too. They need them, they need children. [28:40] Valve: All the links are always active and competing for the lowest latency on a per packet basis, acting as a single robust connection with built-in redundancy that adapts in real time. This means your streaming connection is always as powerful as the strongest link and often better than any single link. [30:04] Highly optimized games and streaming may use less than 10 watts, while some unoptimized or higher spec games may draw up to 25 watts peak. [34:50] a dead battery takes around 2 hours to charge to full capacity. However, the charging is not linear. There's fast charging from 0 to 65% and regular charging from 65 and up. Charging to 65% is expected to take less than 40 minutes. [34:50] From everything shared, it sounds like your experiment is valid. However, we are seeing different results on our end. [42:53] Steve Burke: probably because they're stealing all of your data. [54:00] Mike Gaglione: like a virus. [63:25] Patrick Lathan: I feel like there's a bunch of engineers out there that didn't think that this was going to be the price that they landed on. [66:13] Steve Burke: unusable and intolerable [69:01] Steve Burke: Just because we've been fucking it up, doesn't mean we need to keep fucking it up for the rest of time.

[0:01] The Valve Steam Deck costs $1,060 for [0:04] the 256 GB model and $1,300 for the 1 TB [0:08] model. For this virtual reality headset, [0:10] the only information we had about [0:12] Valve's pricing and release date until [0:13] now was less than $1,000, less than the [0:17] maximum price of the Index, and early [0:19] 2026. [0:21] $1,050 is not less than $1,000 and [0:26] I was just checking, it's September. [0:28] Now, the Quest 3 is the big competition. [0:31] And the Meta Quest 3 costs $600 [0:36] and your firstborn. We paid $600 for [0:38] this Quest 3 last week. Also, we gave up [0:40] our firstborn alongside video and voice [0:42] data from inside of the house, scans of [0:44] the room, body movements, facial [0:45] expressions, reading around 80,000 words [0:47] of agreements, and depending on how much [0:48] you spam agree to, quote, audio [0:50] recordings of your child's voice, [0:51] physical information about or related to [0:53] your child, headset location, your [0:54] child's connections to others, your [0:55] child's activity in Meta Quest devices [0:57] and apps, and quote, like apps [0:58] downloaded, gameplay activity stats, [1:00] content created, as well as the time, [1:01] frequency, and duration of activities, [1:03] and with approval, quote, the size of [1:04] walls, surfaces, and objects in your [1:05] child's room, and quote. We asked Valve [1:07] if it did those things, and they didn't [1:09] need 83,000 words to respond. In fact, [1:12] Valve only needed one word. [1:14] No. [1:16] Maybe a second word, which would have [1:17] been [1:18] What? Why? That's two more words. [1:21] Three still a lot shorter than 83,000, [1:23] so. Just seemed like a good time to add [1:25] privacy testing to our reviews for [1:27] hardware considering the big LG [1:28] investigation and, you know, VR and its [1:31] presence of multiple cameras, [1:33] microphones, sensing information, and [1:35] ability to know lots of things about [1:38] your room. [1:39] But we have other tests, too. Including [1:40] latency testing, where Valve overall had [1:42] a favorable position as compared to the [1:44] Quest 3. However, the Quest 3 had a [1:46] favorable position as compared to Valve [1:47] in battery life testing. Thermal testing [1:49] also came into play. One of the big [1:51] tests we did for this was thermals to [1:53] represent comfort in an objective sense, [1:54] so we used a Fluke RSE 600 thermal [1:56] imaging camera that is stationed on a [1:59] tripod. This is connected via a special [2:02] cable that's adapted into PoE that goes [2:05] through a box that goes into the laptop. [2:07] The laptop captures, and you can see one [2:09] of the captures here. So, the purpose of [2:12] this test is to look at not just the [2:14] external temperature of the headset, [2:16] which that's not too important. What's [2:18] really important is the temperature of [2:20] your face when you're wearing it. And [2:21] so, we get the removal temperature on [2:23] the instant it's removed. What's the [2:25] temperature of different points on your [2:26] face? Then I take that data, I compare [2:28] it between the Frame and the Quest, and [2:30] we get kind of an objective read on the [2:33] human factors skin temperature. And [2:35] thermal results position the Frame as [2:36] having some objective advantages in [2:38] comfort alongside the subjective [2:40] opinions from our team. [2:41] >> The Frame definitely felt a lot more [2:43] comfortable just as a whole. [2:45] >> This out of the box is a lot more [2:47] comfortable. [2:48] >> The price is the worst part of the [2:49] Frame, especially considering Valve [2:50] broke its only promises, which were [2:52] related to price. But the Frame itself [2:54] is not the only part that's relatively [2:56] expensive for Valve. The accessories kit [2:58] is $60 for straps that we think are [3:01] mandatory, seeing as you need them to [3:03] make full use of some of the advertised [3:04] features. The charging cable $30, not [3:06] included. The cushion replacement with [3:08] gasket and light blockers $50. So, the [3:10] price is the rough part with the Frame, [3:12] but the price is also the rough part [3:14] with the far cheaper Meta. It's just [3:16] that you're paying for it [3:18] with like literally everything about [3:20] you. [3:21] This is like [3:23] you thought LG's terms were bad? Jesus [3:26] [ __ ] [3:27] That's just the all I have to say about [3:28] this one. [3:30] The market started to get bad right when [3:31] Valve announced these in November of [3:33] last year. And in the time since memory [3:35] prices have gone up on average 350 to [3:38] 600% depending on which spec of memory [3:40] you look at, and storage has increased [3:41] by more than double. It's changed a lot. [3:44] So, price is going to be the contention [3:46] point, but we also have a lot to talk [3:48] about on technology, benchmarks, and and [3:50] else. Let's get into it. Before that, [3:52] this video is brought to you by Lexar's [3:53] Play series of microSD cards, available [3:55] in capacities 256 GB and up, and acting [3:59] as a portable memory card that can [4:00] easily move games between compatible [4:02] Linux systems that support portable [4:04] hot-swap game installs, especially on [4:06] handheld and small gaming devices. Why? [4:08] Because per device storage needs during [4:11] our dystopian times of consumer [4:12] shortages and pricing require creative [4:15] solutions, like allowing fewer portable [4:17] cards at a lower total cost that enable [4:19] moving games from one device to another. [4:21] Again, particularly with Linux builds [4:23] that are compatible with portable games. [4:25] We've been using Lexar's Play series [4:27] microSD cards for some of our smaller [4:29] high FPS action cameras that we use for [4:31] counting frames in benchmarking [4:33] processes and haven't had any issues [4:35] with them. Learn more at the link in the [4:36] description below. All right, so there [4:38] are other VR headsets than these two, [4:40] obviously. There's the Index for Valve's [4:42] predecessor. We actually did some of the [4:43] original VR benchmarks with the Vive and [4:46] the Oculus Rift like 10 years ago. And [4:48] of course, there's companies like [4:50] Bigscreen and Pimax and others. Today, [4:53] we're focusing on these two for the one [4:55] key reason being that by numbers, the [4:57] most popular VR solution on Steam's own [4:59] hardware survey, and time. [5:02] Because in the couple weeks we had to [5:04] work on these, there was an enormous [5:06] amount of test to do, and we wanted to [5:08] go for depth on each of these [5:10] individually rather than kind of [5:12] spreading ourselves shallow across a [5:13] whole lot of devices and rather really [5:15] drill into two. Maybe we'll look at [5:17] those in the future if you're [5:18] interested. If so, leave a comment let [5:20] us know which one you'd like to see cuz [5:21] each one of these takes a massive amount [5:23] of time. All right, Meta completely [5:24] dominates the VR market, even on Valve's [5:26] own PC hardware survey, and the Quest 3 [5:28] is its flagship. The specs are public, [5:30] but here's a rundown. The frame has a [5:32] Snapdragon 8 Gen 3 8650 SOC, 16 GB of [5:35] LPDDR5X, which is expensive now, serving [5:38] as both the CPU and the GPU memory [5:40] combined, 256 GB or 1 TB of onboard [5:44] storage with a microSD card slot for [5:46] expansion, 2160 by 2160 LCDs with [5:50] pancake lenses, refresh rates from 72 to [5:53] 144 Hz with 144 Hz being marked as [5:56] experimental, field of view or FOV up to [5:58] 110°, four external IR sensitive [6:00] monochrome cameras, two internal eye [6:03] tracking cameras, an expansion port with [6:05] both of single PCIe Gen 4 lane and quote [6:08] eight lanes at 2.5 gigabits per second [6:10] MIPI and quote Wi-Fi 7, Bluetooth 5.4 [6:14] and a 21 .6 watt-hour battery on the [6:16] head strap. The whole thing weighs about [6:18] 445 g plus or minus a gram or so by our [6:21] own measurements. The core module [6:23] containing the computer, optics, screen [6:25] and all the other expensive stuff locks [6:28] into the modular head strap which only [6:30] contains the battery and the speakers. [6:32] The frame comes with the headset, a [6:34] spacer for glasses, two controllers with [6:36] lanyards, the special wireless adapter, [6:39] a USB type-C cable and a cleaning cloth. [6:42] You don't get this AC adapter included. [6:44] This is a $30 optional accessory that [6:47] matches the Deck Charger. Although any [6:50] USB power supply will work such as the [6:52] one that comes with the Quest 3. A [6:53] replacement kit for the stock cushions [6:56] and rubber nose piece run $50 in the [6:58] event they deteriorate or if you just [7:00] think they get nasty over time. The [7:02] single USB type-C port is used for [7:03] charging and data but only at USB 2 [7:05] speed so 480 megabits per second. Unlike [7:08] the Quest 3, the frame lacks a 3.5 mm [7:11] headphone jack. The type-C port however [7:13] can be used with 3.5 mm adapters. The [7:15] stock headphones were designed to be [7:17] lightweight and portable Valve said, but [7:19] an enthusiast head strap kit will be [7:21] launching in a few months with index [7:23] style headphones and a hot-swappable [7:25] battery. This will be an add-on kit not [7:27] another frame SKU. So the kit will only [7:30] include the accessories. Valve has a [7:32] number of DLC add-ons like the ergonomic [7:34] accessories kit for $60 US with [7:36] controller hand straps, a headset top [7:37] strap and an extended light blocker. [7:39] Some all feel necessary. The hand straps [7:41] are important for finger tracking since [7:43] they let you fully open your hands to [7:45] actually utilize that feature. And the [7:47] default light blocker may leave a [7:49] significant open gap depending on your [7:51] face shape. It's very dependent on the [7:53] person. Patrick found the light blocker [7:55] accessory unnecessary while I personally [7:58] found it necessary because the extreme [8:00] amount of light coming in at least for [8:02] me. The optional top strap just got in [8:04] the way for some of us. For me, it [8:06] helped with headset bounce when walking [8:08] possibly because it also helps keep the [8:10] strap secure over long hair, which is [8:12] something I've always had trouble with [8:14] for VR headsets. We think some of [8:16] Valve's first-party accessories should [8:17] be included with the Frame. For example, [8:19] the hand straps that are necessary to [8:21] actually fully utilize the advertised [8:22] finger tracking feature on the [8:24] controllers beyond taking one or two [8:26] fingers off the controller. Up next, [8:28] thanks to our LG deep dive, we have a [8:29] new section that we've never had in [8:31] reviews before. A review of the privacy [8:33] settings, the extent of the device's spy [8:35] capabilities, and the privacy policies. [8:38] After all, both the Frame and Quest are [8:40] loaded with cameras, microphones, and [8:42] motion tracking, and Meta's involved. [8:45] Starting from checkout, Valve requires [8:47] users to accept two agreements when [8:48] purchasing its hardware. Valve's privacy [8:50] policy and the Steam Subscriber [8:51] Agreement totaling a collective 12,616 [8:54] words. If you have a Steam account, [8:55] you've likely already accepted these. [8:57] Valve's privacy policy explains how it [8:59] collects account data, gaming activity [9:01] and statistics, device information, [9:02] transaction payment data, posts, [9:04] comments, chats on Steam, interactions [9:06] on the Steam client, and tracking [9:08] cookies and pixels, which it claims it [9:10] uses to provide both its services and [9:12] content recommendations. We asked Valve [9:13] about the Frame specifically, and it [9:15] told us this, quote, "Camera and eye [9:17] tracking data is processed on device, [9:19] and we don't collect it, nor do we [9:21] collect byproducts of it. Steam Support [9:23] has long had the ability for users to [9:25] choose to submit a system report to help [9:27] us troubleshoot problems they may be [9:28] having. For a Steam Frame, that report [9:30] would contain text logs that include [9:32] some camera and eye tracking systems [9:34] data. As always, at Valve, it never [9:36] sells user data to third-party [9:38] advertisers." End quote. Now we get to [9:39] Meta, which will take longer. Meta's [9:42] Quest 3 checkout, on the other hand, had [9:43] seven user agreements totaling over [9:45] 36,000 words. That includes Shopify's [9:47] terms and privacy policy that Meta [9:48] attempts to getting users to [9:49] unnecessarily agree to. Then the Meta [9:51] Horizon app that Meta claims is required [9:53] to complete the Quest 3 setup requires [9:54] users to accept an additional five [9:56] agreements containing 46,579 words. So [9:58] you're now at almost 83,000 words of [10:00] agreements before using the headset. [10:01] Meta's app store page says, quote, "The [10:03] following data may be collected and [10:04] linked to your identity, health and [10:06] fitness, purchases, financial [10:07] information, location, contact [10:08] information, contacts, user content, [10:10] search history, browsing history, [10:11] identifiers, usage data, sensitive [10:13] information, diagnostics, and other [10:15] data." End quote. In addition to [10:17] collecting all the same types of [10:18] information as Valve, Meta also records [10:21] and processes all voice interactions [10:23] using its voice services, quote, [10:25] "including conversations with AI-powered [10:27] features." End quote. And they say, [10:28] quote, "including if you mistakenly [10:30] invoke our voice services, as well as [10:33] any background sound that happens." End [10:35] quote. So they're pulling an LG, but [10:37] they've actually written it a little bit [10:38] better. Meta claims that voice dictation [10:40] recordings aren't stored. Meta also [10:42] collects all video and audio content you [10:45] create, including, quote, "video or [10:47] audio content and information about this [10:48] content, such as the date and time you [10:50] created it." End quote. And also, quote, [10:52] "environmental information and [10:53] dimensions." End quote. Like when you [10:54] scan your room, and what it calls, [10:56] quote, "abstracted" and quote, "hand and [10:57] body data, gaze data, facial expressions [10:59] data, and fit adjustment data." Meta [11:01] claims that raw image data is processed [11:03] locally, but says it processes what it [11:05] calls the abstracted data, and that it [11:06] collects and retains certain data [11:08] collected from those interactions, [11:09] explaining in its natural facial [11:10] expressions privacy notice, quote, "When [11:12] you choose to enable natural facial [11:14] expressions on the headset or in a [11:15] specific app, software on the Meta Quest [11:17] Pro headset analyzes images of your [11:19] face, raw image data, to create an [11:20] estimate of how your face is moving, [11:22] producing a set of generic facial [11:24] expressions, like a broad smile or [11:25] frown. The abstracted facial expression [11:27] data is generated and overwritten in [11:28] real time on your headset and processed [11:30] on device or Meta servers [music] to [11:32] animate your avatars expressions. The [11:34] abstracted facial expressions data may [11:36] be processed on meta servers in order to [11:38] animate your avatars facial movement in [11:40] a multiplayer scenario. Meta notes, [11:42] quote, "Additionally, when you use [11:43] worlds, the last few minutes of your and [11:45] other users most recent audio, video, [11:47] and other interactions in worlds are [11:49] recorded [music] in case you want to [11:51] report anything you've encountered. [11:52] These recordings may be stored on our [11:54] servers." End quote. But, they do say [11:55] that they don't review them unless you [11:57] report it. And, they say that the [11:58] recordings are deleted on a rolling [12:00] basis. Meta also assigns users [12:02] advertising IDs because obviously, [12:04] enabled by default, and also has [12:05] significantly more user agreements and [12:07] includes a forced arbitration clause in [12:10] its terms. Valve actually removed its [12:11] mandatory arbitration clause previously [12:13] and class action waiver from its Steam [12:15] subscriber agreement in late 2024 after [12:17] getting, quote, "threatened with an [12:18] avalanche of individual dispute [12:19] resolution demands." End quote, as [12:21] reported by Reuters. In Meta's parent [12:23] privacy disclosure, the company tells [12:25] parents what information it collects [12:26] about their children aged 10 to 12 years [12:28] old when using Meta VR products, [12:30] listing, quote, "Audio recordings of [12:32] your child's voice." End quote, if voice [12:34] services are enabled. And, quote, "Video [12:36] and audio information recorded when [12:37] cameras and microphones are enabled. [12:39] Physical information about or related to [12:40] your child." End quote, if hand and body [12:42] tracking's enabled. Headset location, if [12:44] location services are enabled, quote, [12:45] "Your child's connections to others, [12:47] your child's activity and Meta Quest [12:49] devices and apps." End quote, like apps [12:51] downloaded, gameplay activity and stats, [12:52] content created, as well as the time, [12:55] frequency, and duration of activities. [12:56] And, with approval, quote, "The size of [12:59] walls, services, and objects in your [13:01] child's room." End quote. We wouldn't [13:03] want literally anybody having access to [13:06] that much of a child's information. And, [13:08] certainly not a company with Meta's [13:09] track record. And, this isn't [13:11] theoretical. Only a year ago, two former [13:13] Meta researchers, quote, "studying the [13:14] welfare of underage users on Meta VR [13:16] products." End quote, told a Senate [13:18] subcommittee that Meta's VR products [13:20] exposed children to adult content and [13:22] claimed that, quote, "Meta disregarded [13:23] their findings, censored their attempts [13:25] to do further research, and in at least [13:27] one instance deleted evidence of youth [13:29] being harassed, and quote with words [13:31] that YouTube doesn't like, as reported [13:33] by NBC. As for why Meta needs so much of [13:36] your child's information, one of the [13:38] researchers explained, [13:39] >> Children drive profits. If Meta invests [13:42] more in safety to get kids off of there, [13:44] engagement goes down, monetization goes [13:46] down, ad revenue goes down. They need [13:48] them, they need them young, and we know [13:49] that younger audiences drive later [13:51] engagement across many types of [13:53] products. But Meta's smart, they know [13:54] that too. They need them, they need [13:56] children. [13:56] >> And back to Valve, as we said, it's the [13:58] two Steam agreements. They don't [14:00] exfiltrate data for room, object, or [14:03] body off of device, and they don't sell [14:05] data to third-party advertisers. As far [14:08] as we can tell, the only time Valve uses [14:10] data is in its optional cases where [14:12] users submit support tickets. For [14:14] example, if you complain about eye [14:15] tracking or head tracking issues, or the [14:17] data that it already uses for your Steam [14:19] account for things like content [14:20] recommendations, and things that are on [14:22] the Steam store, rather than just the VR [14:23] headset. So, pretty big difference [14:25] between the two products, and maybe a [14:27] bit of an explanation as to how Meta can [14:29] price their headset low. Out of the box, [14:32] almost our entire team found the Frame [14:33] easier to put on and more comfortable on [14:36] the face than the Quest 3, a massive [14:38] improvement over the original Vive. [14:40] Multiple members of our team consider [14:41] the Frame the most comfortable VR [14:43] headset they've used. The openness of [14:45] the nose area on the Frame makes it [14:46] easier to, for example, drink from a can [14:48] while using it, versus a Quest 3 with [14:51] its nose overhang and enclosure. [14:53] However, some people may find this [14:54] positive to be the reason light bleeds [14:56] into the headset, depending on face [14:58] shape. The comfort and casual nature of [15:00] the Frame is probably its biggest [15:01] advantage over the Quest, we think, but [15:03] we also haven't tried any of the [15:05] ultra-light weight headsets, like the [15:07] Big Screen Beyond 2. So, leave some room [15:09] for, obviously, what those may provide. [15:12] We haven't looked at them. The face [15:13] gasket and eye spacer use magnets, and [15:16] the rear cushion uses hook and loop [15:18] patches. The magnets take some getting [15:20] used to, as the face gasket pulls off [15:22] extremely easily. The stock position of [15:24] the lenses is potentially an eyelash [15:26] brushing range, meaning that the inner [15:28] glass gets dirty relatively easily, even [15:30] if not touching it, just from eyelash [15:32] contact. Although, just to be clear, we [15:34] couldn't physically feel that contact, [15:37] it did involve more frequent cleaning. [15:39] We left the spacer on even without [15:42] glasses for this reason. Maximum claimed [15:44] width for glasses is 140 mm, but we were [15:47] able to squeeze in a pair with arms out [15:49] to 145 mm. We'll cover this in the [15:52] thermal section, but anecdotally, we [15:54] didn't feel unpleasantly hot, for what [15:55] it's worth, or uncomfortable with the [15:57] weight and balance of the frame. There's [15:59] a very obvious exhaust vents on the top, [16:01] which we prefer to the hidden cracks on [16:03] the Quest 3. For cleaning, Valve says to [16:05] use their microfiber cloth and water, no [16:08] chemicals. They were clear about that. [16:10] We found it difficult to keep the lenses [16:11] clean, even if not touching them with [16:13] our hands, although this didn't [16:15] noticeably affect vision. So, our next [16:17] test is going to be for thermal imaging, [16:18] and for this test, and we're actually [16:21] basically going to be using a thermal [16:22] camera for one of its only actual real [16:25] uses that works really well, which is [16:28] imaging people. So, originally, I was [16:31] trying to test this with thermal [16:33] couples, which um [16:35] you you almost got a uh [16:38] like a makeup tutorial with me, except [16:40] it was going to be me attaching these [16:42] all over my face. [16:44] And then they wouldn't stay on. Uh and [16:46] so, what we resorted to instead is [16:48] thermal imaging. And the test [16:50] methodology for this is pretty simple. [16:52] So, we have a Fluke RSE 600 camera. [16:56] Um these are pretty high-end, by thermal [16:58] imaging standards, high-resolution [16:59] cameras. I played games on the Quest and [17:02] in the Steam Frame, stationary only, so [17:04] that I could stay basically fixed on the [17:06] camera, and did that for 30 minutes. So, [17:09] basically, set up here, [17:11] uh played through the games, [17:12] and then, after the 30-minute test [17:15] period, [17:16] I was pretty controlled with how I [17:17] removed the headset and doing two [17:19] things. One was making sure everywhere [17:22] my face was visible immediately on [17:23] removing the headset so we can get a [17:25] couple frames there right on removal to [17:28] see where were the hottest points on the [17:29] face [17:30] and then flipping the headset around at [17:31] the same time so we can see internally [17:33] where the hottest points in the headset [17:35] and map those two things together. So [17:38] basically what this does is it allows us [17:40] to get kind of a an objective [17:42] measurement of comfort from a thermal [17:44] standpoint on the wearer's face and by [17:47] taking that measurement instantly on [17:49] removal we're able to capture sort of [17:51] those last moments before your skin [17:52] starts cooling down again to see what it [17:55] feel like inside the headset. To create [17:57] the most thermal load for it we ran all [17:58] the test locally and then we also ran it [18:01] while charging so it's a worst case [18:03] scenario of local run not streamed and [18:05] charging and we used moss for the [18:07] capture. I basically did an as like for [18:09] like version of the test as I could [18:11] across the 30 minute span and then we [18:14] matched the game's settings so it was [18:16] running at 90 hertz on both devices so [18:19] that the Quest and the Frame were as [18:20] close as we can reasonably get them to [18:22] each other. We're not normalizing for [18:24] power draw we're normalizing for [18:25] basically how does it actually just [18:27] function out of the box. So let's take a [18:29] look at the results. Here's the chart. [18:30] These numbers represent the immediate [18:32] removal of the headset. This is within [18:33] 1/3 of 1 second after taking the headset [18:36] off. We captured brow line numbers which [18:38] land right beneath the gasket for each [18:40] headset at five different points. Those [18:42] had the Frame at 33 degrees to the Quest [18:44] 35 at the hottest point. This is our [18:46] most useful figure because the larger [18:48] gap seems more likely a real impact [18:50] rather than some kind of variance that [18:51] could be influenced by human factors. [18:52] Relating to those I did all these tests [18:54] on the same day. I had a brief cool down [18:56] period in between the two tests. The [18:58] room temperature was the same and we [19:00] also have a control test shown in blue [19:01] where I captured basically my face [19:03] temperature without any headset mounted [19:06] and after a period of cool down leading [19:07] up to it. Brow line marks two to four [19:09] also favored of Frame by about half a [19:11] degree in some cases. The brow line 5 [19:13] measurement had the frame at 34.2 to [19:15] 35.0 of the Quest 3. So, objectively, [19:17] it's cooler here. Again, with the kind [19:19] of asterisk of larger variance due to [19:21] human factors. Eye temperature also ran [19:23] warmer on the Quest 3 at 34.7 to 35.1° C [19:27] versus 33.9 to 34.2° C for the frame. [19:31] The Quest 3 ran noticeably warmer on the [19:33] cheek right below the eye and at a [19:35] physical contact point with the headset. [19:37] 33.6 is an advantage against the Quest [19:39] 35.1. The nose hotspot was also [19:41] noticeably higher with the Quest at 34.6 [19:44] to 32.8 on the frame. Given the headset [19:46] shape, that makes sense. The nose is [19:48] less covered with the frame. We also [19:50] captured palm temperature hotspots by [19:52] holding hands up to the camera [19:53] immediately after setting the [19:55] controllers down, but the imperfect [19:56] nature of this measurement makes it [19:58] unreliable. The headset lenses ran [19:59] warmer on the frame, which indicates to [20:01] us that its internal ventilation must [20:03] just be working better than the Quest [20:04] given the internal face temperature was [20:06] lower. HMD or head-mounted display front [20:09] external temperatures were also higher [20:10] on the frame, but those are areas that [20:12] don't contact the skin. Also, those are [20:14] the areas you want it to be warmer [20:16] because otherwise, I mean, it's got to [20:17] go somewhere. It's better for it to be [20:19] on the very front of the headset than [20:20] further back. For latency testing, so as [20:22] I mentioned in the intro, we normally [20:24] would like to wire into the controller [20:26] physically. The reason we do this, so [20:28] this is a lead and uh these leads are [20:33] used for basically giving us a a signal [20:35] response from just the actual literal [20:38] electrons when you push the button, it [20:40] will send that signal both to the system [20:42] where as the player it's being received, [20:44] but also to our capture device, uh which [20:47] allows us to then measure the response. [20:49] And latency's very challenging to test [20:51] generally cuz there's different [20:52] understandings of the meaning. What we [20:54] want today is the VR system latency and [20:59] eliminate all other sort of unnecessary [21:01] latency in the pipeline. The point of [21:03] testing is from press to that signal [21:07] then going to the system if it's [21:09] streaming for example so it goes to the [21:11] host computer it hitting the headset and [21:13] the headsets displays showing some kind [21:16] of physical response with light. So you [21:18] normally call this a click to photon [21:20] latency. The problem we ran into was [21:23] this uses adhesive and uh we only have [21:26] one Steam Frame. Normally I'm not really [21:29] worried about pulling stuff apart for [21:31] purposes of testing it is our job in [21:33] this instance we started to pull it [21:34] apart and then we just became worried [21:36] that we would damage something because [21:39] uh we need this stuff to finish the rest [21:40] of the testing. So we had some [21:42] solutions. One of them was to wire into [21:44] a controller instead forget the Steam [21:46] Frames included controllers and use this [21:49] and then run this via USBC directly into [21:52] the headset. We faced problems with that [21:54] as well and so we gave up on both those [21:57] ideas and then what we did was this this [21:59] microphone rig. [22:01] So this is not quite as fast as the [22:04] speed of electrons in wires but the [22:07] speed of sound is still pretty [ __ ] [22:08] good for purposes of latency testing for [22:10] video games. It's like 343 m per second [22:12] I think. And so uh you can push the [22:15] button next to a microphone and if it [22:17] were an LG microphone it'd be even [22:19] better. That's our next upgrade. You [22:21] push it here the the sound obviously is [22:23] picked up by the mic that transmits down [22:25] to our L dot solution [22:28] uh and then we also measure the photonic [22:29] response in the headset meaning when [22:32] something flashes in the headset we know [22:35] that the input from the press has been [22:37] received by the system processed and the [22:39] user can now see it uh and the only [22:41] thing we don't measure there is human [22:43] latency cuz it doesn't matter for this [22:44] test. The host system it's streaming [22:45] from is our GPU test bench this is our [22:48] standard bench it's published on the [22:49] website if [22:50] it has a 9800X3D overclocked and a 5090 [22:53] so this should be about the best option [22:55] you can get uh and we're isolating down [22:58] to basically the headset at this point [23:00] that's obviously still in the chain and [23:02] then for the settings for the headset [23:04] we're testing at 120 hertz for the [23:05] games. Technically the frame can do 144, [23:07] it's marked as experimental. The Quest [23:10] can technically also do pretty high [23:12] refresh rates that are optional through [23:14] other means, but we matched them both to [23:16] 120 for testing purposes. And then as [23:18] for additional notes on the software [23:19] side, I'll throw it over to Patrick. [23:20] >> The easiest test would be to stream this [23:22] regular L.D.A.T. test application in [23:24] theater mode in VR, but the pipelines [23:27] for VR streaming and non-VR 2D streaming [23:30] are different things. Uh for example, [23:32] multi-link streaming doesn't work with [23:34] non-VR software. So, we did some initial [23:37] tests like this, but the numbers don't [23:39] apply to that VR. [23:40] >> Here's the first chart with controlled [23:42] testing using the L.D.A.T. software [23:43] Vsync to 120 hertz on the PC side. [23:45] Otherwise, it was hit or miss whether [23:47] we'd stream the white frame needed to [23:48] trigger the L.D.A.T. These tests were [23:50] done at 1920 by 1080 with video quality [23:52] set to the beautiful setting, unlimited [23:54] bandwidth, a 240 FPS cap set for remote [23:57] play, and a 350 megabit per second cap [23:59] in Steam VR. Inputs were issued by a [24:01] mouse connected to the host PC, so the [24:03] controllers aren't part of these [24:05] numbers. This is controlled testing. [24:07] Throughout testing, we recorded three [24:08] baselines with 100 samples each on the [24:10] 360 hertz monitor, so 300 samples, which [24:13] fell between 24.3 milliseconds and 25.6 [24:15] milliseconds. At 120 hertz, the frame [24:18] over the wireless adapter averaged 46.1 [24:20] milliseconds, and the frame over Wi-Fi [24:22] averaged 47.1. That these figures are so [24:26] close is an achievement on its own, and [24:28] the adapter is technically better here. [24:29] The Quest 3 on Wi-Fi averaged 72.5 [24:32] milliseconds in the same test. The Quest [24:34] 3 therefore required 57% more time in [24:37] the latency measurement to make the trip [24:39] as compared to the Frame's best result. [24:41] That's an incredibly strong start for [24:43] the Steam Frame and a rough one for the [24:45] Quest, but VR game streaming is the real [24:47] test. We're using Thumper as a test [24:49] scenario since it can run natively on [24:51] both the Frame and the Quest, and it can [24:52] be played locally on the PC to stream. [24:55] We're also using it because Thumper has [24:56] clear actions that are photonically [24:59] observable. The disclaimer of test [25:01] limitations is that we're testing one [25:02] action in one game. So, these are not [25:05] the set in stone latency numbers for [25:06] each device, and it will change based on [25:08] the game and the game's load. The point [25:10] here was trying to keep it as [25:11] like-for-like as possible between the [25:13] two devices. Here's the chart. Big [25:14] picture, the Steam Frame's latency in [25:16] this specific test is superior to the [25:18] Quest 3 with the exception of the Steam [25:20] Frame on Wi-Fi without the wireless [25:22] adapter or Multilink. We kept the same [25:24] 120 Hz refresh rates and maxed out [25:26] stream settings since the point of the [25:27] Frame is supposed to be that it can [25:29] reliably handle high-quality streaming. [25:31] Our control test with the monitor [25:33] measured 24.8 ms. Running Thumper local [25:36] to the Frame at 120 Hz, we measured a [25:38] 56.5 ms response time to what the player [25:42] would see. At 72 Hz, predictably, that [25:44] got worse with fewer cycles and climbed [25:47] to 61.2 ms. With Multilink streaming [25:50] from a host non-local and at 120 Hz, we [25:52] measured a 65.7 ms latency. Versus [25:55] local, the streamed test added 9.2 ms, [25:59] increasing the time measured by 16%. [26:01] Using only the wireless adapter pushed [26:03] that to 68.4 ms, another 2.7 ms or 4% on [26:07] top of Multilink. Multilink does seem to [26:10] be better than just the adapter in at [26:11] least this specific test and by a [26:13] measurable amount. The Quest shows up at [26:15] this point locally, which uses a [26:17] different build via the Meta Store, but [26:18] we have no control over that, so there [26:20] may be some variance from them. The [26:22] response with a default 72 Hz refresh [26:24] was 75.2 ms. Like-for-like against the [26:27] 72 Hz Frame locally, that's about a 23% [26:31] increase in measured latency versus the [26:33] 61.2 ms result of the Frame. Noting [26:35] again the build difference. Running from [26:37] the host computer rather than locally, [26:39] we obviously see some slowdown. We opted [26:41] into the beta version of the Steam Link [26:43] app on the Quest to get plug-and-play [26:45] streaming over USB using the cable that [26:47] shipped with the headset. Assuming the [26:49] beta feature is was correctly, it didn't [26:51] have much of an effect and landed at [26:53] 82.9 milliseconds. That was with the [26:55] cable that ships with the Quest 3, which [26:56] isn't good for high-speed transfers, and [26:58] Meta wants to sell you an $80 link cable [27:00] instead, which we didn't have in time [27:02] for testing. With the Quest 3 over Wi-Fi [27:05] and 120 Hz, we measured 83 and 1/2 [27:07] milliseconds. The Steam Deck on Wi-Fi [27:09] and 120 Hz ran at 86.2 milliseconds. [27:12] Keep in mind that specific results for [27:14] Wi-Fi will depend on your router [27:16] solution. Our tests were conducted with [27:17] like-for-like hardware and in the same [27:19] spot. So, the Wi-Fi solution was fixed [27:21] for all tests. The low-persistence [27:23] displays of the VR headsets make it hard [27:25] to use the LDAT or other latency [27:27] measurement devices, since they rely on [27:29] changes in luminance and the screens are [27:31] rapidly strobing. You can see this [27:33] effect easily by filming from within the [27:35] headset. In the original announcement, [27:37] Valve told us that the duty cycle for [27:39] its LCD panels was locked at 8% at all [27:42] frame rates, meaning that at 120 Hz, the [27:44] panels are illuminated for approximately [27:46] 0.67 milliseconds out of every 8.3 [27:49] millisecond refresh. This reduces blur [27:52] and is part of the reason Valve opted [27:53] for LCD over OLED. Up next, we'll [27:55] explain how foveated streaming and [27:57] multi-link work. The Frame is presently [27:58] wireless only. Valve's included wireless [28:00] adapter gives the Frame an advantage [28:02] over the Quest 3 in households with bad [28:04] wireless signal strength or old routers. [28:06] We found the wireless adapter mostly [28:07] suited to streaming in the same room. [28:09] Walking about 8 m away from the test [28:11] bench and facing directly away after [28:13] that made the signal start dropping out. [28:15] While facing towards the bench set it [28:17] back to excellent status. So, it's [28:19] decent range, but you'd have trouble [28:20] piercing through walls. The Quest 3's [28:22] solution to bad Wi-Fi is that it can be [28:24] used wired. As of August 31st, Valve [28:27] updated the beta version of the Steam [28:28] Link app on the Quest to make it plug [28:31] and play. The Frame can connect to [28:32] normal Wi-Fi, and it can also connect to [28:34] the included wireless adapter. It has [28:36] two radios. When you connect the Frame [28:37] to Wi-Fi, it gives the option to reserve [28:39] its 6E connection specifically for the [28:42] wireless adapter. If you connect the [28:43] Frame to the wireless adapter, but not [28:45] to Wi-Fi, you can still stream games [28:47] locally through the adapter, but your [28:50] headset won't be connected to the [28:51] internet. If you wanted to, this would [28:53] be the way to isolate your headset away [28:55] from the internet while still being able [28:56] to use it wirelessly. The frame can do [28:58] multi-link streaming as well with the [29:00] simultaneous 5 GHz Wi-Fi and 6 GHz [29:02] wireless adapter connections. Valve told [29:05] us, quote, "All the links are always [29:07] active and competing for the lowest [29:08] latency on a per packet basis, acting as [29:11] a single robust connection with built-in [29:13] redundancy that adapts in real time. [29:16] This means your streaming connection is [29:17] always as powerful as the strongest link [29:20] and often better than any single link." [29:21] End quote. Foveated streaming is the new [29:23] concept Valve's been pushing that [29:24] utilizes the frame's eye tracking to [29:26] prioritize high bit rate streaming [29:28] directly in front of your pupils, but [29:30] there's not much we can do to [29:31] demonstrate it unfortunately in video [29:33] recordings because of the nature of how [29:34] it works. There is a debug tool that [29:36] highlights the foveated area that gets [29:38] the highest quality stream, so we know [29:40] it's active, we can show that, but by [29:42] design it's hard to notice what's going [29:45] on and you'd really need your eyes in [29:47] the headset, not looking at a YouTube [29:48] video on a screen, to see it. Valve [29:50] says, quote, "The angular size of the [29:52] foveated area is a function of game [29:54] resolution and the encoded video size." [29:57] End quote. We kept the encoded video [29:59] size slider maxed out at its default [30:01] value of 1344 pixels for everything in [30:04] this review. Battery life and charge [30:05] testing's next. Meta doesn't have a [30:06] battery capacity listed on its main [30:08] Quest page, it just says it should last [30:10] 2.2 to 2.4 hours overall. The most [30:13] common number we found is 19.4 watt [30:15] hours from an iFixit teardown, but the [30:16] label on the battery in that video looks [30:17] like it says 19.12 watt hours. Either [30:19] way, it's probably somewhere around 19 [30:20] to 20. Valve's battery claim is 21.6 [30:23] watt hours, although there could be [30:24] differences in measurement between the [30:25] companies. Valve told us that, quote, [30:28] "Highly optimized games and streaming [30:29] may use less than 10 watts, while some [30:31] unoptimized or higher spec games may [30:32] draw up to 25 watts peak." End quote. [30:34] That works out to about 2.16 hours of [30:36] streaming at 10 watts, so on par with [30:38] the Quest on paper by claim. Valve also [30:41] claims that idle screens off power draw [30:43] should be about 3 watts and the [30:44] long-term low power state draws about [30:46] 250 mW. For our first battery test, we [30:48] picked Moss, whose developers sadly shut [30:50] down 2 days before the Frame launch. [30:52] Starting with the Quest 3, we have [30:53] control over very few variables on the [30:55] headset itself, basically brightness and [30:57] volume. There is a battery saver mode, [30:59] but we kept it toggled off for all [31:01] testing and we used the L dots luminance [31:03] measurement to roughly equalize the [31:05] Quest 3 to the Frame's default [31:06] brightness, which is about 50% on the [31:08] Quest slider. So, these tests are [31:11] brightness normalized. For streaming, we [31:12] set a 120 Hz refresh rate through Steam [31:15] Link and left the per eye resolution [31:16] slider at its default of 150% or 2528 by [31:20] 2704. We then loaded into the first [31:22] level and let the headset spin around in [31:24] circles until it died. This is a [31:25] relatively abusive test because of the [31:27] constant motion, but it allows us to [31:28] perfectly repeat the test and it doesn't [31:30] have human factors. Now, where human [31:32] factors would affect this would [31:33] definitely largely be heat generated by [31:36] a person wearing it, so we'll look at [31:37] that next. Here's the chart. Running the [31:40] meta build of Moss locally, we lose [31:42] direct control over refresh rate, [31:44] resolution, and in-game graphics [31:45] settings, but we know it's running at 90 [31:47] Hz. The Quest 3 lasted 120.2 minutes [31:49] locally and 118 when streamed, but with [31:51] a higher refresh rate of 120 Hz. We did [31:54] multiple rounds of tests for the Frame. [31:56] First, a high-end streaming test with [31:58] system profile set to performance, GPU [32:00] maxed at 903 MHz, all refresh rates game [32:03] and display set to 120 Hz, and [32:05] resolution per eye still at default 150% [32:07] or 2644 by 2644 on the Frame. With those [32:10] settings, the Frame died in 84.1 [32:12] minutes, so the Quest 3 lasted 40% [32:14] longer in as close to a like-for-like [32:17] comparison as we can get without the [32:18] real-world test, which is next. Running [32:20] at 120 Hz in performance mode on battery [32:22] is obviously harsh and creates a stress [32:24] test. We then ran Moss locally with the [32:27] same high settings on the Frame other [32:28] than resolution, which defaulted to 100% [32:30] or 2160 by 2160. The headset died in 47 [32:33] minutes flat. If you're playing games [32:34] locally, you should switch to the 72 Hz [32:37] playtime mode instead, and preferably [32:39] stick to lightweight games. We also [32:40] tried running locally with a 90-Hz [32:42] refresh rate to match the Quest 3, and [32:44] the frame lasted 55.8 minutes. Streaming [32:46] in playtime mode with the default [32:47] resolution, so still 2160 x 2160, and a [32:50] bandwidth cap of 230 megabits per second [32:53] at the time for default, resulted in a [32:55] 111-minute run time. We saw some [32:58] tracking glitches towards the end of [32:59] this run, but overall, it basically 111 [33:01] minutes for that one. With playtime [33:03] settings running Moss locally, the frame [33:05] lasted a more impressive 124.6 minutes, [33:07] a lot longer than its best streaming [33:09] time. Streaming can be fairly intensive [33:10] on its own, but playtime as a setting [33:12] affects multiple settings, not just the [33:14] refresh rate. So, it's not purely [33:16] like-for-like on the settings. The Quest [33:18] broadly has better battery life than our [33:20] Steam Frame. We have limited control [33:22] over Quest 3 settings for local play, so [33:24] we can't get a perfect match for them on [33:25] that, but we can look at one more test. [33:27] All right, so that was the torture [33:28] workload with a lazy Susan. We needed [33:30] some real-world testing, so we paid Mike [33:31] to play video games for a entire day, [33:33] and stream Half-Life Alyx to the [33:35] headsets. We wanted a reasonable middle [33:37] ground that would give both the headsets [33:38] a decent chance at beating each other. [33:40] So, we set the Frame to playtime mode, [33:43] locked both headsets to 72 Hz, which is [33:45] on the low side, capped the target [33:47] bandwidth to 200 megabits per second, [33:49] and kept resolution per eye at the [33:51] default 150%. We set the Frame on the [33:53] turntable again as a control. It lasted [33:56] 123.6 minutes. With Mike wearing the [33:58] Frame and playing his way through the [34:00] story, it lasted 115.8 minutes. Close [34:02] enough to make us confident in the [34:04] turntable method, and actually slightly [34:06] more intensive, which we think is [34:07] probably because of the human factors [34:08] involved, where now it's possible that [34:10] something like heat from being worn by a [34:12] real user likely affects the battery [34:14] life. The Frame was in single link on [34:16] Wi-Fi for Mike's test, but if that had [34:19] any effect, it would be to lengthen the [34:20] battery life. Mike's session with the [34:22] Quest 3 lasted 139.2 minutes, 20% beyond [34:26] the Frame in a like-for-like scenario. [34:28] This makes us pretty confident that the [34:30] Quest 3 has better battery life, at [34:32] least as we've tested it and for the [34:33] unit we have for the Frame. The charging [34:35] test where we ran into some maybe issues [34:37] with the Frame. The first test involves [34:38] waiting until headsets shut themselves [34:40] off due to low battery and then plugging [34:41] them into a meter without turning them [34:43] on. We also tested the Frame in several [34:45] other conditions including fully powered [34:47] off, idle and standby charge, charging [34:49] while gaming and below the fast charge [34:50] range, charging while gaming and above [34:51] the fast charge range, charging with the [34:53] Quest charger, charging with other Steam [34:54] Deck chargers. The reason we did so many [34:56] extra tests is because of what you're [34:57] about to see. Recharging the Quest 3 [34:59] took 146 minutes stock 18-W charger, 9 V [35:02] at 2 A. [35:03] Starting with a rate that oscillated [35:04] between 12 W and 17 W, [35:06] increasing to a point where it was [35:08] charging at 13 to 19 W. At 77 minutes [35:10] in, the rate smoothed out and decreased [35:13] in a curve down to under 2 W before [35:15] finishing. There was no real difference [35:16] in charging behavior for the Quest when [35:18] using the Frame's higher capacity [35:19] charger and 146 minutes is pretty close [35:22] to the 2.3 hours or 138 minutes that [35:24] Meta claims. Valve sent us an official [35:26] Frame and Deck power supply that maxes [35:28] out at 45 W. This is sold separately, [35:31] it's not included. The Frame used just [35:33] under 11 W of that capacity to start [35:35] with, which only increased up to a peak [35:37] of 12 W, then curved down to 3 W before [35:39] finishing. This process took 215 [35:42] minutes, significantly more than the [35:44] Quest, and the bulk of the charging took [35:46] place over roughly the first 2 and 1/2 [35:48] hours versus roughly 1 and 1/2 on the [35:50] Quest. We confirmed the charging [35:52] behavior with another known good 45-W [35:55] Steam Deck charger and tested again with [35:57] a Quest charger. We don't have those [35:58] numbers on this chart cuz they look [36:00] identical, it would just make it a mess. [36:01] Valve stated that, quote, "A dead [36:03] battery takes around 2 hours to charge [36:05] to full capacity. However, the charging [36:07] is not linear. There's fast charging [36:09] from 0 to 65% and regular charging from [36:11] 65 and up. Charging to 65% is expected [36:14] to take less than 40 minutes." End [36:16] quote. We didn't see that. With that in [36:17] mind, we drained the battery below 65%, [36:19] did some spot checks, manually shut down [36:21] the Frame instead of letting it die, and [36:22] then plugged it in and it drew 11 W. We [36:24] then put it to sleep and it drew even [36:25] less at 9 W. We also tested it simply [36:27] leaving the headset idle with the [36:28] displays inactive and that drew 13 to 14 [36:30] watts. We're confident in our equipment. [36:31] We've tested every variation, [36:32] permutation, combination and the [36:34] equipment has been professionally [36:35] calibrated and the methods are known. [36:37] So, that means either Valve is wrong, [36:38] there's a bug with the charging behavior [36:40] they don't know about, or there's a [36:41] defect with our review unit. We only [36:43] have one and we can only represent what [36:44] they sent to us. So, we have no way to [36:46] verify beyond that. We emailed Valve [36:47] about all this. You're seeing the [36:48] response on the screen now as we were [36:50] trying to compare notes and we're [36:51] confident in our findings and methods [36:52] and we ran everything. Valve said, [36:54] quote, "From everything shared, it [36:55] sounds like your experiment is valid. [36:57] However, we are seeing different results [36:58] on our end. Below are test results from [36:59] last month where we measured 100 charge [37:01] cycles across four Steam Frame headsets [37:03] with panels off and Steam VR stopped [37:05] simulating standby and they took between [37:07] 90 and 120 minutes to fully charge. One [37:10] hypothesis our team has for this [37:11] discrepancy is that it could be an [37:13] effect of battery capacity learning. In [37:15] this battery tech, a few full charge [37:17] cycles and discharge cycles are required [37:18] to have an accurate estimate of max cell [37:20] capacity after initial power on." And [37:23] quote there for a second. We did have a [37:25] lot of full charge and discharge cycles. [37:27] So, not sure this is the case. We're [37:29] well past the learning period. But [37:30] anyway, let's keep going. Quote, "Prior [37:31] to going through such a cycle, we have [37:32] seen Steam Frame hit over 90% after the [37:35] usual 90 to 120 minutes, but then creep [37:38] up slowly after that until hitting [37:39] termination at the new higher max [37:40] capacity. It's also possible, of course, [37:43] that there's been a regression in our [37:44] charging software and we'll try to [37:46] double check that next week. But we're [37:48] confident the hardware can do a full [37:50] charge in standby or off in the [37:52] originally quoted times." End quote. [37:54] Valve also shared this chart. Quote, [37:56] "These are plots of the raw data from [37:58] our most recent charging tests. You can [38:00] observe the inflection point where fast [38:02] charging ends after 35 minutes versus [38:04] 143 minutes you observed and charging [38:06] asymptotically completes in the 90 to [38:08] 120 minute range. Also, note that the [38:10] battery charge levels in these graphs [38:11] are not directly visible to the end user [38:13] in the UX. Our charging protocol uses [38:15] the standard approach to leave a bit of [38:18] headroom at the top to extend battery [38:19] lifetime." End quote. Now, we think [38:21] Valve's information and testing as they [38:23] present it makes sense. Valve thinks our [38:25] information and testing makes sense. Our [38:27] numbers disagree. Without a second unit, [38:29] we can't verify and we're left with [38:32] either a regression in the software that [38:34] Valve did leave as a possibility based [38:36] that they'll check this week or a [38:38] defective unit and it's not the charger [38:40] because we tried several. So, it's not [38:42] the biggest deal in the world. Something [38:43] screwy is going on. At this point, I'm [38:45] very confident it's not us. We've spent [38:48] an ungodly amount of time trying to [38:49] verify this. So, as for the controllers [38:52] for both the Quest 3 and the Frame, the [38:53] controller battery life is so far beyond [38:55] the headsets that we don't have an easy [38:57] way to test them. We got through the [38:58] entirety of this review and many headset [39:00] charge cycles with the stock disposable [39:02] double A's for both brands of [39:03] controller. These were running for about [39:05] 2 weeks or so during the work hours, [39:07] work days. So, in our opinions, it's a [39:09] non-issue. Valve claims 40 hours [39:11] depending on the batteries used and we [39:13] only just started considering replacing [39:15] the double A's. That's true for both the [39:16] Quest and for the Frame. Time to get [39:18] into the software and user experience. [39:19] The Frame uses SteamOS running on ARM [39:21] using the FEX emulation layer with ARM [39:24] native builds of DXVK and VKD3D. It's a [39:27] completely new frontier for Valve, so [39:28] the most impressive thing that we can [39:30] say about it is we didn't really notice. [39:32] The Frame SteamOS interface is familiar [39:34] for anyone who's used a Steam Deck or [39:36] Big Picture Mode within the last couple [39:38] years. As of this writing, Valve has [39:39] only confirmed 130 games to be [39:41] standalone verified great on Frame, so [39:43] there are compatibility limitations. [39:44] However, Valve is marketing this as a [39:46] streaming headset and all 2D games and [39:49] many VR games should stream equally [39:51] well. Standalone verified is the program [39:53] and great on Frame is the library filter [39:55] that shows those verified games. Valve [39:57] previously also told us, quote, "APKs [40:00] can also be sideloadable just like any [40:02] non-Steam applications on Steam Deck. We [40:04] expect VR APKs that don't leverage [40:06] proprietary APIs to just work." end [40:09] quote. We haven't tested standalone APKs [40:11] yet. For VR games that have separate [40:13] windows and great on Frame ARM builds, [40:15] you can force download the Windows [40:16] version by going into the per game [40:18] settings and overriding the [40:20] compatibility mode to proton. This may [40:22] be worth doing for games like Moss where [40:24] the frame build removes the graphics [40:26] quality settings that we see in Windows. [40:28] This is already the method needed for [40:30] downloading Windows builds on Linux and [40:32] it's messy. This needs to be integrated [40:34] into the game versions and betas tab and [40:36] devs need to actually make use of that [40:38] tab. Valve stated that streaming from [40:40] the Steam machine wouldn't be enabled [40:42] until after launch. We tried it anyway. [40:44] Streaming 2D games works currently. [40:46] Streaming VR does not. Desktop mode is [40:48] still present. By default, it's the same [40:50] Linux desktop environment as the Deck [40:52] and machine, just in VR. Being able to [40:55] plug in a keyboard and work on a virtual [40:56] desktop is neat, but like the Deck, [40:59] you'll need a hub if you want to plug in [41:01] more than one peripheral at a time. We [41:03] downloaded a game on our Steam machine, [41:05] moved it to a micro SD card, and then [41:07] stuck that micro SD card into the frame, [41:10] and it just launched. It worked fine. [41:11] So, that was pretty cool. Like with [41:13] inputs, Valve needs to streamline its [41:14] UI. The frame is different from past VR [41:17] headsets in that it can stream from [41:18] Steam while also running Steam. So, [41:21] there are two full sets of options to [41:22] check. Just from battery tests, we had [41:24] to check a confusing web of host Steam [41:26] settings, host Steam VR settings, host [41:27] PC settings, host game settings, client [41:29] Steam settings, client Steam VR [41:30] settings, and headset settings. While [41:32] streaming, there's a toggle in the Steam [41:33] VR menu on the headset that switches [41:36] from showing the local frame Steam VR [41:38] settings to the remote PC Steam VR [41:40] settings. The audience that's kept up [41:42] with VR over the last 10 years may be [41:43] familiar with all of this. For [41:44] newcomers, this will be a learning curve [41:46] and we do think that the UI and software [41:49] layout here is a mess, just like some of [41:51] the past Valve launches. The frame [41:52] wireless adapter shows up as a Realtek [41:55] 8832CU Wi-Fi 6 USB NIC in Windows, and [41:59] it can technically connect to home Wi-Fi [42:01] like any other wireless NIC, except when [42:03] Steam is using the adapter. Moving on to [42:05] the bugs, on the frame, the most common [42:06] issue was that exiting games from the [42:08] headset while streaming wouldn't close [42:10] the game on the host side. When running [42:11] games locally, sometimes the launch [42:13] screen would open, then go away after a [42:15] few seconds without feedback about what [42:17] went wrong. Sometimes unplugging and [42:19] reconnecting the wireless adapter [42:20] required rebooting, but as long as we [42:21] left it installed, it worked [42:23] consistently. We also experimented with [42:25] using the wireless adapter to connect to [42:26] Wi-Fi and then opening Steam, which [42:28] caused a conflict that might be [42:29] confusing. Several times when we tried [42:31] to stream Open Brush, it worked normally [42:33] on the host PC, but turned into a [42:34] nauseating flickering smear on the [42:36] headset side. But, it looks like this [42:38] was patched with a note from Valve [42:40] stating quote, "Fixed video corruption [42:42] in some cases while streaming VR." End [42:44] quote. On to the camera quality on the [42:45] accessory camera. We think Valve may [42:47] have made a miscalculation with the [42:48] cameras. In Valve's lineage of VR [42:50] headsets, the passthrough cameras exist [42:52] to tell you when you're about to smash [42:53] your shins into something and not much [42:55] else. The original Vive defaulted to [42:57] showing ghostly outlines of objects, for [42:59] example. With the Quest 3, the cameras [43:01] are at the forefront. It acts as an AR [43:03] headset by default, constantly showing [43:06] you the real world with menus [43:07] superimposed. The frame creates that [43:09] same feeling when you first put it on, [43:11] except the cameras suck. The frame rate [43:13] is low and choppy, which left some of us [43:15] feeling sick when using it in [43:16] passthrough mode, not helped by the [43:18] monochromatic coloring, although that is [43:20] useful when the lights are out. [43:21] Likewise, real world watch, phone [43:23] screen, and text are illegible from more [43:25] than a couple inches away, even for [43:27] those of us with 20/20 vision. There's [43:29] also distortion as objects come towards [43:31] the front of the headset between the [43:32] cameras with a clear area where they get [43:35] distorted. Monitors showing the headset [43:37] viewport represent this pretty well for [43:39] a two-dimensional YouTube video, where [43:41] you can see the same type of flickery [43:42] warbliness that the wearer is actually [43:44] experiencing. Depending on how used to [43:47] VR you are, this can range from [43:49] sickening to simply just a minor [43:51] annoyance. The Quest 3 has some of this [43:53] as well, but whatever image correction [43:55] math Meta is doing in post-processing [43:57] works a lot better than Valve's right [43:58] now, probably because they're stealing [44:00] all of your data. It affects the exact [44:02] area a couple feet in front of the frame [44:04] where you'd naturally hold an object to [44:05] look at it. The Quest 3 takes a big step [44:08] towards making VR feel less isolating [44:09] and burdensome by just letting you [44:11] [ __ ] see. While the Frame goes back [44:13] to locking you into the VR zone as long [44:16] as you have it on. Besides, with the [44:17] Quest 3, you're never truly alone [44:19] because Meta's right there with you. [44:22] >> It happens. [44:22] >> We stuck a camera inside the lens and [44:24] recorded at 120 FPS to get an estimate [44:26] of frame rate for the passthrough [44:28] cameras by counting the frames. Moving [44:30] objects appear at a sickening 30 FPS, [44:33] which is extremely obvious when moving [44:35] your hands. It's a lot different than 30 [44:36] FPS on a screen. It feels worse when [44:38] it's strapped to your face. The coolest [44:40] feature of the stock cameras is that [44:42] they can see and track in complete [44:43] darkness using the infrared illumination [44:45] from the headset, which also means that [44:47] they can pick up infrared light from [44:49] security cameras, phones, and remote [44:51] controls. As stupid as this sounds, it'd [44:54] be a good device for hidden camera [44:55] detection. This has no effect on camera [44:58] frame rate. Both the Frame and the Quest [45:00] showed flickering in situations with [45:02] mixed LED lighting. According to Meta's [45:04] product page, the Quest 3's dual RGB [45:06] cameras give 18 PPD, or pixels per [45:09] degree, which at 110° horizontal and 96° [45:12] vertical would calculate to 1920 by 1728 [45:16] each. Meta's API page states that [45:18] passthrough can do 60 Hz, which we [45:20] validated at 1280 by 1280 per camera [45:23] with 20 to 40 ms of capture latency. All [45:26] of those numbers are painfully low for [45:28] reproducing reality, but we find it [45:31] easier walking around in AR using the [45:33] Quest 3 than in the Frame. And AR gaming [45:36] is feasible on the Quest, if that's your [45:39] thing. Valve made us aware of two [45:40] third-party accessories before launch, [45:42] prescription lenses from Zenni Optical [45:44] and an RGB camera from Arcturus. The [45:46] camera, technically two cameras, is [45:48] $150. And besides being color, it [45:50] significantly improves image resolution. [45:53] The Arcturus camera made it [45:54] substantially easier to look at a phone [45:56] while in VR and solved some of my issues [45:58] with Valve's cameras. Although it still [46:00] has a relatively low frame rate, it just [46:03] feels a lot better. Of course, this will [46:04] also draw on the battery a little bit. [46:06] As for the color dual Arcturus cameras, [46:08] they take advantage of the high-speed [46:09] MIPI interface, four lanes at 2 GHz on [46:12] the headset, and operate up to 72 Hz at [46:14] the maximum resolution of 2464 by 2464 [46:16] each. The fine print notes that [46:18] passthrough is 60 FPS by default, 72 to [46:21] 144 with special config. End quote. All [46:24] processing is handled by the frame. The [46:25] cameras are just cameras, and the only [46:27] function they replace is camera [46:28] passthrough. Passthrough works [46:30] immediately without configuration. The [46:32] cameras interface with the OpenXR API [46:35] implemented by Valve, so anything that [46:37] uses onboard cameras for passthrough, [46:38] like AR games, should work exactly the [46:41] same. The Arcturus cameras can't see in [46:43] the dark like Valves can, although it is [46:45] possible in the future that Valve could [46:47] update the headset to switch to the [46:49] onboard cameras in low-light conditions. [46:50] We asked Valve about hand or gesture [46:52] recognition as seen on other headsets. [46:55] The company told us that it's not a [46:56] priority due to the computational costs, [46:59] but they did say that they're interested [47:01] in it. Let's talk controls. The external [47:02] controls for the headset are similar to [47:03] the Quest 3. Volume, power, and IPD [47:06] adjustment. The Frames IPD adjustment is [47:08] superior, we think, as the UI shows [47:09] finer adjustments, and the wheel can be [47:11] locked. The Frame also has a separate [47:13] aux button, which brings up a cursor [47:15] that can be used for basic menu [47:17] navigation. The Quest 3 combines this [47:19] with the power button, and also has [47:20] controller-free hand gestures. The [47:22] addition of a D-pad also gives an [47:23] alternative for those who don't like the [47:25] usual point-and-shoot VR menu [47:26] navigation. I actually prefer the D-pad [47:28] personally. Unfortunately, VR tracking [47:30] for the Steam controller wasn't ready in [47:32] time for this review. Valve has not [47:34] finished that feature. We don't know how [47:36] well it'll work in general since the [47:37] Steam controller has a single tracking [47:39] LED versus dozens on the dedicated VR [47:42] controllers. The controller must be [47:44] connected via Bluetooth or through the [47:46] Frames USB port currently. If you're [47:48] streaming, it may make sense to have the [47:50] Steam controller if using it connected [47:52] directly to the host PC via the puck [47:54] instead. The finger tracking worked [47:55] better than we expected with gradients [47:56] for fingers and on and off for the [47:58] thumb. The biggest limitation here will [48:00] be dev support. Finger tracking isn't [48:02] universal across VR hardware, so it'll [48:05] rely on developers to build games around [48:07] it. The biggest positive is that the [48:09] tracking makes it very convenient to [48:11] flip someone off in VR chat, which is [48:12] the most we'd really ask of it. The [48:14] rebinding situation is insane and not in [48:18] an easy to understand way. Maybe it has [48:21] to be, but we don't recommend engaging [48:23] with it unless you need to. Consider a [48:25] situation where you're streaming a VR [48:27] game from PC to headset and the game is [48:29] using an Oculus controller profile [48:31] because your SteamVR controllers emulate [48:33] Oculus controllers by default. Your [48:34] options are to change bindings in the [48:35] in-game settings, your headset per game [48:37] settings, your headset controller [48:38] settings, your PC Steam client [48:39] controller settings, PC Steam client [48:41] game settings, or PC SteamVR controller [48:42] settings, and it's not clear where to do [48:44] that. You also might not know whether [48:45] you want system-level bindings or VR [48:46] link bindings or app-specific bindings, [48:48] and the buttons don't even match up to [48:49] the Oculus anyway, and a click can mean [48:50] a literal click or just pressing the [48:51] button even if it doesn't click, and [48:53] everything has a capacitive touch [48:54] sensor, and you have to do half of this [48:55] with the VR headset on, and we were [48:57] frequently greeted by a game pads do not [48:58] yet work in VR mode tool tip, but we [49:00] were able to get a Steam controller [49:01] hardwired to the headset and working in [49:02] VR at least once, and then we went to [49:04] reproduce it and we couldn't easily [49:06] figure out whatever the [ __ ] combination [49:07] of factors we used and/or mistakes we [49:09] used that allowed that to begin with. If [49:11] you don't open the controller menus, [49:14] generally, things kind of just work. The [49:17] biggest problem we foresee is with VR [49:19] games that already expect a normal [49:21] gamepad. For example, Star Wars: [49:23] Squadrons wants gamepad inputs, but the [49:26] Frame VR controllers show up as Oculus [49:28] controllers by default when streaming, [49:30] so even though they should be able to [49:31] work by default, they don't, and we [49:32] don't think it's intuitive to figure out [49:34] how to resolve these issues. So really, [49:36] the binding situation is a mess. If [49:38] you're [49:39] a bindings enthusiast, you have lots of [49:40] experience messing around with bindings [49:42] beyond just whether you're using a [49:43] keyboard or gamepad, then this is a [49:45] little more navigable, but it's still a [49:47] mess. For mainstream users, this is a [49:49] big mess, and you really just want to [49:51] use everything as default as you can. [49:53] All right, home stretch. We're in the [49:55] conclusion now. It's been a long [ __ ] [49:56] week with the two ginormous LG videos [49:58] investigation straight into one of the [50:00] biggest review videos we'll make this [50:02] year. Ultimately, VR headsets are very [50:04] subjective and so to do something a [50:07] little different, I asked Mike and [50:08] Patrick to give some of their subjective [50:11] thoughts for the conclusion here just to [50:13] get some more voices in the video, some [50:15] different perspectives, things like [50:16] that. I specifically said to them, [50:18] "Don't worry too much about like being a [50:20] reviewer. Just give me the consumer [50:22] perspective purely." And so if that [50:24] means they've got some opinions that [50:26] they don't know what the internet will [50:27] think. I was like, "Just just say it. [50:28] Just tell everybody cuz someone out [50:30] there will agree with different pieces." [50:31] So we're going to hear from them for a [50:33] second here and then we'll jump back to [50:35] the close out. Mike, the battery on this [50:37] headset is dead so I'm only seeing [50:38] blackness. [50:39] >> [laughter] [50:39] >> Same here. [50:40] >> Okay. [50:41] >> Mine just died. [50:41] >> We should probably take these off. [50:43] Um there is however someone [50:45] who who is still wearing his headset. [50:48] >> Hey, wait, I still need that. [50:50] >> Wait, what does that say on your privacy [50:52] testing [50:53] and you have four cameras pointed at me. [50:58] What do you have to say for yourself? [51:00] >> Uh [51:01] the privacy's for me. [51:03] >> [laughter] [51:04] >> What are your general thoughts here [51:06] starting with the frame in terms of [51:07] maybe start with comfort? [51:09] >> Um the frame definitely felt a lot more [51:11] comfortable just as a whole. Um it [51:14] didn't really get hot. We've had a [51:16] consistent temperature and humidity in [51:17] here all day since this morning. So [51:20] um [51:21] it was pretty consistent across the [51:23] board. [51:24] >> That's definitely the GN testing [51:25] methodology coming into play. [51:26] >> Yes, see I just I needed I needed to [51:28] preface that. [51:29] >> Yeah, I don't think you're going to hear [51:30] other [51:30] >> climate. But it was yeah, it I the my [51:33] face definitely felt cooler in the the [51:36] frame and it was just in general a [51:38] little bit more comfortable. It's not as [51:40] heavy. [51:41] Um I like the the top strap and this [51:44] like I just kind of forgot it was on [51:46] there a little bit more often. [51:48] Obviously, you don't forget that [51:49] something's strapped to your face all [51:50] the like [51:51] >> Right. What do you think uh what do you [51:53] think the comfort mostly is coming from [51:55] in terms of what's making it feel better [51:57] than the Quest? [51:58] >> To me, it's the weight. [51:58] >> Weight, okay. [51:59] >> Yeah, it's either the weight or weight [52:00] distribution. I don't know what they [52:02] have going on in the backpack, but um [52:04] the Quest has always felt a little front [52:06] heavy. Um and it it always seemed a [52:09] little bit harder to get it to sit um [52:11] exactly where I wanted it on my face, [52:13] whereas [52:14] uh I was very able to or I was very [52:16] quickly able to get this positioned [52:19] uh optimally on my glasses and just on [52:21] the front of my face so that it was [52:23] sitting comfortably. Whereas, at the end [52:25] of the playtest on this, I could feel a [52:26] lot of kind of pressure on the bridge of [52:27] my nose through the glasses and stuff. [52:29] So, [52:29] >> have the marks on your forehead. [52:30] >> Yeah, [52:31] >> [laughter] [52:31] >> yeah, I could just [52:33] playing games [52:34] >> It's a proof of proof of work. The uh [52:37] so, I noticed you put the top strap on [52:38] this, which is an optional thing they [52:40] sell separately. [52:41] >> Yes. [52:41] >> How do you feel about it without the [52:43] strap? [52:43] >> I've got pretty strong strap opinions. [52:45] So, um on my wife's Quest 2, we ordered [52:49] some second-hand Amazon one that had [52:51] [clears throat] more substantial [52:52] strapping up top and more adjustment um [52:55] because I really prefer having the top [52:58] strap. For me, personally, like being [53:00] able to kind of suspend the headset and [53:02] put the pressure over kind of the crown [53:04] of my head is so much more comfortable [53:07] than just kind of like ratcheting it [53:09] down across, you know, the sides of my [53:11] face. [53:12] >> What if you take the strap off of it, [53:14] the top strap? [53:15] >> Uh I didn't even want to play with it [53:16] like that. It felt very unsecure. [53:19] >> Okay. [53:19] >> Right. We when we were prepping for the [53:21] test uh yesterday, uh Patrick had the [53:24] the top strap off and I put it on to [53:27] look where settings were, you know, for [53:29] to setting up the test environment, and [53:31] um he it was it felt very unsecure and [53:34] there was a lot of pressure on the [53:35] bridge of my nose. And that might be [53:36] because of like the force driving down [53:38] to the glasses. [53:39] >> Mhm, right. [53:40] Right. So, for you, it [53:42] uh [53:43] you you would really only want to play [53:44] it with the strap if you have one. [53:46] >> Yeah, and I it's credit to both of them. [53:49] Uh, I didn't really have to mess with [53:51] the adjustment much at all, actually, to [53:54] make it fit my glasses. So, as a glasses [53:56] user, they both they both worked great. [53:59] >> Right. Including the controllers, like [54:00] the full system, do you notice [54:02] differences between the two? [54:03] >> Yeah, for Half-Life Alyx specifically, [54:06] um, the Steam frame was definitely uh, a [54:10] lot easier to use because you were able [54:12] to [54:13] string your hand through it and like [54:15] almost like let go of [54:18] of the controller, right? When you're [54:20] doing grabbing motions, whereas, um, [54:23] both controllers have like that finger [54:24] detection, but obviously, I can only [54:26] like I have to still like hold on to a [54:28] pinky, so it's a little bit more [54:31] obnoxious and unnatural to try to grab [54:33] things in the game with the with the [54:35] Quest. [54:35] >> Not sure if Patrick told you, but these [54:37] are also optional accessories that you [54:38] have to buy separately. [54:41] >> That's a bummer. [54:41] >> Yeah. [54:42] Especially considering, you know, it's [54:44] like a a major feature. [54:45] >> Yeah. I would say the feature of VR [54:48] games. [54:49] >> Yeah. [54:49] >> Unless you're into sim racing, but [54:51] >> Performance, tracking, you know, stuff [54:53] like that. [54:54] >> Uh, I was just impressed with the [54:55] experience overall. Uh, [54:56] >> For both? [54:57] >> For both. So, so doing the battery test [54:59] was definitely the longest I've ever [55:00] spent in VR in like one like given [55:03] stretch where I didn't take a break. [55:04] >> Right. [55:05] >> Um, and I would say they were the Quest [55:08] was tolerable. The frame was definitely [55:11] still enjoyable. Or like when it started [55:14] giving me the battery warnings, I was [55:16] like, "Oh, okay, you know, like I'd like [55:17] I could take this off soon." whereas [55:19] like I was kind of waiting a little bit [55:21] for the battery to die with the Quest, [55:22] so. [55:24] >> Right. Okay, cool. Any other, uh, [55:26] closing thoughts? [55:27] >> I think that's it. I mean, it's it's [55:29] it's unextraordinary in a good way. [55:31] There was no like major major flaws and [55:36] or downfalls that I experienced while [55:38] while playing with it. So, it was a good [55:39] experience. [55:40] >> 600 for the setup you were testing today [55:41] with the Quest, 1050 or more for the [55:44] Steam Frame. If you're buying today with [55:46] the information, you know, if you went [55:48] and did this in a Best Buy and then you [55:49] made your decision there. [55:51] >> Ignoring Facebook privacy [55:53] >> No, everything you know about both [55:55] companies. [55:55] >> Oh, yeah, I mean, I I would never [55:58] I'm going to talk to my wife about her [56:00] Quest 3 [56:01] tonight [laughter] [56:02] due to due to the privacy concerns and [56:04] stuff and everything going on with that. [56:06] Yeah, it's worth it. The the extra extra [56:08] money is definitely worth it to me to [56:10] have a little bit more security over my [56:12] information. [56:13] >> Yes, Mike was the one who I don't know [56:15] if we showed in the video or not, but [56:17] you immediately got a ton of ads for [56:20] Meta devices as soon as on your personal [56:22] device. [56:23] >> Yeah, I I this was my first involvement [56:25] in the Frame review today. And as well [56:28] as the like any of the Meta Quest stuff [56:31] we've been doing and my phone's been [56:33] getting pumped full of Meta glasses and [56:35] Meta [56:37] you know, Quest ads, so [56:39] um they got me. [56:40] >> Yeah. [56:41] >> [laughter] [56:41] >> I'm in there. They know where I live and [56:44] yeah, I'm getting them at home, too, so [56:46] it comes with me [56:47] right through my phone. [56:48] >> It's like a like a friend that's there [56:50] all the time. [56:51] >> Or a virus. [56:51] >> Right. [56:52] >> Yes. [56:53] >> All right, so talking with Patrick now [56:55] about subjective opinions. Patrick's [56:58] been working on the testing and the [56:59] script, so a lot of your words are [57:01] already in the rest of this video, [57:04] but we kind of filter for some of the [57:06] opinion stuff. So, [57:08] from an opinion, you know, subjective [57:10] user [57:11] experience scenario, not necessarily [57:13] strictly the review style, [57:15] >> Mhm. [57:16] >> what what's your big picture feeling [57:18] between the two, you know, the [57:21] Quest and the [57:22] Frame? [57:23] >> Well, I have a Quest 3 at home, so I [57:26] have some experience with it. [57:28] Um [57:28] I think this out of the box is a lot [57:31] more comfortable. [57:32] I think the head strap for the Quest 3 [57:35] is uncomfortable enough that a lot of [57:37] people, myself included, just [57:38] immediately swap it out for something [57:40] else. [57:41] That said, the the Bobo VR strap that [57:45] Battalion and I have, um, [57:47] is kind of bulky for me and I honestly [57:50] this is the most comfortable. Like this [57:52] this is the best [57:53] feeling VR headset that I've had [57:56] including the, um, [57:57] aftermarket head strap. [57:58] >> What [57:59] um what aspects of it or where [58:01] specifically is it like noticeably more [58:03] comfortable? [58:04] >> Uh, I I know Mike likes the top strap. I [58:07] don't I don't really like it. Um, [58:09] I kind [snorts] of [58:10] like it more just the way it comes. [58:12] >> Yeah, you haven't even been using it, [58:13] right? The top strap. [58:15] >> Yeah, I've been leaving it off. Um, [58:17] I think even if this back part wasn't a [58:19] battery, I would like the way that it [58:22] just kind of like you you can just grab [58:24] it and put it on and then it [58:27] it takes you can do it with one hand if [58:29] you want to and then you can tighten it [58:31] here and then it's pretty much done. If [58:33] you're walking around, I know it's kind [58:34] of bouncy, but, um, [58:36] it feels fine for like [58:38] seated VR and stuff and it's pretty [58:40] lightweight. [58:41] Um, [58:43] I think that's the best part of it. [58:46] Um, this takes some getting used to. [58:48] Grabbing it and having it slip off. [58:50] >> easily. Yeah. [58:51] >> Uh, [58:53] I would say that actually looking at [58:55] stuff in the headset, like using it for [58:57] a long time, this the Quest 3 gets hot [59:00] and it's kind of heavy. [59:02] But, actually looking at games, I I [59:05] don't know that I could tell the [59:06] difference. Like if you if you put this [59:07] on me without telling me which one it [59:09] was, [59:10] um, [59:11] you know, they have similar max [59:13] >> From a game standpoint, not from a fit [59:14] standpoint. Yeah. [59:15] >> the refresh rates, um, [59:17] this technically is goes up to 144. This [59:19] technically goes up to like 270 or [59:21] something, but 120 is the realistic max [59:23] for both of these and at that refresh [59:25] rate it, [59:26] you know, they look pretty similar. [59:29] >> He You mentioned he So, how does it feel [59:32] in the headset for somewhat extended [59:35] use? [59:36] >> I I guess sweaty in the Quest 3. This [59:38] This I feel like is [59:41] lighter. It's potentially less hot, but [59:43] it's also kind of open at the bottom. I [59:45] mean, I I like this shape more. [59:47] >> that nose hole. [59:48] >> Yeah, and I I this doesn't seem like it [59:52] would be as big a deal as it is because [59:53] it's the headset itself is not that much [59:56] smaller than the Quest 3, but just [59:57] having this open like even just being [60:00] able to scratch your nose or whatever is [60:01] like it just feels like less of a thing [60:04] over your face. [60:05] >> The material is pretty different, too. I [60:07] feel like that is the same material. [60:08] >> softer for sure. Maybe they can get away [60:10] with that if they're saying like, "Oh, [60:11] you can replace it by just, you know, [60:12] yanking on it." [60:13] >> Right. That's true. I hadn't thought [60:15] about that. Yeah, that that might be [60:16] less durable, I guess. But, we're not [60:19] Obviously, we haven't had it long enough [60:20] to know. I don't know exactly how long [60:21] we've been actually working on this, [60:23] maybe around 2 weeks. [60:25] >> Yeah. [60:25] >> It's a relatively long review window. [60:27] Normally, we don't get that much time, [60:28] but [60:30] So, obviously, all the opinions here, [60:33] it's like max a couple weeks of kind of [60:35] on and off use. Pretty much daily, [60:37] though. So, I guess tracking, do they [60:39] feel about the same for head tracking to [60:42] you? [60:43] >> Yeah. [60:44] Um [60:45] they have very clearly like [60:48] in terms of tracking and the [60:51] the device and the inputs and what it [60:53] is, they [60:56] like these are so similar. [60:57] >> [laughter] [60:57] >> Yeah. [60:58] >> They they they very clearly like the [60:59] design of the of the Quest stuff [61:02] influenced the way the frame looks. [61:05] Uh so, [61:07] in terms of tracking, I haven't I [61:09] haven't actually tried turning out the [61:10] lights and using the Quest 3 in the [61:12] dark. I mean, theoretically, the [61:15] uh Valve headset is, you know, you can [61:17] see in the dark. You can game in the [61:19] dark. Um [61:20] >> Well, we did test that right in the [61:21] sound chamber, and you can [61:22] >> Yeah, you you can definitely see in the [61:24] dark with the with the frame, yeah. [61:26] >> It's kind of funny, yeah. It's a little [61:27] a little weird. The input we talk about [61:29] in the script for the greater review, [61:31] it's probably worth mentioning again, so [61:33] the input setup and um [61:35] you know, managing the control options. [61:38] >> Yeah, so that stuff gets complicated [61:40] because this is it I mean, as far as [61:43] I can think of, this is the first time [61:45] that there's a VR headset that is [61:47] running Steam [61:49] and streaming Steam. [61:50] >> Mhm. [61:51] >> So, you have the Steam client and Steam [61:55] VR on the headset, and you have the [61:57] Steam client and Steam VR on the host [61:59] PC, and you got a full suite of settings [62:03] for both of those plus the in-game [62:04] settings plus the computer's settings, [62:07] and [62:08] uh that complication carries over to the [62:10] bindings as well, because if you think [62:12] about it, you know, you've got your [62:14] controllers that if you're streaming a [62:16] VR game, they're pretending to be Oculus [62:18] controllers because that's widely [62:19] compatible, but if you're streaming a 2D [62:21] game, they're pretending to be a [62:22] gamepad, and then [62:25] you're streaming, so it's you know, [62:27] you're passing through inputs to a [62:30] remote device. Like, you could be [62:32] plugging input devices into the host PC [62:35] as well, so there's a lot of layers of [62:38] the um binding stuff as well as the fact [62:41] that like [62:42] all of this stuff is rebindable, like [62:44] touching any part of this or grabbing or [62:48] squeezing it or squeezing it a lot, like [62:50] all of those are separate binds. [62:51] >> And to be clear for people, too, [62:54] it's not the having options to change [62:57] bindings [62:59] that we have uh any any complaints [63:02] about. [63:02] >> No. [63:02] >> It's the complexity of the places that [63:05] you can do that. [63:05] >> Yeah. [63:06] >> And you do it per game even, so [63:08] >> Right. [63:08] >> And and if you're in the headset, you're [63:10] changing bindings for [63:12] uh like the VR link streaming, and if [63:16] you're on the PC, you're changing [63:17] bindings for [63:19] the specific piece of software, like [63:21] like Half-Life Alyx or whatever, uh uh [63:23] it gets complicated. [63:25] >> Any um any kind of closing thoughts on [63:27] the frame in general, like just [63:29] broad thoughts on how Valve did with it? [63:32] >> Um [63:35] >> [laughter] [63:37] >> I I feel like I need to like [63:40] I feel bad for them over the price. Like [63:43] I feel like there's a bunch of engineers [63:43] [laughter] out there that didn't think [63:45] that this was [63:47] going to be the price that they landed [63:48] on. [63:49] >> They told us less than the Index. [63:51] >> And it's not less than the [laughter] [63:52] Index. [63:53] >> not even less than the Index bundle. So, [63:55] they're off by [63:57] >> Yeah. [63:57] >> hundreds of dollars. [63:58] >> So, [64:00] um [64:01] I'm I'm happy that it's out, and I hope [64:03] it doesn't stop them from [64:05] further development, but I'm a little [64:08] pessimistic about where this is going. [64:10] >> Yeah, depends on adoption, which will be [64:12] a lot lower at over a thousand dollars. [64:14] >> Yeah. [64:15] >> So. [64:16] Okay, cool. We'll go over to the next [64:17] section. [64:19] All right, that should be good. [64:21] >> And he's [64:24] He's gone back into the Matrix. [64:26] He's a [snorts] He's [64:28] Patrick has had enough talking to [64:29] humans. He's gone back to [64:32] VR land. [64:33] >> Actually, subjective opinions for me, [64:35] they boil down pretty simply. On the [64:36] hardware side, I found the frame to be [64:38] very comfortable, just like the other [64:39] guys. I was impressed with the weight [64:41] and the materials, the material [64:43] specifically of the pad that contacts [64:45] your face, I found to just breathe a lot [64:47] better than stuff like, for example, the [64:49] Meta one or the past VR headsets that [64:51] I've used, which have mostly been the [64:52] older models [64:54] dating all the way back to the HTC and [64:56] the Oculus Rift. Those were the first [64:57] two that I tried. [64:58] So, definitely the best material for the [65:00] face contact that I've felt. Thermally, [65:02] it it did feel like overall cool. I [65:05] would say that head strap definitely [65:06] affects the temperature that my forehead [65:09] was getting, where adding the strap does [65:11] help keep some of the heat off your [65:12] head. I also find the strap, personally, [65:15] to be necessary. And Mike felt the same [65:17] way Patrick did not. So, Patrick has [65:19] been using the headset without the strap [65:21] on the top for basically his entire [65:23] period of testing, no problems. Mike and [65:25] I were basically like, "This is [65:26] mandatory." Uh just because it either [65:28] pushes down on the bridge of the nose or [65:30] the glasses for him. And then for me, I [65:32] mean, it's I think it's just it slides [65:34] too much on my hair or something. I [65:35] don't know. But anyway, I found it [65:36] necessary. Um Valve's privacy policy and [65:40] privacy practices, by comparison to what [65:43] they could collect on the device, are [65:45] excellent. They were very clear with us [65:47] in saying, "No." When we asked them if [65:50] they do, here's this big list of [65:51] horrible things that Meta does. Uh and [65:54] then they were really clear in answering [65:55] our questions about what they do with [65:57] the data and to what extent they can use [65:59] it. They even got ahead of our questions [66:01] by saying, "And we don't Meaning [66:06] some form of fingerprint or hash or [66:07] whatever. Uh and so that was awesome. [66:10] That was That was really happy to see [66:11] that coming off the LG story. [66:13] Things I don't like. I absolutely hate [66:16] the cameras that are in the frame. This [66:18] is going to be the most opinionated [66:19] thing [laughter] you'll hear me say for [66:21] pretty much any review. [66:23] Um because it's a purely subjective [66:25] thing. I find it unusable and [66:27] intolerable. The add-on camera was [66:29] better, but man, for me, [66:33] it's a low speed. It has these weird [66:35] sort of re-projections in space where I [66:36] feel like if I'm looking at a fixed [66:38] object like a shelf in a room, it just [66:41] There's just enough lag there where it [66:43] feels like it's kind of warping and [66:45] shimmering and teleporting. Some of [66:47] that's because of the [66:48] uh [66:49] the math issues we were talking about [66:51] earlier where it just gets some stuff [66:53] wrong center screen. But beyond that, um [66:56] I mean, I just I found it bouncy to walk [66:58] around in without the head strap and [67:00] with the passthrough cameras active. I [67:02] didn't have the bounciness problem in [67:04] games. And I didn't have a camera [67:06] problem in games cuz you're in games, [67:08] but with the passthrough, I wouldn't use [67:10] I would sooner take the headset off than [67:12] I would like look at something through [67:13] the passthrough. I just found it [67:14] completely useless personally. [67:16] Um it's a pretty strong opinion, I [67:18] recognize that. Not everyone's going to [67:19] agree with me. Some people are going to [67:20] be like, "Wow, this guy's overreacting." [67:21] It's fine. Just telling you what I [67:23] think. Um for some reference, I I have [67:25] perfect vision. Nothing makes me dizzy. [67:27] I don't get any kind of vertigo. I've [67:29] used a lot of VR over the years, [67:30] including some of the worst ones ever [67:32] made with the worst video cards at the [67:33] time. Uh and I didn't get sick from any [67:36] of those. Um I don't get sick playing [67:38] the games. I don't feel dizzy playing [67:39] the games. But man, walking around with [67:41] that passthrough camera, that gave me a [67:44] headache pretty fast. So, um some [67:46] physiological reason, I couldn't tell [67:48] you. I just have a strong reaction to [67:49] it. Uh [67:50] but anyway, it was improved [67:51] substantially with the optional head [67:52] strap for the bouncing and then the [67:54] camera was improved by the add-on, which [67:56] unfortunately, of course, costs more [67:57] money. So, that's my thoughts on that. [67:58] Strongest opinion I have uh on the [68:00] negative side. In VR mode, that goes [68:02] away. All right, so pricing is the big [68:04] discussion. The value is bad. bad [68:05] period. The price is high. The frame is [68:07] expensive. Valve objectively failed to [68:09] hit its only target it gave on price, [68:11] which was cheaper than the Index, and [68:12] it's not. Uh and also, optional [68:15] accessories that don't feel so optional [68:16] to us cost even more money. So, they've [68:18] objectively failed to hit that price [68:20] marker. Now, yeah, the market's changed. [68:22] They knew that at the time they said [68:23] that it was starting to change. I don't [68:24] think anyone, including Valve, including [68:26] us, knew it would get this bad for the [68:28] memory, but you know, it did. And so, [68:32] here we are. I think we're pretty [68:33] empathetic to the price situation given [68:35] our coverage over the last year. Uh we [68:37] get it. I know exactly what's going on [68:39] with memory and storage prices. Um and [68:43] that's that's fine, but at the end of [68:45] the day, it doesn't change the fact that [68:46] the device is expensive for a consumer, [68:48] and uh you know, that's [68:51] it looks worse because of things like [68:54] this. [68:55] The Meta Quest, where depending on which [68:57] one of these you buy, they run [68:58] substantially cheaper than Valve. [69:01] Except for the one key factor, which is [69:04] the privacy, that no one ever really [69:06] seems to talk about. And Meta is the [69:08] company, out of all the companies, that [69:10] you need to be talking about it with. I [69:12] don't think you can review this device [69:15] that I'm holding, the Quest, without [69:17] thinking about what Meta does with [69:19] privacy, [69:20] uh just because of their history. So, [69:21] the contrarian comment you'll see to [69:23] that sometimes is, "Yeah, but your [69:24] phone's tracking you, too." One, maybe [69:25] not to this extent. [69:27] So, you shouldn't just throw it all away [69:28] because of that. Uh and two, to quote a [69:31] viewer of ours who was at a town council [69:33] hearing I went to about flock cameras, [69:35] talking about privacy, he said, "Just [69:37] because we've been [ __ ] it up, [69:39] doesn't mean we need to keep [ __ ] it [69:41] up for the rest of time." [69:42] Uh so, you can always draw a line [69:43] somewhere. And we wouldn't recommend the [69:45] Quest at all purely on the privacy [69:46] grounds, because the things that they're [69:47] telling you that they collect and do are [69:51] disqualifying on their own for me. So, [69:54] um now, that doesn't mean the frame is [69:55] worth over a thousand dollars. One of [69:57] the biggest problems with VR is that [69:59] there remains a sort of chicken and egg [70:00] problem, where there still has to be [70:02] hardware adoption before developers will [70:04] really commit to making games. As a [70:05] result, the categories of VR games on [70:08] the frame are something like VR Chat, [70:10] racing game, spaceship game, 2D game [70:12] with token VR mode, uh adult content, [70:14] and Meta exclusives you can't play, and [70:16] then, of course, Half-Life Alyx. Those [70:18] are the main options. There's some gold [70:20] out there. Superhot VR, as an example, [70:23] is one of those, and is a big part of [70:26] the reason we think that Valve [70:28] emphasized the ability to stream 2D [70:30] games on the frame. For people super [70:33] into VR, you already have your games you [70:35] like, uh or applications you like, and [70:37] you have your reasons to like them. For [70:39] people who are new to it, what I would [70:41] caution is that because these devices [70:42] are so expensive, specifically now with [70:44] the [70:45] hardware crisis, [70:47] um really look into the games that are [70:49] available, and make sure they're going [70:50] to work for you. Because VR, it's got [70:52] kind of a different selection of [70:54] categories that are done really, really [70:56] well. And then a lot that's really [70:58] unpolished still. So, just make sure [71:00] that's compatible with what you're [71:02] looking for in a game before you commit [71:03] to over a grand, I think. Um and uh [71:06] that'll pretty much wrap it up for us. [71:07] So, we'll leave the review there for [71:09] now. [71:10] The thousand plus dollar price category [71:12] of VR headsets has more options than the [71:15] sub $1,000 category in terms of things [71:18] we haven't looked at yet. So, uh I've [71:20] never used a big screen VR solution or a [71:23] Pimax solution. I would love to hear [71:25] recommendations from those of you who [71:26] are really into this stuff to know what [71:29] maybe we should consider testing next. [71:31] Um if there's something that you think [71:32] would make sense that closer to the [71:33] Valve frame price probably makes the [71:35] most sense for the next one for us to [71:37] look at. Um just because of how I I feel [71:39] about the Quest with its privacy issues. [71:41] You can turn some of that stuff off. Uh [71:43] there are a lot of [71:45] I'm not going to get into it in this [71:46] video. Um we're going to do a full deep [71:47] dive on this. So, I'll leave that there. [71:49] Anyway, thanks for watching. Subscribe [71:51] for more as always. You can go to [71:52] store.gamersnexus.net to support our [71:53] efforts like this. This has been a very [71:56] hard week rolling from LG into LG [71:59] response into gigantic review of very [72:03] complicated and technical but also very [72:05] interesting hardware. So, [72:07] uh thank you for the support as always. [72:09] And you can go to store.gamersnexus.net [72:11] or patreon.com/gamersnexus. [72:12] We'll see you all next time.