Although he is no longer Oculus CTO, John Carmack’s Keynote speech at the Connect conference seems to have become a tradition. As a practitioner in the VR industry for many years, Carmack has a deep understanding of VR technology and Oculus strategy, so his speech this year has also received attention.
Unlike previous years, Carmack’s speech this year focused on many perspectives on the meta-universe in addition to VR and other technologies. Compared with the beautiful vision, he analyzed the future development direction of the meta-universe from reality, which may bring some for VR practitioners. Inspired.
This reminder will involve the following key points:
1) Before Oculus was acquired, it was actively opposed to exploring the meta universe, because it would require a lot of time and capital investment, and the future prospects are uncertain;
2) Before the realization of Metaverse, the scale of simultaneous online VR users needs to be increased;
3) It is believed that there is no need to explore new VR hardware design for Metaverse. Quest 2 is sufficient. In the future, Quest 2 should continue to explore the cost-effective product path;
4) Have tried to open the Oculus ecosystem, but due to different business models, it is difficult to cooperate;
5) Only by reducing costs and user resistance and introducing existing 2D content can it be expected to promote the popularization of VR;
6) Quest 2 still has user churn, and VR needs greater application value.
7) Meta is optimistic about the application prospects of WebXR;
8) The internal project approval process is complicated, and efficiency needs to be improved.
The following is a more complete content of this Keynote organized by Qingting.com :
Carmack said: In the past, I was often dissatisfied because of the slow development of VR. Fortunately, there are many things in the game this year that I am happy about. For example: Quest 2 has achieved great success, the market feedback is very positive, and the effect of Airlink wireless streaming mode is also beyond. expected.
In addition, cloud rendering technology has also developed and begun to be integrated into future Metaverse planning, and has shown a certain value in use.
He also pointed out that App Lab has already achieved initial results after its launch, and he is optimistic about this free and open application market. He also hopes that the slow review of App Lab can be resolved. Compared with the preliminary screening, Carmack is more optimistic about the later review , which means that developers can freely publish VR applications, and Oculus can conduct follow-up reviews and remove unqualified content after it is found.
It is worth noting that theoretically speaking, the 72Hz refresh rate of Quest 1 can actually be increased to 90Hz , which means that by optimizing the calculation of some VR games, it may be possible to run on Quest 1 with a higher refresh rate of 90Hz. However, this upgrade has not been approved internally by Oculus, because the clock frequency needs to be changed, and the FCC needs to be re-applied for equipment certification. Considering the product life of Quest 1, it is not necessary to carry out such an upgrade.
One of the most happy developments for Carmack this year is Oculus Go’s opening of Root permissions, and next he hopes to retain Go’s content ecology, which requires not only Oculus internal approval, but also the support of third-party developers.
How to look at Metaverse
Talking about the topic of Metaverse, he said: The development of Metaverse is an ethical responsibility. I care about this technology and believe in the vision of the metaverse. Nevertheless, before I acquired Oculus from Facebook, I have been actively arguing and opposing every attempt made by Oculus in the meta-universe field.
This is because, although Carmack hopes to achieve yuan universe, but he has filled grounds to believe that points to Metaverse to development as the goal, it may not be the best way to build the platform. Generally speaking, he believes that Metaverse is easy to attract unsophisticated programmers or designers. Their characteristic is that they don’t want to consider the specific details of GPU micro-architecture and merged and shared networks, but talk about Metaverse from a macro and abstract perspective.
Considering that Facebook has changed its name to Meta and has entered the Metaverse field vigorously, Carmack believes that this will cost a lot of money and resources, and can only hope that these investments can bring positive short-term practical value. In fact, he is worried that after investing several years and thousands of talents, the popularity of Metaverse will still not be comparable to the current mainstream hardware platforms (PCs, mobile phones) and so on.
In this regard, Carmack’s suggestion is to focus on the actual product rather than the technical architecture or idea. Take game development as an example. Previously, Carmack wrote the engine for the purpose of developing games at id Software Studio. After the game was launched, it was discovered that this engine can be reused in other scenarios. Consider whether it can be used for other purposes.
In fact, technologies that prematurely consider the popularity of technology and make long-term plans rarely can really bring value, and people need to be aware of this. He also pointed out that usually premature planning may hinder later development, and even the final plan cannot be realized.
If you just release the SDK and development kit, but no developers use it, you lose their value. Carmack believes that some interesting proof-of-concept technologies can prove their importance if they can be used by hundreds of thousands of users and used for millions of hours.
In contrast, “Horizon World” and “Horizon Workrooms”, the definition of a product is that you can evaluate it in many ways, such as: how many users, user behavior, number of transactions, comparison with other products, and so on.
Technical upgrade required
Carmack also mentioned the problems that Horizon Workroom currently needs to solve. He found that compared with Oculus Go, Quest 2’s voice call delay is more serious (over 100 milliseconds), because Quest 2 is running echo cancellation, DSP processing, etc. at the same time. Other tasks. Therefore, the next goal is to reduce the speech delay to less than 50 milliseconds.
By solving these detailed problems, the related technologies that will be upgraded in the future can also be used in other products such as “Horizon World”.
At the same time, although “Horizon” is a very promising product, the existing Metaverse technology is still far from the expectations. For example: “Horizon” has a limited number of people online at the same time, and an event can only accommodate 16 listeners, while an ordinary offline speech may have an audience of at least 50 or 60 people. Carmack said: If the problem of the number of people online at the same time cannot be solved, it will be difficult to realize the vision of the meta-universe. Therefore, I hope that this problem can be solved next year, and maybe the Keynote of the Connect conference next year can be conducted in a large-scale meta-universe.
In addition, even if there are only 16 people communicating in the “Horizon” room, problems such as low fidelity and screen jitter have already occurred. Especially for office applications such as “Horizon Workrooms”, the number of people online at the same time greatly affects its application scenarios, and it is difficult for large companies to use it for meetings.
Carmack pointed out that the above problems may be solved through cloud rendering, but this requires high bandwidth and may not be able to support all users. Alternatively, the “Horizon” cloud service can also be developed independently, which means that users can choose to run “Horizon” locally on an all-in-one machine, PC, or run in the cloud. Another solution is that the room interface and handle interaction can be rendered locally, while the avatars of other people can be rendered in the cloud. This hybrid approach may reduce the bandwidth requirements.
As for the computing power of VR hardware, even though PC VR can run 10 times faster than all-in-one VR, when PC VR runs existing “Metaverse-like” VR applications, it still cannot achieve a large number of online people or ideal rendering effects. This shows that computing power is not the only focus. As far as the existing technology is concerned, it is impossible to achieve a realistic avatar like the Codec Avatar and a large number of online people at the same time, and the cost is very high, even if it is rendered in the cloud.
In the future, cloud rendering may be able to stream Metaverse on various devices, and this will require optimization of Metaverse content. For Carmack is concerned, the rationale for want of form Metaverse will not rely on high-end hardware, but through a series of optimization, can run on mobile devices. He said: Now we should not talk about the infinite possibilities of Metaverse technology, but should think about how to simplify and filter out the most important content in order to run it on the existing hardware platform.
For Carmack, VR is like the experience of 3D screens and people, while Metaverse is more like the experience of many screens and many people, so VR games such as “Beat Saber” are just a small part of this platform. In addition, the initial users of Metaverse may not understand 3D modeling, so there may be scenes where 2D photos, videos and 3D content are mixed at the same time.
In order to provide users and creators with a high-quality platform experience, Carmack will then push the Meta team to optimize the compatibility of diversified applications and multiple cloud services. In short, various tasks can be performed in the VR meta-universe, and the effect is beyond Existing hardware technology.
For the future of VR hardware, Carmack’s goal is to allow it to include and replace other existing hardware devices, rather than to make VR headsets and TVs, game consoles, computers, notebooks and other mainstream hardware complementary additional equipment. That is to say, like mobile phones, it can replace TVs and other devices, while also bringing additional value.
Carmack said: Almost everyone agrees that a closed platform cannot be counted as a meta universe platform. Nevertheless, the open scope of the meta universe also has different levels. Some meta universe platforms are completely open and free, and there are also completely closed single application platforms. For Meta, in order to better enhance the value and experience of the platform , the Meta Universe platform needs a centralized system to formulate standards, so it is almost certain that Meta’s Meta Universe will not be fully open.
Whether it is a centralized meta-universe or an open meta-universe, every transaction of people involves a variety of different commercial interests. He laughed and said that to judge the openness of a platform, adult entertainment content can be used as the standard. If a platform has adult content, it means that it is a very open business platform.
Although the open metaverse allows more free transactions, the risks that follow include unprotected fraud, advertising, and so on.
Carmack said: In the next 18 months, open and closed meta-universe platforms may appear at the same time. Which form is more suitable is worthy of our attention.
On the other hand, the most significant element of the universe development path, may be from one through- used app derived. Just like Roblox, if an application has a Turing- complete extension language and enough interfaces, then in theory, any function and scenario can be implemented in such an application. At the same time, the Oculus operating system can run a large number of VR applications. As the types and numbers of applications become more abundant, the boundaries between the Oculus platform application scenarios are becoming more and more blurred, that is to say, it is more versatile.
Nevertheless, Carmack does not believe that a single software or application can cover all meta-universe scenarios. If the central system of this software makes a wrong decision, the third party cannot fix it, so many users may give up using this software. In other words, in the future, Metaverse will not only be developed by one company, but will compete through multiple companies to bring users a sufficiently good experience.
About content development
Carmack pointed out that Metaverse will support Unity plug-ins in the future. Anyone can develop applications based on the large-scale basic functions of Metaverse. In fact, “Horizon Workrooms”, “Horizon Worlds”, and “Horizon Venues” are all Unity-based applications, but the existing technology is not well integrated with Unity, so it is difficult for different Horizon applications to communicate with each other. Inter-integration is chaotic.
In order to solve this problem, Meta will unify the overall code, improve the basic functions of the meta universe, and simplify the threshold for content development. Some simple functions of the Horizon platform will include: sharing and displaying existing functions such as virtual avatars and pictures. The deeper meta-universe will not only have simple social functions such as game score lists and inviting friends, but also include the Horizon Home scene based on C++ language and Shell commands. The 3D virtual avatar used in this scene not only supports Unity, but also supports C++ language. .
Meta universe hardware carrier
Carmack revealed that Meta is exploring some new AR/VR hardware designs around the meta universe. I object to this, and I don’t think it is necessary. In fact, Quest 2 is sufficient for future AR/VR office and meta-universe scenarios. Indeed, the high-end VR headset Project Cambria that will be released next will be equipped with facial expression tracking, upgraded environment recognition and other functions, but these new features will not be the core experience of the meta universe. At the same time, the upgraded hardware also means significant Increased cost/selling price.
He believes that high-cost VR headsets may bring resistance to the development of VR content, and then VR developers’ feedback on Cambria will be worthy of attention.
On the other hand, VR hardware development takes a long time, so we did not release new products at this year’s Connect conference, only revealed the prototypes under development. On the one hand, the Quest 2 product cycle will be very long, on the other hand, Meta is developing a variety of VR prototypes, and there are many functions that require long-term testing.
For Meta, the VR headsets they will develop in the future may only cover a part of the market. The mature VR market will include low-cost, low-threshold models, as well as tens of thousands of high-end models, like the current mobile phone market. In order to expand its market share, Meta had previously considered opening up the application ecosystem and supporting other third-party VR headsets, but found it difficult. Because Meta sales of headsets actually have no hardware profit (maybe only Meta in the world can do this), mainly relying on content to make money. If other manufacturers also develop low-cost VR headsets and access the Oculus content ecosystem, they will also need to pay Oculus the application transaction share. No one is willing to make such a loss-making transaction.
And if high-end VR headset manufacturers want to access the Oculus ecosystem, it will be difficult. Because VR headsets may support functions such as eye tracking, facial expression tracking, and environment recognition, this will require deep core system software integration, and Meta cannot guarantee cooperation with third-party vendors to develop such technologies.
Carmack believes that it may be less difficult to jointly develop certain VR headsets. For example, VR headsets with sensors and modes similar to those of Quest 2 have features such as ultra-wide field of view and ultra-high resolution. However, there is currently no such product on the market. Whether it is the Pimax pie with a large field of view or the Varjo with a “retinal resolution”, the form is basically PC VR, which is not the same as the technical path of Quest 2.
Review Gear VR and Oculus Go
In this speech, Carmack mentioned Oculus Go again. As two low-cost C-end VR headsets, Gear VR is based on the computing power of Samsung Galaxy phones, and Go is based on the built-in chips of all-in-one phones. In contrast, Galaxy has a faster processor and a better screen, so the visual effects, The content experience may be better than Go. Nevertheless, Go’s user retention rate is several times that of Gear VR.
Carmack believes that, compared with Go, the disadvantage of GearVR is the need to put the phone in the headset box, which increases the user’s resistance to use. Therefore, it can be seen that some optimizations of VR hardware design will help improve user experience and make it more convenient to use.
Carmack believes that “out of the box” VR products will be expected to greatly enhance the user experience. Although Quest 2 is already close enough to this goal, every time you put on the headset again, you probably need to reset the borders, confirm the pop-up reminder, and wait for the homepage to load. Even if the VR application has been opened, you may still need 30 seconds. Or above loading time.
Imagine that if it takes so many steps to open the phone, it takes 2 minutes to load, the experience must be very poor, and most mobile phones now have this problem. Therefore, for VR, there is still much room for improvement in reducing user resistance in the future.
Analyze Quest 2 from user behavior
Carmack said: Cost/price is very important for VR headsets, but I am worried that when it comes to the trade-off between cost and technological innovation, people often pay more attention to low prices than quality. As for the scientific research team, they are usually willing to pursue the ultimate technological innovation, rather than paying attention to cost performance. Moreover, they have seen the latest technology and may not be able to appreciate that some basic VR hardware can actually bring great value to users.
Indeed, there is evidence that the demand curve of many consumer electronics products is non-linear, which means that after a price cut of $50 or $100, the sales volume may increase more than the linear growth curve. But the price alone may not be the reason why Quest 2 sells well. Quest 2 is not only lower than Quest 1 by $100, its operating speed, resolution and other performance have also been improved, so its demand curve is largely non-linear.
There is also evidence that even if it’s money-free VR, it may not necessarily bring value to users. In fact, many users do not use Quest 2 regularly, and some Quest 2 is even idle for a long time, causing user loss. In contrast, you rarely meet the long-term without the phone person, one might replace the phone, but hardly a long time without. This shows that for some users, VR has not yet provided enough value, even if they already have a VR headset, they are unwilling to use it.
Therefore, VR still needs to find more application value at present. This will need to be resolved from two aspects: software upgrades and hardware upgrades.
In terms of software applications, after careful observation of Meta, many game consoles will experience a cycle, that is, some users will be temporarily lost, and these users will return after the release of new high-quality games. Therefore, Oculus released the major production VR game “Resident Evil VR”, and will further observe the impact of new high-quality content on user activity, whether it will attract some existing users to return to VR and start to experience other new games. Upgrade of VR experience.
In terms of hardware technology, perhaps some lost users hope that VR can add new features such as facial expression recognition and high-quality AR perspective, but this will require further technological improvement to prove that the experience of these new features is much improved than before.
As for Carmack, Quest 2’s functions are sufficient. In the future, if Quest 2 is used to further increase speed and reduce costs, there are also great prospects.
Future technology development
So on the basis of Quest 2, where will Meta’s VR headsets be upgraded in the future? Carmack believes that some basic upgrades include optimizing the software system, improving camera resolution, expanding memory and flash memory, speeding up calculations, and so on. In addition, he believes that it is not necessary to increase the camera sensor, even with the existing mobile phone technology, it is impossible to balance the cost and the number of cameras at the same time.
Carmack said: The number of hardware modules will determine the bottom line of the product’s price, and software can actually improve the VR experience in many ways. In the future, we can continue to pay attention to the importance of software and hardware to the improvement of the VR experience, such as how much value the Cambria headset can bring to VR in the future.
There is little information about Cambria at present, it may be equipped with eyeball, face sensor, RGB perspective camera module, and Pancake lens . Carmack pointed out that the higher cost and more compactness of the Pancake lens will help reduce the thickness of the VR headset. In addition, the Pancake lens is expected to be clearer and more optically efficient.
In addition, Carmack believes that heat dissipation is also a major factor limiting the performance of VR. Perhaps the next-generation SoC has higher computational efficiency and can make up for the heat dissipation problem. However, in order to avoid battery fire and other dangers, the design of heat dissipation measures is too conservative.
In fact, he feels that the existing Oculus headset has not fully utilized the cooling system. Even if the Quest 1 is not as well configured as the Quest 2, it can theoretically run most of the content of the Quest 2. In other words, he believes that the amount of calculation that Quest 1 and Quest 2 can run should be higher than expected.
Especially during the test, the limits of the hardware were not explored , so I don’t know under what circumstances the Quest will explode and the SoC chip will catch fire. Carmack does not approve of this conservativeness, believing that if the hardware is tested to the limit, more valuable optimization space may be found.
Carmack also said: According to different hardware forms and application scenarios, VR headsets can be equipped with different cooling solutions, such as an efficient active cooling system, or a quiet and lightweight passive cooling module, or a cooling solution suitable for all-weather AR glasses.
He also mentioned the future VR interactive solution. At present, the dual 6DoF controllers do show great value in the game field, and they are expected to be improved in many aspects in the future. For example, the controller supports autonomous positioning, reduces costs through passive positioning, and uses gestures. Map handle buttons and so on. In addition, there are many scenes in the future will not need to handle, but with gestures, brain waves, voice, note view to controlling the like.
In the future, Carmack is also optimistic about the flexibility of VR headsets, such as the use of modular design and so on.
In terms of software, future VR also needs to optimize the stability of the system and virtual memory to run multiple windows and applications in the background at the same time. At present, Meta is optimistic about the prospects of WebXR internally, and Carmack believes that VR should first natively support Android applications, so that the process of content migration is omitted, so that VR can quickly support millions of existing content.
Carmack said: We should make full use of existing 2D content and proactively provide support instead of waiting for developers to port content to VR. In addition to Android applications, he hopes that VR can also run any existing applications such as Windows programs in the future.
Even VR handles can be optimized for different application scenarios, so that VR can realize the functional experience of TVs and laptops. Or, Carmack envisions that in the future VR may support text terminals in games, which can control various functions by entering codes.
In short, for the future VR scene, Carmack has many ideas. At present, various research projects have also been incubated inside Meta. For example, some people have begun to explore how to shoot stereoscopic videos in VR and so on. However, he also said: Meta’s internal approval process is more complicated, it needs to be reviewed by the design team, and new features need to be translated into multiple languages. The whole process requires a lot of communication, and even simple functions may take 6 months to approve, so efficiency needs to be improved.
Posted by:CoinYuppie，Reprinted with attribution to:https://coinyuppie.com/carmack-make-complaints-meta-metaverse-is-not-empty-talk-technical-experience-and-cost-performance-are-the-key/ Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.