IMR is a company specializing in the development of wireless technology for high-end VR headsets. Their secret weapon is an algorithm designed for VR. Compared with traditional video compression technologies, IMR technology can compress large amounts of data at a faster rate. I (Ben Lang) visited IMR's office in Silicon Valley a few days ago and experienced their systems.
Daniel Fitzgerald, executive director of IMR, is preparing the wireless system.
There are currently a range of different wireless solutions. In order to manage the large amount of data required for 90Hz, 2160 x 1200 resolution VR heads, some companies have used newer ultra-high bandwidth 60GHz technology. Other companies are looking at the existing video compression codec and adapting it to the current bandwidth by compressing the data.
IMR's strategy is closer to the latter, but instead of using existing compression codecs, they have developed proprietary algorithms. IMR said that current codecs are designed for the compression of traditional digital video, and their algorithms are designed specifically for VR. Why is this important? An example given by the IMR is a codec such as h.264. They stated that this codec uses frame-to-frame compression, but this technique will exhibit a delay of at least 11ms at 90Hz frame rate. IMR claims that their proprietary compression technology does not use frame-to-frame compression, and that it can encode and decode within 1ms and achieve 95% compression. The company said that the delay caused by the entire end-to-end wireless VR system, including various hardware interfaces such as HDMI, is about 2-3ms.
The 95% compression rate is remarkable because it opens the door for VR headsets to transmit data over existing wireless technologies such as 5GHz 802.11ac.
So this involves technical issues. So, what happens when IMR's solution really applies to VR headlines? That's why I went to IMR's offices in Silicon Valley.
I saw a proof-of-concept system in IMR's office. They showed their core technology through HTC Vive (although IMR's system can support Rift and other headers). I found a small white box on the desktop, and the device connected to the Vive head. You can notice a yellow cable on the side. The effect of this cable is actually to connect the decoding and encoding hardware so that the system can easily debug the compression algorithm without the need for a wireless link.
To demonstrate the wireless performance of the system, the IMR removes the yellow cable, inserts the wireless transmitter into one end, and the receiver plugs into the other end to wirelessly transmit the same data. What you see in the picture is the wired version of the HTC Vive, but the data is actually transmitted over the air, but it looks a bit different. But this is very reasonable for a prototype, and it still demonstrates the main functions of the system. (Note: Most of the pictures in this article show yellow cables, but this is only used in AB tests that compare wired and wireless delays)
The ultimate goal of the system is to attach one half of the white box directly to the Computer and the other half to the head-up itself, and this solution is currently under development.
When the system was able to run wirelessly, I tried a series of familiar games, including "The Blu", "Space Pirate Simulator" and "Job Simulator". For the most concerned delay issues, I didn't actually feel any difference with the tiered version of the HTC Vive.
As for the visual quality, IMR indicates that the decompressed image is the same as the original image. I agree that the image quality is very high (especially considering IMR's stated 95% compression rate), but strictly speaking, it is not "there is no difference". If you want me to do blind testing to determine which one belongs to the decompressed image and which one belongs to the original image, I believe I can identify it. In other words, if you want to get a wireless VR experience, the IMR system can achieve enough quality to make you think it is a feasible solution. I think ordinary users will probably not notice that there is a big difference between them.
IMR said that their algorithm is highly adjustable. According to the bandwidth used by wireless technology, they can adjust the compression rate, say 75%. At this time, the difference in image quality may even be comparable to what the naked eye can see.
I personally saw this at this year's CES 2017 conference. Even at the same compression rate, image fidelity has improved dramatically. This improvement is so great that I thought it was necessary to make a trade-off between image quality and IMR wireless technology, but now I think it is possible to seriously consider whether it should cut out the Vive cable.
Of course, as with all compression technologies, different regions have different optimal compression ratios. IMR compression technology still has some challenges. The primary issues are gradients, and dark/low contrast scenes. Both may cause a certain amount of streaks and artifacts. IMR claims that they have a way to deal with these issues, and they will continue to adjust algorithms to avoid artifacts by intelligently compressing different parts of the scene. But you can imagine how difficult it is to ensure that the algorithm can compress a large amount of different VR content with 100% accuracy. After all, each content has its own optimal compression ratio.
In order to help the algorithm use the appropriate compression method, IMR means that the game may send parameters to the algorithm, allowing the system to make adjustments based on the specific game. This way, developers may be able to fine tune themselves or crowdsource them to users.
IMR Wireless VR Accessories Possible Concept Rendering
The white box I see is just a proof of concept. The next step for IMR is to integrate the white box into a reference design so that the device can be carried on the head and achieve true wireless operation. IMR is currently developing a reference design and will soon be making a public demonstration. At the beginning it may seem very heavy, but IMR says the system can be easily miniaturized in the future and can be integrated directly into the VR headset.
IMR said that their compression technology can support current VR headsets through existing wireless technologies such as 802.11ac, such as Rift and Vive. For higher resolution head-ups in the future, IMR hopes that the more advanced 802.11ad technology can be popularized because the bandwidth of this technology can support monocular 4K resolution.
It is understood that the IMR algorithm can already be used for eye tracking. IMR shows that when eye tracking data is input into the algorithm, the system can further optimize the compression of the eye gaze area while saving the resources of other areas (acting similar to fixation rendering).
IMR said that they are currently looking for production partners to help manufacture wireless accessories suitable for Vive and Rift, which may be available in 2017. The company will also share its reference design hardware with partners to facilitate the integration of IMR's wireless technology into the VR headset.
Wenzhou Hesheng Electronic Co., Ltd. , https://www.heshengelec.com