Foreword
Domestic manufacturers are always dismissive of the delay of VR head display. At each conference, they all claim that the delay of the product is controlled at 18ms, 17ms, 16ms, as if they are punching oculus and kicking VIVE. The Chinese manufacturers are really so powerful? What is the reason for the delay of VR and the cause of dizziness?
What is the delay?
VR "delay" refers to the entire latency of the optical signal from the person's head movement until the head-mounted device (HMD) is mapped to the human eye. The full steps that need to be passed are shown in the figure below.
.png)
Delays can affect the player's attention, and when the delay is large, there will be motion sickness, also known as motion sickness. The principle is that the visually accepted state of the body is inconsistent with the middle ear vestibular organ responsible for sensing the state of the body. The feedback of the central nervous system to this state is nausea to remind the abnormal state of the body. Simply put, when wearing the VR head to move the head, due to the delay, the visual observation changes more slowly than the body feels, causing a conflict and then causing dizziness.
According to the research, this delay is only controlled within 20ms, and the human body will not have rejection. At the same time, more than 20ms will not necessarily cause dizziness and nausea. Just like some people have motion sickness, some people are not dizzy, and each person's physique has different sensitivity to delay and rejection. For the author, it is easy to get sick and seasickness, so when using VR equipment, the delay will be dizzy sooner, and some people will not faint even if they use the equipment with higher delay.
Therefore, some manufacturers claim that the so-called dizziness is only an individual case, and it is not correct that the product equipment itself is completely ok. Their products do not meet the 20ms delay requirement, so the wearer will not be dizzy. Only a lot of wearers have a strong anti-motion sickness, so they don't have dizziness.
The American legendary programmer, the father of FPS and 3D, John Carmack, wrote an article in 2014 to introduce the delay problem of VR head display. Based on this article, combined with the development of related technologies in the VR industry in the past two years, Let's talk about the issue of VR latency.
Sensor and signal transmission cause delay
Starting from the head movement, the VR head display built-in gyroscope or space positioning instrument will capture the movement of the human body and receive the signal transmission to the computer processor for processing.
For the sensor itself, the better sensor frame rate is basically around 1000hz, so the delay caused by the sensor receiving the human body moving signal is about 1ms, plus the processing signal of the single-chip computer and the work transmitted to the computer, the process is about 3ms. .
In this process, the delay caused by data transmission and filtering is basically fixed. But because the sensor's frame rate (a better sensor can reach 1000HZ) and the frame rate of the rendering are inconsistent, it will bring a variable delay. On the sensor side, the prediction of head activity has the potential to reduce certain delays. But even with a very complicated human head model, this can only deal with the situation where the head moves at a relatively uniform speed, which is difficult to deal with when the head starts to move and the movement changes dramatically.
Properly arranging the sensor for data sampling can reduce the delay to a certain extent. Most of the tasks of analog input are not directly dependent on the user's input data, which means that the input of this data has little effect on the delay of one frame. If you sample when you need data from the user, instead of buffering the input data offline at the beginning, you can reduce the overall latency. This method is called late frame scheduling, which has the disadvantage of requiring a large amount of scheduling and improving the overall power consumption.
Display delay
Then there is the delay caused by the image display. Today's head-mounted screens are mainly divided into LCD screens and OLED screens. Under modern technology, it takes about 10ms for a pixel on an excellent LDC screen to change from one value to another. After optimization, this value can be reduced by half. In contrast, OLED pixel conversion takes less time, typically within 1ms.
In addition, there is a problem: the traditional display is to display images line by line, so for a 60HZ screen, the bottom image will be 16ms later than the top image. This doesn't matter in a still screen, but it makes a big difference in any HMD where the image is rendered in real time. It looks like the edge portion of the display is slowed down, and the screen looks like it is shaking due to the delay when the head moves. Especially for HMDs that use OLED screens with a high refresh rate, it is very obvious.
However, with the development of screen technology, the current active matrix screens such as AMOLED and TFTLCD have solved the problem of screen display delay. Each pixel on these screens can control the switch, and the image on the screen can be displayed at the same time, especially using the AMOLED screen, which can control the delay of the image display step to within a few milliseconds. However, the core technology of AMOLED screen is mainly in South Korea, and the overall shipment volume is relatively low. There are few VR head-mounted devices in China that use this screen.
Delay caused by image processing
The time of image processing refers to the time it takes from the time the input signal is stored in the digital form into the memory until the next frame is transmitted to the next frame.
The process of image processing is mainly divided into digital signal denoising, scene rendering and subsequent anti-distortion inverse dispersion processing. There are three main ways to reduce the time spent and improve the final picture quality.
The first is dual GPU stereo rendering. Since the VR head is composed of two screens, if the graphics processing system is split into two and each is equipped with a GPU, each GPU only needs to render a single screen image, which can achieve the best performance and the lowest. Latency comes at the cost of requiring the device to manage two separate caches of rendered content. If the device is unable to maintain the cache of the two GPUs, the device's data throughput performance will be degraded, and in the case of a cache overload, the frame rate will be more severely degraded.
The second way to reduce latency is to allow the render layer to modify the parameters passed by the game layer (VIEW BYPASS) based on the latest sampled data. Calculate the difference between the current user input and the last sample, which is used to modify the view matrix difference processing submitted by the game layer to the rendering layer, thereby reducing the processing workload.
The third is Time Wraping, which post-processes the rendered image based on the newly entered data. After drawing a high-quality image (using View bypass), this image is not displayed directly, but allows the developer to continue to capture the latest data entered by the user, generate an updated visual matrix, and then calculate and input data. The matching rendered image needs to be shaped and placed where. After this conversion, the rendered image can be more consistent with the image that should be seen after the user actually operates.
The combination of the latter two is equivalent to minimizing the image rendering work caused by the user's operation. With continuous time wraping, the delay in image rendering will be minimized. The view bypass ensures that the image can avoid the parallax of the final image to the greatest extent. The advent of such an algorithm can reduce the delay of tens of milliseconds spent on image rendering.
Old buns
If you look at it, reducing latency should be a very easy task. However, in the middle of the process from the sensor to the screen to the algorithm, as long as there is an error, it will increase the delay of the product, and eventually cause the user to stun. The delays claimed by many manufacturers are often theoretical calculations or estimates that omit some of the steps that are required to move from head to final display. I hope that the domestic manufacturers of VR equipment can start from the algorithm and polish the products a little bit, and finally solve the minimum threshold of VR head display with 20ms delay.
Yonghao Wire Company is committed to providing high quality electrical products, taking YC02 photovoltaic cable connectors as an example, whose superior performance makes them a reliable choice. We meet our customers' needs with innovative technology and superior quality, providing reliable power connectivity solutions for the renewable energy industry.
Applicable scenarios include photovoltaic power plants, solar power generation systems, outdoor photovoltaic equipment, etc., to provide reliable electrical connections for these environments to ensure the efficient operation of solar power generation systems.
YC02 photovoltaic cable connector has the following advantages:
High reliability: The YC02 connector design is carefully optimized to ensure a stable and reliable electrical connection in the photovoltaic system, reducing the failure rate.
Strong water resistance: with excellent waterproof performance, adapt to outdoor, bad weather and other complex environments, to ensure long-term stable operation of the connector.
Corrosion resistance: Strong corrosion resistance, can maintain the stability of the connector in harsh weather conditions, extend the service life.
Anti-ultraviolet: It has anti-ultraviolet characteristics and is suitable for outdoor photovoltaic power plants and other scenes that need to be exposed to sunlight for a long time.
High temperature resistance: YC02 connectors maintain stable performance in high temperature environments and are suitable for devices such as photovoltaic battery modules and inverters.
Applicable scenarios include photovoltaic power plants, solar power generation systems, outdoor photovoltaic equipment, etc., to provide reliable electrical connections for these environments to ensure the efficient operation of solar power generation systems.
solar connect, cable connector, connector terminal,pv connector, dc solar cable connector
Suzhou Yonghao Cable Co.,Ltd. , https://www.yonghaocable.com