Augmented Reality and Virtual Reality HMD Testing

HMD performance and user experience is about achieving immersion and presence

AR, VR or MR (augmented, virtual and mixed or mixed reality) HMD (Head Mounted Display) performance is all about achieving immersion and presence. Ultimate presence means that the user is immersed into the virtual reality and that they are genuinely enjoying the experience. Poor performance affects the viewing experience and can even cause motion sickness and nausea. Achieving presence is difficult when system latencies are poor. Failing to achieve any of the presence factors is enough to ruin the whole experience.

What is immersion?

Achieved immersion is the foundation of the AR, VR and MR user experience. There are four types of immersions which are:

1. Spatial-Temporal - the simulated world is convincing. Spatial-Temporal can also be called presence. 
2. Sensorimotor - skill challenge
3. Cognitive - mental or thinking challenge
4. Emotional - user is immersed in the story

How does OptoFidelity measure different presence and quality factors?

OptoFidelity’s systems focus on measuring presence, as it has the greatest impact on hardware quality and performance. Presence is the most qualitative of the immersion categories. OptoFidelity has divided the presence into two categories: Spatial and temporal. Spatial presence refers to the perceived image quality. The main factors for image quality are optical quality and resolution. Display sets the theoretical limit for resolution and optics. Display components, such as lenses and waveguides are creating the virtual image to match with human vision.

Tracking, latency and persistence are factors of temporal quality. Temporal quality requires a well-optimized and integrated HMD system. Latency and tracking quality measure end-to-end performance of sensors, cameras, display and software all together.

Measuring spatial quality

Normal image quality measurables are used to evaluate the image quality of near eye displays in a way that is similar to the way in which normal flat panel displays are evaluated. The biggest difference is that in the near eye displays all the values are presented relative to the field of view (angle). The performance of the device must also be similar for different people with different physical dimensions and qualities.

The most common physical quality that affects the image quality is different interpupillary distance between the human eyes. In order to accommodate for this phenomenon, an HMD device needs to be able to either physically change the distance of the lenses and displays, or if this is done optically, by having a large area where the full image can be seen. This area is called the eyebox. The size of the eyebox should be large to enable ease of use for the end device and still provide equal image quality in all settings. These two factors are extremely hard to implement at the same time. Therefore it is extremely important that an HMD image quality solution has both extremely accurate positioning of the DUT and integrated high-absolute accuracy motion, preferably with five degrees of freedom integrated. These two factors enable accurate measurement of eyebox location related to DUT and makes it possible to execute the rest of the measurements within multiple locations in the eyebox.

Virtual Reality HMDs

Head-Mounted VR Display quality is a sum of the optics and the display used. Imaging of the VR headset is similar to viewing the displayed image through a magnifying glass. In standard flat displays the single pixel defects are less important. A mobile phone single pixel defect in the middle of a display won’t affect normal use. When viewing the same display through an HMD Eye Piece, it will be magnified so that it easily exceeds human eye resolution. Using the enclosed HMD system also increases sensitivity to brightness variation. A panel as a light source in VR goggles will emphasize all grayscale variation.

.

Augmented Reality HMDs  

See-through near eye AR displays' quality challenges are very different since the resolution and image source defects still affect perceived quality. The biggest challenge is making a small form factor system with a large FOV and eyebox. In addition to image generation challenges, everything should work nicely with the real world.

Sunlight brightness is 1.6 billion nits. Mobile phone display brightness is between 500 and 1,000 nits. A large eyebox generally means that the throughput efficiency will be low, which will set demands for light source and waveguide. These are just some of the challenges related to ambient light in eyewear design.

Measuring temporal performance

The complete set of VR and AR HMD design parameters comprises almost 20 individual properties. Only one or two of these measure temporal performance. All others measure spatial quality. Despite this fact, measuring temporal performance is extremely important, because lacking adequate performance will cause the users nausea and motion sickness. The most essential temporal performance properties are:

• Motion to photon latency (M2P)
• Duty time (display persistence)

M2P describes the length of time between the user performing a motion and the display showing the corresponding content for that particular motion. For VR systems, the M2P is typically below 20 milliseconds. For AR, the required latency is below 5 milliseconds. This is due to the fact that the user has the surrounding environment as a reference.

Display persistence describes the amount of time that the display is on for each individual frame. Typically this time is around 1-2 milliseconds. Short persistence time is required in order to minimize perceived motion blur during the user’s head and eye movement.

HMD developers need to utilize the so-called motion prediction in the headsets in order to achieve the required low latencies. These prediction algorithms are prone to errors, especially in special motion cases, like unpredicted fast movements, that users may do. Additionally, motion prediction collects the input signals from various sensors (IMU, camera trackers, etc.), and any anomaly in the sensor data can potentially confuse the algorithms.

Measuring M2P latency, for example, is not an easy task. This is due to its end-to-end nature:

• Motion: requires accurate, repeatable and synchronized movements of the HMD
• Photon: requires display-synchronized imaging directly from the head mounted display

Successful temporal performance measurement requires that the above steps are seamlessly combined together.