Integrated Systems Europe: Light-emitting diodes, AI and immersion

Contents

In the exhibition halls at this year's Integrated Systems Europe (ISE) in Barcelona, the trend towards direct-view LED displays became abundantly clear. These are seamlessly assembled from handy LED modules. They are therefore very flexible in terms of target formats and still have no bars in the image.

Here too, manufacturers are aiming for smaller pixel pitches in the range of 1 to 1.5 millimetres, which still have to be reasonably affordable. This is where LED displays for televisions still fail. In addition to large-format eye-catchers, all-in-one displays with mini-LED or even micro-LED technology were also on show at ISE. These are not suitable for wall installation, but are also suitable for conference rooms with rollable mounts and formats of 120 to 180 inches, i.e. diagonals of 3 to 4.65 meters.

LED display instead of projection

As a replacement for classic projectors, LED displays produce a comparatively deep black, even in bright rooms. Projection screens react very strongly to ambient light, as their reflection factor is usually closer to 95% – they are intended to reflect the light from the projector to the viewer. In contrast, the reflection factor of an LED display surface is only around 5 percent, which is why much less ambient light is reflected on it. This noticeably improves the image contrast in bright surroundings.

As the image-producing RGB LEDs in LED displays can be significantly smaller than the area assigned to each pixel, the technology can also be used to produce almost transparent displays. These can also be curved or flexibly bendable on an appropriate carrier (transparent film or mesh). This makes them real eye-catchers.

Display blindness

Speaking of eye-catchers: digital signage is all about capturing the attention of passers-by for as long as possible. One problem is the effect of what is known as "display blindness": as a lot of content is not tailored to the viewer, they tend to only take a quick glance at the screen without taking any notice of the content.

Bright colors and strong motion stimuli in the image could change this, but they are often perceived as a pressure to communicate and therefore unpleasant. This can be transferred to the advertised product, which is of course undesirable from the advertiser's point of view. The exception here are media art installations, such as those pursued by LG with its Kinetic LED modules.

Instead, advertisers are trying to create added value through individual content and/or interaction options. In future, this is to be achieved with large-scale language models and generative AI, which are linked to sensors. In a first step, generative AI can also be used to produce inexpensive still and moving images without having to work with real actors.

Digital people

Pantheon Lab from Hong Kong goes much further in the use of AI. The company works with corporate clients in Asia and the USA, such as KFC in Taiwan and the Hong Kong metro. Pantheon focuses primarily on digital people and their integration into signage systems. The company offers a SAAS tool with which customers can generate their own video clips with pre-defined digital people.

Digital people, presented as almost photo-realistic, interactive animations on the displays, can transform classic information displays into interactive contact points. In airports, train stations or shopping centers, they can answer questions, make recommendations, but also take orders and carry out payment transactions. Interactive touchpoints only establish a connection to service staff when human interaction is absolutely necessary.

Such applications require a stable operating system and a combination of cloud AI and edge computing, i.e. processing close to the point of origin in order to reduce latency. High-resolution cameras and noise-canceling algorithms are also needed to ensure that natural language interaction also works reliably in public spaces. If, for example, the virtual barista looks at his customers for a few seconds without saying a word until he carries out a spoken order, they quickly feel uncomfortable.

AI agents

All of this deals with applications that work purely reactively: Passers-by can ask questions, for example at an information display in the metro, or order something in a fast food restaurant and pay immediately – but the digital signage system only responds to the customer's request.

Proactive AI agents, on the other hand, work differently. According to Christina Tse from Pantheon Lab, this is where the future lies, which she believes will come very quickly. Such AI-powered digital signage devices will be able to take action themselves.

Example: Someone goes to a subway station and asks for the connection. Given the time, the AI could then ask whether they are on their way to dinner and have already reserved a table? If not, the information system could help you reserve a table. The whole thing happens seamlessly, you just have to ask a question and the AI in the background does the rest. Pantheon Lab is currently testing proactive agents and hopes to be able to use them in productive operation in the course of the year.

Distanceless

Immersion was in demand at ISE both for events and in the field of video conferencing. Here, it is intended to give participants who are physically separated the feeling of being close together. Cisco calls this a "zero distance" experience.

Typically, the person opposite is shown on a display and, for example, a group of people is shown as a long shot, but this does not actually correspond to human perception. People always scan their surroundings and focus on the person they are currently paying attention to. Conference systems try to replicate this with the help of AI algorithms. To do this, they record different perspectives with several cameras, automatically capture the people speaking with one of the cameras and automatically track them as they move through the room.

With "Spatial Meetings", Cisco offers a firmware extension that enables Cisco video bars to transmit a 3D stereo image. The person connected can then experience the room view in three dimensions using an Apple Vision Pro. They appear as part of a virtual user interface in a window that can be expanded to any size. If the connected person is to be visible themselves, they must be replaced by a photorealistic avatar due to the VR headset.

3D without glasses

An immersive and ideally a 3D display makes such avatars superfluous. Google proved this at ISE with the demo of the Starline project. It took place a little under the radar, but was nevertheless quickly fully booked for all days. During the one-to-one video conference realized in it, the two people very convincingly got the impression of sitting opposite each other in person.

The Starline system consists of several cameras and an autostereoscopic 3D display that produces a three-dimensional representation of the person at the other end. Although there is no camera directly in the line of vision, you can look each other in the eye and have a certain range of movement, which also leads to a realistic parallax of movement. If the display is large enough and has sufficient resolution, this illusion is not destroyed by the limits of the display technology.

Economical e-paper

In many applications, color-intensive, energy-hungry displays are not really necessary; a paper poster would do instead. An e-paper display comes closest to meeting this requirement: it only needs power to change the content and does not necessarily need a mains connection, but can also get by with rechargeable batteries. Reflective e-paper displays with a 32-inch diagonal and e-paper walls in 75-inch format were among the products on show at ISE.

Such electrophoretic e-paper displays cannot produce as many colors as a modern LCD, which is why photos have to be rasterized using diffusion dither. However, their image resolution is sufficient for typical applications such as advertising spaces. Only fast moving images are not yet possible with e-paper displays. (nie)