Sony XR, next-generation organic EL, and player demise theory-Honda and Yamanouchi AV talks in the summer of '21-AV Watch

Will Sony Bravia's "XR processor" create a new trend in image creation?

--One of the features of the latest TVs in the first half is HDMI 2.1 support. You can.

Honda: Certainly, I think many people are interested in the HDMI version. However, it is not true that HDMI 2.1 is important for everyone. This is because there is a limit to the content that can be supported. Gaming, mainly PlayStation 5, is probably the biggest content at the moment. If your goal is to watch movies, dramas, and anime, I don't think HDMI 2.1 is important.

However, with the improvement of computing power, the game world quickly becomes high definition and high frame rate. HDMI 2.1 support is important if you're into the latest computer games. It's a function that will naturally be needed in the future, but for now, I think it's good to be conscious of "I hope it's compatible with the next time I buy a new one."

Yamanouchi: At the moment, I don't think there are many people who say, "Don't buy a model that doesn't support HDMI 2.1", but the number of compatible models is increasing in the 2021 model. If you think that "it is safe if it is compatible", it would be wise to choose and purchase.

Functions such as 4K120p input and VRR of HDMI 2.1 are currently limited to content such as games

Honda: Unlike HDMI 1.2, 1.3, and 1.4, where the audio formats that can be transmitted differ greatly depending on the version, 2.1 is There is no such element, and there is no need to think about sound quality.

Rather, the topic of visual technology in the first half of this year will be Sony's introduction of a new concept with its Bravia XR processor and LG's introduction of a next-generation organic EL "OLED evo" TV. .

The latter has changed the luminous material, so if you master it in the future, there will be a big difference. Currently, it is only used by LG, but it can be assumed that other companies will also adopt it in the future. OLED TVs have matured over the past few years, but at this point in time they are taking the next step.

In May, LG announced the 'OLED G1' series, which uses next-generation organic EL panels. The photo shows the 65-inch 4K organic EL TV "OLED 65G1PJA"

The top 4K TV "G1" of the next-generation panel "LG OLED evo". Wall pita installation

――Mr. Yamanouchi, what are your impressions of the XR engine?

Yamanouchi: I didn't get the impression that this technology was an extension of the image quality improvement technology that Sony has been aiming for, and that it was doing something completely new. To put it simply, I understand that we have improved the precision of movement with the aim of realizing the know-how of how to show with XR in a more sophisticated form than before.

The technique of optimizing the definition and gradation expression for the gaze point depends on how accurately the area where the person gazes can be detected.

Where the human brain reacts and what kind of information is important is not limited to TV images, but it is almost a constant law not only in the natural world, in everyday life, but also when you go to an art museum and look at a picture. , and distance and brightness also have important meanings. The idea of ​​actually creating rules and focusing on signal processing makes sense in its own way, and if the accuracy is improved, certain effects can be achieved. Looking at the actual LCD and organic EL devices equipped with the XR processor, I was able to confirm that some results have been achieved.

Sony Bravia's "XRJ-85X95J" equipped with the cognitive characteristic processor "XR"

Yamanouchi: When the screen gets bigger, it becomes relatively clear where to look on the screen. Some people may focus their attention on the background, but when it comes to people, they pay close attention to facial expressions and react sensitively to moving objects.

However, the parts that are considered important in creating images vary depending on the creator. For example, there are cases where a person who is not the main character, but who is very important as a supporting character appears in the photo and contains important information. Of course, it will not be possible to detect such a part. To some extent, I think it's an interesting attempt as a method of analyzing universal gaze points.

Cognitive Characteristics Processor "XR"

Honda: The XR engine actively changes the image. While this is a good point, it may be a question for AV fans, so I would like to talk about the current trends first.

For displays, the most important thing has been the ability to give a detailed impression of premium images. It is the so-called "director's intention", and it means whether you can feel the impression that the director intended from the image. Even recently, there are Netflix mode, IMAX mode, etc., and there is also THX mode. This point has not changed since then. Movie mode is basically based on that idea.

However, the images displayed on TV are not always created by the director who thinks carefully about every aspect of the image production. There are news, variety shows, stage performances, music concerts, sports broadcasts, and online content such as YouTube. Documentaries and things that reflect nature should just want to feel nature as it is. Until now, the idea was that it would be better to display content that does not have a strong director's intention as it is, like a master monitor.

However, in our technologically advanced age, new tools have emerged to try out different approaches. That is the method of processing video with an AI-like approach. The images captured and recorded through the camera are different from the actual reality, the real thing. This is because you are forced to make compromises at the stages of recording, delivery, and display. Therefore, attention has been paid to the use of neural network processing to correct differences from the scenery and textures seen with the naked eye, or to compensate for slight problems with exposure, etc., to bring the images closer to reality.

Processing image of the cognitive characteristics processor "XR". The XR processor performs optimal image quality processing according to the person's point of view

Even before that, each company has been actively trying to make it look more beautiful and closer to the natural scenery in memory. has changed in recent years. The XR engine was born in that context and performs “cognitive” image quality enhancement processing.

"Cognitive" is a word that IBM began to use. Although computers cannot perceive, they can "move as if they are perceiving", so they can perceive subjects, backgrounds, materials, etc. and take appropriate action.

In a sense, it's the exact opposite of the director's intention, but neither is wrong, and both are necessary. I think the XR engine is currently the most aggressive AI-like approach to image processing.

As Mr. Yamanouchi said earlier, 70-inch and 80-inch screens have a wider field of view. Instead of staring blankly at the entire screen, they often focus on a specific spot on the screen. The method of making images has changed from the old days, and the number of shots where only one person is bust-up has decreased, and the number of scenes in which multiple people are shot at the same time has increased. In other words, since there are a lot of background scenes, I predict that "the viewer should be watching this" in a wide angle of view, and thoroughly pursue reality for the identified subjects. That's the point of the XR processor. Although it is an extension of the conventional technology, I feel that it is more detailed and delved into it.

The XR processor divides the screen into hundreds of zones and recognizes details such as focus, color and contrast for each object within the zone. Just like how the human brain works, it is possible to inter-analyze about hundreds of thousands of different elements that make up an image in one second

So what about other companies? It turns out that other companies are doing the same thing, but with a different approach. Rather than recognizing each object too much, the approach is to discriminate the scene and try to optimize it for each scene. Do you have an image of doing things individually, like this, this, this, instead of treating them in a uniform way?

Smartphone cameras are probably the closest thing to an XR processor.

For example, if you shoot a composition that has a model, a building with a backlit wall, and a sky, the sky will be blown white and the shadows on the walls will be a little strong. However, when shooting with the iPhone 12 series, the appropriate white balance is applied to the face, the shaded building, and the blue sky, and the exposure is appropriate for each subject. This is tone mapping, which is impossible with a normal camera, but this is because the subject is cut out for each area, recognized, and image processing is performed. For that reason, we are using an AI-like processing approach, and we are trying to do the same thing in real time when displaying images.

The optimal processing of the XR processor creates a natural and real-world image

――I watched it too, but perhaps because the effect of the XR processor was too strong, the sense of depth that should have been lost was lost. There were also scenes that looked flat.

Honda: It would be better to think of “picture quality” and “picture quality” separately. When the amount of information is originally small or there is more than a certain amount of noise, how much correction is needed to make it easier to see. In the case of terrestrial digital, which originally has a lot of noise and distortion, you may feel that it is overkill. But it doesn't work the same for high-quality images.

In other words, if you want to adjust the quality without affecting the quality of the video, I think it's better to do it until the last minute. REGZA is doing the same thing, and each company is trying to make it easier to watch by intervening as aggressively as possible with low-quality video, although the degree is different.

4K organic EL REGZA "X9400S series" equipped with cloud AI high image quality technology

Honda: On the contrary, there are times when he behaves as if his dignity has improved. For example, in a common demo video showing high-definition scenes from around the world, the blur from the background to the foreground, that is, the change in depth of field, looked very natural. I feel that this brings a sense of depth and creates a natural three-dimensional effect.

Yamanouchi: I didn't find anything in the video I saw that made me feel so uncomfortable. Verdi's "Il Trovatore" seen on UHD BD is centered on the scene where the whole scene of the stage is captured by the pulled camera and the screen shot by getting a little closer and taking a full shot. Comparing the two, I was able to depict the facial expressions of the people and the changes in the brightness of the background in a natural range without any sense of incongruity. While carefully reproducing the facial expressions of the characters, I felt that the texture of the realistic stage setting was also reproduced.

Sony Bravia's 4K liquid crystal "XRJ-85X95J" (left photo), 4K organic EL "XRJ-83A90J" (right)

"Process like a brain" Bravia XR organic EL. 4K120p compatible with PS5