And welcome to episode 12 of Lost in Immersion, your weekly 45-minute stream about innovation. As VR and AR veterans, we'll discuss the latest news of the immersive industry. So let's go. And I guess that this week is a special week dedicated to the Google I.O. conference. So let's start with Fabian as usual, please. Thanks. So, yeah, indeed, there was a lot of announcements last week at Google I.O. conference. And so on my side, I was really interested by what they call a special API. So it's a tool for augmented reality experiences. And basically what it is, it allows AR developers to anchor objects basically anywhere where Google Studio, Google Earth is supported. So if you want to have a quick look at this video here. So. You see that by using your location, using on your phone, you could anchor 3D objects, 3D scenes and such like that into the real world. So these images are obviously fake, but you can see. Real images here, you know, they have two supported SDK, which is Adobe Arrow and Unity. So it's a mobile app, a dedicated experience. And of course, they partner with a lot of different brands and studio to release games that are using this technology. So, yeah, so here this is a very good, very good illustration of what it could look like. So you can put objects really onto buildings on top of buildings and stuff like that. So it's a very exciting technology. Yes, so they have a collaboration with Singapore, TouristBorne, street art, of course, will be a big one. Architecture and real estate, of course, will be a big one as well. So, yeah, it's a very impressive release. And I've seen test videos appearing in the last few days, and it seems to be very accurate, even at distance. So you can look at something quite far away, but still this 3D object that would be laid on top of it be correctly positioned. So there was on the web side of things, there is 8Fold was doing a similar GPS, which is a visually positioning system on the web. But here it's dedicated to app and games. So, yeah, I'm very impressed by the technology, which seems to be super accurate. And I'm curious to know what you think. Just to correct you, I think it was Niantic SDK, and they bought S Fold, so they turned that from app to web lately. But the GPS was already available in the app with Niantic SDK. But the main thing that changed here, and that's really interesting, is the fact that you can prepare your scene with Google Street View on your PC. So you can position things on the 3D rendering map of the place you want to add the element on. And you can directly adjust your positioning inside that 3D render of the environment where you want your real element to be displayed. So that's, for me, a game-changer, because for Niantic, you have to almost be on site to position your element, launching the app, and directly position your element. But with Google now, there is really a tool dedicated to prepare the scene before on your PC. Yeah, I wonder if this tool is available also for Unity, or if it's only for Adobe Aero. But yeah, as you can see on the screen I'm sharing, you can have a tool to position correctly the images, I mean, the media. This is very impressive, yeah. Yeah, I saw that a few weeks back, or months back, that Gorillaz had a new concert in Times Square. And back in the days, I was wondering what kind of technology they were using, and apparently this is this technology. And again, a few months back when I saw the experience, it reminded me some kind of, like, this kind of experience has already been tested and it was not that successful. However, the Gorillaz concert seemed to work, and people were, like, enjoying it, and I couldn't find what kind of technology was behind. So it answers some of my questions here. And yeah, when we are seeing these kind of tools, we really believe that the only step lacking to a global adoption is the device, of course. Because when it represents every kind of usage, everyday usage, we can think of the AR, you can have the advertisement, the entertainment, and all the professional side, especially with the building and configuration of streets and architectural sites. So yeah, once again, we are very, very productive on the tools and the rendering and what we can do with AR, and it's still frustrating that we have to use our smartphone or tablet to see that. So maybe in June or September for someone, depending on your bet. I also wonder if this is the next step for advertisements. Will there be, like, will this allow a new layer of out-of-home AR advertisements? I don't know if Google has plans to monetize this, and, like, I don't know, if someone wants to put something on an iconic building, does anyone have the right to put anything on that building in AR? I don't know. That's something that is very interesting to explore. It comes to the very old video of what could AR be with your everyday life and with advertisement everywhere. So it's a great solution when you have only a few users, but yeah, if you have hundreds of advertisements on the same building, it would become a mess. But don't you think this is a bit strange that you talked about Snapchat opening their AR for retail just a few days after that Meta did the same, and then we have this kind of tool? Don't you think this is – maybe they have information we don't, but it seems like we are witnessing an AR rebirth. We talked about the thing that AR was kind of falling in the shadow, and we can see that it's becoming – it's getting back on the front scene again. So maybe something is happening, and we don't have all the keys yet, but it's very exciting to see AR coming back right now. Maybe just announcements before the headset from Apple is released. Yeah, the latest person that tasted it, like Palmer Luckey, who is very critical usually about the headsets, he seems to have enjoyed it. So let's see. It's still 3K. Yeah. So Seb, we can continue with your subject here, but maybe Fabien has some last words about this. Will you be using these tools in your company or project? Did you try it, or will you use it? Yeah, actually, for us, it opens really a new capability that was kind of already there with A4 and Niantic, as you mentioned, Seb, but having Google really pushing for that and the power of the – I don't know – years of data that they have with Google Earth and Google Street View, it really opens up a new market, I think. And I don't know if you read anything about that, but do they say that it's making augmented reality available absolutely everywhere? Because for Niantic, it was only on small cities that they scanned, but it was limited. Here, it's really everywhere. I don't have the answer, but I would think that it's linked to Google Street View, so I would say they use the Street View data, so if Street View is available. Sorry, it's not everywhere, but it's better than it was. It covers much more than what Niantic was offering, which is nice and makes sense to use it for the next project. Nice. So, Seb, what's your Google subject? For me, it's what they announced about the Google Maps tiles, so the fact that they will release a way for using in Unreal directly their source 3D media of Earth, basically, to be able to navigate and use that in games or simulation. And I think the results start to be really impressive. I'm sharing my screen right now. Yeah, we can see that. I know it's the first version, so it's already amazing. I can't imagine what they will release in the future with artificial intelligence to even make the 3D model better than that, but it's already quite impressive what they managed to get. And what are their use cases, despite the fact that you can access the Google Maps? They have the metaverse world in this description. Do you think this could be a way of having Earth digital twin? It could, definitely, yeah. And what would be the point? We could meet somewhere. Yeah, I don't know about the metaverse, but I think the fact that you can, at least more for industrial use case, I think that's really interesting even to see if a truck can go through a construction site through the normal route, or do they need to think about something else? I think there is a lot of use case in the industrial use case. Yeah, I think that... Let's go Fabien, please. Sorry. I saw as well that using that technology, you can have a preview of the road on Google Maps. If you are going from A to B, you can have a drone-like view of where you will go, which seems to be a use case that they are thinking of for this, to improve the immersion, I would say, and the ability to see the surroundings during a view. But yeah, it's really impressive, and it's really amazing how quick the developer community was to use this. I don't know if they had some preview access or something, but in just a couple of days, you had what you are showing, the flight simulator, coming out. So, yeah, it's, I don't know, maybe less work for the three artists to create cities. That's what we were saying. Yeah, I'm impressed by the quality. I thought that it would be more blocky than it seems to be. Of course, when you are on the street, the resolution shouldn't be enough to get a good view, of course. But for an aerial view like this one we are seeing right now, it seems to work very efficiently. The only thing that I'm thinking of is when you are doing the building application, for example, that you really need to have the updated version of the Google Maps view. And we know that they don't, they are not doing a refresh of that view every day or every week. It can take a few months or years, depending on where you are living. So, this is maybe, as we have this kind of technology in hand, maybe the next effort from Google would be to have a more frequent scan or view of what they are sharing with Google Maps. One idea that I'm having right now is, I don't know if this raises some privacy concerns. I know that it happens sometimes to me when I'm looking at Google Street View, some buildings are blurred. I don't really know why. I didn't bother searching why. I wonder if this can raise, yeah, I don't know, security issues maybe for governments, or if they have some, I don't know, secret access to something, I don't know. But I guess, as I am saying this, I guess if someone wants to do it, they just look at Google Earth and they can already look at that right now. So, maybe it's not a real concern. Okay, do you have anything more to add, Seb, about this? No. Do you have any use case you would like to try with this? Not yet. Not yet, no. Fabien, use cases? Real estate. I think if the data are up to date, as you were mentioning, it can be a really nice way to have a real 3D preview of what could the landscape be. I think that's, and also for, I don't know the word, but city planning and stuff like that, being able to, instead of, Google Earth is great, but having a real time application in Unreal Engine, you can add buildings and interactions and navigations. So, I think it can be a great application with that. Okay, great. So, I'll move to my own subject, which is a Google one as well. Okay, let's do this. So, I would like to talk about something that I discovered just a few days or hours back. It's the Project Starline. Did you hear about this already? Yeah, okay. Is this recent? Because I found out that the first video is dated on 2021, so it's becoming a bit old. Okay, nice. Yeah, it was done for COVID, yeah. During COVID start. So, for those like me who are not up to date with this kind of news, they are working on some kind of hardware coupled with software-based, where basically you can interact with a person, like in a Zoom or Teams video meeting. But instead of having just a 2D view of your interlocker, you can have some kind of autographic 3D view. The main idea is to have the feeling of presence between the two speakers. So, in 2021, they had some kind of bulky design to this with 3D captors, 3D scans, and of course video ones. As you can see, there are three different tracking or capture devices around the screen. And they had this, some kind of scale one-to-one screen so that you have the feeling that the person is right in front of you. And what they did for the Google I.O. a few weeks back is that they updated this first bulky design to something that is way lighter and easy to use. So, they already deployed these prototypes to several companies like Salesforce or T-Mobile, and they are trying it right now. So, basically on the technological part, I guess they are doing some kind of volumetric capture and they found out some kind of algorithm to compress this data and send it to another person for them to see. From what I heard and saw about this, because people were actually trying it during the Google I.O., they had two different boxes. They weren't allowed to take pictures, but they could give their review on it. And as they are seeing the things, the feeling is good. However, the 3D effect is not that impressive. People were more thinking of holographic. So, I guess that the 3D is more behind the screen than on top of it, so you have the depth feeling of the person. Someone talked about their friend giving them an apple and they could feel this kind of holographic effects, but still always behind the glass. So, I guess it confirmed that it's only on the negative stereoscopic part. And yeah, I don't know about the resolution because we know that the volumetric capture is not that good right now. Maybe they found out something that can improve that. I guess they are just using the 3D mesh to give this kind of volume or 3D aspects and mapping the 2D video rendering on it. So, it would give them the resolution of the video, but the depth of the 3D scan. So, what do you think about this? Fabien, for example. So, yeah, it looks very nice. And I think maybe it's still in development and they will continue to improve it if they keep working on it. Google sometimes drops things without warning, but I think presence is very important and maybe we can imagine in a few years with the compression algorithm getting better and better, the 3D capture getting better. We could maybe do this podcast with a better feeling of presence with motion capture, depth capture on our laptops and stuff like that. I don't know. I wonder also if there is a usage for VR. Like if we think far away in the future, can this be used to have real time avatars, volumetric avatars? I don't know. It looks very interesting as a starting point, I would say. It looks nicer than what we have seen with the Azure Kinect. There was some testing done with the Azure Kinect on that using the same kind of setup with three Azure Kinect around a person and showing that to another person in another room. I don't know how fake this video you are sharing is compared to what you are saying about the person that was able to see the holographic effect. I guess it's like a blur on the border, but the main issue is to get where the user is and the background is and make a 3D model that is sharp on the border. You can see on the demo video that I don't know if they added this to blur the lines, but at some point when you are seeing the hair, it's kind of flicking like when you are doing the 3D scan. I don't know at what point, as you said, which part is fake or not. We don't have any more information about this. For me, it's only useful if the quality is there, because otherwise you decrease the quality of the video. What is the point if you could have only a video of the other person? What is the point of adding the text? I'm not sure. The use case that Fabien was describing about having a VR or an hologram of the other person in VR or in a mixed reality, then that would make sense to be able to bring someone next to you and have him giving you some information or some training. Then that would make sense. But for this kind of call, I think we are so used now to call this video only, that adding the depth into it doesn't add that much, but for a lot of expensive devices, I would say. Well, on my part, I was still a bit surprised that Google is doing this kind of project, but as we know, they are not very successful with this kind of project like Project Tango or Google Glass or any of this. It smells like all these other projects right now with very good demo videos and showing some stuff to some people, but the last step to get it in the customer's hands is not that easy for them. I understand the algorithm part, getting the volumetric data compression and all that stuff, but I don't really understand the hardware part where they want to get with this, because as we are seeing, it's kind of bulky. It's a very huge device, and I guess the price is also very interesting. Well, it's fun and it's exciting on the paper, but I really don't know where they are getting with this. The use case is still pending for me, despite the fact that you can have the presence, but is this step enough to get people buying this kind of hardware system? I don't know. I was wondering if they are not doing what Microsoft did with Kinect, which is to iterate on the device through the sensor to get better and better to be able to embed it in the HoloLens afterwards. Is that a way for them to get better to compress the data, like you said, make a 3D model of the person and send it away to another person? Is that their main goal is to get better at that with a temporary device, maybe, and get the feedback and get better at that before making something that would be for multi-reality, et cetera, for making training videos, 3D videos that you can embed it with your VR experience? Yeah, I'm sure the volumetric video part is a big key for the future, especially, as Fabien said, for VR or AR. We talked about the, again, the Apple headset that is willing to provide some kind of holographic meeting or discussion app in their AR. Once again, with this kind of application, people are, I don't know if you saw this video when people are talking to each other in an holographic way with AR glasses, but if you are doing so, you are wearing glasses, and in the video, they don't have them. So, it would mean that someone is not having the AR and filming it in volumetric. So, there's something that is not working in this video, because if you have the glasses, of course, it would be on the 3D scans, unless they are finding a way to erase your headset device and bringing your whole face in the volumetric scan. We can do that, I guess, with the latest AI innovation. Well, I guess I'm finished with this part. Do you have any more subject or topic to bring to the table? Maybe a bit more news about Google. During the Google I.O., they also announced a lot of, of course, AI stuff. So, Google Bar, the competitor to ChatGPT, is available to anyone now, in limited countries, but it's still available. It's getting available. They also announced and showcased generative AI in Gmail, so to assist with writing emails, and also in Google Workspace, so it will come through Google Docs and slides and stuff like that. So, there was a lot, a lot of announcements. They have APIs similar to OpenAI. One thing that was interesting on that is that they try to go to the safety road by, on Google Image, they will start to label images, so the user will know if the image is generated by an AI, or if it's a real image. So, that was interesting news, but yeah, I'm sure you saw it as well, it's AI, AI, AI on Google's side. So, I think it would be interesting to see, because so many users are on Google services, will that drive more usage onto the Google AI stuff, and decreasing the usage of the OpenAI, like ChatGPT? I don't know, but it's still very interesting to see the response to OpenAI with Microsoft. I guess the battle is on, and do you know that if there are some subscription to get the BARD on Google Workspace features, or are they offering them like any Google services? That's a very good question, I don't know. I have access to BARD and it's free, but maybe the Google Workspace, there will be like an additional plan, I don't know, that's a very good question. Yeah, because as we talked, the OpenAI side, they are getting a bit more aggressive on the subscription part, the free part is getting less and less interesting right now. So, I don't know, maybe Google is trying to get back on the front line as well, and offering these services for free right now, for now, sorry, and we'll see if they are getting a new subscription or paid service like the others. But yeah, you talked about Google, NVIDIA is the same, they are all AI-oriented right now, because it's a trending term, and they are showcasing all the work they've done in the past years, of course. Okay. So, anything to add, Seb, Fabien, we are done for today? Yeah, sorry, just one very quick thing is the OpenAI and ChatGPT, they released what they call plugins a few weeks back, so it's a way to give eyes, and like a connection to internet, basically, to APIs, to ChatGPT, so you can, you know, book tickets using ChatGPT, for example. And from the start, Google has already a lot of similar plugins, so they work with a lot of companies to connect the chat with these plugins, so it's very interesting to, like, again, a new development, and a new battle, as you said, between the two companies. So it's like the Chrome extensions, or Chrome plugins, you can have some partners that are using AI inside the Google, or, okay. Nice, nice one. I was wondering why I was seeing more and more travel advisor, where you put your destination, your location, and give you an itinerary of things to do, but I guess it's already connected to ads, and to hotels, and stuff like that, to advise on things that you should pay for on your trip. Follow the money. Yeah. Okay, perfect. So, that will be a wrap up for today. So, good morning, good evening, and good afternoon to everyone, and we'll see you.
