Nvidia's NEW 'AI NEURALANGELO Takes the Industry By STORM! (NOW ANNOUNCED!)

so Nvidia just once again it changed the game completely with their new AI tool this is nvidia's neural Angelo so what this tool does is it essentially turns 2D videos into 3D sculptures and this is what we have ladies and gentlemen this is rapidly advancing Ai and this actually allows high quality 3D models to be made simply from a video now if you don't understand why this is about to change the game in many different Industries this is the video for you where we're going to dive into the several different points as to how this is absolutely incredible so let's Jump Right In into why this is possibly one of this year's biggest AI announcements on the surface it does just seem like this is a normal AI announcement which is among the many different AI announcements that we've had this year but the reason I want to take part in explaining why this is such a crucial detail in the evolution of AI is because this actually allows us to have a very high level of immersive detail what you're seeing on screen right now is what we call Maps now if you don't know what maps are essentially when you are having a 3D models what you want to do if you want to have more detail without increasing the amount of Ram or essentially the amount of compute that your computer has to do is of course you want these normals and this software is able to generate normals able to generate high quality 3D models simply from a 2d video now many of you might not be thinking that doesn't sound crazy but you have to understand that there's a lot of data that we simply don't have from a 2d video that is simply being extracted by these neural networks now this has tons and tons of different applications and I'm going to show you more of the stuff that they didn't include on the main video which everybody saw on the research paper which they dive into more examples and exactly why this is truly incredible but when you see the strange purple stuff on your screen that you're about to see now that kind of looks like the 3D model that is something that the AI is actually generating to make the 3D model look a lot more realistic and it's something that helps with the increased realism and this is why this is truly truly incredible because there's a a lot of things that went into making sure that this is very effective you can see right here that this is the input video and then of course we do get the final 3D surface reconstruction now what makes this really important is because in a moment after this video does end I will show you some of the key details as to how this compares to many of the other models and many of the other models that can actually get 2D video to 3 deconstruction don't actually allow for this level of detail now you might understand this technology does actually on the surface seem quite similar to some technology that were released in the past year now what I'm talking about is something that you may have heard of called lumalabs AI now essentially this is what lumalabs AI is so essentially what you just need to do here is all you have to do is simply take a few images or I see a few a bunch of images of a certain scene and then you are able to get a complete 3D scene in which you can manipulate it and record different things maybe you want to have Drone footage maybe you want to have a 3D object but this is something that I did did see in the AI community that was highly touted as something that was going to change the industry and so far it has but you need to understand that there are differences in this technology so lumalabs AI actually takes advantages of the nerves so Nerfs are essentially neural Radiance fields which are techniques that generate 3D representations of an object or a scene from 2D images using Advanced machine learning now the thing is what Nvidia has done is they've done something completely different so essentially the PDF that Nvidia have released does mention references to Nerf neural Radiance fields in the introduction sections but the authors state that Nerf is a recent technique although showing impressive results it requires a large amount of training data and computational resources and they mentioned that their proposed neurological technique is very different from Nerf as it focuses on Surface reconstruction rather than the view synthesis and it can handle a wide range of scenes with less training data and computational resources so in short nvidia's technique is far more effective and less computationally expensive although I would say that lumalabs AI is something that you can use on your phone and to be honest lumalab's AI is actually pretty good especially for beginners because you can get started just using it right away but anyhow this is an Nvidia video and although it is very interesting to compare the different AI tools that are constantly being released when you actually compare nvidia's new tool to the previous tools that are actually able to do the same thing we do see nvidias actually excelling in terms of the details so one thing that you can see right here is that you can see that there are fine details such as the small lines in the roofs of this building and on other 3D models that this neural Angelo AI is able to generate there are other small details that there's other softwares aren't able to Simply pick up and we're about to see this in many of the other examples so here we take a look at neural Angelo Versus the prior work so you can see that there's noose right there with normal detail then a neural warp and then we can see neural Angelo which includes all of the 5 find tune detail and that's why I said including these maps and the surface normals are actually really really important when it comes to providing those small extra details that add to that level of immersion and this is something that you don't honestly realize when you are playing video games or maybe just interacting with a 3D environment but as someone who has worked in the 3D space before I do know exactly how important the majority of these details and how effective neural Angelo is so right here you can see that neurologangelo's technique is obviously far more effective and far more detailed than the previous techniques outlined so I wanted to quickly just take this time to show you some of the key details that you might have missed earlier on in the video because although it was explained it still is quite hard to see despite there being no side-by-side comparison so you can see right here that this is the old technique which is new s then eurolangelo then you continue a war now what I want you to do is pay attention to the windows because this is where the vast level of detail is separating these AI softwares from one another so we can see that new s on the windows you can see that there's not much detail but then on your Angelo you can see that we can actually see all of the bricks on this roof and then of course we can see the individual pieces of wood or whatever material it may be that separates these windows and then of course we can see neural warp as well and you can see that this material right here if you didn't know simply looks like a smooth material and of course that's not very good if we're trying to extract 3D data from a scene so right here you can see that neurolangelo is the only 3D model software that actually allows you to get the data which actually has those small other details now what's also good about this is it actually does get every single piece of the 3D model so not just those small surface normal Maps which add to the effective detail you can see right here that neural warp doesn't actually manage to gauge that this is an object which has those spirals on it whereas new s also doesn't have that and you can see there's some overlapping of these objects whereas near a wall whereas nearest Angelo which you can see right here from Nvidia actually does have all of that precise detail and when you really do this side by side analysis you can truly see the different levels of detail you can also see another example here with a brick as to how even on details that you might think aren't really needed there's still a very high level of detail on these bricks and that's why it's important to show you for example right down here you can see that there's a bit of overlapping once again and down here there's a bit of overlapping once again and of course you can see the neural engineer does it well it's important to see that even on these blocky assets where you might think that there's not really going to be much different on the small scale when you're looking at these close-up you're going to notice these crazy differences and then of course lastly here we can see that this fruit you can see that neural Angelo does perform this very very well now if it wasn't already obvious an educated guess as to why they call this Nvidia nerangelo was because there was a famous sculptor called Michelangelo and of course if you merge the names you do get Nvidia neurolangelo and of course nvidia's neurological essentially works by re-sculpting what they've seen in that 2D video so I guess essentially it's a callback to that specific name I do want to add a snippet from another video that we made essentially where we talked about how Nvidia actually did include a 3D generated model from a live 2D image so take a look at this clip because this was released in one of nvidia's earlier papers which not that many people saw so I just wanted to include this also because it is very very relevant to the tech that we just saw today this is very interesting this once again is showing where Nvidia leads the way this is live 3D portraits from real-time Radiance fields for single image portrait view synthesis or in simpler terms they present a One-Shot method to infer and render photorealistic 3D representations from a single image in real time so they can just pretty much just get an image of something and then essentially get a 3D realistic photo realistic representation of that image now right here you can see the inputs versus the outputs and for those of you who haven't seen the competitions level of detail you want to understand why this is truly groundbreaking because it's very hard to infer 3D data from an image because there are many different complex things at stake when you're trying to depict what goes where and it's something that you can easily get wrong but this is something that Nvidia has mastered and this research paper slash video that they talk about it goes over in detail how they've managed to do this and all the different techniques that they've used I mean it's honestly quite interesting as to how they managed to do this because this wasn't something a couple of years ago that you thought you'd be able to get with this level of accuracy and like I said before as AI continues to develop at this rapid Pace we're going to start seeing Innovations which we weren't thought possible develop at a ridiculous level now this was something that I found really really interesting and where you can see in real time you're able to get these different effects you can see that with the input compared to the output novel view you're able to get a completely different angle on what speaker is saying and you can see right here that we have a driving video then we have an input single image and then of course we have the 2D talking head and the 3D lifting with their method now this can be comparable to did in some aspect if you don't know what that is essentially it's where people convert an image to something moving but this is on a completely another level you can see right here that you can see in videos at the top left compared to the others in the other areas and we can see clearly that nvidia's is far more accurate with far more detail and this is clearly Superior to anything else that is currently being developed so I would say hat stuff to Nvidia for doing this because constantly with every single blog with every single research paper that I read compared to the other research papers that I read literally just a couple weeks before we're seeing Nvidia push the boundaries on what's truly capable now what's also cool is that essentially as we talked about previously in real time you can see that this is a live demo of someone essentially using a phone to RGB video to 3D now of course this is running on an RTR 490 which is essentially one of their top tier graphics cards but this also goes to show the potential applications of this imagine maybe you wanted to see different angles of something or maybe you wanted to see someone in essentially a more 3D way this is going to present more lifelike and more realistic video calls so I'm not exactly sure what these applications are going to be of course right now it's just pure speculation but just goes to show that being able to generate this amount of depth and detail from a simple 2D image is truly incredible

Channel: TheAIGRID

Views: 33,477

Keywords:

Id: lC2uPDfaTcE

Length: 11min 28sec (688 seconds)

Published: Tue Jun 06 2023