Unreal Engine 5.5 - It Gets More Incredible!

Unreal engine 5.5 is officially finally here. It is an all in one solution for creating animations, virtual characters, video games, movies, you name it. We looked at a few features that it offers, but it turns out there is more to it than we talked about. And before we start, don't forget, all this is available for free for most people and applications. And now it offers better tools for animation, better ways to filter information on the user interface. But look. Oh my. Haha. While you're moving the leg, what do I see there? Beautiful reflections computed in the viewport and in real time. You can also create these incredible virtual metahumans and animate them convincingly. And if even that is not good enough, they are also getting an absolutely amazing new feature. Listen. Metahuman Animator enables everyone to create stunning facial animations from an actor's performance. Oh yes, just chuck in the audio and you get a virtual version of yourself speaking animated properly. This is a research area that is about finding important key points on the face and moving these little dots around to make sure that it really resembles your emotions and what you just said. Other languages also work so so you can easily localize what you just said in different languages, one click for each and just in general sounds that you make work really well. A stunning feature. Goodness. So cool. The character customization system is also fantastic. Lots of things can be adjusted in real time. I love how you can customize this and make it your own with just a few clicks. But there is more. Dear fellow scholars, this is Two Minute Papers with Dr. Zsolnai. You can also set up choosers which help you pair an animation to something happening in the game. For instance, movements to perform when taking damage. To the best of my knowledge, for the longest time you needed a separate plugin for this and now part of the system. And if you want the most photorealistic results that are achievable and are willing to wait in return, it can perform a full light transport simulation for you. That is ray tracing if you will. However, if you don't feel that patient, their real time stuff is also nearly as good. Which is something that as a PhD student I thought would not be possible in my lifetime. And yet, here we are. What a time to be alive. And the craziest thing is that the two systems are getting closer and closer to each other. We will talk about this in a video very soon. Subscribe if you don't want to miss it. Render layers are also better in which you can separate different parts of your output, like the sky landscapes, virtual effects Smoke and cloud simulations and so on. And ooo about this. Something really caught my eye. I mean, what in the name of papers is going on here? Spatiotemporal noise filtering in production systems. I am out of words. Ok Caroj, what does this mean? Now hold on to your papers for this fellow scholars, Noise filtering is something you need for a slower ray tracing based system. As the initial images are full of noise and over time it cleans up. We can give it a little nudge by using noise filtering algorithms. However, most of these are designed to just denoise one image. But spatiotemporal noise filtering is a game changer. Imagine that we took a long time to finally compute a clean image. Yes, no more noise. And then what happens in the animation? We get the next frame and we start from scratch. That is criminal. Imagine a scene with a slowly panning camera where lots of information we could reuse from previous frames and. But we don't do that. Why? Because when I wrote my PhD thesis, these algorithms took literally days to compute. And today you can get early access to a production version. I am completely stunned by this. Wow. In the meantime, Megalights gives you hundreds and hundreds of these beautiful lights in a black market scene. Each of which are casting their own shadows computed in real time. And we are talking proper soft shadows, still experimental. The virtual reality side of things is also improving at an incredible pace. Virtual stages are possible. You can also walk around in virtual scenes and get better productivity tools like color correction too. Improvements have also been added for mobile game development. You can emulate different devices to see and test what your game will look like on different phones. That is huge. I mean, look at the difference between an older, less capable device and the shiny new stuff. Huge difference that you can test for in real time. And don't even get me started about the simulation engine in Unreal. It is unreal. Just look at that. Simulating every single part of a car. Liquid simulations, muscle simulations, you name it. Most if not all of them at interactive speeds. The things that we talked about 400 papers ago are finally coming. The papers we all love so much are finally coming to life. I am loving this. So don't be shy. Go ahead and give it a try through the link in the video description. And note that we are not affiliated with Unreal Engine in any way. But what is good is good. So what do you think? What would you fellow scholars use this for? Let me know in the comments below. And for now fellow scholars, let the experiments begin. And if you feel inspired and you wish to run your own experiments on Nvidia GPUs. Check out Lamda Lambda is a cloud service for large scale AI projects and it is my absolute favorite. Seriously. Try it out now@lambdalabs.com papers or click the link in the description. If you use Lambda you can run your own experiments on super powerful hardware and you also support two minute papers.