Beautiful Glitter Simulation…Faster Than Real Time! ✨

Dear Fellow Scholars, this is Two Minute Papers with Dr. Károly Zsolnai-Fehér. This incredible paper promises procedural and physically-based rendering of glittering surfaces. Whoa. Okay, I am sold… provided that it does as well as advertised. We shall see about that! Oh, goodness, the results look lovely. And now, while we marvel at these, let’s discuss what these terms really mean. One, physically-based means that it is built on a foundation that is based in physics. That is not surprising, as the name says so. The surprise usually comes when people use the term without careful consideration. You see, light transport researchers take this term very seriously, if you claim that your model is physically-based, you better bring your A-game. We will inspect that. Two, the procedural part means that we ourselves can algorithmically generate many of these materials models ourselves. For instance, this earlier paper was able procedurally generate the geometry of climbing plants and simulate their growth. Or, procedural generation can come to the rescue when we are creating a virtual environment and we need a hundreds of different flowers, thousands of blades of grass, and more. This is an actual example of procedural geometry that was created with the wonderful Terragen program. Three, rendering means a computer program that we run on our machine at home and it creates a beautiful series of images, like these ones. In the case of photorealistic rendering, we typically need to wait from minutes to hours for every single image. So, how fast is this technique? Well, hold on to your papers, because we don’t need to wait from minutes to hours for every image. Instead, we need only 2. 5 milliseconds per frame. Absolute witchcraft. The fact that we can do this in real time blows my mind, and, yes! This means that we can test the procedural part in an interactive demo too! Here we can play with a set of parameters, look, the roughness of the surface is not set in stone, we can specify it and see as the material changes in real time. We can also change the density of microfacets, these are tiny imperfections in the surface that really make these materials come alive. And if we make the microfacet density much larger, the material becomes more diffuse. And if we make them really small…ohoho! Loving this. So, as much as I love this, I would also love to know how accurate this is. For reference, here is a result from a previous technique that is really accurate, however, this is one of those methods that takes from minutes to hours for just one image. And here is the other end of the spectrum, this is a different technique that is lower in quality, however, in return, it can produce these in real time. So which one is the new one closer to? The accurate, slow one, or less accurate, quick one What?! Its quality is as good as the accurate one, and it also runs in real time. The best of both worlds. Wow! Now, of course, not even this technique is perfect, the fact that this particular example worked so well is great, but it doesn’t always come out so well. And, I know, you’re asking, can we try this new method. And the answer is a resounding yes, you can try it right now in multiple places! There is a web-demo, and it was even implemented in Shadertoy. So, now we know what it means to render procedural and physically-based rendering of glittering surfaces in real time. Absolutely incredible. What a time to be alive! And if you enjoyed this episode, we may have two more incredible glint rendering papers coming up in the near future. This, and this. Let me know in the comments below if you would like to see them. Write something like, “yes please! ”.

Thanks for watching and for your generous support, and I'll see you next time!