What is Nvidia’s Ray Tracing technology and what does it mean for the gaming community

Often when running after computer graphics technology, we always know our destination. It is about recreating the world around it perfectly – or at least choosing to create one. The theory behind creating such a thing has been around for decades, and it’s called raytracing (or pathtracing).

At a high level, imagine that computer graphics modeling tries to reconstruct the world around us. As photons travel everywhere, they are emitted from light sources such as the sun, light bulbs, or computer screens. And what we can see is a collection of photons as they come into contact with the receptor of the eye and are processed by the brain as an image.

The point is that all of this happens to a similar degree – photons are always emitted, not necessarily at a uniform rate, and at different wavelengths. By rough estimate, a billion photons will come into contact with our eyes every second, while a light bulb can emit 8×1018 photons per second. It’s not realistic at all to simulate how those photons move through a room, but what if we just simulated but the photons made contact with the eye?

And that is the essence of raytracing. Start with your computer monitor running at 1920×1080 resolution. Now reverse the projection of each of those pixels to a model of an environment, like an office, house, etc. Return a bunch of these rays around and you’ll almost get a good look. physicality of the environment.

If you’ve ever run benchmarks like Cinebench, you’ll probably have no idea how long it takes for these maths to be performed. In filmmaking, rendering an image is often complex – higher resolution, higher texture resolution, and a more detailed model of the reconstructed world. It usually takes hours to render frame by frame in a CGI movie.

Playing games at a frame rate over 10 hours sounds pretty awful, and of course we need a way to estimate things and get them to work in real time. This process can be collectively referred to as rasterization, and that’s how computer graphics in games work. And it works pretty well, and gets better and better every year. But rasterization has made use of somewhat complicated techniques, including screen-space reflection, global lighting, environment redrawing, and more.

The difficulty with rasterization techniques is that they are often time consuming. If the end goal is to make everything as close to the real world as possible, we’ll often have to go back to raytracing.

Today, at GDC, Microsoft is announcing the next generation of DirectX API called DirectX Raytracing (DXR). The new API will be used to bridge the gap between rasterization and raytracing, allowing for more realistic rendered images.

First, DXR doesn’t really require radical hardware changes. DXR workloads can be handled on top of existing DirectX engines, because that’s essentially compute workloads. The same capabilities that allow your GPU to be used for things like cryptocurrency mining can be used for gaming-related things. In the long run, graphics chips are becoming more common – more programmable. All fixed function units are being replaced by generic hardware that run shader programs.

What does this mean for gaming? At first, don’t expect titles to suddenly switch to full raytracing. However, DXR will enable new rendering techniques with improved image quality, and over time things like screen-spatial reflections and global lighting should migrate to the new API. Furthermore, raytracing could replace rasterization as the standard method for creating 3D worlds, although we may need new rendering technology before we stop using rasterization.

The good news is that DXR will be able to run on existing hardware – so you won’t have to ditch the RX Vega or GTX 1080 models just to experience DXR. However, future hardware will likely be engineered to handle the new functionality, and Nvidia has at least talked about how its Volta architecture could work better with the DXR. Nvidia went one step ahead with RTX, which builds on top of DXR but still includes additional hooks that require Volta hardware.

As for AMD, it announced that: “AMD is currently working with Microsoft to help define, refine, and support the future of DirectX12 and raytracing. AMD remains at the forefront of new programming paradigms and improved application programming interfaces (APIs). We look forward to discussing with game developers their ideas and feedback on raytracing.”

The great thing about DXR is that using raytracing for gaming is possible in the future. As well as Microsoft just announced it will support EA’s Frostbite engine, SEED’s Futuremark, Epic Games’ Unreal Engine, and Unity 3D’s Unity Engine.

If and when we get to the point where everything can be completely simulated with raytracing, instead of spending time and money creating shadowmaps and calculating different hacks to simulate the environment, raytracing can do most of the heavy lifting like handling light, shadows, reflections. So artists can focus on creating great models from their world and let the engine do the rendering.

In short, although we still can’t play games with graphics on par with Pixar cartoons at the moment. But only for now, with technology constantly evolving everything will be within reach.

Source: PCGamer