AMD's upcoming FSR package could solve the noise issue on path-traced frames with AMD's proprietary neural supersampling and denoising techniques.
AMD will reportedly follow NVIDIA (DLSS Ray Reconstruction) in its course for developing AI-based denoising techniques & neural supersampling
NVIDIA has come far with its upscaling techniques, which have solved the performance issues in demanding scenes for modern titles when ray tracing is turned on. As soon as NVIDIA started implementing various ray tracing and upscaling techniques, AMD and Intel also followed with their own FSR and XeSS upscaling methods.
While DLSS has matured quite decently in the past years, AMD and Intel are still catching up by releasing newer and better upscalers. AMD has released its FSR 3.1 this year, which is much more stable than the previous versions, but there are still issues when we talk about denoising the visuals in the path traced images.
NVIDIA is way ahead of its rivals as it introduced Ray Reconstruction to improve the visuals by reducing the noise, but AMD depends on the denoisers present in modern games. While the in-game denoisers do a satisfactory job, the AI-based neural network i.e., NVIDIA's Ray Reconstruction is much more powerful, which denoises the images more accurately.
To do this job, NVIDIA uses its own Tensor Cores but AMD relies on the WMMA(Wave Matrix Multiply Accumulate). In a recent blog post, AMD announced that
...actively researching neural techniques for Monte Carlo denoising with the goal of moving towards real-time path tracing on RDNATM GPUs.
This means that AMD is now on its journey to develop a technique that will use AI to deliver better visuals and denoising that will work on RDNA GPUs. It's not clear whether these techniques will work on the existing RDNA-based GPUs from previous and current generations or only on the upcoming RDNA 4 or future GPUs. AMD's FSR is known for its universal compatibility, which gives it an edge over DLSS, but the upcoming techniques may work only on RDNA GPUs.
AMD also added,
Our technique can replace multiple denoisers used for different lighting effects in rendering engine by denoising all noise in a single pass as well as at low resolution.
It's also possible that the FSR 4.0 will have some limited compatibility with NVIDIA and Intel GPUs but will work at full potential for RDNA GPUs outside the box. It's no surprise that AMD might consider using AI for developing its neural network because this year, AMD's CTO, Mark Papermaster, announced that they are enabling AI-based upscaling on their gaming devices.
News Sources: GPUOpen, PCGamer
About the author: Sarfraz Khan is a hardware reporter with a focus on PC components and the builder community. With years of experience writing about PC hardware and laptops, his work has been featured on several reputable technology publications. Sarfraz's hands-on experience is demonstrated through his first-person accounts of using and comparing different hardware configurations, providing practical and relatable insights for everyday users. His technical analysis is respected by peers in the enthusiast community and has been cited by specialized hardware sites such as Germany's Igor's Lab.
Follow Wccftech on Google to get more of our news coverage in your feeds.
