Nvidia Deep Learning Super Sampling (DLSS) is the technology has been scaling up for over two years, but a new challenger is on the way. Ghostwire Tokyo showcases a relatively new technique in Unreal Engine 5 called Temporal Super Resolution (TSR), which looks and performs almost as good as DLSS, and it has one big advantage: it works with any graphics card.
DLSS has been in the spotlight as a proprietary supersampling technique that delivers much better image quality than competitors like AMD FidelityFX Super Resolution (FSR). However, companies such as AMD have not stood still and common scaling solutions such as FSR 2.0 and TSR will make DLSS obsolete.
TSR is a feature of Unreal Engine 5, but developer Tango Gameworks was able to get it to work in the UE4-based Ghostwire Tokyo† Unlike DLSS, no special AI accelerators are required to work. Instead, it feeds temporal (time-based) data into a supersampling algorithm to scale up the image.
While TSR is an emerging feature, it is already being used in other places. AMD’s upcoming FSR 2.0 revision is a good example, which uses transient inputs that feed a supersampling algorithm. Ghostwire Tokyo offers a glimpse into the future of PC graphics: a game where every game has high-quality upscaling that works on all GPUs.
The image below shows TSR, FSR 1.0 and DLSS side by side, in that order. DLSS and TSR look identical. Even zoomed in greatly, I can’t find any significant differences. Compare that to FSR 1.0, which has black spots in the blue Tottoko Cine board, as well as a dirty border around the green board below.
The same is true in a scene with sharp details. TSR and DLSS look the same, and FSR 1.0 has issues. Note the TV hanging to the left, which is much more blurry with FSR 1.0, as well as the dimmer lights going down the hallway. With FSR 1.0, these lights flickered as the scaling algorithm struggled to keep up. With TSR and DLSS they were stable.
The main draw of DLSS has been the excellent image quality, which Nvidia has attributed to the dedicated Tensor cores in RTX 30-series and 20-series graphics cards. Ghostwire Tokyo shows that special hardware doesn’t do much. TSR looks just as good, and if FSR 2.0 is indeed comparable, it should be.
However, we cannot ignore the achievements. At 4K with ray tracing enabled and all sliders maxed out (minus motion blur), I averaged 40 frames per second (fps). TSR was able to more than double my frame rate, boosting it to 84 fps.
It’s a huge improvement, though not as great as FSR 1.0 and DLSS. FSR 1.0 jolted at an average of 90 fps, while DLSS topped the list at an average of 100 fps. While 16% better performance for DLSS is significant, it doesn’t seem that important when TSR can already double your frame rate.
We may see a repeat of Nvidia G-Sync here. DLSS has been a walled garden since its launch and TSR shows that a restrictive approach may not have been necessary. While other companies push their collective knowledge to build better products for gamers, we get comparable image quality and performance without having to pay for a GPU with a particular brand on it.
FSR 2.0 and TSR are enough to destroy DLSS alone, and with Intel’s upcoming XeSS technology in the mix, the future doesn’t look bright for Nvidia’s scaling technology. Also think of game developers. If a solution like TSR can offer comparable performance and image quality to DLSS, and it works on all GPUs and consoles, it just makes more sense.
The future may not be bright for DLSS, but it is for PC gamers. if Ghostwire Tokyo is a sign of things to come, PC gamers are looking for more upscaling options that work with more hardware while still offering near native picture quality.