Introduction

Call of Duty: Modern Warfare 2 was released on October 28th and on launch day the PC game had support for NVIDIA’s Deep Learning Super Sampling (DLSS), AMD’s FidelityFX Super Resolution 1.0 (FSR 1.0) and Intel’s Xe Super Sampling (XeSS). Nowadays it’s very weird to see an outdated FSR 1.0 option, instead of FSR 2.x as an upscaler, especially when the game already has DLSS and XeSS support, so we didn’t test these upscaling options on launch day, instead waiting for the upcoming Season 1 update and Warzone 2.0, which, as we were hoping, should have brought FSR 2.x support, but with the Season 1 update and Warzone 2.0 the developers have added support for NVIDIA’s Deep Learning Anti-Aliasing (DLAA) and nothing else. So in this mini-review, we take a look at temporal upscalers, comparing the image quality and performance gains offered by DLAA, DLSS and XeSS in Call of Duty: Modern Warfare 2 | Warzone 2.0.

Below, you will find comparison screenshots at 4K, 1440p, 1080p, and in different DLSS and XeSS quality modes; the DLAA and Filmic SMAA T2X screenshots are also available in the dropdown menu. For those who want to see how DLSS and XeSS perform in motion, watch our side-by-side comparison video. The video can help uncover issues like shimmering or temporal instability, which are not visible in the screenshots.

All tests were made using a GeForce RTX 3080 GPU at Ultra graphics settings; motion blur and depth of field were disabled for better image viewing. DLSS in this game shipped with version 2.4.12.

Screenshots

Side-by-Side Comparison Video

Conclusion

In Call of Duty: Modern Warfare 2 | Warzone 2.0, the in-game Filmic SMAA T2X solution, DLAA and XeSS implementations do not use any sharpening filters in the render path, and do not support the ability to add the sharpening filter through separate sliders. However, the DLSS implementation does use some level of sharpening filter in its render path, and it also has the ability to tweak the sharpening filter value through a separate slider. The DLSS sharpening filter slider has the option to set the sharpening value from 50 to 100, in which the 50 value completely disables the sharpening, we used a value of 50 in our testing. The inclusion of a separate sharpening filter for DLSS is a great option to have, but there are a few important issues of note. At lower internal resolutions, such as 1080p DLSS Quality mode for example, the value of 51 and higher for sharpening filters can cause negative side effects such as excessive shimmering in motion, so we recommend to set the sharpening filter value to 50 for 1080p resolution, when using upscaling. Also, the DLSS implementation in this game is heavily conflicting with the depth of field and motion blur effects, resulting in heavy artifacts around player characters, so we recommend to disable these effects when using DLSS.

XeSS comes with three upscaling kernels that are optimized for various architectures. The first is the kernel that gets used on Intel Arc GPUs with XMX engines. This is the most advanced model too, that not only performs better in terms of FPS, but also offers the best upscaling quality, Intel calls this “Advanced XeSS upscaling model.” Intel also provides an optimized kernel for Intel Integrated Graphics, and another compatibility kernel, used for all other architectures that support Shader Model 6.4, e.g. all recent AMD and NVIDIA cards. These use the “Standard XeSS upscaling model,” which is somewhat simpler, with lower performance and quality compared to what you get on Arc GPUs (we use the compatibility model on our RTX 3080). If DP4a instructions aren’t available, as on the Radeon RX 5700 XT, slower INT24 instructions are used instead.

Speaking of overall image quality, there are also a few important issues of note. Call of Duty: Modern Warfare 2 | Warzone 2.0 is a fast paced first person shooter, so when using any temporal upscaling solutions, the temporal stability of the image is key to enjoyable gameplay. When using DLSS, the image was stable in motion in Quality modes, the level of detail rendered in vegetation and tree leaves is improved in comparison to the in-game Filmic SMAA T2X solution, and small details in the distance, such as wires or thin steel objects, are rendered more correctly and completely, and the DLSS implementation greatly reduces the shimmering issues, which are clearly visible on the in-game Filmic SMAA T2X solution. The XeSS DP4a implementation comes with noticeable compromises in image quality—in favor of performance in most sequences of the game. We spotted excessive shimmering and flickering on thin objects and especially steel objects; they are shimmering even when standing still, and when going down from XeSS Quality mode to Balanced or Performance mode, the whole image will start to shimmer even more. Only the DLAA solution will completely eliminate any of the shimmering issues in this game.

Speaking of performance, when using XeSS, there are some major differences in performance gains, compared to DLSS. As we are testing XeSS with an RTX 3080 GPU, which does not have the XMX instruction set, designed to accelerate XeSS workloads on Intel’s Arc GPUs, the performance gains are less than what we can expect on Arc GPUs, so keep that in mind. That said, the actual performance increase difference between XeSS and DLSS is about 13% at 4K and 1440p Quality modes, in favor of DLSS. However, compared to native 4K resolution, XeSS manages to deliver up to 20% more performance in Quality mode while using the DP4a instruction set that’s compatible with all GPU architectures, which is still a quite decent performance uplift. Interestingly, when using the DLAA solution, the performance at 1080p and 1440p were the same, compared to the Filmic SMAA T2X solution, but at 4K the DLAA solution has a performance cost of around 5%.