Anti-aliasing is an integral technology to the newest generation of games with the advent of higher detailed textures. While anti-aliasing technology makes games look good at any resolution, experimentation with its implementation has made the top triple-A titles maintain performance while still looking stunning.
Anti-aliasing smooths the “jaggies” or pixels that form the asset on the screen. Depending on your GPU, multiple versions of anti-aliasing work at different capacities to best impact performance vs. clarity. The type of anti-aliasing plus the sample number per pixel determines the smoothness of the textures for a specific resolution.
This article covers what anti-aliasing is in gaming, the most popular types, and progress in GPUs to elevate back-end processing.
What is Anti-Aliasing in Games?
Anti-aliasing in video games is the process of smoothing pixels on an on-screen asset. Without it, in-game objects have jagged-looking edges.
Anti-aliasing cures the pixel-oriented divide when assets are loaded and projected into 3D or 2D space. These lines can be quite attention-grabbing when zoomed in on a character or background asset, with these steps of assembly moving as the asset is rendered. Here's an example between No antialiasing and sub morphological antialiasing(SMAA).
No Anti-Aliasing SMAA x1
This is an uphill challenge for developers who need to accommodate 5+ years of hardware. Overall the range of these options spread from performant to visually intensive in the realism scale.
What Anti-Aliasing Does
While the result of anti-aliasing makes a game’s visuals smoother, there are still some performance issues to consider. High sampling means different images displayed in drastic quality depending on the resolution. Higher-end GPUs can accomplish this task with ease on moderate to high-resolution monitors, while mid to low-end GPUs often struggle with this task.
The 3 Types of Anti-Aliasing/Assisted Image Enhancement
There are 3 common types of anti-aliasing when it comes to gaming; Spatial Anti-Aliasing, Post-Processing Anti-Aliasing, and GPU Bound AI Image Sharpening. Of course, each of these have their own sub-types, so let's break them down.
Spatial anti-aliasing relies on image data to process smoother display images during the render process.
Super Sample Anti-Aliasing (SSAA)
This is another form of spatial anti-aliasing that can work on higher-end GPUs due to memory requirements. While intensive, using an in-depth pixel scanner creates a much smoother picture. Some argue it is too soft for realism, but just enough to eliminate most, if not all, jaggies.
No Anti-Aliasing (SSAA 8x)
Multisampling Anti-Aliasing (MSAA)
Multi Sampling Anti-Aliasing is a step up from no anti-aliasing because of its basics organization and color blending compared to other intense options. Multisampling attempts to break down the jaggies in aliased frames and smooth them over.
No Anti-Aliasing (MSAA 2x)
While not predominant with jaggies, Multisampling uses the environment around each pixel to better blend pixels for viewing assets. Looking at this lower sample, the light post on the right looks drastically better.
Fast Approximation Anti-Aliasing (FXAA)
Unlike Super Sampling or Multisampling, fast approximation anti-aliasing uses algorithms to organize pixels to display without being sampled for color correction or proximity. While messy, FXAA is still performant on lower-end systems or consoles.
No Anti-aliasing FXAA
Post Processing Anti-Aliasing
Post Processing Anti-Aliasing takes advantage of the render pipeline after the image has been rendered and applies a default setting to how the image is displayed. This incurs some performance hits, but the trade-off for small to medium monitors is viable performance in gaming.
Temporal Anti-Aliasing (TAA)
When generating a smooth image in-game, temporal anti-aliasing takes advantage of sampling the entire rendered image and uses the data to transition to the next smoothly. This anti-aliasing is excellent for still picture examples when playing photoesque games; however, the results can vary depending on the contrast of detail and how fast the gameplay is for processing the following image.
Temporal Anti-Aliasing No Anti-Aliasing
Sub-Pixel Morphological Anti-Aliasing (SMAA)
Sub-Pixel Morphological Anti-Aliasing is one of the most performant smoothing algorithms by detecting edges and smoothing them post-render, resulting in its popularity in today's games.
This is convenient for non-exposure-based anti-aliasing methods like TAA or general organization methods like FXAA, making this the ideal anti-aliasing method for post-rendering. To see the difference, look at the details, like the antenna.
No Anti-Aliasing (SMAA x8)
GPU Bound AI Image Sharpening
Nvidia GPU lines have taken to utilizing their AI assisted cores (preferably known as Tensor Cores) in enhancing the images even further.
Deep Learning Anti Aliasing (DLAA)
This anti-aliasing method is homebrewed by Nvidia for their recent RTX series of cards. This mid-render process focuses on establishing the quality of each rendered image. This process is more comparable to TAA in the sampling method but is more performant and clearer depending on the Nvidia card used.
Deep Learning Super Sampling (DLSS)
While not directly related to anti-aliasing, DLSS is a revolutionary technology that appeases most monitors in terms of fitting the most quality into a pixel space. The GPU renders a higher-resolution picture, then downscales it to the monitor's resolution while attempting to retain the quality.
In addition to the upscaled image, a parry image is also generated and cross-checked with the downscaled image providing a check rate for every frame rendered.
Contrary to what it may seem, these checks do not add any significant frame lag between the GPU and the monitor and instead increase the overall FPS experienced per game. If your GPU has spare resources, going with DLSS is your best option.
Determining Which One to Use
While the available options may seem complicated, each choice is available for a specific system after distinguishing between the differences. Here is an assorted recommendation based on graphics card abilities commonly seen in the past decade.
Anti Aliasing/GPU Setting Recommendation
No AA, FXAA
DLSS 2.0/3.0, DLAA, SMAA(xx)
Does Anti-Aliasing Affect Performance & FPS.
Anti-aliasing by itself does take some resources from the GPU. However, it depends entirely on the post-image process or during the rendering of images. With as little as 2GB of VRAM, anti-aliasing options are open to those looking to sharpen their graphics.
The demands will scale as the resolution increases, and any imbalance in GPU resources will become apparent when setting higher textures or other settings. The best way to get the best results is to play around and see which setting is best for you, with the most notable frame differences being 5-10% across all anti-aliasing settings, except for DLSS and Nvidia-assisted options.
Summary of Anti-Aliasing in Video Games
Whether you have an entry-level computer looking to bring some crispness to your gaming experience or hardware enthusiasts looking to optimize your gaming experience, Apex Gaming PCs have you covered with our predominantly Nvidia line. We support all anti-aliasing solutions recommended in our custom gaming PCs with the GPUs we offer. No matter your image-enhancing needs, we hope to fulfill them at Apex!
Written By William Wilson
Special thanks for the information breakdown to Jorge Jimenez, Jose I. Echevarria, Tiago Sousa, Diego Gutierrez, and Alex Tardif