DLSS 3 is a new technology that is being developed by the Japanese company J-Wave. It is a type of digital signal processing that can be used to improve the quality of television shows and movies. DLSS 3 is said to be more accurate and efficient than current technologies, and it is hoped that it will help to improve the overall quality of television programming. DLSS 3 has been designed to work with existing TV sets, and it is said that it will be able to provide a better experience for viewers. It is hoped that DLSS 3 will help to make television programming more engaging and interesting, and it may also help to improve the overall quality of television broadcasts.
You may have heard that NVIDIA’s new DLSS 3 technology for its RTX 40-series GPUs is no different from the dreaded “soap opera” effect on TVs, but the truth is it’s a qualitatively different version of the same idea.
How NVIDIA DLSS 3 Works
DLSS 3 (Deep Learning Super Sampling) is a feature exclusive to NVIDIA’s 40-series graphics cards and (presumably) later models. It’s a technology designed to boost game frame rates while still producing a high-resolution image.
DLSS 3 consists of three components: DLSS 2, NVIDIA Reflex, and frame interpolation.
DLSS 2 uses a machine learning algorithm and special acceleration hardware on the GPU to take a low-resolution frame rendered normally by the GPU and “upscale” it to the target resolution, such as 1440p or 4K UHD. This boosts frame rates because the GPU does not have to render such a detailed image but still provides a final image that looks like it was rendered at a high resolution.
NVIDIA Reflex is a technology that optimizes the render pipeline of a game to reduce the latency between providing input to the game (e.g., pulling the trigger in a shooter) and seeing that input reflect on the screen.
Frame interpolation takes two frames before they are sent to the display. It then generates an entirely new frame that fits between them using machine learning and special hardware that accelerates how quickly that in-between frame can be generated.
Taken together, these three technologies produce the results we see in games that support DLSS 3, but why are people comparing it to the hated “soap opera effect” on TVs?
Frame Interpolation and the “Soap Opera Effect”
Frame interpolation is an option on just about any modern flat-panel display. Flat panel “sample and hold” displays, including all the LCD and OLED panels, suffer from a perceived “smearing.” There are different ways to combat this, such as Black Frame Insertion, but a more common solution is frame interpolation. It goes by various names depending on the brand of your TV, but it’s usually called something like Smooth Motion or Motion Plus.
It’s referred to as the “soap opera” effect because soap operas have a certain smooth look thanks to their use of high-frame-rate video cameras rather than film cameras with the more “cinematic” 24 frames per second rate.
Motion smoothing can ruin content not created to be viewed at frame rates other than it was filmed at, but motion smoothing has its place. Mainly it improves content such as sporting events by clearing up the smearing these high-motion activities produce.
Because many TV manufacturers used to have this feature on by default, it quickly developed a poor reputation, but motion smoothing is a great technology when you use it correctly. Unfortunately, since DLSS 3 also uses a form of motion smoothing, some of that negative sentiment toward the soap opera effect seems to have rubbed off on it.
How Is DLSS 3 Different?
There are a few key differences between TV motion smoothing and DLSS 3. If you turn on motion smoothing to try and make a game look smoother, you’ll end up with lots of input lag. That is, the time from when you press a button on the controller to when you see your action reflect on-screen, as we mentioned above.
This happens because it takes the TV a long time to create those interpolated frames. This does not matter when you’re passively watching a movie or show, but if you’re trying to interact with something in real time, it can render a game unplayable.
This is why most modern TVs have some sort of “game mode” where image post-processing effects such as smoothing are stripped away. This means the image quality can take a hit, but your games will feel snappy and responsive.
DLSS 3 specifically tries to address this weakness of motion smoothing by cutting down on input latency. Since NVIDIA Reflex is a mandatory part of DLSS 3, this already compensates for the additional latency of the interpolation process. On top of this, DLSS 3 GPUs have special hardware acceleration that speeds up how fast the interpolated frame can be generated while maintaining an acceptable level of quality.
Based on the tests and benchmarks we’ve seen the end result is that, unlike TV motion smoothing, turning DLSS 3 on reduces input lag compared to rendering the game natively, while generally only having slightly more lag than DLSS 2 with Reflex activated.
This probably makes it unsuitable for competitive gamers who play titles such as Counter-Strike, but for everyone else these small difference in latency probably won’t matter in the face of smoother motion.
