Recommended Free Tools
Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
NVIDIA announced DLSS 2.0 on March 23, 2020, as a new way to render games at lower internal resolutions and reconstruct a higher-resolution image using AI. A key part of that process was motion vectors: data supplied by the game engine to show how objects and the camera move between frames. Combined with information from earlier frames, those vectors helped DLSS place image detail more consistently over time.
Motion vectors were important, but they were not the whole upgrade. DLSS 2.0 also brought a generalized model intended for multiple games, Quality, Balanced, and Performance modes, and a more reusable developer integration path. NVIDIA’s launch performance and image-quality statements were vendor claims, not guarantees for every game or PC.
Why DLSS exists
Rendering a game at a higher resolution usually requires the GPU to draw more pixels. That can improve detail and reduce visible jagged edges, but it also increases the workload—especially when a game uses demanding effects such as ray tracing. Rendering at a lower resolution can improve performance, at the cost of a softer or less stable image.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Deep Learning Super Sampling, or DLSS, is NVIDIA’s approach to that trade-off. The game renders fewer pixels, and an AI-assisted reconstruction process produces an image at the selected output resolution. NVIDIA positioned DLSS as a way to create more performance headroom, including for higher resolutions and ray tracing. The actual benefit depends on the GPU, game, settings, and whether the system is limited by graphics rendering rather than the CPU or another component.
#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
What NVIDIA announced in DLSS 2.0
NVIDIA’s March 23, 2020 announcement described four main improvements:
- A generalized AI model: NVIDIA said DLSS 2.0 used a model trained on non-game-specific content and designed to work across multiple games, rather than requiring a separate trained model for each title. A shared model did not mean identical results in every game; integration and rendering data still mattered.
- More quality choices: The launch modes were Quality, Balanced, and Performance. They traded internal rendering resolution against reconstruction quality and performance.
- Improved image quality: NVIDIA said DLSS 2.0 could approach native-resolution image quality while rendering roughly one-quarter to one-half as many pixels in relevant modes. That was a launch claim, not a universal result across all scenes and games.
- More efficient execution: NVIDIA said the new network ran up to twice as fast as the original implementation. This referred to the network’s execution, not a promise that a game’s frame rate would double.
NVIDIA also described Performance mode as supporting up to 4× super resolution—for example, reconstructing a 4K output from a 1080p internal render. “4×” refers to the output-to-internal resolution relationship, not four times the frame rate or image quality. The resulting performance gain varies with the game and system bottleneck.
How DLSS 2.0 used motion vectors
A single low-resolution frame may not contain enough information to reproduce every fine edge and texture at a higher resolution. Previous frames can offer additional clues, but only if the system can determine where that earlier information belongs now. That is the temporal part of the problem.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Motion vectors are data describing how rendered elements move from one frame to the next. The game engine can produce them because it knows about camera movement and object transforms. They are not AI-generated predictions; they are inputs to the reconstruction process.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
In NVIDIA’s explanation, DLSS 2.0 used the current low-resolution frame and its motion vectors, along with temporal feedback from the previous high-resolution output. In simplified terms, the process works like this:
- The game renders the current scene at a lower internal resolution.
- The engine supplies motion vectors that describe movement between frames, along with the rendering information needed by the implementation.
- DLSS uses those vectors to align relevant information from temporal history with the current frame.
- Tensor Cores on supported RTX hardware run the trained neural network to reconstruct the higher-resolution output.
Motion data helps the system distinguish moving objects from stationary detail and reuse useful information without treating every frame as an unrelated still image. It can support more stable edges and detail during camera movement or animation. But it cannot supply detail that was never rendered or correctly represented in the inputs.
Why implementation quality matters
Temporal reconstruction depends on accurate information from the game. If vectors are missing or wrong for an object, the system may place historical detail incorrectly. Areas newly revealed when an object moves—called disoccluded regions—have no valid history and must be reconstructed from current information. Fine or subpixel features such as foliage, hair, wires, fences, and particles can also be difficult to keep stable.
Free tools Windows power users keep installed
One-click scans. No signup required.
These are general challenges for temporal image reconstruction, not proof that motion vectors alone cause every visible artifact. Depending on the scene and implementation, a player may notice ghosting behind moving objects, smearing of fine detail, unstable patterns, or problems around transparency and particles. UI and HUD elements also need suitable treatment in the rendering pipeline; they are not necessarily reconstructed like ordinary 3D geometry.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
That is why DLSS was not simply a driver feature that could be switched on in any game. Developers had to integrate it and provide appropriate rendering data. NVIDIA’s launch announcement said its network was trained on DGX supercomputers against offline-rendered, ultra-high-quality 16K reference images, then delivered to GeForce RTX systems through drivers and updates. Those are NVIDIA’s descriptions of its training and distribution process.
DLSS 1.x and DLSS 2.0
Early DLSS implementations varied, so “DLSS 1.x” should not be treated as a single identical version. Broadly, NVIDIA presented DLSS 2.0 as a move toward more reusable integration and temporal reconstruction.
| Area | Early DLSS implementations | DLSS 2.0 |
|---|---|---|
| AI model | More game-specific training and behavior | Generalized model intended to serve multiple games |
| Temporal data | Less flexible early approach | Explicit use of motion vectors and temporal feedback |
| Quality controls | More limited or implementation-dependent | Quality, Balanced, and Performance modes |
| Developer path | More game-specific work | NVIDIA promoted a more reusable SDK and broader integration |
| Hardware | Supported RTX hardware | Still relied on supported RTX hardware and Tensor Cores |
Launch games and developer availability
NVIDIA’s announcement said DLSS 2.0 was already available in Deliver Us The Moon and Wolfenstein: Youngblood. MechWarrior 5: Mercenaries was launching with it on March 23, 2020, while Control was scheduled to receive a patch on March 26. NVIDIA also said DLSS 2.0 was available to Unreal Engine 4 developers through its DLSS Developer Program. These dates describe the 2020 announcement; game support has since changed through patches and later integrations.
Hardware and practical limits
DLSS 2.0 was designed for supported GeForce RTX graphics hardware, which uses Tensor Cores for the AI work. It is not a feature that works on every GeForce card, or a technology that an AMD or Intel GPU can use simply by enabling a setting. A compatible graphics card is also not enough by itself: the game must include DLSS support.
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
DLSS is most useful when the GPU is the performance bottleneck and reducing the rendering workload helps. If the CPU, simulation, asset streaming, or frame pacing is limiting performance, lowering internal resolution may do little. Likewise, if a game already runs comfortably above a display’s refresh rate, a player may prefer native rendering or a less aggressive mode.
Quality mode generally uses a higher internal resolution than Performance mode. The right choice depends on output resolution, display size and viewing distance, the game’s implementation, and how sensitive the player is to reconstruction artifacts. More aggressive upscaling can provide more performance headroom but may make image instability easier to see.
DLSS 2.0 did not guarantee native-quality output in every game, eliminate all ghosting or aliasing, or make unsupported games compatible. Nor did it mean every part of the frame—including menus and HUD—was rendered and reconstructed in the same way; developers can handle those elements separately.
DLSS 2.0 versus later DLSS features
DLSS 2.0 was primarily a super-resolution reconstruction technique: it used a lower-resolution render and temporal information to produce a higher-resolution frame. It should not be confused with later frame-generation features, which create additional frames rather than only reconstructing the current rendered frame.
Best Value
- Powered by the NVIDIA Blackwell architecture and DLSS 4 OC mode: 2640MHz/Default mode: 2610MHz (Boost Clock)
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.125-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
NVIDIA’s current DLSS overview describes a broader family that includes Super Resolution, Frame Generation, Ray Reconstruction, and DLAA, alongside newer model generations. Those later features and models are not part of the March 2020 DLSS 2.0 announcement. Later updates also continued to refine motion handling; for example, NVIDIA’s 2021 discussion of DLSS 2.3 described further motion-vector-related improvements.
For developers, NVIDIA’s current DLSS resources cover later versions and integrations; they should not be read as a description of the exact 2020 SDK. Unity’s Unity 2021.2 documentation, for example, records later native DLSS support in that engine line, not a launch feature of DLSS 2.0.
The takeaway
DLSS 2.0 made temporal information central to NVIDIA’s AI upscaling approach. Motion vectors gave the reconstruction process a way to track scene movement and align useful information from earlier frames, while a generalized model and selectable quality modes made the feature more practical to integrate across games. The quality and performance outcome still depended on the game’s implementation, the chosen mode, and whether the GPU was the system’s limiting factor.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
