DLSS Super Resolution
Renders fewer pixels internally, then reconstructs a higher-resolution output using temporal data, motion vectors and an AI model.
NVIDIA rendering technology explained
DLSS explained without the marketing fog
DLSS is not one single setting. It is a family of AI-assisted rendering technologies that can reconstruct a higher-resolution image, generate additional displayed frames, improve ray-traced image reconstruction and help compatible GeForce GPUs trade rendering cost for smoother gameplay.
Super ResolutionFrame GenerationRay ReconstructionMulti Frame GenerationLatency
DLSS in one sentence
DLSS uses AI-assisted rendering to reduce how much work must be conventionally rendered while attempting to preserve or improve the final image, and supported DLSS features can also create additional frames or reconstruct ray-traced effects.
The important distinction
DLSS is a family of features
When a game says it supports DLSS, that does not automatically mean every DLSS feature is available. Support depends on the game, renderer, GPU generation and the feature itself.
Frame Generation
Creates additional frames between conventionally rendered frames to increase displayed smoothness in supported games and hardware.
Ray Reconstruction
Uses an AI model to replace or improve parts of the traditional denoising pipeline for supported ray-traced effects.
Multi Frame Generation
On supported hardware and games, more than one generated frame can be inserted around conventionally rendered frames to raise displayed frame rate further.
What happens every frame
How DLSS Super Resolution works
Calling it “AI upscaling” is convenient, but incomplete. DLSS Super Resolution uses current-frame data plus information accumulated across previous frames to reconstruct the final image.
Step 1Render fewer pixels
The game creates a lower-resolution internal image, cutting conventional shading work.
Step 2Collect engine data
Motion vectors, depth and jittered samples describe where objects and pixels are moving.
Step 3Use previous frames
Temporal information can reveal detail that is not fully visible in one low-resolution frame.
Step 4Neural reconstruction
The model combines inputs to estimate a higher-resolution result while preserving edges and detail.
Step 5Display the output
The reconstructed image is presented at the target resolution, such as 1440p or 4K.
Interactive recommendation
Which DLSS setting should you try first?
Use this as a practical starting point, then check the exact game on your own monitor.
Suggested starting point
Start with DLSS Quality
At 1440p, Quality mode is usually the sensible first test. Compare fine detail, motion stability and your new frame rate before moving to Balanced.
Super Resolution modes
Quality, Balanced and Performance are different trade-offs
Exact scaling behaviour can differ by title and implementation, so treat the modes as quality-versus-performance presets rather than universal fixed ratios.
DLAAImage quality first
Uses DLSS-style reconstruction at native output resolution rather than primarily reducing internal resolution for extra performance.
QualityBest first choice
A strong starting point at 1440p and 4K when you want extra performance while keeping image quality as the priority.
BalancedMiddle ground
Reduces internal rendering further. Useful when Quality is still too slow and the visual difference remains acceptable.
PerformanceFPS first
More aggressive reconstruction. Usually a more natural fit for high output resolutions or very GPU-heavy workloads than 1080p.
Conceptual quality-versus-workload trade-off
These bars are illustrative, not benchmark data.
Illustrative only. This does not represent universal FPS or fixed internal-resolution ratios.
Deep dive
What DLSS actually changes in a game
DLSS Super Resolution: the feature most people mean by “DLSS”
DLSS Super Resolution is designed to reduce the cost of rendering at the final display resolution. Instead of conventionally shading every output pixel, the game renders at a lower internal resolution and supplies additional information that helps DLSS reconstruct the final frame.
The important word is reconstruct. A simple spatial upscaler only looks at the current image. DLSS can use temporal information from previous frames, motion vectors that describe how objects move, depth information and jittered samples. That gives it more evidence than a single low-resolution image.
In a good implementation, the output can look much closer to the target resolution than the internal render resolution suggests. Difficult scenes can still expose problems around particles, thin geometry, transparent effects, disocclusion or unstable engine inputs.
Judge DLSS in motionDo not rely on a paused screenshot alone. Look for motion stability, ghosting, shimmering and fine detail while moving the camera.
What DLSS Frame Generation does
Frame Generation is separate from Super Resolution. Instead of reconstructing a higher-resolution version of a conventionally rendered frame, it creates an additional displayed frame between conventionally rendered frames.
This can make motion look smoother and raise the FPS counter substantially. But a generated frame is not another complete game simulation, input and render update. That distinction matters for responsiveness.
A game conventionally rendering at 50 FPS with generated frames can look much smoother than 50 FPS, but the input feel should not be assumed to equal a game genuinely rendering every frame at the higher displayed rate.
Frame Generation and latency
Frame Generation adds work and relies on surrounding conventionally rendered frames, while NVIDIA Reflex is commonly used alongside supported implementations to reduce parts of the render queue and latency pipeline.
The practical lesson is that Frame Generation works best as a smoothness multiplier, not as a rescue tool for an extremely low base frame rate. If performance is poor before Frame Generation, improve the base frame rate first with settings, Super Resolution or faster hardware.
Competitive gaming noteFor esports and latency-sensitive games, prioritise a high conventionally rendered frame rate and low latency first. Generated frames improve visual smoothness, but they do not replace real simulation throughput.
What is Multi Frame Generation?
Multi Frame Generation extends the same idea by creating more than one additional displayed frame around conventionally rendered frames in supported games and hardware. Its main appeal is very high displayed frame rates in extremely expensive workloads such as heavy ray tracing.
The latency principle remains the same. A higher displayed FPS number does not mean the game simulation is operating at that same rate. Base performance still matters.
What is Ray Reconstruction?
Real-time ray tracing often works with a limited number of ray samples per pixel. Those samples are noisy, so traditional pipelines use denoisers to turn incomplete ray-traced information into a stable image.
DLSS Ray Reconstruction uses an AI model to reconstruct supported ray-traced effects instead of relying only on a chain of hand-tuned denoisers. In compatible games, this can improve reflection detail, lighting stability and temporal consistency.
Ray Reconstruction does not make ray tracing free. The rays still need to be traced. It changes how the noisy result is reconstructed.
Is DLSS better than native resolution?
There is no universal answer. Native rendering avoids reconstructing from a lower internal resolution, which can preserve some details and avoid certain reconstruction artifacts. DLSS can produce excellent anti-aliasing and temporal stability in other scenes, while the performance saved can fund higher settings or ray tracing.
Modern “native” rendering also often uses temporal anti-aliasing, dynamic resolution, sharpening or other reconstruction. The useful comparison is the complete rendering path in the actual game.
| Rendering path | Main advantage | Main trade-off | Best use |
|---|---|---|---|
| Native + game AA | No lower-resolution SR step | Highest conventional render cost | When performance is already comfortable |
| DLAA | DLSS-style reconstruction at native output resolution | Little or no SR performance saving | Image-quality-first gaming |
| DLSS Quality | Strong balance of image quality and performance | Can introduce reconstruction artifacts | 1440p and 4K starting point |
| DLSS Balanced | More GPU performance headroom | Lower internal render detail | Heavier games or ray tracing |
| DLSS Performance | Largest common SR performance push | More demanding reconstruction | High output resolutions or very GPU-heavy workloads |
Should you use DLSS at 1080p?
You can, but 1080p gives the reconstruction system fewer source pixels than 1440p or 4K at the same named mode. Aggressive modes can therefore be easier to notice.
Start with Quality if you need more GPU performance. If the game is already running well, native rendering or DLAA may be more attractive.
Should you use DLSS at 1440p?
1440p is a strong use case for DLSS Quality. The output resolution provides useful reconstruction detail, while modern games can still be demanding enough for the performance saving to matter.
See our best GPU for 1440p gaming Australia guide if you are deciding whether to solve the problem with settings or hardware.
Should you use DLSS at 4K?
4K is one of the strongest use cases for DLSS because native 4K asks the GPU to shade a very large number of pixels. Quality can reduce that workload while still targeting a 4K output, while Balanced or Performance can provide more headroom in very demanding games.
For a broader hardware view, read our best GPU for 4K gaming Australia guide.
DLSS artifacts to watch for
- Ghosting: faint trails around moving objects.
- Shimmering: unstable fine detail on foliage, fences or distant geometry.
- Disocclusion artifacts: temporary errors where hidden areas become visible.
- Particle instability: smoke, sparks, hair or transparent effects behaving inconsistently.
- Over-sharpening: halos or an unnaturally crisp look.
These issues vary by game. Judge the exact implementation rather than the DLSS label alone.
Does DLSS help when you are CPU-limited?
Super Resolution mainly reduces GPU rendering work. If the CPU, game simulation or another non-GPU bottleneck is already limiting the frame rate, lowering the internal render resolution may provide little improvement.
If lowering resolution barely changes FPS, use our GPU bottleneck guide.
What do you need to use DLSS?
You need a supported GeForce RTX GPU, a game or application implementing the relevant DLSS feature, and compatible software. Exact feature support depends on the GPU generation and game.
Do not assume every RTX card supports every DLSS capability. Super Resolution, Frame Generation, Multi Frame Generation and other features can have different requirements.
A sensible DLSS setup order
- Measure the game without Frame Generation first. Know your real base frame rate.
- Check whether the GPU is actually the bottleneck.
- Start with DLSS Quality. Compare image quality and FPS.
- Move to Balanced only if needed.
- Use Performance when the workload justifies it.
- Then test Frame Generation. Decide whether the extra smoothness is worth the trade-off.
- Check motion, not just screenshots.
The best DLSS setting is not the one with the highest FPS counter.It is the setting that gives you enough performance while keeping image quality and responsiveness where you want them.
DLSS vs FSR
NVIDIA DLSS and AMD FSR both aim to improve performance and image reconstruction, but they use different technologies, models and hardware requirements. The better option can vary by game, mode, GPU and implementation.
For a full breakdown, read our DLSS 4 vs FSR 4 comparison.
Should DLSS affect which GPU you buy?
Yes, but it should be one factor rather than the entire decision. DLSS can materially improve supported games, especially when ray tracing or 4K makes conventional rendering expensive. Not every game supports DLSS, and not every DLSS feature is available on every RTX generation.
Compare raster performance, ray tracing, VRAM, power draw, software support and Australian pricing as well. Start with our best GPU Australia guide if you are choosing a new card.
Keep learning
Related MyGPU guides
DLSS questions
Frequently asked questions
What exactly is DLSS?
DLSS is NVIDIA's family of AI-assisted rendering technologies. Super Resolution reconstructs a higher-resolution output from a lower internal render resolution. Other DLSS features can generate additional displayed frames or reconstruct ray-traced information.
Does DLSS make games blurry?
It can in aggressive modes or poor implementations, but DLSS Quality can look very good, especially at 1440p and 4K. Judge motion stability, fine detail and your monitor rather than assuming every DLSS mode looks the same.
Is DLSS Quality better than native?
Sometimes it can look cleaner in certain details because of temporal reconstruction and anti-aliasing, while native rendering may preserve other details better. There is no universal winner across all games.
Does Frame Generation double real performance?
It can substantially increase displayed frame rate, but generated frames are not additional full game simulation updates. Do not treat displayed FPS as equivalent to the same conventionally rendered FPS for latency.
Should I use DLSS at 1080p?
Start with Quality if you need more performance. Aggressive modes have fewer source pixels to work with at 1080p, so visual compromises can be easier to notice.
Should I use DLSS at 4K?
4K is one of the strongest use cases for DLSS because native 4K is expensive to render. Quality is a sensible starting point, with Balanced or Performance useful when the workload is especially demanding.
Does DLSS work on AMD GPUs?
NVIDIA DLSS requires compatible GeForce RTX hardware. Other GPU users should check the reconstruction technologies supported by the game and their hardware.
Is DLSS useful without ray tracing?
Yes. DLSS Super Resolution can improve performance in rasterised games as long as the title supports it. Ray Reconstruction is specifically tied to supported ray-traced rendering.