How does dlss work
Last updated: April 1, 2026
Key Facts
- DLSS renders games at lower resolution (like 1440p instead of 4K) then uses AI to intelligently upscale to target resolution
- NVIDIA's tensor cores on modern graphics cards process the AI upscaling extremely fast with minimal performance cost
- DLSS offers multiple quality modes: Performance, Balanced, and Quality, with different resolution ratios and quality tradeoffs
- DLSS 3 introduced frame generation, using AI to predict and generate entirely new frames between rendered ones
- Game developers integrate DLSS through NVIDIA's SDKs, requiring specific implementation for each game
What is DLSS?
DLSS (Deep Learning Super Sampling) is a revolutionary technology developed by NVIDIA that uses artificial intelligence to improve gaming performance while maintaining visual quality. Instead of rendering at full resolution, which requires significant GPU power, DLSS renders at a lower resolution and uses neural networks to intelligently upscale the image to your target display resolution.
How the Upscaling Process Works
The process involves several steps: first, the game renders at a lower resolution (for example, 1440p instead of 4K); second, NVIDIA's tensor cores on the GPU process AI algorithms trained on millions of images to predict what the higher-resolution image should look like; finally, the upscaled result is displayed at your target resolution. This hybrid approach reduces GPU workload while maintaining sharp, clear visuals through intelligent prediction rather than simple image enlargement.
DLSS Modes and Quality Tiers
DLSS offers three primary quality modes with different performance/quality tradeoffs. Performance mode renders at the lowest resolution for maximum FPS, Balanced mode offers a middle ground, and Quality mode renders at higher resolution for the best visual fidelity. Additionally, DLSS 3 introduced DLSS Super Resolution which can upscale even beyond native display resolution for enhanced sharpness.
Frame Generation Technology
DLSS 3 added frame generation, a groundbreaking feature that uses AI to generate entirely new frames between rendered ones. By analyzing motion vectors and temporal data, the AI predicts what the next frame should contain, effectively doubling frame rates. This technology has transformed high-resolution gaming performance, enabling 4K gaming at previously impossible frame rates.
Hardware Requirements and Adoption
DLSS requires NVIDIA GPUs with tensor cores, available in RTX series cards (RTX 20, 30, 40 series) and newer. Game developers must integrate DLSS through NVIDIA's SDKs, and adoption continues growing with hundreds of games supporting the technology. DLSS has become a standard feature for demanding games, particularly at high resolutions where performance gains are most significant.
Related Questions
What is the difference between DLSS and traditional upscaling?
Traditional upscaling uses simple mathematical interpolation to enlarge images, resulting in blurry or jagged results. DLSS uses AI neural networks trained on real images to intelligently reconstruct detail, producing sharper, more natural results.
Is DLSS only available on NVIDIA graphics cards?
DLSS is proprietary to NVIDIA hardware. AMD offers an alternative technology called FSR (FidelityFX Super Resolution) with similar goals, and Intel has XeSS, but they're separate implementations.
How much performance improvement does DLSS typically provide?
Performance gains vary by settings and game, but DLSS typically provides 1.5x to 4x performance improvement depending on the quality mode used, making previously unplayable settings feasible.
Sources
- Wikipedia - Deep Learning Super Sampling CC-BY-SA-4.0
- NVIDIA - DLSS Technology Copyright NVIDIA