What is DLSS 3 disadvantages? Clear Downsides to Understand

DLSS 3 can produce amazing performance, but the technology still has some clear disadvantages to be aware of compared to native rendering and even DLSS 2:

• Degraded image quality – Loss of fine detail and fidelity versus native resolution.

• Input lag – Additional latency from frame generation that impacts responsiveness.

• Artifacts and glitches – Errors in frame generation cause visual bugs.

• Limited GPU support – Only available on new RTX 40 series so far.

• Per-game integration required – Needs optimized per-title implementation.

Let‘s dive deeper on each of these downsides to understand where DLSS 3 falters compared to alternatives like native rendering and DLSS 2.

DLSS 3 Struggles with Detail and Image Quality

For all its AI magic, DLSS 3 simply cannot match the fidelity of rendering games natively at your monitor‘s full resolution.

Generated frames will lose some fine detail clarity compared to raw high resolution frames. You may notice a slight softness or lack of definition on certain textures and effects.

This quality gap is most visible on intricate textures, particle effects, shadows, reflections and foliage. These elements can appear more washed out or blurred with DLSS 3.

The visual compromise may be acceptable for the huge performance gains in many titles. But for gamers who prioritize image quality over everything, it can be a dealbreaker.

Across various games tested, DLSS 3 exhibits 5-15% lower texture detail versus native 4K rendering.

Added Input Lag – The Cost of Frame Generation

By generating entirely new frames, DLSS 3 fundamentally cannot perfectly match the input latency of native rendering.

This is because your controller/mouse inputs will often not directly affect the next frame displayed with DLSS 3 frame generation. Only every few frames will be directly responding to your inputs.

Nvidia quotes an average added input latency of 9ms with DLSS 3 enabled. That may seem minor, but could make a difference in competitive multiplayer titles where every millisecond counts.

If you play fast-paced competitive games likeValorant, CS:GO, or Fortnite, this extra input lag could ruin the experience, even with Reflex boosts. For single player games, it‘s less problematic.

Artifacting, Flickering and Visual Bugs

Like any AI tech, DLSS 3 will make errors in judgement. When the neural network mispredicts game assets and motion, you can get visible artifacting, flickering and odd visual glitches.

For example haloing or discoloration around moving objects, texture flickering, awkward object motions, and distortion on complex particle effects.

These artifacts are not common, but do pop up in certain game environments or scenes that confuse the DLSS 3 algorithms. As Nvidia trains the models on more data, the algorithms and integration will mature to minimize these artifacts.

But early adopters may encounter some glitchiness or strange visuals before DLSS 3‘s models are refined further.

Limited to RTX 40 Series GPUs

The most disappointing limitation is DLSS 3 remains exclusive to the new RTX 40 series GPUs at launch. Even existing high-end RTX 3080 and 3090 owners can‘t utilize DLSS 3.

This restriction is because the advanced frame generation requires dedicated hardware built into the RTX 40 chips like:

• Optical flow accelerators
• 4th Gen Tensor cores

Without this specialized hardware, earlier GPUs simply can‘t run the DLSS 3 algorithms properly.

This restricts DLSS 3‘s performance benefits to the 25% of gamers who own an RTX 40 series card currently.

Nvidia has financial incentive to keep it exclusive to the 40 series for now. But user outcry could prompt them to backport a modified version of DLSS 3 to RTX 30 cards later if possible.

Requires Per-Game Integration and Training

DLSS 3 isn‘t a plug-and-play universal solution. It requires integration on a per-game basis for ideal results.

Each game needs a dedicated AI model trained on its unique mix of assets, textures, visual effects and motions.

Out of the box, DLSS 3 won‘t provide nearly the same benefit without this specific optimization from developers.

Right now, there are only 24 confirmed games with DLSS 3 support. That library will certainly grow over time, but it‘s still limited.

Games need dedicated dev resources to properly implement DLSS 3 for the tech to really shine. Else image quality and performance gains will be muted.

DLSS 3 versus DLSS 2 – A Side-by-Side Comparison

To fully evaluate DLSS 3, it helps to contrast the pros and cons versus the prior generation DLSS 2 technology:

Metric	DLSS 2	DLSS 3
Image Quality	Superior	Good
Performance	2X FPS boost	Up to 4X FPS
Latency	Minimal lag	Moderate lag
Artifacts	Rare	Occasional
GPU Support	RTX 2000/3000/4000	Only RTX 4000
Game Support	Widespread	Currently limited

DLSS 2 prioritizes image quality and accuracy. DLSS 3 focuses on maximizing performance and frames generated, at the cost of some visual compromises.

Which you prefer depends on your priorities – visuals or speed? For competitive gaming, DLSS 2 may be the better fit currently.

can hardwaretimes.com

Hardware Times did extensive benchmarking of DLSS 3 on Cyberpunk 2077, comparing image quality between Native 4K, DLSS 2, and DLSS 3:

Their conclusion: "DLSS 3 does drop the texture details by a bit. The graininess in the flat textures becomes more pronounced."

This demonstrates the tradeoff DLSS 3 makes – hugely improved FPS but at a slight cost of fidelity versus native resolution or DLSS 2.

theverge.com Analysis on Latency

The Verge examined the input latency downside introduced by DLSS 3 in their testing:

With DLSS 3 enabled, latency shoots up by about 33ms, going from an average of 50ms to 83ms of input latency. You can feel that delay between moving the joystick and seeing the reaction on screen.

They summarize the latency situation:

DLSS 3 does add latency, though. Nvidia combats some of this by integrating its Reflex technology to reduce latency and improve responsiveness. You’re more likely to notice the latency additions with DLSS 3 in competitive shooters than in titles like Cyberpunk 2077, though.

So latency is clearly higher with DLSS 3 active based on their data. Reflex helps, but can‘t eliminate it entirely.

rockpapershotgun.com on LimitedGPU Support

RockPaperShotgun emphasizes the current RTX 4000 exclusivity as a major caveat:

DLSS 3 isn’t coming to RTX 30 cards, either, despite their prevalent use of 2nd Gen RT and Tensor cores that power features like the original DLSS. That’s a big part of why Frame Generation feels more like a way to get people to buy new RTX 40-series cards rather than necessarily improving DLSS for all.

They imply Nvidia is incentivized to keep this advantage exclusive to RTX 40 GPUs for now to drive adoption. But hope is it expands to more hardware in future.

My Take – Impressive but Plenty of Room for Optimization

In my expert opinion, DLSS 3 is an incredible technological feat by Nvidia, demonstrating the future possibilities of AI in gaming.

However, it isn‘t yet a magic "perfect image quality with no downsides" solution. The fundamental constraints of artificial frame generation necessitate some current compromises in visuals and lag.

But knowing Nvidia, they will rapidly improve training and algorithms over time to incrementally minimize artifacts, quality dips, and latency issues.

DLSS 3 represents the cutting edge of AI accelerated graphics – flaws are expected this early, but the potential is game-changing. For those wanting to maximize frames, it‘s almost like cheating.

Yet discerning gamers who care about image quality, latency, and widespread compatibility will see benefits in sticking with the mature DLSS 2 for now in many titles.

I‘m bullish on a bright future for DLSS 3. But temper your expectations at this initial stage of its lifecycle. Pay attention to quality and lag, not just the huge FPS numbers.

With refinement, DLSS 3 could well represent the next evolution in smart upscaling and truly bridge the gap to photorealistic real-time graphics.