Cyberpunk 2077 DLSS Quality Comparison vs. Native, Benchmarks, & Blind Test

Read more about Cyberpunk 2077➜

In spite of NVIDIA’s best efforts to coerce media into talking about DLSS and Ray Tracing, we’re doing it because we actually want to. Bad timing from NVIDIA. This is #Cyberpunk2077’s DLSS.
Sponsor: Thermaltake Core P3 on Amazon (

Watch our Cyberpunk 2077 GPU benchmarks:
Watch our #Cyberpunk 2077 CPU benchmarks:

In this content, we’re comparing the many types of DLSS for Cyberpunk 2077, using both an RTX 3080 and an RTX 2060 KO. We’ll also briefly talk about FidelityFX and CAS. The goal of DLSS isn’t to increase the quality past native resolution, but rather to get to a point of “good enough” that it’s actually viable to run higher settings elsewhere. In some cases, the quality does actually exceed native resolution, and impressively so; at the same time, we also found a few spots where DLSS makes things worse in Cyberpunk’s Night City. Overall, it’s improved significantly since the Battlefield V vaseline smear and the Final Fantasy XV flickering mess, and DLSS is actually a usable feature that provides meaningful uplift. In this content, our objective is to show you the actual qualitative differences. We also propose a few blind tests.

Find the RTX 2060 KO on Amazon:
Find the RTX 3080 TUF on Amazon:

The best way to support our work is through our store:

Like our content? Please consider becoming our Patron to support us:


00:00 – If Steve Blinks, NVIDIA is Holding Him Hostage
05:21 – Blind Comparisons
06:34 – Methodology
07:16 – Detail Loss & Upscaling Amounts
08:54 – Comparison Shots & Chromatic Aberration, Film Grain, DOF
09:54 – Shot 1 – DLSS Performance, Ultra Performance, Quality, vs. Off
10:40 – Broken DLSS in Some Areas Useful for Validation
11:10 – Pixel Counting to Prove the Resolution
12:10 – NVIDIA’s Unfortunate Email Timing
13:53 – Shot 2 – Subtle Differences, On vs. Off Comparisons
14:19 – DLSS Sometimes is Better Than Native Resolution, But…
15:52 – Ghosting Caused by Temporal Anti-Aliasing
16:31 – Shot 3 – Broken Lighting Bug, Texture Detail Loss
17:54 – Shot 4 & 5 – DLSS Easily Noticed in Flora
18:31 – Shot 6 & 7 – On vs. Off Flickering & Unexpected Bonuses
19:07 – Shot 8 – Final DLSS Impact
19:25 – The NVIDIA Money Hand
20:00 – Conclusion, Our Thoughts, & Blind Comparison Recap

** Please like, comment, and subscribe for more! **

Links to Amazon and Newegg are typically monetized on our channel (affiliate links) and may return a commission of sales to us from the retailer. This is unrelated to the product manufacturer. Any advertisements or sponsorships are disclosed within the video (“this video is brought to you by”) and above the fold in the description. We do not ever produce paid content or “sponsored content” (meaning that the content is our idea and is not funded externally aside from whatever ad placement is in the beginning) and we do not ever charge manufacturers for coverage.

Follow us in these locations for more gaming and hardware updates:


Host, Additional Reporting: Steve Burke
Editorial, Testing: Patrick Lathan
Video: Keegan Gallick, Andrew Coleman


49 thoughts on “Cyberpunk 2077 DLSS Quality Comparison vs. Native, Benchmarks, & Blind Test”

  1. Interestingly, in the side by side comparison image at 5:50, on my 1440P Monitor, while C and A were obvious and dramatic outliers, the other three were hard to differentiate, but I ultimately chose the reverse order. I gave a slight edge to E (Performance), thinking it might have been a littler sharper in places, and identified D (Quality) as slightly more blurry in the details. Thinking about upsampling images for graphics, this might actually make sense for still images: 1440 -> 2160 is 2:3 so it requires weird pixel mapping but the other two are both just upscaling by x2 or x3. On the other hand, it's also a 4k screen shot zoomed in by some factor for a 4k video (hopefully 200%?), then resampled to 1440p, and compressed for YT, so a 1:1 or 2:1 zoom on the actual screenshots would be a better comparison.

  2. Great content. This is what I wanted to know: should I be using 4K performance DLSS or 1440p quality to play this and Watch Dogs Legion. This video came the closest to answering that question for me. And if I go study those blind comparisons a bit longer then I'll have my answer.

  3. DLSS Balanced seems to be the sweet spot. Not much different looking then quality but with a 10-15% better performance then the quality setting. Also, it still renders at 1440p at 4k. If you use digital foundries optimized settings with DLSS and ray tracing it's a pretty smooth experience. I won't say it runs well because we all know that isn't true but at least FPS won't be a problem.

  4. nvidia wants more cover of dlss they got it but not from the 1 channel they really want to play around with. who would have thought to use 3×3 pixels box to confirm a resolution….

  5. 15:55 You can actually disable TAA using config file "Cyberpunk 2077/Engine/Config/Base" General.ini adding following lines:

    Antialiasing = 0

    AntialiasingSuppressed = 1

  6. I love Jay and I love Linus but when I want meat and potato cold hard un biased facts I come to GN and good old Steve. Always down to brass tac and a no bs approach. Thankyou all of you for what you do and offer for all of us in the PC and gaming community. much love and respect guys seriously keep it up!

  7. How am I capable to use DLSS with 1070 and no ray tracyng not that I want to. But is seems what DLSS really does is lower the resolution. Fun fact, when I change the "resolution" in the game it seems like it uses DLSS, since my monitor running in 1080p and running the game at 720p in Fullscreen does not switch the resolution at all to 720p….
    Fuck it, so in order for Raytracing to work all we have to do is lower down our resolution in to obismo, great job nVidia.

  8. I don't understand why you are spending so much time benchmarking 4k ultra performance mode DLSS when nobody uses that. You said yourself it was not intended to be used yet so many of your comparison shots to show flaws in DLSS are using it. WTF steve.

  9. On a 2080 at 1440p, dlss on blanced, some sharpening from the nvidia panel to get a bit more clarity, lowered cascaded shadow and remove motion blur and film grain and the game can run in ultra with full rtx.
    I would be totally unable to run it that way without dlss. I would say it's now a game changer setting.

  10. In the blind comparison, I could tell C looked the best followed by D, and the rest all looked noticeably worse. Not sure if it's placebo or not, but I feel that scenes in motion are more noticeable with DLSS than static images – perhaps due to the ghosting effect? Nevertheless, 40% performance gain for a negligible loss in detail using DLSS Quality is tempting enough that I may give it another go, despite being very close to a steady 60 FPS without it.

  11. 9:35 the sample [mis]captioned Ultra Performance DLSS is clearly sharper than the sample [mis]captioned Performance DLSS, see most notably "CHINA" … lay of the booze, editor 😐

    (the progressively unnerving and evidently self-confusing switcheroo between ascending and descending order of DLSS quality levels further contributes to both above implied assumptions)


Leave a Comment