Combining DLDSR and DLSS

[MrFuzzy 2,118]

36 minutes ago, mpo910 said:

Why you get upset when someone makes a video? Because it is evidently clickbait? Where exactly is your evidence? You spoke to the owner of the video and he gave you the evidence?

It is who makes a statement that has to bring evidence actually, and in these videos there is none.

Sorry but sharing misinformation adds nothing to the community.

[mryan75 281]

4 hours ago, MrFuzzy said:

Scale up the resolution, then scale it down when the game is launched to return to the original resolution of the screen; then apply DLSS to scale it down internally and again scale it up with AI to match the resolution of the screen. How TF would that accomplish anything?

If that’s what was being done I would agree, it wouldn’t make sense. I find it interesting that people who don’t even try something know it doesn’t work. I was about to say “read some of the comments to the videos,” but never mind. You won’t. There are people with 50% performance improvements, people that were thinking about upgrading already very high-end systems or buying new ones entirely because they didn’t get the performance they expected who have completely unlocked their system performance. Some people using 2-series cards have seen massive improvements, some none at all. The variables are of course endless. But it’s not a matter of “if I help one person I’ll be happy,” we’re on to thousands by now. 

Enjoy your simming.

[mryan75 281]

36 minutes ago, Ixoye said:

Will DSR only work with DX12? I would like to try it without DLSS if possible instead of increasing render scale in game settings.

DSR works with DX11 as well, but I’m not 100% sure about DLDSR, which is the AI version. DLDSR is 200% more efficient than DSR per nVidia. And that’s a great idea, it’s worth trying for sure. 

[mryan75 281]

10 hours ago, SKEWR said:

So maybe I was misunderstanding. I am able to turn on DLAA in MSFS and adjust the settings in NCP. (That works fine) I was thinking I had to have DLSS and DLAA in the sim.(that has the blurry glass screen displays) 

So for this particular setup it’s DLDSR on in NVCP and then DX12-DLSS on in the sim. I personally used to use DLSS-DLAA (standalone, before i even knew what DLDSR was) and it worked the best for me with glass displays, but with DLDSR+DLSS I have found the performance hit on DLAA is pretty hard, whereas Quality is absolutely perfect. Literally no ghosting on my class cockpits at all. Like, none. So you really have to try it and see what works best, but it’s DLDSR + DLSS and then whatever DLSS “subsetting” (for lack of a better word) which works for you. 

[Luis Hernandez 571]

4 hours ago, MrFuzzy said:

I am upset with these videos because they are evidently clickbait videos that are not based on empirical evidence

However, this guy did something very helpful: he wrote down all the settings in the timestamps, and he left the video for explaining why each setting.

Spoiler

Quote

00:00 Intro
03:50 The What
06:11 NVCP DSR-FACTORS (DSLR): 2.25 AND 1.75
15:20 The How
17:37 NVCP DSR SMOOTHNESS: 60%
18:35 NVCP TEXTURE FILTERING:  HIGH QUALITY
19:16 NVCP ANISTROPIC FILTERING: 16X
19:35 NVCP LOW LATENCY MODE : ULTRA
20:20 NVCP POWER MANAGEMENT
21:30 MSFS ANTI-ALIASING: NVIDIA DLSS SUPER RES.
21:43 MSFS select a resolution: highest (for 2k monitor 3840x2160 = 2.25X) (3rd from bottom is native resolu.)
22:00 MSFS NVIDIA DLSS SUPER RESOLUTION: QUALITY
23:00 The Setup
23:17 MSFS GLOBAL RENDERING QUALITY: ULTRA
24:12 MOTION BLUR: OFF
29:40 The Details
32:10 The Flight

 

Clickbaity? I don't know, but, because of the way he summarized his video, I didn't have to watch the whole 55 minutes for applying these settings.

Edited December 12, 2023 by Luis Hernandez

[MrFuzzy 2,118]

35 minutes ago, mryan75 said:

If that’s what was being done I would agree, it wouldn’t make sense. I find it interesting that people who don’t even try something know it doesn’t work. I was about to say “read some of the comments to the videos,” but never mind. You won’t. There are people with 50% performance improvements, people that were thinking about upgrading already very high-end systems or buying new ones entirely because they didn’t get the performance they expected who have completely unlocked their system performance. Some people using 2-series cards have seen massive improvements, some none at all. The variables are of course endless. But it’s not a matter of “if I help one person I’ll be happy,” we’re on to thousands by now. 

Enjoy your simming.

People comments are not proof. There are millions of posts stating that the earth is flat on the internet.

People comments are not proof. Scientific methods of testing and reproducible empirical results are proof. 

People have messed up systems and generally have no idea how to conduct a test in reproducible conditions. For example, this was posted on Reddit today, what do you think about it? 

P.S. I tried the HOLY GRAIL and didn't get nicer graphics quality or better performance. I'm gonna stick with native resolution + TAA.

Edited December 12, 2023 by MrFuzzy

[AnkH 2,268]

3 minutes ago, MrFuzzy said:

 P.S. I tried the HOLY GRAIL and didn't get nicer graphics quality or better performance.

 

I totally agree on proof, but nicer graphics is so much a subjective experience, you cant measure it... For me, using DLDRS 2.25x on 1440p improved the graphics slightly over using TAA with 130 resolution scale, which is basically totally logic considering the fact that I now use a slightly higher rendering resolution than before. FPS stayed the same, so I stick with it, holy grail or not is basically irrelevant 🤣

[Epgd 136]

is this HOLY GRAIL setup working in VR?

is there any benefits?

brds

[FederFlyer 27]

I have an i7-13620H  2.40 GHz and an RTX 3050, and had done a variety of MSFS and Nvidia tweaks that had greatly improved the experience. Yesterday I made changes based on the Holy Grail DLDSR + DLSS video. No difference in how MSFS runs, except for LOD settings. Wow!

Before making the changes, I had LODs at the highest settings that were just below where stutters, lags and other misbehaviors would occur. Now I have them at their maximum settings, with no issues.

Terrain LOD was 150, is now 400

Objects LOD was 125, is now 200

There are countless variations in hardware, accessories, aircraft complexity, settings and tweaks, etc. Obviously, individual results will vary. In my case, it was well worth the time experimenting with it.

 

 

 

[MrFuzzy 2,118]

40 minutes ago, AnkH said:

I totally agree on proof, but nicer graphics is so much a subjective experience, you cant measure it... For me, using DLDRS 2.25x on 1440p improved the graphics slightly over using TAA with 130 resolution scale, which is basically totally logic considering the fact that I now use a slightly higher rendering resolution than before. FPS stayed the same, so I stick with it, holy grail or not is basically irrelevant 🤣

But you didn't downscale it/reupscale it again with DLSS which is the whole point of the video... Of course if you use DSR alone, the quality will be objectively better.

Edited December 12, 2023 by MrFuzzy

[mryan75 281]

37 minutes ago, Epgd said:

is this HOLY GRAIL setup working in VR?

is there any benefits?

brds

VR has run the gamut, some people have seen no change but a few have reported enormous improvements. 

[mryan75 281]

1 hour ago, MrFuzzy said:

I'm gonna stick with native resolution + TAA.

Excellent, I wish you the best with your simming!

[Stroman81 18]

Hi all,

I've been simming for 20 years now, and seen a billion topics on performance improvements on dozens of different FS forums going from FS98 over P3D up to MSFS. FS, in my opinion, has always been and will always be a resource hog just due it's nature.

As such, and since there are more than a billion different hardware and software setups possible there is no one size fits all solution and so also no "HOLY GRAIL". So I think the title is somewhat badly chosen, but...

I have always been triggered into trying some of these tips to see where it brings me (haven't we all been?) The go/no go decision on trying such a proposed solution is for me based on following factors:

1. the complexity 2. my hardware 3. the first few responses on the topic/youtube vid/... 

So, in spite of the few (very rude) sceptic replies to this topic I decided to give the fix a go because implementing (and reverting if no positive outcome) is simple as buying a bread, my hardware (RTX 3070 & i7 powered laptop connected to an external HD monitor) seemed to support it and the first responses to the video seemed to be quite positive...

For me at least, and after a few shorter flights last evening in the PC12, this indeed seems to improve my experience in a truly effective and positive way. Meaning: no more stutters while panning around or during the approach, being able to turn up my LOD settings in the sim and last but not least and enormous improvement in image quality/sharpness even on my older external monitor (partly because of improved LOD settings, but also due to the scaling I believe).

I still need to have a final test in a performance breaking scenario like landing at KLAX or so, but so far a huge kudos to the OP for finding this. If anything, the goal of this post is to convince people that suffer from stuttering or subpar image quality to try this (if there hardware allows it). It's absolutely no effort and the return until now has been brilliant!

Kind regards,

Nico

 

[Ron Lefebvre 113]

I do not usually do these as they are mostly a waste of time and can really screw up your system.

However this one is pretty straight forward. I made a list of the changes so I could revert back without to much trouble.

The difference is subtle, but is is there. Have not notice a change in frame rate or stutters.

Anything that makes the panel sharper is a winner for me.

Hard to tell if it is worth it or not,but no harm done so far on my end.

Will keep it for now.

Ron

 

 

 

[cianpars 1,170]

Tried 2.25 and the graphics looked great but reduced me to a crawl.  1.75 not so bad but after a couple of days of testing, the performance hit was not worth the visial gains or the screaming fans on my graphics card for me.

Have now reverted back to DRS off.

It was interesting to experiment with though but results will be very system dependent.