Reduce input latency by using fakenvapi with LS

[swift lichen]

fakenvapi github page: https://github.com/FakeMichau/fakenvapi

For unknown reason enabling Nvidia Reflex in LS using RTSS or SK doesn't do anything, but enabling LatencyFleX using fakenvapi does

I use Special K to load the fakenvapi dll

Known issue: when fakenvapi is used, the FPS can be super low after multiple scales, the only solution I've found so far is to restart LS

[golden mica]

May I ask a few questions?

1- does this only work with Nvidia GPUs or AMD too?

2- is there a guide on how can we setup this?

[swift lichen]

It should work on all GPUs. LatencyFleX supports all GPUs

[golden mica]

Oh that's good to hear, thanks.

[golden mica]

What about this? ^

[swift lichen]

I think the easiest way is to put both the Special K and fakenvapi dll files in the LS directory, and change the SK config file to make SK work via the guide in the Special K thread in the forum
#1279794350980595734 message

I've tried simpler methods like using Koaloader to load the fakenvapi dll directly, but it won't really work, I've tested it several times and the latency doesn't change. It looks like using SK to load fakenvapi is the only way I found so far.

I uploaded my SK + fakenvapi zip file, theoretically just unzip them into the LS directory and it will work, but I haven't tested it on other PCs

[golden mica]

Alright thanks, I'll test this out when I'm back home

[golden mica]

do I have to put the fakenvapi dll every time I launch LS or is it a one time thing?

[golden mica]

also I'm tryna do this rn, where can I get this special k dll? I only found the app and installed it

[ivory socket]

So nvidia reflex doesn't work with LS?

[golden mica]

He's saying it doesn't actually do anything if it was injected via RTSS or Special K

[ivory socket]

I'm not sure about that but it could be true

[median lily]

how do i know if it’s working?

[golden mica]

Can you help me set this up?

I haven't ever used Special K

[ivory socket]

sure

It's easier than you think

[median lily]

nvm i see it now

[golden mica]

thanks, Where can I start?

[ivory socket]

What game do you want?

[golden mica]

Fortnite

I can't inject anything in the game itself btw

cuz ban

[ivory socket]

Yes, the anti-cheat won't like that very much

[golden mica]

Yeah, but isn't this guide to LS itself?

didn't know there was injecting stuff in the game, thought only to LS

[ivory socket]

Lossless Scaling will put a window on top of your game, but SK will be injected into the game underneath that window

[golden mica]

ah, ig I can't do this then, unfortunate

[ivory socket]

If I'm not mistaken, there is a reflex in Fortnite, right?

[golden mica]

yes there is

Thought this was better than Reflex but seems like it isn't

[potent plover]

So maybe i'm doing something wrong, but i don't quite understand what are the steps to make it work. When you say "load Special K in LS first" you mean to inject special K into LS and run it through Special K? also, the fakenvapi dll goes to the root directory of LS or the directory that Special K creates when it injects?

[ivory socket]

Downsides of LatencyFlex:

Limited latency reduction compared to AL2
Low fps just after launching a game that stabilizes over time
(this is from the github page)

[late compass]

I tried adding those files to the ls directory. Strangely the toggle lsfg key combo doesn't work anymore with special k enabled.
Using the scale button works, and shows the special k startup info on the screen, but i can't interact with it or open the special k hud with ctrl-shift-backspace.

[potent plover]

same, but also screen goes crazy like a cascade pattern of a fraction of the screen

[late compass]

I don't have that, but i also have a small "HDR calibration" special-k window on the top right of the screen that won't go away.

[ivory socket]

If you open the SK menu you can turn it off

[craggy frigate]

coooooool i will try that soon

but isnt special k will only inject in lss ? i thought i read that in the special k +lss forum here somewhere

[late compass]

Thanks, i managed to turn it off in the special k app.
Is there any indicator to know that LatencyFlex is running with special k?

[craggy frigate]

ohhh shiiiiiiiiiit i think it worked , by feeling when i try it now on marvel rival , i tried too feel it with and without

[craggy frigate]

maybe the log file in ls directory fakenvapi.log

[late compass]

It has a line "LatencyFleX initialized" in the logfile, so i guess that means it's working?

Does the scale hotkey work for anyone when using this special k method? I tried starting LS as admin, but it didn't help.

[craggy frigate]

i guess so thats the way to know

hotkey works for me

[late compass]

If there were a command-line option to toggle scaling i could work around it, but i don't think there is one.

[swift lichen]

Use old version of Special K

I'm using version 24.11.23.2

I don't know which version exactly, but the shortcut doesn't worked since some version between January and February

[swift lichen]

Put both the Special K and fakenvapi dll files in the LS directory, and change the SK config file to make SK work via the guide in the Special K thread in the forum

#1279794350980595734 message

[swift lichen]

You can set enable_trace_logs to 1 in fakenvapi.ini, and if fakenvapi.log looks like this it means it's working

If it looks like this then it's not working

[next junco]

Can u plz make a pre configured files just copy past and upload it thanks 😊

And I think you can try optiscaler files only dxgi and ini confug file inject reflex , try config it reflex on and take 2 files put them in ls folder and try if it work 100% with latency test

[swift lichen]

I've tried optiscaler and LS will crash when unscale

And I've uploaded my pre-configured SK and fakenvapi

[next junco]

Old version 2.11 I was able to use with ls but after or maybe I edit config now work once 2nd time crash maybe ths or .. can help 👍

[potent plover]

right, that guide is what i was missing, will try later, thank you!

still same error as before, hotkey stop working and the scale button makes the screen goes crazy

definitely i'm doing something wrong

i don't know if it should say fakenvapi dll instead of d3d11 dll

[swift lichen]

Use Special K v24.11.23.2, and try renaming it to dxgi.dll instead of d3d11.dll. also d3d11.ini needs to be renamed to dxgi.ini

Or you can try my pre-configured Special K and fakenvapi

[potent plover]

nope, the moment i put any kind of dll into the ls folder the error appears

i tried usinbg your config, i'm going to try with the v24...

[reef nexus]

Tried to use special k but couldn't get the shortcut to function

[potent plover]

okay IT'S definitely the v25 version doing weird stuff, i managed to make it work now with v24

now i'm gonna test the latency

[potent plover]

thank you btw

now my problem is that the display menu of SK doesn't open when i press ctr space backspace xD

[reef nexus]

Same issue agaib

Can't actually open special k

Clicking the banner is impossible

[potent plover]

i just want to enable latent sync or something to prevent tearing,isn't a method to do it through an ini files or something?

[late compass]

Thank you so much for your help, according to the trace logs it looks like it's working.

I've now tested several versions of special k and this seems to be the last version where the LS Hotkey still works:
https://github.com/SpecialKO/SpecialK/releases/tag/SK_25_1_3_3

I just renamed SpecialK64.dll to dgxi.dll and combined it with your files and put it in the LS folder.

[swift lichen]

#1308625919157604403 message

It's a little tricky to open the menu

[potent plover]

this version doesn't work for me, it reverts the error where the screen goes like melting imagen from top to bottom

nah, it's really unstable for me, some versions doesn't work at all, others do but can't open the menu and crash LS often, i will leave this while we figure out how to make it stable and working

maybe it has to do with not having MPO support on my pc, 23h2 windows version, etc... who knows

[reef nexus]

Pretty sure you have to do it in the menu

Maybe if you copy from another game

[reef nexus]

still can't get it to open lol

[reef nexus]

@swift lichen what settings are you using?

Reflex start or end boost yes no

[normal minnow]

can you do fakenvapi without SK

[reef nexus]

Dunno

Why

For ls?

[normal minnow]

yeah

for lossless scaling

wanna inject this to LS too

[reef nexus]

Hmm

Yeah I'm wondering the same

Injecting special k to lossless is jank asf

[slow nova]

How to install

[normal minnow]

read the thread

#1339992172807589929 message

[slow nova]

is it better to inject in the game or lossless?

[normal minnow]

do both if you're unsure

[late compass]

I did some 240fps phone-cam input lag tests with very promising results:

I tested with Retroarch and Yuzu Emulator at 60 FPS x2
I also tested DXGI without Special K, as it was said to have higher input lag than WGC, which my tests confirmed.

Retroarch:

Without LSFG: 19.20 ms
DXGI: 39.44 ms
WGC: 35.42 ms
WGC with Special K and LatencyFleX: 28.87 ms

Yuzu:

Without LSFG: 93.91 ms
DXGI: 119.17 ms
WGC: 114.58 ms
WGC with Special K and LatencyFleX: 108.58 ms

So using WGC instead of DXGI reduced latency by 4.31 ms
And using WGC with Special K and LatencyFlex reduced latency by 6.28 ms compared to just WGC

Of course take those results with a grain of salt, that was just a quick cam test.

But i'll keep using this Special K method, thanks @swift lichen!

I'm curious to see what others measure with actual measuring hardware.

Edit: Added measurements without using LSFG.

[slow nova]

Injected to game or lossless?

[late compass]

injected to lossless

[slow nova]

and you use DXGi.dll or d3d11.dll?

WGC double buffering?

what version of specialk?

[late compass]

dxgi.dll, yes i use WGC with double buffering

[slow nova]

i downloaded this one

[late compass]

i posted a link in this thread

exactly

[slow nova]

my lossless is 40/60 now

[reef nexus]

Wgc has less latency?

[next junco]

So at which base latency for both retro and yuzu ?

[late compass]

Good question, i now tested it for both without LSFG and added the result to my post.

[jolly latch]

DXGI has less latency but WGC is not far behind, but if you enable double buffering it will add serious latency

[reef nexus]

okay

[glacial sapphire]

its the opposite

[late compass]

On which windows version did you get lower latency with DXGI than with WGC?

I did my tests with the current windows 11 version and got less input lag with WGC

[glacial sapphire]

wgc by default has less latency, double buffering reduces it further

btw if someone could test antilag 2 with this, it would be great

[jolly latch]

double buffering increases my latency...

[glacial sapphire]

We've been trying to get it to work for months to get it to do something but in the end latencyflex seems like the one that does something

[glacial sapphire]

it doesn't

check the graph for the latency tests

[late compass]

also, in my measurements, double buffering only makes a difference when NOT using gsync, but for lowest input lag you'll want to use gsync/vrr.

[normal minnow]

outdated knowledge

check recent benchmarks and measurements

[swift lichen]

Don't enable Reflex in SK

[reef nexus]

?

Then how does it get activated?

Just the presence of fake reflex is enough?

[swift lichen]

For unknown reasons, SK will automatically load nvapi64.dll in the executable's folder and will make latencyflex work automatically, enabling reflex will instead make latencyflex ineffective and latency will not be reduced

[reef nexus]

hmm

Okay

Makes sense

[glacial sapphire]

okay I didn't know that

might explain a lot lol

[reef nexus]

What if you renamed that dll to dxgi tho

[glacial sapphire]

doesn't work

we tried

[reef nexus]

Zamn

So atm you need something to load the dll?

In this case special k

[glacial sapphire]

only specialk

maybe because it gathers all the data

and latencyflex works off of it

[swift lichen]

Only using SK can load it and make it reduce latency

Using things like koaloader or dll injectors doesn't really make it work.

[reef nexus]

If you don't touch special k it shouldn't intefere right?

Cuz I already inject it into the game itself

[glacial sapphire]

should be fine

[reef nexus]

Any setup required?

[glacial sapphire]

wdym?

[reef nexus]

In special k menu of lossless

Or just let it inject

And then everything works

[glacial sapphire]

just follow orwennes advice

no need to tweak settings

just let it do its thing

[late compass]

I got curious about the rather high base input lag i got in yuzu and tested the same game in ryujinx, 60 FPS x2:

Without LSFG: 107.87 ms
WGC with Special K and LatencyFleX: 129.17 ms

So Yuzu with LSFG (+SK+LatencyFleX) (108.58 ms) has pretty much the same input lag as ryujinx WITHOUT LSFG 😄

Edit: This was with an old ryujinx version, with the newest fork Ryubing/Ryujinx it looks better, but still higher input lag than yuzu:

Ryubing/Ryujinx WGC with Special K and LatencyFleX: 115.89 ms

[jolly latch]

this literally happens to me, if I activate double buffering I have more latency, this on Windows 11 24h2

[normal minnow]

@swift lichen for the initial video, did you test with vsync off

or default 1 frame latency

[normal minnow]

If it's default 1 frame latency, I think you should test it again to make sure it's not placebo

cause 1 frame latency is definitely not working if you use this method

crazy tearing just like vsync off

[next junco]

Nice what about qemu , and whiche better software to test latency added by lsfg "only" like more accurate and in aproximity app like : rtts , capx, nvidia with how much diff vs real oslt , : rtts example 9 ms +1 ms or % differences vs real measurement ty

[next junco]

it's supposed that double buffering increase latency not reduced but with new lsfg test and win 11 does the opposite 👍

[swift lichen]

I always use allow tearing + max frame latency 1 + backbuffer count 3(Backbuffer count can be set in SK, setting it to 3 or above can solve the freezing issue). In fact I tested each case at least 5 times, it's just that the video only shows one test for each case. For using lfx and not using lsfg there is only a 1-2 frames variation. The latencies without lsfg are all between 29-30 frames in 480FPS slow motion, while with lfx the latencies are all between 34-36 frames. Without lfx it varies a bit more, between 42-46 frames, but no matter how many times I test it, it's always higher than with lfx

[normal minnow]

No freezing issue here what i mentioned is tearing issue

[late compass]

don't have qemu. Dont know of any software to measure latency accurately, thats why i'm doing it with a phone cam.

[quartz belfry]

I've followed all the steps and it seems I've managed to get fakenvapi working correctly... it was a bit tedious. My problem is, when activating LS along with SK and obviously fakenvapi, when I click (to do something in the game) Lossless Scaling minimizes and stops working along with fakenvapi... does anyone have any idea how to fix this? The idea would be to test the latency with keyboard and mouse but since I can't even fire a weapon it's hard to tell.

[swift lichen]

The first scale after LS starts will not be able to use the mouse to interact with the game window, from the second time onwards it should be fine

[normal minnow]

Yeah after checking im doubting more and more if this works

[opal igloo]

Is there any reason why [API.Hook] d3d12= is set to False ?

[reef nexus]

You guys get judder too?

Became it's on 216

@normal minnow same

It's capped to 216 for some reason and there is judder

[normal minnow]

yeah it's not as smooth

[reef nexus]

Gonna uninstall it tbh

Can't even use 4x60 as it gives massive judder

[normal minnow]

My guess is the special K inject probably messes something up so the sync is diff

like, normally if I use Default sync mode there is no tearing

with SK injected you can obviously see tearing

when there should be none

also this was already written in the original post

but this mod nukes your perf every time you alt-tab back from another program

try alt tabbing a few times and see your performance fall off a cliff

so actual usability is ehh

[late compass]

I don't get judder, but i'm using gsync, so v-sync doeesn't come into play

[reef nexus]

Don't have it unfortunately

[late compass]

i didn't try it without

[normal minnow]

have you noticed the uhh

insane tearing

g-sync with vsync off will still have insane tearing

you should be using gsync with vsync on anyways

[reef nexus]

Yep

But I'm using lossless vsync atm 🙂‍↕️

No gsync no vrr = pain

[late compass]

gsync without vsync causes some games to stutter, but in lossless scaling i didn't see a difference with it on or off.

[normal minnow]

LS without mod

i dont notice it either

[reef nexus]

Nah he means NVCP vsync

I believe blurbusters recommend it as well

[normal minnow]

but with sk fakenv injected its tearing city bro

[reef nexus]

interesting

Looks like we all get different issues lol

I'm gonna remove it and see if it gets resolved

[late compass]

One thing i noticed with this special k method is that often only the first scaling aber starting LS works smoothly.
Just now got stuttering with and without gsync, but got resolved for both, with just restarting LS.

[reef nexus]

For me that didn't help unfortunately

[late compass]

i have vsync forced on in nvidia control panel, if thats relevant. in ls i have it set to off, as i force it anyway.

[calm zephyr]

i get judder at 100% gpu usage. either with igpu lsfg or dgpu lsfg. feels like latencyflex is dropping lsfg frames and causing judders

[reef nexus]

My GPU usage isn't 100% though

[prisma egret]

So, I inject Special K into LS instead of the game, like Skyrim?

[late compass]

yes

[calm zephyr]

Been testing all SK versions to see whats the most recent one we can use with LSFG without getting a bugged screen.

• last working version is Special K v 24.11.24.2
• version that broke LSFG screen is Special K v 24.12.3

Link to the last working version:
https://github.com/SpecialKO/SpecialK/releases/download/SK_24_12_3/SpecialK.7z

Something was done after Nov 24, 2024, that completely broke LSFG screen, and there are a lot of changes made:
https://github.com/SpecialKO/SpecialK/compare/SK_24_11_24_2...SK_24_12_3

If only we could pinpoint what setting is causing LSFG screen to break, we could fix the setting and make newer versions of SK work with LSFG again

[swift lichen]

The latest version is working for me, the hotkey not working issue is fixed

But the huge VRAM usage problem still hasn't been fixed. I don't know since which version every scale takes up a few hundred MB more VRAM, and it only takes a few scales to fill up the VRAM

[late compass]

Thanks for the update, the latest version is working for me too!
Can confirm about VRAM, every scale adds 700mb VRAM usage in my case.
As a workaround i stopped using auto-scale and use an autoit script, bound to a hotkey to restart LS with every scale.

[calm zephyr]

I forgot im testing sk only on alpha builds. Will test on stable release later. Maybe its why it doesnt work with newer sks

[prisma egret]

The trade-off is too much for a ±10ms reduction

[potent plover]

For me still not working the hotkey issue

and the screen still breaking 😩

[calm zephyr]

I did some tests with an alpha build and 240 fps videos with frame counting (precision of 4ms, 6ms of error margin more or less), and at 39/78 fps and 100% gpu usage, i havent seen any objective latency reduction when comparing stock lsfg vs latencyflex lsfg. In fact, if i use my igpu for lsfg (which is my favorite setup), latencyflex's latency rises sharply. Take these tests you guys are doing with a pinch of salt.

for my usage, its useless

[nova nebula]

Any video for install?

[fervent coyote]

Please, give me a hint. I just need to move the SK + fa archive from the hat to the LS folder. It is not necessary to use RTSS or SK overlay or anything else. It all works only with DirectХ 11

I understand correctly?

[fervent coyote]

I'm not asking for a detailed answer. I would be very grateful if you could just write "yes" or "no."

[keen urchin]

how did u go about doing that exactly? i would love to know 😄

[late compass]

My AutoIt script to (re)start LS and scale:

Local $hWnd = "Lossless Scaling";

If WinExists ($hWnd, "") Then
    WinClose($hWnd);
EndIf

Local $hWnd2 = Run('"C:\Steam\SteamApps\common\Lossless Scaling\LosslessScaling.exe" ""', "", @SW_HIDE)
Sleep(1000);
Send("^!h");```

To use it, adjust the path and shortcut to scale in LS. I use "^!h" which means ctrl-alt-h. This has to be the same as the scale shortcut in LS.

Then compile the script in autoit ~~and create a windows shortcut to it. In the right-click properties of the shortcut you can define a hotkey to run the compiled exe.~~

I also use another script and hotkey to close LS, for this it's basically just the same script without the part below "EndIf".

Edit: The Windows shortcut hotkey method doesn't work right. I defined the hotkey to run the compiled exe in PowerPro, where it works flawlessly, linked a couple posts below.

[keen urchin]

thx man!

[late compass]

you're welcome! 🙂

[keen urchin]

wait so the 2 scripts just restart and close LS?

like one restarts it and one closes it

[late compass]

The script i posted closes and restarts LS, and then sends the scaling hotkey to LS to start scaling

[keen urchin]

gotchya

the shortcut keys dont even work when trying to do so for the compiled scripts, i hate microsoft sometimes

[keen urchin]

i mean the ones to run shortcut that restarts LS

[late compass]

So strange, it doesn't work for me either, why to they have this option when it doesn't even work? 😄 i use an old tool for keyboard shortcuts, PowerPro:
https://www.majorgeeks.com/files/details/powerpro.html

Edit: I managed to get the windows shortcut method to fire, but only when not using something fullscreen, which makes it pointless.

Defining a hotkey in Powerpro to the compiled exe however works 100% in all cases on Windows 11.

You can also use another hotkey tool, if you use one.

[swift lichen]

A simple Python script written by DeepSeek, put it in the LS directory. Running main.py as admin will automatically start LS and minimize it to the system tray, press Ctrl+Alt+D to restart LS, press Ctrl+Alt+C to exit the script

[fervent coyote]

I can't enable enable_trace_logs: 1. After running it returns false to me. Can you tell me what could be the reason?

[fervent coyote]

.

[neon lantern]

Ive never installed thsi before. How well doe it work?

[fervent coyote]

If this question is asked of me, I haven't felt any difference, either on nvidia or AMD. So now I'm trying to find out if it works and what I'm deraying wrong

[neon lantern]

hmmmm alright alright

[fervent coyote]

But lamers are not loved in this thread, that's for sure...

[late compass]

Did some 240 fps cam input lag tests with the new LS version, Special K with LatencyFleX doesn't improve the latency anymore for me with the new LSFG update, in fact it made it worse in some cases.
Curious if others have tested it with the new version (3.1.0.2)?

I only tested this in retroarch in a 60fps limited game.

[regal frost]

Is there have tutorial to teach how to use this?

[swift lichen]

I tested the latest version and it seems that fakenvapi does no longer reduce latency, but instead makes it less consistent, with latency fluctuating over a huge range, sometimes increasing latency by only 10ms and sometimes by 30ms at 30 base FPS and fixed X4 mode

[next junco]

Can you please give latency at 60 fps with all sync off , 0 buffers etc lowest possible settings for latency and 25% flow scale , with special k hight priority "rendering or...
i forget " ty

[nocturne quest]

so I need to put SK files on LS, right?

[fervent coyote]

I think we should do that, too. But there are partisans in this thread who won't even say yes

I didn't test this with, the old version but in the new one it doesn't work for me

[nocturne quest]

New one, you mean SK or LLS?
I'm using the SK that OP uploaded, when I select Scale in LLS, it shows the SK notification, but if it really is reducing input delay I dunno, lol after all it is just 8ms faster according to the tests

[fervent coyote]

I'm using the archive from the hat. Other than moving files to the lossles folder, I haven't configured anything

[nocturne quest]

Me too, but It should be configured From what I saw

The .ini and .log shows that it is "initiatted"

[next junco]

8ms its huge where you find it or did u test yourself ?

[nocturne quest]

Saw it here in this thread, I just saw someone showing some tests, let me see if I find it...

[nocturne quest]

@next junco this post, bazoka

Like I said before, I didn't noticed any notable improvements using the SK uploaded by the OP, and since we don't have a Tool to show/see exactly, we can only test for ourselfs

[late compass]

But like i and Orwénnes said, it doesn't improve latency anymore for both of us with the current LS version, it even makes latency worse, so no point in using it anymore, if you're using the current LS version.

[nocturne quest]

Oh, good to know, thanks
Reading this thread, it was not clear if you guys were refering to newer SK or LLS, thanks again.

[late compass]

With the LS version BEFORE the Adaptive Frame Generation (AFG) mode was introduced, the SK method reduced latency by around 6ms for me, but with the new version with AFG, even when using fixed x2 multiplier using SK makes latency the same or worse as without SK.

[nocturne quest]

I see, that's awesome, I didn't like to use SK that way anyway xP

[next junco]

And old ls vs new how much ms u gain can u give numbers

[late compass]

I wonder if @lucid plank looked into this, why using Special K + fakenvapi + LatencyFleX injected into LS consistently reduced latency when using LSFG up to and including version 2.13.2 Released | 3.0.0.2 Beta? (And not anymore with the new 3.1 version.)

[late compass]

new LS version 3.1 latency is the same as old version without special K, so we now have higher latency than the version before, as we cant reduce it with special k anymore.

[next junco]

And some guy said disabling multicores reduce latency about 3ms !? did he mean smt or real cores , and if I disable them in task manager give the same benifits vs bios or no 🤔

And other thread op said that special k by enabling hardware independent flip reduce latency by a lot in dxgi maybe somebody can try confirm those tweaks

And this help too but break gsync

https://www.reddit.com/r/losslessscaling/comments/1j0vcxc/lower_latency/?share_id=7hbYKbId6ncqHKZM62OuG&utm_content=1&utm_medium=android_app&utm_name=androidcss&utm_source=share&utm_term=1

[next junco]

Weird stuff

[late compass]

I tried the "Tiled Display Topology" thing, but i have no clue if it's even working on my TV, don't notice/measure any difference. There are so many conflicting comments about how to even get it to work and no clear way to test if it's working.

[next junco]

https://www.reddit.com/r/OLED_Gaming/comments/1j6eidw/guide_changing_to_tiled_display_topology_to/

Best guide it work on borderless and window as full-screen perfect for ls 👌 without extra buffer just ome frame for screen ( kind direct communication or real time displaying but may introduce more tearing or gsync not working try and see at lower fps like 25 30 see if there moticible difference one guy try it in c1 and confirm the improvement and its same rendering that vr hardware use as default rendering mmethode in win 11 😀

#🛸│testing-chat message
one person try it camera slow motion light of keyboard 4 time with it 3.5 so about 12% in total unkbown latency 60fps 17ms reduction, work only in win 11

https://youtu.be/yq4zXrIJm44?feature=shared

[late compass]

I used this guide, but as i said, don't have any indicator that it's enabled and working. No difference in latency measurements, no tearing, gsync works as usual. I'm on latest win 11.

[next junco]

Did you get rid of black screen when do full screen as guide