Hi All,

Hoping somebody might be able to help. Whenever I enable frame generation I get these strange black ghosting time artifacts on the edges of the screen. I've tried changing every setting I can think of both in MSFS and NVIDIA control panel but can't for the life of me find what might be causing it. I've tried changing monitors (three different types/brands), changing cables, three different drivers, VYSNC on off, changing refresh rates...nothing seems to fix it. Only happens with FG ON. if I turn FG off it goes away. 

 

The video below demonstrates - In this I'm panning quickly but its present even with a slow pan. Any ideas?

 

 

Cheers,

 

Edited April 19, 20233 yr by KL Oo

Unfortunately the only downside of FG. 
Nothing you can do about this. 
You see it more when the contrast is higher like a blue sky and the tail of the aircraft when panning around it.  

interpolation at play = visual glitch…

Does running MSFS in windowed or fullscreen mode has any effect on this?

Frame Generation has its limitations unfortunately. 
It simply tries to guess the next frame based on the current frame and any current direction of movement it can see on screen.

It is fine at slower speeds and smaller movements, but when you jump around erratically or move more quickly, when a significant portion of the displayed screen can change, it doesn't have much to go on, as it is just trying to guess what the in-between frame should look like.

If you can hold 60 FPS normally, then as it doubles to 120 FPS with FG, the artefacts may not be as noticeable, but at lower frame rates they will be, and then again, if you can hold higher FPS, I can't see the point of FG anyway.

I got shot down for playing down the importance of FG a few months ago, but I stand by what I say, and there is no substitute for raw horse power.
I am looking to purchase a 4090 myself at the next decent price drop, but this will be simply so I can hold 60 FPS solid at TAA (without FG).
I can do this now with my 3080Ti in most situations, but with DLSS balanced mode, which has its own limitations.

I see no real need to move the camera around that quickly (as shown in your video) anyway.
So with FG, you will just need to get used to it and move the camera more slowly. 
FG is a nice thing to have, but it can't work miracles with fast movement.

 

Edited April 19, 20233 yr by bobcat999

7 minutes ago, AnkH said:

Does running MSFS in windowed or fullscreen mode has any effect on this?

I would say not significantly, no.  I can't see a reason why it should have any significant impact.

The difference between running in these different window modes is not noticeable now, at least in W11.

Of course, given the choice, I would always run in full screen, but I have not noticed any FPS gain between the two for example.

Edited April 19, 20233 yr by bobcat999

5 minutes ago, bobcat999 said:

I can't see a reason why it should have any significant impact.

My guess was: it might be that in windowed mode, a single black pixel line around the MSFS window might be already enough to compromise FG during fast pacing lookarounds. Fact is, there are videos of fast pacing lookarounds on YT from people using FG and they do NOT show such artifacts...

10 minutes ago, AnkH said:

My guess was: it might be that in windowed mode, a single black pixel line around the MSFS window might be already enough to compromise FG during fast pacing lookarounds. Fact is, there are videos of fast pacing lookarounds on YT from people using FG and they do NOT show such artifacts...

Yes, that is right - I have seen some also. The thing is, we don't know what their original frame rate was compared to the OP.
It seems to be quite low (mid 20's? - difficult for me to see).  Obviously the FPS counter isn't taking FG into account on the video (or then again, is it?).

Hopefully someone with a 40 series card can confirm this for us.  Might be worth checking at least.

Edited April 19, 20233 yr by bobcat999

Do you have Geforce experience?

If so, try its suggested settings for your setup, and see if there is any improvement.

3 hours ago, bobcat999 said:

Yes, that is right - I have seen some also. The thing is, we don't know what their original frame rate was compared to the OP.
It seems to be quite low (mid 20's? - difficult for me to see).  Obviously the FPS counter isn't taking FG into account on the video (or then again, is it?).

Hopefully someone with a 40 series card can confirm this for us.  Might be worth checking at least.

Not sure about that alt+z add-on but RTSS does account for FG when it displays FPS.

Not sure what FPS is like with three 4k screens on a 4090 but what is the raw rasterization FPS without FG? The downfalls of FG are more apparent the lower FPS you go. The eye is going to catch less of it if you're getting like 100+FPS and 50-70 are actual rendered frames. You get down to 30FPS with FG enabled and there just isn't enough "good" frames generated to "hide" this kind of thing.

Yeah another video of FG artifacts that convinces me that the FG provided by 4000 series GPU's is hardy any different from the FG provided by Samsung 4K TV. 🤣

27 minutes ago, FBW737 said:

Yeah another video of FG artifacts that convinces me that the FG provided by 4000 series GPU's is hardy any different from the FG provided by Samsung 4K TV. 🤣

I've seen some of those videos.

And after actually using FG, my take is that many of the derogatory videos are what another Youtuber calls "Pixel Peepers"; people who blow up images to huge sizes to showcase anomalies, or freeze frame through things to find momentary interpolation artifacts barely possible for the human eye to detect in actual motion.

In other words, there are indications of an agenda.

For me, its just another tool in the arsenal, sometimes more useful than others, and subject to improvement.

 

 

Edited April 19, 20233 yr by HiFlyer

34 minutes ago, FBW737 said:

Yeah another video of FG artifacts that convinces me that the FG provided by 4000 series GPU's is hardy any different from the FG provided by Samsung 4K TV. 🤣

 

22 minutes ago, HiFlyer said:

I've seen some of those videos.

And after actually using FG, my take is that many of the derogatory videos are what another Youtuber calls "Pixel Peepers"; people who blow up images to huge sizes to showcase anomalies, or freeze frame through things to find momentary interpolation artifacts barely possible for the human eye to detect in actual motion.

In other words, there are indications of an agenda.

For me, its just another tool in the arsenal, sometimes more useful than others, and subject to improvement.

Yeah, are we talking about frame interpolation that's been on like every TV for the past decade or is this something new that Samsung is doing? A TV's motion smoothing and FG are both going to look like word not allowed at 30 FPS because, yeah they are basically doing the same thing. Trying to generate a fill-in frame with lots of motion between two rasterized frames at only 30 FPS is never going to be accurate. You need more true frames for accurate interpolation and less time between frames for the artifacts to remain unnoticeable. Historically motion smoothing, or whatever each TV manufacture calls their version, was a horrible solution for gaming because it created loads of input lag (especially at high refresh rates) because the hardware being used to create these frames isn't quick. The 4-series cards having this implemented can create these frames much faster so the technology is actually useful at much higher refresh rates with way less impact on input lag. Frame Generation has it's place in PC gaming.... a flight simulator at 30 FPS is not it.

Hi Guys - OP here

 

10 hours ago, AnkH said:

Does running MSFS in windowed or fullscreen mode has any effect on this?

No, unfortunately it makes no difference - those black streeks and other artifacts (I get them around the engines of AI Traffic aircraft for some reason) are there regardless of window'd or full screen mode.

In this particularly case with the PMDG and FSDT KORD I'm CPU limited and the sim is running at around 30 FPS (both as per SHIFT-Z stats and the MSFS DevMode FPS Counter that I didn't have on at the time) natively - so I'd suspect the FG frames are closer to 50 or so. For what its worth, I'm actually running a triple monitor setup which is why the FPS are so low, but the problem persists irrespective of triple monitor or single monitor. No difference there.

6 hours ago, vrdubin6 said:

Not sure about that alt+z add-on but RTSS does account for FG when it displays FPS.

Not sure what FPS is like with three 4k screens on a 4090 but what is the raw rasterization FPS without FG? The downfalls of FG are more apparent the lower FPS you go. The eye is going to catch less of it if you're getting like 100+FPS and 50-70 are actual rendered frames. You get down to 30FPS with FG enabled and there just isn't enough "good" frames generated to "hide" this kind of thing.

I think you're right. When I turn off the side monitors and run just the main monitor in a situation like the one above at KORD and PMDG I'd maintain close to 70 FPS without frame generation. Turning on FG then introduces the artifacts you see, but they aren't quite as noticeable and pronounced. It doesn't remove them though.

I guess I'm just going to have to deal with it. Its annoying that this is such a power tool in the toolbox thats just not quite there yet. With any luck NVIDIA and/or Asobo can improve on it over time. 

In the meantime, if anybody finds something that helps reduce it I'm all ears!

For what its worth as well, frame generation does give me some weird CDT issues. For example if I'm using FG I often get a crash to desktop when closing the flight and returning to the main menu. RTSS also doesn't play nice when I'm using multi monitors and will often result in the now famous "your graphics card has overheated" message.

I guess its all a work in progress with this newer tech!

 

 

 

Im not a FG naysayer but am unsure why spending $$ on a marketing tech will put me in a better position than 80 fps with my 3060TI.  Since FG will give me oodles of false frames I still don't see the point.

Correct me if I'm wrong but real frames do not equal false frames.  

sp