I don't care if my frames are real or fake, I'll take 90-120 FPS with my 4090 anyday. It is NOT just a "marketing tech". This has been the single biggest advancement in FPS (and thus performance/experience) ever since I started simming in the 90s as a kid.

FG diminishes in quality greatly as you approach 30 fps. Ideally you'd have 45-60. OP had <= 30 in his video, not surprising there's a bunch of weirdness. I haven't seen ANY of that. In fact the only FG artifacts I ever see is on spinning turbines, and some models like the Fenix A320 don't seem to cause the issue at all.

This video by LTT does a good job of demonstrating the strengths and weaknesses of FG. 

 

OP: Do you use gsync with your tvs/monitors? I wonder if that may be sparing me these kinds of artifacts.

Edited April 20, 20233 yr by kiwikat

I freaking love FG - I use Track IR so I'm constantly moving my view around.  In frame gen and TAA mode everything is beautiful!

Triple digits or high double digits in heavy payware is a reality.

Well its sounds strange to me that that problems occur with GPU frame generation at 30fps since at that fps you gonna get 60fps on screen right? I can understand that perhaps it possible that something north of 60fps might mater in some gaming application for some people but definitely not in Flight sim unless you are flying at supersonic speeds at low altitude. Personally, I believe that anything north of 60fps is more about bragging rights than anything else. I've no doubt that the 4090 is a great card and FR is great feature to have in the card. I have my system tuned for over 30fps for flying GA and over 24fps flying Airliners in and out of international airports in big cities and I have everything in the graphics setting menu in MSFS set to the highest setting. I see 60fps on the screen for the former and 48fps on the screen for latter thanks to judder reduction setting on my Samsung TV. It is virtually anomaly free. But look I get it the OPs anomaly is worse than anything I get with my TV frame interpolation. I'm running a i9 10900K at 5.2ghz with 3090 and I'll say that what I see on screen when the GPU is delivering 30fps at 30hz is far inferior to the 60fps at 30hz that I get with judder reduction on. And in the case of 50fps at 50hz which I use for racing games like Codemasters F1 I can get 100fps from that 50hz signal using the TVs Judder reduction setting and it looks a lot better than 60fps at 60hz. At that speed although my reflexes are not what they used to be I'm not aware of any input lag. Heck at 24hz I'm not aware of any input lag unless I go looking for it. I don't doubt that the 4000 series with its FG is a great card but don't be fooled Nvidia could do that with the 3000 series and 2000 series GPU's too if they wanted too. Heck if a samsung TV could it 10 years+ ago so could Nvidia. They had to do it this around because without it the best possible performance increment they could achieve over the 3000 series wouldn't be enough to persuade 3000 series GPU owners to take the hit (and that's what it is) for 4000 series card. Look I'm not saying there is now progress but look at what the marketing is telling you and the progress is a small fraction of that. Maybe in the region of 1/10. I've been messing with cutting edge PC Hardware for at least couple of decades. I've pumped near freezing cold water though water blocks on 3 GTX 680 in SLI and overclocked that word not allowed out of them. I've put a bleeding edge PC in an ambient temperature of -20C and ran a i7 4770K stable at 5ghz.  Two common refrains come to mind. Fool me once shame on you, Fool me twice shame on me. And, a fool and their money are easily parted. The bottom line is that I doubt there is a great deal of difference between Nvidia FG and Samsung FI.

 

Edited April 20, 20233 yr by FBW737

Fake frames in MSFS are pointless.  You can't convince me of otherwise because I don't beleive in the tooth fairey,

sp

 

Edited April 20, 20233 yr by Sky_Pilot071

6 minutes ago, Sky_Pilot071 said:

Fake frames in MSFS are pointless.

sp

 

I disagree.  40 fps vs 80 when using head tracking and the difference is night and day.

Even without...a bumpy final approach in the 737 etc is vastly nicer with frame gen enabled.

With a GSYNC monitor there's zero comparison between 60 and 120 FPS. The experience with FG is considerably better. Sorry you can't experience it yet.

46 minutes ago, FBW737 said:

Well its sounds strange to me that that problems occur with GPU frame generation at 30fps since at that fps you gonna get 60fps on screen right?

Think about it for more than 5 seconds and it'll make sense. There's a much bigger gap between two frames at 30 FPS and 60 FPS. If you are interpolating a frame between the two, it will be a lot easier to make something consistent with movement on the screen when the frames are closer together than farther apart.

46 minutes ago, FBW737 said:

I can understand that perhaps it possible that something north of 60fps might mater in some gaming application for some people but definitely not in Flight sim unless you are flying at supersonic speeds at low altitude. Personally, I believe that anything north of 60fps is more about bragging rights than anything else. I've no doubt that the 4090 is a great card and FR is great feature to have in the card.

This is your personal preference and has no bearing on other peoples' perception of reality. The difference between 60 FPS and 120 FPS on my 120 Hz GSYNC monitor is night and day, on Windows, MS Flight Sim, Overwatch 2, or any other thing I'm running. There's a reason high refresh rate devices are the norm now, even phones are going that direction.

46 minutes ago, FBW737 said:

I have my system tuned for over 30fps for flying GA and over 24fps flying Airliners in and out of international airports in big cities and I have everything in the graphics setting menu in MSFS set to the highest setting.

I am sorry you have to settle for 24 or 30 FPS in 2023. I'm maxed out at 120 FPS @ 4K.

46 minutes ago, FBW737 said:

I see 60fps on the screen for the former and 48fps on the screen for latter thanks to judder reduction setting on my Samsung TV. It is virtually anomaly free. But look I get it the OPs anomaly is worse than anything I get with my TV frame interpolation. I'm running a i9 10900K at 5.2ghz with 3090 and I'll say that what I see on screen when the GPU is delivering 30fps at 30hz is far inferior to the 60fps at 30hz that I get with judder reduction on. And in the case of 50fps at 50hz which I use for racing games like Codemasters F1 I can get 100fps from that 50hz signal using the TVs Judder reduction setting and it looks a lot better than 60fps at 60hz. At that speed although my reflexes are not what they used to be I'm not aware of any input lag. Heck at 24hz I'm not aware of any input lag unless I go looking for it.

I don't know if your TV's frame interpolation is anything like the 4000 series' FG tech. I personally don't care about the technology, only the results. Input lag has never been the problem FG is trying to solve, in fact it tends to increase input lag. But that doesn't matter very much in a simulator where things are moving relatively slowly and you don't need shooter-like reflexes. This is like the PERFECT application for frame generation as a result.

46 minutes ago, FBW737 said:

I don't doubt that the 4000 series with its FG is a great card but don't be fooled Nvidia could do that with the 3000 series and 2000 series GPU's too if they wanted too. Heck if a samsung TV could it 10 years+ ago so could Nvidia. They had to do it this around because without it the best possible performance increment they could achieve over the 3000 series wouldn't be enough to persuade 3000 series GPU owners to take the hit (and that's what it is) for 4000 series card. Look I'm not saying there is now progress but look at what the marketing is telling you and the progress is a small fraction of that. Maybe in the region of 1/10. I've been messing with cutting edge PC Hardware for at least couple of decades. I've pumped near freezing cold water though water blocks on 3 GTX 680 in SLI and overclocked that word not allowed out of them. I've put a bleeding edge PC in an ambient temperature of -20C and ran a i7 4770K stable at 5ghz.  Two common refrains come to mind. Fool me once shame on you, Fool me twice shame on me. And, a fool and their money are easily parted. The bottom line is that I doubt there is a great deal of difference between Nvidia FG and Samsung FI.

A quick google search indicates that these technologies are not exactly the same, and that FG has lower input latency as a result of the extra processing done predicting movement. The benchmarks all show that the 4090 absolutely stomps the 3090/Ti even with FG off in basically every metric, including value, so I am not sure where you're getting the idea there hasn't been much progress made. This reads more like a bunch of 3000 series owners are salty they paid inflated prices during the pandemic compared to cards that are getting far superior performance and providing a much better experience.

I was one of those salty 3080 owners who was disappointed with performance on my 4K/120Hz OLED monitor. The 4090 works wonders with it.

Edited April 20, 20233 yr by kiwikat
dumb spacing

24 minutes ago, kiwikat said:

Think about it for more than 5 seconds and it'll make sense. There's a much bigger gap between two frames at 30 FPS and 60 FPS. If you are interpolating a frame between the two, it will be a lot easier to make something consistent with movement on the screen when the frames are closer together than farther apart.

This is your personal preference and has no bearing on other peoples' perception of reality. The difference between 60 FPS and 120 FPS on my 120 Hz GSYNC monitor is night and day, on Windows, MS Flight Sim, Overwatch 2, or any other thing I'm running. There's a reason high refresh rate devices are the norm now, even phones are going that direction.

I am sorry you have to settle for 24 or 30 FPS in 2023. I'm maxed out at 120 FPS @ 4K.

I don't know if your TV's frame interpolation is anything like the 4000 series' FG tech. I personally don't care about the technology, only the results. Input lag has never been the problem FG is trying to solve, in fact it tends to increase input lag. But that doesn't matter very much in a simulator where things are moving relatively slowly and you don't need shooter-like reflexes. This is like the PERFECT application for frame generation as a result.

A quick google search indicates that these technologies are not exactly the same, and that FG has lower input latency as a result of the extra processing done predicting movement. The benchmarks all show that the 4090 absolutely stomps the 3090/Ti even with FG off in basically every metric, including value, so I am not sure where you're getting the idea there hasn't been much progress made. This reads more like a bunch of 3000 series owners are salty they paid inflated prices during the pandemic compared to cards that are getting far superior performance and providing a much better experience.

I was one of those salty 3080 owners who was disappointed with performance on my 4K/120Hz OLED monitor. The 4090 works wonders with it.

Let me be blunt. Your full of it.😜

13 minutes ago, FBW737 said:

Let me be blunt. Your full of it.😜

If you're going to insult someone, at least do it with proper grammar.

1 hour ago, FBW737 said:

Let me be blunt. Your full of it.😜

Nice rebuttal. My full of it!

Frame rate is only one component of visual 3D animation quality and image quality.  It comes into play mostly during rapid panning and when objects are moving rapidly relative to the pilot's view, for example during takeoff while looking out a side window.  In the course of a 2h flight.  Once in the air it's down to rapid panning only, when you will clearly notice a difference in panning quality for obvious reasons.  In the end it's arguably playing a minor part in the total performance experience, which also should incorporate image quality and freedom from graphic anomalies.  What matters all of the time from gate to gate is the perception of smooth animation and freedom from major and micro stutters, and this only correlates marginally with high frame rates, no matter the method of creation of those high frame rates.  More critical to this larger part of total performance is ultra low frame time variance, not frame rate per se.  This is what on the ground and until up a few thousand feet I will hit the * key to unlock the frame lock I use.   After that locked at 34 with this kind of FTV equates to liquid smooth animation, without undo and unnecesssary power use, and leaving headroom in CPU/GPU to cope w/ transients spikes in load.  No interest in FG at this point in MSFS development.

3 minutes ago, Noel said:

Frame rate is only one component of visual 3D animation quality and image quality...

What matters all of the time from gate to gate is the perception of smooth animation and freedom from major and micro stutters...

Agreed, and that's why I enjoy 4K at 120 FPS with TAA. The best of all worlds.

Fake frames = testosterone buzz IMO.  Not to mention more GPU sales from a marketing standpount.

If you can't cut it with a true fps count then you can't cut it.

sp

1 minute ago, Sky_Pilot071 said:

Fake frames = testosterone buzz IMO.  Not to mention more GPU sales from a marketing standpount.

If you can't cut it with a true fps count then you can't cut it.

sp

Cool opinion bro. Come back when you can defend it with evidence.

12 hours ago, kiwikat said:

Agreed, and that's why I enjoy 4K at 120 FPS with TAA. The best of all worlds.

Sure, $1500+ covers rapid panning better for the couple of times if that I might rapidly pan during a 2h flight, and near field objects during take off but only if you're looking out a side window which you really shouldn't be doing a whole lot of anyway.  After that, you're not seeing anything better than what 34 fps does with this type of FTV.  

Run CapFrameX and do a capture at 120 yourself and post it right here it would be fun to see. Brute force does help obscure poorer FTV....