Input Lag Being Ignored - Learn From Your Competitors

[ThorCoolguy]

A couple weeks from release now and the tone from iniBuilds continues to be that input lag doesn't exist and the flight model is perfect.

 

The flight model may or may not be great; I can't tell, because I'm still getting a delay of 1/2-1 second between stick deflection and aileron response. Yes yes yes, I know, it's a big aircraft, I get how physics works. There is a delay between a roll command and the *start* of aileron movement. That has nothing to do with the mass of the aircraft or even its roll rate or sensitivity. It's a bug.

 

So far the response from iniBuilds has been some variant of "git gud" - the plane is sensitive, you just don't know what you're doing. That's very frustrating.

 

Just to light a bit of a fire, it's worth noting that one of your competitors had a similar issue with input lag. They not only took it seriously (and fixed it), they wrote this about the problem:

 

"A while ago some of our customers complained about input lag and sent in some tickets. We did a first round investigation and observed a measurable lag between a control input and the sidestick animation - but seeing as the animation is not tied to the flight controls, it wasn't a concern. The tickets did not, however, stop - and so we took it to the second round where we started building in some diagnostics and trying to figure out what was going on. And boy, did we find out. For a not insignificant number of users, we found the time it took for the aircraft to calculate and translate control values could be as much as 300 ms. Whilst this didn't affect everyone, it was unacceptable."

 

Please start taking the input lag problem seriously. It's real.

[ThorCoolguy]

Since the last post got absolutely no response from iniBuilds, here's video proof.

https://youtu.be/O6xQR9dC9hE

As you can see, there is a significant lag between making control inputs and a reaction from the plane. There is some variation, but generally there is about .33 seconds of lag between IRL stick input and stick response in the sim, and there is about a .5 second delay between IRL stick input and control surface response in the sim.

 

Again, this has nothing to do with roll rate or the weight of the aircraft. There is a delay between IRL control input and the sim aircraft *beginning* to issue its control command. No real-life aircraft would ever be allowed to fly with a .5s delay between a control input and the *beginning* of a control command.

 

.5 seconds is absolutely noticeable. I believe this problem is the root of many of the handling characteristics complaints that customers have been reporting.

 

My system:

MSFS2024 (recorded with the release SU1 - but the problem existed in SU1 beta as well)

Windows 11 12700K RTX3080 64GB RAM

Virpil Constellation Alpha + Saitek X55 Rhino throttle (not recorded, but the same input lag problem occurs using an Xbox 360 controller for pitch/roll control)

A350 1.0.4 (but the problem has been occurring since release)

 

Edit: Fixed link.

Edited March 17 by ThorCoolguy
Fixed link.

[haza2]

I will report back the feedback you have given. Thank you.

 

[ThorCoolguy]

Still not fixed, and the recent dev stream doubles down on "git gud."

I made another video with even better evidence of the problem.

https://youtu.be/hcrJAUncEps

I used a high-speed camera and an Xbox 360 controller this time (because the Xbox controller return-to-center is so fast).

Conclusions:

1. It's not a problem with my stick.

2. It's not a problem with USB hubs. The only peripherals installed while running this test were a mouse, keyboard, and the controller, all plugged directly into the mobo.

3. Again, I understand that it takes time for control surfaces to *fully* deflect and I understand that because physics a large aircraft will take time to *complete* a maneuver. That's not the issue. The issue is that if I issue a control input, the control surfaces don't even *start* to move until up to .6s later. This is demonstrably proven in the video.

4. Despite arguments that this is correct to the real aircraft because it takes time for the PRIMs to compute control surface commands, consider the stick movement. In the real aircraft, the stick moves when it moves. Logically, "waiting for the stick to move" will never be part of any command delay. Here we see "waiting for the stick to move" delays of up to .6s.

5. Transport category aircraft input lag is measured as part of the certification process. FAA calls it "phase lag" and transport aircraft are expected to have Level 1 phase lag, which is not more than .1 seconds. I would be extremely surprised if the phase lag on the real A350 is anything higher than .1 seconds, but even if it is, see #4 above - that real-world phase lag is being added to the demonstrated sim input lag.

6. I didn't bother uploading them, but I recorded tests in MSFS 2020 that showed about the same input lag.

It is obvious that the problem is real and that instead of addressing it they are blaming the user. That is, again, disappointing.

[BillyBobJoe0987]

Hey, I've got a couple questions for you since I've been trying to replicate this but don't have nearly as much delay:

1. Are you using any sort of frame gen? The built in one, frame gen mod, lossless scaling, etc. These all add input lag and may be part of the problem.

2. Have you tried changing the graphics settings and testing again? Lower frame rates will have more input lag although not as much as you are experiencing.

3. Could you try these tests in some other aircraft? Some other Airbus aircraft and some from other companies.

All of the controls seem to visually move correctly for me although the controls do feel slightly more delayed than other aircraft. Inibuilds may genuinely not be able to replicate these issues because of different hardware or controls, or it may be how the real aircraft acts since it is easy to compensate, just move your stick to stop a roll slightly earlier than expected. The A350 does a lot to try and make the flight feel as smooth as possible for passengers so it may limit the roll acceleration/deceleration making it take a while to stop a high roll rate and feeling like input lag.

 

Also as a side note, neither the FAA nor EASA have input latency requirements for commercial aircraft. The "Phase Lag" you talk about is a phenomena found in helicopters. Now, if the controls were bad enough to cause pilot induced oscillations (PIO) it would not be certified. Several people have reported experiencing PIO's in this aircraft but for whatever reason I've not experience this. It may just be control settings or overcontrolling of the plane causing this.

[IniSteve]

3 hours ago, ThorCoolguy said:

Still not fixed, and the recent dev stream doubles down on "git gud."

I made another video with even better evidence of the problem.

https://youtu.be/hcrJAUncEps

I used a high-speed camera and an Xbox 360 controller this time (because the Xbox controller return-to-center is so fast).

Conclusions:

1. It's not a problem with my stick.

2. It's not a problem with USB hubs. The only peripherals installed while running this test were a mouse, keyboard, and the controller, all plugged directly into the mobo.

3. Again, I understand that it takes time for control surfaces to *fully* deflect and I understand that because physics a large aircraft will take time to *complete* a maneuver. That's not the issue. The issue is that if I issue a control input, the control surfaces don't even *start* to move until up to .6s later. This is demonstrably proven in the video.

4. Despite arguments that this is correct to the real aircraft because it takes time for the PRIMs to compute control surface commands, consider the stick movement. In the real aircraft, the stick moves when it moves. Logically, "waiting for the stick to move" will never be part of any command delay. Here we see "waiting for the stick to move" delays of up to .6s.

5. Transport category aircraft input lag is measured as part of the certification process. FAA calls it "phase lag" and transport aircraft are expected to have Level 1 phase lag, which is not more than .1 seconds. I would be extremely surprised if the phase lag on the real A350 is anything higher than .1 seconds, but even if it is, see #4 above - that real-world phase lag is being added to the demonstrated sim input lag.

6. I didn't bother uploading them, but I recorded tests in MSFS 2020 that showed about the same input lag.

It is obvious that the problem is real and that instead of addressing it they are blaming the user. That is, again, disappointing.

Here is raw video from the sim with input windows shown
For 1:1 comparison - No FG, No vsync, sustained fps of >40.
Joystick is a TM warthog with zero input curves and no null zone. 

As you can see there is no discernable input lag. 

[Michael Wallace]

4 minutes ago, IniSteve said:

Here is raw video from the sim with input windows shown
For 1:1 comparison - No FG, No vsync, sustained fps of >40.
Joystick is a TM warthog with zero input curves and no null zone. 

As you can see there is no discernable input lag. 

 

I'm really not sure how this post is relevant. If it's fine for you, it must be fine for all?

[IniSteve]

10 minutes ago, Michael Wallace said:

I'm really not sure how this post is relevant. If it's fine for you, it must be fine for all?

Michael, the relevance is I am showing the user who is having the issue results from the simulator without any factors that may be inducing lag. 
I'm not sure why you feel need to make assumptions on my intention or jump to the conclusion that we feel there isn't an issue. 

[BillyBobJoe0987]

2 hours ago, Michael Wallace said:

I'm really not sure how this post is relevant. If it's fine for you, it must be fine for all?

This post is trying to show that the plane controls are moving as quick as they are receiving the input from the game. Any additional input lag comes from some other source such as vsync, low fps, USB hubs, frame gen, etc.

[geckuz]

So, to share my personal experience on the topic. I previously had an older CPU and low FPS, and input lag was a nightmare on this plane. Now I moved to an i9 14900k and raised settings just a bit, I have good frames and no more input lag.

[ThorCoolguy]

Steve was good enough to reach out to me on Discord as well, and I provided him some extra info there.

I think folks are still missing the point, though. It's clear that not everyone is experiencing this issue. I never claimed that *everyone* had an input lag of up to .6s. Obviously it runs fine for plenty of people.

But on some systems and some configurations there is clearly an issue, and I'm not the only one experiencing it. Numerous users on Discord and Reddit have reported the same thing, and I continue to think that at least some, if not many, of the "handling sucks" complaints are a result of widespread input lag issues.

I don't have this problem with any other aircraft. Frame gen, DLSS, and Vsync are all off. No USB hubs, no FSUIPC, no funky controller software, various controllers tried (Xbox 360, Saitek X56, Virpil Constellation Alpha). My system is not bleeding edge but it runs the sim fine: Windows 11 12700K RTX3080 64GB RAM and a 980 Pro for the FS install. It's also noteworthy that I get the same issue in FS2020 - I don't know anything about the backend of how the sim ingests control inputs and feeds them to WASM aircraft, but whatever the issue is, it's not isolated to 24.

Also note that devmode FPS counter in FS24 actually shows main thread delay and rendering thread frame times - both are at about 20-40 ms. In the A350 I get 45-60 FPS - yes, that number could be higher, but that is not going to cause a .6s input lag. For reference - frame time at 45 FPS is 22 ms, or .02 s. This is less than 5% of the lag I demonstrate in the video. Add to the main thread time of 30ms and worst case scenario we are talking 50-70 ms - not 600 ms.

I generally run FS24 on a mixture of medium and high settings. Just for fun, I bumped all settings to low and got the same input lag, so I really don't think it's a frame rate issue.

Edited Friday at 11:24 PM by ThorCoolguy

[EpixXx]

I can confirm that i also got these input lags and dont experience it on other Planes. And i run the sim at average 60-80 FPS withouth Frame Gen or Vsync. Specs: 13900k and RTX 4080S with 64gb DDR5 RAM.