MULTI Linus: Dualshock 4 "is the BEST game controller"

[CerebralTiger]

That ~180ms measurement is for "readying the weapon" which takes into account the animation priority.

It's until the animation startup (see measurement). For the shooting tests, he waited until gun discharge, therefore included the startup animation. The first person latency profile has less prioritized animations, hence the noteworthy drop in latency when pushing higher frame rates.

 

[Necrokiller]

So if you're looking for the exact input latency and not input latency + animation priority, the best way to test that in a shooter is the shoot button.

RDR2, running at 60fps, feels very responsive using this method. ¯\_(ツ)_/¯

 

[CerebralTiger]

RDR2, running at 60fps, feels very responsive using this method. ¯\_(ツ)_/¯

Based on Alex's test, running at 60fps still results in a baseline latency of 181ms (not counting animations), which indicates there is likely some induced latency to ensure animations get enough of a window to startup, as he notes as well. What reading did you get?

 

[Necrokiller]

Based on Alex's test, running at 60fps still results in a baseline latency of 181ms (not counting animations), which indicates there is likely some induced latency to ensure animations get enough of a window to play out, as he notes as well. What reading did you get?

Alex did not use the test case that you suggested is the best. And I just carried it out as well. I will average out and post my findings. Safe to say though, that it's nowhere near 181ms.

 

[CerebralTiger]

Alex did not use the test case that you suggested is the best.

That was literally his first test. For his other tests, he mentioned taking animations into account, as he was testing the time taken for the action to be performed. Test 2 and 3 were input latency + animation frames.

Also, I did not say the approach in test 1 is best. It depends on what you're trying to find out. Are you looking to find input latency or input latency + animation priority? It depends on that.

 

[Necrokiller]

That was literally his first test.

Hes using the LT button to start up the aiming animation. "Weapon draw"

I'm testing the input lag from RT trigger depress to gun discharge. You know, the shoot button.

So if you're looking for the exact input latency and not input latency + animation priority, the best way to test that in a shooter is the shoot button.

 

[CerebralTiger]

Hes using the LT button to start up the aiming animation. "Weapon draw"

I'm testing the input lag from RT trigger depress to gun discharge. You know, the shoot button.

You should also test the LT-to-startup-aiming animation to see if you get the same result as him at 30 or 60fps.

Also, for the shooting test, remap the input to the shoulder button (RB) for more accurate results. Trigger input leads to reading errors.

Edit: yeah, this is inaccurate.

There's no telling at what threshold the trigger accepts your input. The gunfire animation begins 2 frames before you even fully press it.

 

[Necrokiller]

You should also test the LT-to-startup-aiming animation to see if you get the same result as him at 30 or 60fps.

Also, for the shooting test, remap the input to the shoulder button (RB) for more accurate results. Trigger input leads to reading errors.

Edit: yeah, this is inaccurate.

There's no telling at what threshold the trigger takes your input. The gun fires before you even fully press the trigger.

But this limitation is true for anyone who tests using the controller. There is no indication whether NX gamer or DF use a different method.

But I realized this and tested with my mechanical keyboard, using its back light mechanism to guide when the key was pressed. Mapping both LT and RT to keyboard buttons. This is by far the most accurate method.

The "weapon draw" reading I got was ~125ms.
Shooting while already aiming with weapon = ~116ms

This is using triple buffering and vsync from NV control panel. Lower display lag (monitor) and keyboard will also account for lower reading. So yeah, actual input lag is much lower than people realize.

I'm now really curious to find out how TLOU2 performs under these similar tests (even though the trigger threshold problem will still be there).

 

[Chandoo]

How did a thread about dualshock 4 turn into a game engine response time thesis (lmao)

I'll only contribute this regarding TLoU 2, via Eurogamer.

The Last of Us Part 2 is a more responsive game than Naughty Dog's previous works

Also, God of War (2018) has a performance mode which will have big improvements over the input latency on graphics mode if it bothers anyone that much.

 

[Necrokiller]

Also, God of War (2018) has a performance mode which will have big improvements over the input latency on graphics mode if it bothers anyone that much.

lmao, I should check this on my base PS4 again. You know, the console that the vast majority owns.

 

[Chandoo]

le bhai thats like me saying how my laptop should be used as an indicator for all PC gaming regardless of who has what machine

God of War has an option for better latency if the base one feels bad to you, games like RDR2 don't at all regardless of platform.

 

[Necrokiller]

Your laptop should be used as an indicator for anyone who has similar specs. Yes.


You're always championing how stable 30fps is better than variable 30-60fps. Because God of War definitely doesn't run at a stable 60fps in performance mode.

 

[Chandoo]

I'm also never complaining about input latency just posting relevant pointers to the discussion going on

 

[Necrokiller]

I'm also never complaining about input latency just posting relevant pointers to the discussion going on

Your analogy was terrible and made in haste. No need to stick with it

 

[CerebralTiger]

But this limitation is true for anyone who tests using the controller. There is no indication whether NX gamer or DF use a different method.

Buttons can be re-mapped from the PS4 OS. Plus RDR2 might have an in-game option to remap aim/shoot to L1/R1. Can't remember if it did. Either way, this is a non-issue. DF and NX gamer were most likely aware of this and tested accordingly.

The "weapon draw" reading I got was ~125ms.
Shooting while already aiming with weapon = ~116ms

Did you make gifs? I'd like to see them.

So yeah, actual input lag is much lower than people realize.

Yeah, well, it's known that animation priority is the main culprit in the case of RDR2, and Rockstar's stubbornness when it comes to running animations in their full "glory".

How did a thread about dualshock 4 turn into a game engine response time thesis (lmao)

Expect threads to take wild U turns when Necrokiller's around

 

[Necrokiller]

Did you make gifs? I'd like to see them.

Heres one.

The flickering is due to the high frame rate capture. Not an issue since frame by frame you're only looking for the first time it lights up.

 

[CerebralTiger]

Heres one.

Assuming it's the CTRL key that triggers gunfire, I'm seeing visible signs of the gun firing at the 13th frame, and of the gunshot hitting the target at the 15th frame.

13 * 16.67 = ~217ms (using a 60fps camera). That said, I believe Alex conducted his test using a 120fps camera, so it's probably 181ms.

 

[Necrokiller]

Assuming it's the CTRL key that triggers gunfire, I'm seeing visible signs of the gun firing at the 13th frame, and of the gunshot hitting the target at the 15th frame.

13 x 16 = 208ms. That said, I believe Alex conducted his test using a 120fps camera, so it's probably 181ms.

My capture is also at 120fps. This is important for the measurement.

Ctrl lights up at 17th frame.
Gun fires at 31st (this is probably not clear to you in the gif).

That comes out to be 116.6ms.

 

[CerebralTiger]

That comes out to be 116.6ms.

Interesting. Though, ideally, one ought to be repeating the test a number of times and average out findings to get more precise data. Regardless, it gives a good rough idea.

Also,

Different actions in-game may also take variable timing before animating on-screen. As a result, I am going to do latency testing based on the action that would typically require the quickest response, such as shooting a gun, braking a car, or jumping a platform.

Console Latency: Exploring Video Game Input Lag | DisplayLag

<strong>Affiliate Disclosure</strong> This post may contain affiliate links. Clicking these links will redirect you to Amazon.com, and we will receive a small commission on purchases made through this link. Time sure does fly. The PlayStation 4 and Xbox One have firmly cemented themselves as an...

displaylag.com

This calls for GoW being re-tested with the dodge, quick attack button, or axe throw animation startup (whichever is faster). Would be interesting to see how close NX Gamer's findings are to yours. He tested 138.7ms of latency.

 

[Necrokiller]

This calls for GoW being re-tested with the dodge, quick attack button, or axe throw animation startup (whichever is faster). Would be interesting to see how close NX Gamer's findings are to yours. He tested 138.7ms of latency.

As a game like RDR2 is proof enough, the quickest or best case scenario is not indicative of real world perceived lag. As with RDR2, we might see a lower figure depending on a different action but Kratos still doesn't feel responsive to control and data backs that up.

The only readings I'm interested in now are for different actions in TLOU2. Destiny is considered to be the best feeling 30fps shooter and on Xbox One the firing lag is 96ms.