Noodalls input lag testing (unique method - games/controllers/etc) (DBZF X1 Beta MVCI PS4 added)


#1

I wanted to create my own thread on this topic, as the method that I’ve come up with overcomes several of the disadvantages of other methods. I feel that I have a lot of information out on twitter and in other threads, but it would be nice to consolidate it all into one thread that can be updated as I test.

The concept for how this test works is to watermark the screen image at the exact time the controller is being pressed. This is done using optocouplers, so that there can be no electrical cross talk between different components of the testing circuit. This should ensure mean that the signal arrives at the controller at the same time as it interrupts the video signal.

I have made several videos explaining this here.

https://www.youtube.com/playlist?list=PLevqZFjlWIddDGUeFbznuZbwDECBm4luj

New one explaining the results format

https://youtu.be/Li8sdHDWapY


#2

This thread will be for controllers

New method. ~100 tests, 1F long, 30F apart.

Tested with BBCPE for PS4/PS3/X1. Tested with BBCS for X360. WiiU tested with Smash WiiU. Time includes response of controller + game.

16/4 update - first PS4 brook converter tested
15/4 update - added website links for controllers.

PS4
UFB FW 1.80 42.9ms http://www.brookaccessory.com/detail/06960737/

Spoiler

https://twitter.com/noodalls/status/843317150358024192

DS4 Old wireless 43.9ms

Spoiler

https://twitter.com/noodalls/status/843384352952135680

DS4 Old wired 43.9ms

Spoiler

https://twitter.com/noodalls/status/843384422682386432

DS4 New wireless 44.1ms

Spoiler

https://twitter.com/noodalls/status/843384531532959745

PS360+ FW 1.66 HW 2.10 44.4ms

Spoiler

https://twitter.com/noodalls/status/843317325893849088

HRAPV Hayabusa 47.4ms http://www.hori.jp/products/p4/p4_rap_vh2/

Spoiler

https://twitter.com/noodalls/status/846159307267698688

HRAPV4 48ms http://www.hori.jp/products/ps4/ps4_rap_h/

Spoiler

https://twitter.com/noodalls/status/846159378927370240

UFB in X360 mode, X360-PS4 Converter 48.1ms

Spoiler

https://twitter.com/noodalls/status/853495190627037184

Madcatz FS Alpha 48.2ms http://www.madcatz.com/fightsticks/fsalpha.php

Spoiler

https://twitter.com/noodalls/status/848305122690031617

Hori FC4 48.7ms http://www.hori.jp/products/ps4/ps4_ftc/

Spoiler

https://twitter.com/noodalls/status/846159509290598400

Hori VLX 48.8ms http://www.hori.jp/products/p4/p4_rap_vlx_HAYABUSA/

Spoiler

https://twitter.com/noodalls/status/846159421180821506

PS3PS4FB FW 1.80 49.2ms http://www.brookaccessory.com/detail/58690501/

Spoiler

https://twitter.com/noodalls/status/843317256218017792

DS4 New wired 54.8ms

[details=Spoiler]https://twitter.com/noodalls/status/843384610159435776
https://www.youtube.com/watch?v=zNSq42s9Lps&feature=youtu.be&t=309

  • video of Cerny saying it will be lower latency, better for fighting games. Note time encoding doesn’t seem to be working, skip to 5:09.
    [/details]

DS4 Retest for FW 4.55 wired - no change 54.9ms

Spoiler

https://twitter.com/noodalls/status/852689258989068288

Venom 58.9ms

Spoiler

https://twitter.com/noodalls/status/843317413319860224

PS3
UFB FW 1.80 PS4 mode 43.6ms http://www.brookaccessory.com/detail/06960737/

[details=Spoiler][/details]

UFB FW 1.80 43.8ms http://www.brookaccessory.com/detail/06960737/

Spoiler

https://twitter.com/noodalls/status/843646566472720384

MC Cthulhu 1ms firmware 44.5ms

Spoiler

https://twitter.com/noodalls/status/845878533524094976

PS360+ Old hardware FW 1.66 44.5ms

Spoiler

https://twitter.com/noodalls/status/845921535818809344

PS360+ HW 2.10 FW 1.66 44.4ms

PS3PS4FB FW 1.80 45.1ms http://www.brookaccessory.com/detail/58690501/

Spoiler

https://twitter.com/noodalls/status/845555192787492864

PS360 47.9ms

Spoiler

https://twitter.com/noodalls/status/845921777331056641

MCZ FS Alpha PS4 mode 50.7ms

Spoiler

https://twitter.com/noodalls/status/848366261029425152

Hori V3SA 53.4ms

Spoiler

https://twitter.com/noodalls/status/845921993002123264

Hori Wireless Stick 54.9ms

Spoiler

https://twitter.com/noodalls/status/845922102435733505

MC Cthulhu 10ms firmware 54.7ms

Spoiler

https://twitter.com/noodalls/status/845878603044741120

Zero Delay 55.1ms

Spoiler

https://twitter.com/noodalls/status/845863324478717952

Hori Fighting Commander 3 55.1ms

Spoiler

https://twitter.com/noodalls/status/845921924786012160

MCZ FS Alpha PS3 mode 56ms

Spoiler

https://twitter.com/noodalls/status/848366185678749696

Q4 58.9ms

Spoiler

https://twitter.com/noodalls/status/845261612214726656

Xbox One
UFB FW 1.80 46.1ms http://www.brookaccessory.com/detail/06960737/

Spoiler

https://twitter.com/noodalls/status/843645041910284288

TE2 53.1ms

Spoiler

https://twitter.com/noodalls/status/843645337243807744

X360
UFB FW 1.80 48.1ms http://www.brookaccessory.com/detail/06960737/

Spoiler

https://twitter.com/noodalls/status/844640886075998208

Q4 49.2ms

Spoiler

https://twitter.com/noodalls/status/845261547836338176

PS360+ HW 2.10 FW 1.66 52.2ms

Spoiler

https://twitter.com/noodalls/status/844640961107918848

PS360+ Old HW FW 1.66 59.8ms

WiiU
Wii Controller 73.9ms

Spoiler

https://twitter.com/noodalls/status/848660304301092868

Gamecube controllers 74.7ms and 74.9ms (old and new)

Spoiler

https://twitter.com/noodalls/status/848660189746126848

Spoiler

https://twitter.com/noodalls/status/848660246172008448

WiiU Gamepad 76.9ms

Spoiler

https://twitter.com/noodalls/status/848660383971762176

Brook UFB in PS4 mode via WiiU P3/P4 adapter 79.5ms http://www.brookaccessory.com/detail/06960737/

Spoiler

https://twitter.com/noodalls/status/848672730975092736

PC BBCPE Assume Vsync on unless mentioned. 780TI. 60fps. 1080p.
UFB X360 mode, Vsync off 54.4ms

Spoiler

https://twitter.com/noodalls/status/853571266376196096

UFB PS4 mode 55.8ms

Spoiler

https://twitter.com/noodalls/status/853568816118247424

UFB X360 mode 56.2ms

Spoiler

https://twitter.com/noodalls/status/853568766873030657

Previous results

Spoiler

https://twitter.com/noodalls/status/826531580495540225


#3

This post will be for games. Tests are with Brook UFB FW 1.80 unless specified. Time includes response of game + controller.

7/5 update - X1 UMvC3 added. Now we have all five versions.
1/5 VF5 added (4 versions). Tekken (8 versions)
29/4 update - NRS PS4 games and GGAC for Wii added
15/4 update - 6 new ps4 games added

PS4
BBCF 41.9ms

Spoiler

https://twitter.com/noodalls/status/852692434588807168

BBCPE 42.9ms

[details=Spoiler]https://twitter.com/noodalls/status/843317150358024192
[/details]

Nitro Blasters 55.1ms

Spoiler

https://twitter.com/noodalls/status/853185582616289280

GGXRD Revelator 58.2ms

Spoiler

https://twitter.com/noodalls/status/853185365611388929

Demo 58.7ms

Spoiler

https://twitter.com/noodalls/status/866458204498124800

USF4 59.1ms

Spoiler

https://twitter.com/noodalls/status/845261753852166146

Injustice 65.6ms *note command history occurs 1F before animation starts

Spoiler

https://twitter.com/noodalls/status/858134657400950785

SkullGirls 2nd Encore 73ms

Spoiler

https://twitter.com/noodalls/status/853185469403742208

Garou 74.6ms

Spoiler

https://twitter.com/noodalls/status/853185630892728320

Last Blade 2 74.6ms

Spoiler

https://twitter.com/noodalls/status/853185675033575424

Dead or Alive 5 LR 75.3ms

Spoiler

https://twitter.com/noodalls/status/853500920042184704

Mortal Kombat X 75.5ms *note command history occurs 1F after animation starts

Spoiler

https://twitter.com/noodalls/status/858146132341309440

KOFXIV ver 2.00 76.5ms

Spoiler

https://twitter.com/noodalls/status/852696770861686784

MVCI 88.5ms

Spoiler

https://twitter.com/noodalls/status/909887692900466688

MVCI demo 88.8ms

Spoiler

https://twitter.com/noodalls/status/874904573789560833

SFV 88.6ms

Spoiler

https://twitter.com/noodalls/status/843664351126929408

UMvC3 89.1ms

Spoiler

https://twitter.com/noodalls/status/853185406027812867

Injustice 2 92.4ms command history appears one frame earlier

Spoiler

https://twitter.com/noodalls/status/864569562221559809

Tekken 7 120.1ms *command history appears one frame faster

Spoiler

https://twitter.com/noodalls/status/870293237969240065

Tested at 720p. 119.9ms

Spoiler

https://twitter.com/noodalls/status/870296013306712064

Tested on PS4Pro 120.5ms

Spoiler

https://twitter.com/noodalls/status/870510075122556929

PS3
KOFXII 26.9ms

Spoiler

https://twitter.com/noodalls/status/853576147736461312

SF3 42.7ms

Spoiler

https://twitter.com/noodalls/status/844849977574637570

Note - tested with different filter options, quick check doesn’t seem to make any difference.

BBCPE 43.8ms

Spoiler

https://twitter.com/noodalls/status/843646566472720384

Virtua Fighter 5FS 60.3ms *note command history appears 1 frame before animation starts

Spoiler

https://twitter.com/noodalls/status/858884742300565506

Virtua Fighter 5 60.4ms *note command history appears 1 frame before animation starts

Spoiler

https://twitter.com/noodalls/status/858884742300565506

Tekken 5 DR 63.0ms

Spoiler

https://twitter.com/noodalls/status/858206763501314048

Tekken 6 65.8ms *note command history appears 1 frame before animation starts
Repeat 66.5ms

Spoiler

https://twitter.com/noodalls/status/845261682746150912

Spoiler

https://twitter.com/noodalls/status/858206763501314048

UMvC3 66.6ms

Spoiler

https://twitter.com/noodalls/status/855208382025785344

Tekken Prologue 66.7ms

Spoiler

https://twitter.com/noodalls/status/858207011451789313

USF4 77.4ms

Spoiler

https://twitter.com/noodalls/status/844880921920602112

Tekken Tag 85.5ms

Spoiler

https://twitter.com/noodalls/status/858206763501314048

TTT2 94.0ms *note command history appears 1 frame before animation starts
Repeat 95.1ms

Spoiler

https://twitter.com/noodalls/status/844140889589010432

Spoiler

https://twitter.com/noodalls/status/858206881831018497

Tekken Revolution - incomplete but looks same as TTT2
Repeat - 93.7ms

Spoiler

https://twitter.com/noodalls/status/844140959956791296

Spoiler

https://twitter.com/noodalls/status/858206940853182464

KOFXIII 105.2ms * note, command history appears 4 frames before animation starts

Spoiler

https://twitter.com/noodalls/status/853576085291782145

X360
SF3 47.4ms

Spoiler

https://twitter.com/noodalls/status/844849884746301441

BBCS 48.1ms

USF4 57.4ms

Spoiler

https://twitter.com/noodalls/status/844880969114931200

Virtua Fighter 5FS 64.4ms *note command history appears 1 frame before animation starts

Spoiler

https://twitter.com/noodalls/status/858884742300565506

UMvC3 64.4ms

Spoiler

https://twitter.com/noodalls/status/855208328997122048

TTT2 99.1ms *note command history appears 1 frame before animation starts
Repeat 99.9ms

Spoiler

https://twitter.com/noodalls/status/844641041550524416

Spoiler

https://twitter.com/noodalls/status/858206881831018497

X1
BBCPE 46.1ms

Spoiler

https://twitter.com/noodalls/status/843645041910284288

UMvC3 62.0 ms

Spoiler

https://twitter.com/noodalls/status/861123564203683840

KI 64.1ms

Spoiler

https://twitter.com/noodalls/status/843645549974769664

Garou 67ms

Spoiler

https://twitter.com/noodalls/status/858938372198289408

MVCI Demo 75ms

Spoiler

https://twitter.com/noodalls/status/874953070194470913

Virtua Fighter 5FS 78.9.3ms *note command history appears 1 frame before animation starts

Spoiler

https://twitter.com/noodalls/status/858884742300565506

DBFZ Beta 86-91ms

Spoiler

https://twitter.com/noodalls/status/909157945467748352

Tekken Tag Tournament 2 117.2ms * note command history appears 1 frame before animations starts

Spoiler

https://twitter.com/noodalls/status/858206881831018497

PC Windows 10 i4770K @ 3.50GHz 16GB Ram GTX780ti

GGXRD 45ms

Spoiler

https://twitter.com/noodalls/status/855208531229683712

BBCPE 56.2ms

Spoiler

https://twitter.com/noodalls/status/853568766873030657

UMvC3 61.8ms

Spoiler

https://twitter.com/noodalls/status/855208447213682689

SFV 73.9ms

Spoiler

https://twitter.com/noodalls/status/855210441089310720

Wii
Tatsunoko vs Capcom 60.8ms (done through WiiU)

Spoiler

https://twitter.com/noodalls/status/856294740995620864

Guilty Gear Accent Core 56.0ms (done through WiiU)

Spoiler

https://twitter.com/noodalls/status/858110337790263296

WiiU
Tekken Tag Tournament 2 128.3ms*command appears 1F before animation

Spoiler

https://twitter.com/noodalls/status/858206881831018497

The first game I have tested using my new setup is Ultimate Marvel vs Capcom 3.

[details=Spoiler]

https://twitter.com/noodalls/status/840536944098471936[/details]


#4

Online testing

The way I did this is to have two PS4 systems, which we can call PS4A and PS4B. For reference PS4A is a standard PS4, PS4B is a PS4Pro. Both are wired connections to a R7000 router. My ping on each system is about 30ms over a few trials. However this is going to a major town about 70km away, so not sure if the ping will be as bad p2p.

A button is pressed (via the arduino) on PS4A, and the video output is captured on PS4B, via the standard HDMI --> component with a colour band to indicate when the button press occurred.

What is interesting about the results to me is the 1. BBCPE is blindingly fast, 2. T7 is actually a little faster than offline and 3. the range of input timing for SFV ranges from 66.6ms to 133.3ms.

BBCPE 57.8ms

Spoiler

https://twitter.com/noodalls/status/876573772345520128

SFV 88.2ms

Spoiler

https://twitter.com/noodalls/status/876573646734540800

T7 113.6ms

Spoiler

https://twitter.com/noodalls/status/876573509543120897


#5

And this post will be reserved for answering questions.

To predict some questions.

Why not just use the video method like everyone else?
I guess the best answer is, if everyone else can do it, what could I add? I have played extensively with that method, and find that it is too finnicky in terms of being sure that either a button has been pressed, or an LED has started lighting up, as well as having to reset the camera between takes to ensure that different synchronisations with the LCD monitor have occurred etc. etc.
I only have access to a 60fps camera. In the future, when Sony’s 1000fps mobiles come out, this method may be more useful again.

Doesn’t the HDMI --> Component converter introduce lag?
This was one of my major concerns. However, using the PS3 I was able to output via either HDMI via converter or directly via component into my testing system, and depending on how the results are analysed, it would seem to suggest that it adds 0.25-0.5ms of lag.

Why don’t you just test via an oscilloscope?
Unfortunately, I don’t have the know how to do this. I have bought an arduino USB adapter and cheap oscilloscope but never made much headway in using them. I would be more that happy for someone else to do so.
Also, this method allows us to test the entire chain up until before the monitor, so changing anyway one component can tell us the contribution of that component. Also, this is the real environment of the controllers, so issues like polling rate will be revealed.
Furthermore, it would only take one oscilloscope test by someone else to give us the information required to remove the contribution by the game/system. For example, this post by undamned

gives us a baseline. In looking at the recordings for the MC Cthulhu on PS3, I also find a tight band, located just above the end of the frame.

How are you able to give millisecond timings?
This is explained in one of the videos, but by dividing the 720p or 1080p image into 16.66 equal segments (accounting for non-displayed lines) we can determine where within the frame an input is occurring.

Did you come up with this method yourself?
The idea of interrupting a video signal is not my own, and has been used by people on this forum as well as other places on the net. However the extension this far, and so far as I am aware the idea of just using 1ms instead of a longer input to target the input window is my own.


#6

Reserved.


#7

https://1drv.ms/u/s!AizjHalveL7Pj6B9oLViqEXjGrRnWw

This should link to a zip file containing the images I used to review PS4.UFB.BBCPE if anyone is interested as to how I go about it.


#8

Input lag in PC gaming varies a lot, as detailed in https://displaylag.com/reduce-input-lag-in-pc-games-the-definitive-guide/ , so you should probably specify what settings you used, which include settings not available in the game itself (e.g., managed on the NVIDIA Control Panel / AMD Catalyst)… :slight_smile:

I’m a G-Sync fan… :stuck_out_tongue:


#9

Yep. That’s one of the reasons I’ve not done as much testing on PC. Consoles are a very set environment, so my experience should be the same as everyone’s. PC seems to keep throwing up new frustrations each time I try to use it.


#10

V-Sync off or G-Sync / FreeSync, if you have it, should be the priority, IMO, as they yield the best results. I suppose V-Sync on out of curiosity would be an ok contribution…


#11

Pretty sure my 780TI doesn’t do gsync, and even if it did, I run the HDMI out to an HDMI to component converter, which definitely doesn’t support G-Sync.

Updated with at least 10 new tests, mainly old PS3 controllers. Still have another 4 PS4 controllers to test, and a few on the X360.


#12

Could you possibly test UFB in forced mode for some PlayStation games, I mean force PS4 mode on a PS3 and vice versa? Curious to see if there is any difference. Thank you for your thorough testing!

Btw, Gsync should work fine for your 780Ti, but yeah DP not HDMI… = /


#13

Noodalls is there any chance you could test the razer panthera with the updated firmware? Thanks


#14

I’ll test anything I can get my hands on. Not sure if anyone local has one though.

In the mean time


#15

I will look to do this at some point. From memory Nikogel360’s conclusion was that setting the controller to PS4 on a PS3 system worked best, and they questioned why you even need a PS3 mode (presumably compatibility).


#16

I don’t remember it clearly now, but I suspect PS4 controllers on PS3s don’t have exactly as many functions as PS3 controllers…


#17

Thanks! That was my feeling just from playing (placebo maybe), but on pc, I need to set it to PS3 and use the confusing inputmapper 2.0 as a wrapper. That can get 1ms polling but it was a struggle to setup. If I use PS4 mode, the lowest it goes for me was 4ms = /. inputmapper 1.5 can set an actual DS4 to 1ms easily though FYI, in case you do get into testing pc lag, though it doesn’t seem to work with the UFB or much else.


#18

Large update - I’ve devised a way to convert the raw data into an average, I have updated this on the games posts and ordered them from fast to slow. Will apply this to controllers next.


#19

@noodalls — can you please test Brooks ps2 convertor to ps4 .https://www.focusattack.com/brook-super-converter-playstation-2-to-ps3-ps4-adapter/
I have heard lot of good things about this convertor… Thanks in advance


#20

If I can get access to it I will, but that goes for most things. The other thing to mention would of course be, no one has done input lag for PS2 controllers, so that would need to be sorted out.