comic | episodes & e-books | store | about | forum 




Post new topic Reply to topic  [ 3 posts ] 
crude input lag test, SNK etc fighting games, DC PS2/3/4 PC 
Author Message
User avatar

Joined: Fri Mar 13, 2009 4:18 pm
Posts: 4233
    


I wanted to know how responsive some fighting games I was thinking about playing were, in their various console and PC incarnations, so I used 60 fps cell phone footage of me pressing their Light Punch button, mostly with Kyo from King of Fighters as the active character seen on my monitor in the background, to count, visually, about how many frames it took for the action activated by the button press to appear on the screen; I did this by stepping through the recorded footage frame by frame using the "Enhancements > Play speed settings > Next frame" button in the Windows 10 version of Windows Media Player. You can't really do this yourself with the YouTube version of the footage, I guess, so the actual video here will simply be a pretty annoying proof of concept--but anyway you'll find approximate time stamps to each of the 70+ tests, and the input delay frame count I came up with for each by eyeballing them--further down in this description.

Real tests of this "input delay" or "input lag" in games are made much more precisely using a very high speed camera filming standard controllers wired to an LED, so you can see exactly when their button circuit activates. Not having such cool equipment, I just counted as frame "zero" the frame in which the punch button was fully pressed down. This probably shrunk the frame delay count by about one frame, on average, because the button actually activates its signal when only partially pressed, whereas I didn't start counting until it was pressed all the way down, which tended to be about one frame later...probably. : P

That inaccuracy was possibly more or less offset in my counts by my inadvertently having a processing feature active on my Acer S271HL monitor--Acer's "Active Contrast Management" (ACM)--which seems to have added about one frame of delay to everything I was seeing. You can see a couple comparisons of ACM on and off in a few of the final segments of the video. At any rate, in theory the button press timing boost coupled with the monitor processing delay cancelled each other out as far as how many frames it took for actions to appear after a button press--except maybe for the two segments showing a DualShock 2, since their button throw is much shorter; the DS2's frame delay count thus probably came out about a frame too high, so I subtracted one frame from those few DS2 tests I ran.

All video output was also passing through the video pass-through of an Elgato Game Capture HD or HD60 external video capture box. I intended to test KOF 98 on Dreamcast (where it is called '99 Dream Match or something :p) but its composite video output through the Elgato pass-through did not work on my monitor.

At any rate, please note that this is a very crude test really only meant for my own comparison purposes on my own gaming set-up; it should by no means be considered a definitive test of input delay in these games!

It is important to note that games can have intentional input delay. After I read on http://www.orochinagi.com/2016/07/5-5-f ... ned-kofxiv that older KOF games had 4 frames of intentional delay on their normal attacks, like Light Punch, and just 1 on their movement, I counted movement in NeoGeo mode KOF 98 UM on PS2 (the final test in the video) as having 3 frames of movement delay, vs 6 frames of attack delay in Kyo's "Light Punch."

My other test results--from PS3 and PC--seem to indicate that 5 frames of Light Punch delay is the best my set-up can expect with the older KOF games (2000 and earlier, at least), whereas by at least 2003 they started to remove that attack delay; I measured Kyo as being able to punch in just 2 frames in KOF XI on PS2, for instance.

Other games can differ in their attack delay design as well. While I measured Ryu as starting to animate in response to a Light Punch button press in just 2 frames in Street Fighter III: 3rd Strike on PS3, for instance, his reaction is to hold his wrist cocked for 4 frames before throwing the punch; in KOF 98, by contrast, for which I counted an attack response of minimum 5 frames (on PC), Kyo's initial reaction is to hold a lunging position for just 2 frames before throwing his punch.

(Games may also have input delay in order to check for combination inputs. For instance, the game may wait for a frame or two after the player pushes button A in order to see if they also push button B at nearly the same time, in order to count it as a A+B button combo.)

The counted Light Punch attack delay, in frames, per test, listed chronologically by machine release date, then chronologically by game or emulator release date per machine:

Dreamcast
- NA model chip modded to play JPN games as well
- Official DC "Dreamcast Arcade Stick" controller for input
- Composite video out

0:00 - Marvel vs. Capcom 2 (Mar 2000) - 3 frames
0:11 - Street Fighter III: 3rd Strike (Jun 2000) - 4 frames
0:20 - Guilty Gear X (Dec 2000) - 3 frames
0:29 - Capcom vs. SNK 2 (Sep 2001) - 3 frames
0:37 - The King of Fighters 2000 (Aug 2002) - 6 frames
0:47 - The King of Fighters 2002 (Jun 2003) - 7 frames


PS2
- Final JPN slimline model
- Capcom vs SNK 2 (SNK version) ASCII Stick FT2 controller for input (DualShock 2 used for a couple tests for the sake of comparison since I wasn't sure how the ASCII Stick FT2 stacked up in terms of its own input lag)
- Games with a progressive video output option were mostly tested in both interlaced and progressive mode (activated by holding Triangle and X buttons during game boot)--it turned out not to matter for frame counts
- Component video out (Elgato custom cable)

0:56 - Capcom vs. SNK 2 (Nov 2001) - 3 frames
The King of Fighters 2003 (Oct 2004)
- 1:07 - interlaced, with DualShock 2 - 3 frames
- 1:20 - interlaced, with ASCII Stick FT2 - 3 frames
- 1:30 - progressive, with DualShock 2 - 3 frames
- 1:41 - progressive, with ASCII Stick FT2 - 3 frames
The King of Fighters 2000 (in NESTS Collection, Apr 2007)
- 1:51 - main "Dreamcast" version - 6 frames
- 2:00 - "NeoGeo" version - 8 frames
2:09 - Kizuna Encounter (in Fu'un Super Combo, Jun 2007) - 4 frames
The King of Fighters XI (Nov 2007)
- 2:24 - interlaced - 2 frames
- 2:33 - progressive - 2 frames
2:42 - The King of Fighters '96 (in Orochi Collection, Oct 2008) - 6 frames
The King of Fighters 2002 Unlimited Match (Mar 2009)
- 2:52 - NeoGeo Mode, progressive - 6 frames
- 3:02 - Unlimited mode, progressive - 6 frames
The King of Fighters '98 Ultimate Match (Mar 2009)
- 3:13 - NeoGeo mode, interlaced - 6 frames
- 3:21 - NeoGeo mode, progressive - 6 frames
- 3:30 - Unlimited mode, interlaced - 6 frames
- 3:39 - Unlimited mode, progressive - 6 frames


PS3
- NA slim model
- Hori RAP V Hayabusa 2017 Ver. controller for input, modified with Sanwa buttons; tested games with the stick set to both "PS3" and "PS4" modes
- PS2 Classics games that originally had a progressive video output option were mostly tested in both "interlaced" and "progressive" mode (activated by holding Triangle and X buttons during game boot)--it turned out not to matter for frame counts
- Component video out (Elgato custom cable)

Marvel vs. Capcom 2 (Aug 2009)
- 3:51 - stick set to PS3 - 3 frames
- 4:02 - stick set to PS4 - 3 frames
Street Fighter III: 3rd Strike (Aug 2011)
- 4:11 - stick set to PS3 - 2 frames
- 4:21 - stick set to PS4 - 2 frames
The King of Fighters '96 (now-delisted NEOGEO STATION version by M2, Sep 2011)
- 4:30 - stick set to PS3 - 5 frames
- 4:40 - stick set to PS4 - 5 frames
Capcom vs. SNK 2 (Sep 2012)
- 4:50 - stick set to PS3 - 5.5 frames : p
- 4:56 - stick set to PS4 - 6 frames
The King of Fighters '98 Ultimate Match (PS2 Classics, Nov 2014)
- 5:06 - NeoGeo mode, interlaced, stick set to PS3 - 8 frames
- 5:17 - NeoGeo mode, interlaced, stick set to PS4 - 9 frames
- 5:28 - NeoGeo mode, progressive, stick set to PS3 - 8 frames
- 5:38 - NeoGeo mode, progressive, stick set to PS4 - 7 frames
- 5:51 - Ultimate mode, interlaced, stick set to PS3 - 7 frames
- 6:01 - Ultimate mode, interlaced, stick set to PS4 - 8 frames
- 6:12 - Ultimate mode, progressive, stick set to PS3 - 8 frames
- 6:21 - Ultimate mode, progressive, stick set to PS4 - 8 frames
The King of Fighters XI (PS2 Classics, Dec 2014)
- 6:28 - interlaced, stick set to PS3 - 3 frames
- 6:37 - interlaced, stick set to PS4 - 3 frames
- 6:45 - progressive, stick set to PS3 - 3 frames
- 6:53 - progressive, stick set to PS4 - 3 frames
The King of Fighters 2003 (PS2 Classics, Apr 2015)
- 7:02 - interlaced, stick set to PS3 - 5 frames
- 7:14 - interlaced, stick set to PS4 - 4 frames
- 7:25 - progressive, stick set to PS3 - 5 frames
- 7:35 - progressive, stick set to PS4 - 4 frames
The King of Fighters 2000 (in PS2 Classics NESTS Collection, Jun 2015)
- 7:42 - Dreamcast version, stick set to PS3 - 8 frames
- 7:50 - Dreamcast version, stick set to PS4 - 8 frames
- 8:02 - NeoGeo version, stick set to PS3 - 10 frames
- 8:08 - NeoGeo version, stick set to PS4 - 10 frames
- 8:18 - duplicate, oops : P
- 8:29 - another duplicate :o
The King of Fighters '96 (in PS2 Classics Orochi Collection, Jul 2015)
- 8:38 - stick set to PS3 - 9 frames
- 8:49 - stick set to PS4 - 10 frames


PC
- 2.4 GHz dual core running Windows XP
- same Hori controller as above used for input; in "PS4" mode (DirectInput) unless otherwise noted
- HDMI video out (1080p)
- Game ROM from Humble Bundle Neo Geo collection running in the MAME and FBA 0.2.97.40 emulators, using the Neo Geo BIOS file from the Humble Bundle Neo Geo version of Twinkle Star Sprites (BIOS files from most other HB NG games were not MAME/FBA compatible); MAME config: -nofilter -bios euro

9:02 - The King of Fighters 2000 (run in MAME 0.147, Sep 2012, vsync off) - 6.5 frames
The King of Fighters Ultimate Match Final Edition (Dec 2014)
- 9:09 - NeoGeo mode, vsync off - 5 frames
- 9:18 - NeoGeo mode, vsync on - 7 frames
- 9:48 - Ultimate mode, vsync off - 5 frames
- 9:57 - Ultimate mode, vsync on - 8 frames
- 10:12 - Ultimate mode, vsync off, stick set to "PC" (XInput) - 5 frames
The King of Fighters 2000
- 10:21 - run in Final Burn Alpha (Jul 2017), vsync off - 6.5 frames
- 10:31 - run in MAME 0.189 (Sep 2017), vsync off - 5 frames
- 10:40 - run in MAME 0.189, vsync off, stick set to "PC" - 6.5 frames


PS4
- NA launch version
- same Hori controller as above used for input
- HDMI video out (1080p)

10:51 - The King of Fighters 2000 (PS2 Classics, May 2016) - 7 frames
11:01 - Kizuna Encounter (in Fu'un Super Combo, PS2 Classics, Dec 2016) - 5 frames
11:09 - ACA NEOGEO The King of Fighters '96 (Aug 2017) - 7 frames


Addendum:

After filming all the above, I added a few more:

- Went back and checked attack delay in two other progressive video-enabled PS2 fighting games I dug up:

11:19 - NeoGeo Battle Coliseum (Dec 2007), progressive - 2 frames
11:31 - SNK vs. Capcom: SVC Chaos (Apr 2005), progressive - 4 frames


- Found my monitor was using a 1-frame-delay processing filter (Acer's "ACM"), and turned it off:

The King of Fighters '98 Ultimate Match (PS2)
- 11:43 - ACM on
- 11:53 - ACM off
The King of Fighters XI (PS2)
- 12:11 - ACM off
- 12:22 - ACM on


- Found out older KOF games had intentional (4?) delay frames added to their normal attacks, and fewer (1?) to their movements, so made a movement test to demonstrate that 3-frame difference (with ACM off):

12:35 - The King of Fighters Ultimate Match (PS2), progressive


Thu Sep 07, 2017 11:31 pm
Profile
User avatar

Joined: Fri Mar 13, 2009 4:18 pm
Posts: 4233
    


Got 5 frames in Marvel vs. Capcom: Infinite on PS4.


Wed Sep 20, 2017 3:16 pm
Profile
User avatar

Joined: Fri Mar 13, 2009 4:18 pm
Posts: 4233
    
Haohmaru's light kick in Samurai Shodown II:

PC - MAME - 2 frames
PS4 - ACA NEOGEO Samurai Shodown II - 2 frames


Wed Oct 11, 2017 11:45 pm
Profile
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 3 posts ] 


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
cron
Powered by phpBB © phpBB Group.
Modified from the "Hestia" theme designed by Vjacheslav Trushkin.