About the “filthy casuals” remarks, this is what I was getting at if it wasn’t clear:
If you’re a casual player, most likely you don’t play the game at the level where input lag is something you care enough to learn about or attempt to correct, and that’s completely fine.
If you’re a competitive player, input lag is a part of the game that you have to deal with, whether it be on your own monitor, stick, or network connection.
If you do compete, you can use this information to gain a competitive edge (like SRK’s old slogan… forever ago)
For example, I recently found out that Ogawa, Evo2015 GGXrd champion, and widely-considered overall #1 player in the history of Guilty Gear, used my site as a reference my site in determining which PCB to have put into his new stick which he brought to Evo and used to win the event:
It was very cool to see the results used in this way and I hope more top players start to take notice.
calling someone a casual because they don’t care about this is of course misguided. plenty of strong players play on laggy sticks for any number of reasons.
but when you respond to a topic with “Please keep details of Teyah’s irrelevant lag information in his irrelevant thread here” you immediately torpedo any chance for a worthwhile conversation. can’t think of a lot of nice ways to respond to that.
Because indirect mention I might as well state my opinion officially at this point, even though I wanted to avoid coming into someone else’s thread to do so.
Do I respect the actual work put into the testing procedure? Sure. For that, I give @Teyah credit. He has put a tremendous amount of work into this information. Anyone gathering that much information in any sort of scientific situation is worthy of praise.
What I DO NOT like, is the fact that this information is being used by players (probably all over the world now) who do not understand the data or testing methods; to make judgement calls on what to buy, or to sell or trade their existing arcade sticks which work perfectly fine for something else that is rated higher on the list and using this data as an excuse on why they didn’t win X or Y situation.
This isn’t Teyah’s fault, but bumping the topic is his fault after it’s been on page 4 or 5 so it gets back up to the top of the list is when there is much more relevant information that should have a higher focus in Tech Talk.
Why do I think this information is irrelevant?
-Testing flaws stated many times by members in this thread and outside of it
-Combined factors of input latency from the human brain (average of 220ms), the system (0ms-32ms), the game (variable), and the monitor (10ms-12ms)
-Top players have proven you can win on any stick. Use the same stick, on the same system, with the same monitor as what’s played in a tournament, and your skill will win. Worrying about a 10ms difference between your arcade stick and someone else’s is straight up stupid and a gigantic waste of someone’s time and effort.
I don’t speak for all of SRK Tech Talk, and I’m not trying to, but I can guaranfuckingtee that the majority of the SRKTT members think the results of this data is irrelevant and consider it wasted valuable space on the front page of the SRK TT forum.
If you’re one of them, great, if you’re not, great, but don’t get butthurt over my 1 word opinion.
EDIT ADDITION - In the name of Cthulhu, I praise you to please please please let this thread die and not go on for years wasting space that could be used for valuable Tech Talk information.
Pretty much everyone in this thread on any side of the discussion has pointed out all of the above repeatedly. Practically on every page. It doesn’t need to be done any more.
These have been addressed.
This is a non-argument and the same as saying that the monitor lag or game lag are also irrelevant, both of which typically fall into the same window of time or more as the PCB lag measurements. Human average response time (detection and physical response) is not how fast the brain works and does not equate to the inability to detect lag.
Fun anecdote: Did you know the final “EVO monitor” was selected by having John Choi play on it and pick the least laggy one available? Nobody was measuring lag yet. Not a bad choice.
You can say the same about a frame of converter lag, or a frame of display lag, or a frame of extra game lag. All three of those examples are things fighting game players have cared about for many years and that is unlikely to change.
As a completely non-gameplay related example, people have used sticks as the input device to measure in-game or system lag (e.g. 360 vs PS3 input lag for SF4) and gotten misleading results. If they knew their stick was lagging a frame, they would have known to compensate for it in their findings.
You are saying that you don’t like it, but that doesn’t mean it is a waste of space or not valuable.
People want to discuss this stuff and all of the derision from members of this forum is embarrassing.
I just want to say I have no opinion on this topic but as a scientist one thing is bothering me. Your repeated mentioning that multiple test (1000) of a stick and its averaging of values is proof of the accuracy of your method. This does not measure the accuracy of your method, it measures its’ precision. Precision and accuracy are not the same. And to determine its’ precision you should go further and calculate the %RSD, to avoid big deviations in values collected, from averaging out and giving you the illusion of precision. Accuracy would be determined by performing a test to determine if your results are hitting a known value when all other variables are taken into effect. I am still unsure of how testing sticks simultaneously removes the variable of how game handles inputs from 2 users. For example a game like MK9 does not handle player 1 and player 2 inputs equally.
Just wanted to say thanks to Teyah for this extremely valuable information. I hope it stays on the front page of tech talk for a long time and motivates manufacturers to make PCB response time a priority.
Arguments that this info is irrelevant are a little bit ridiculous. When a stick like the VLX is approaching an extra frame of lag in comparison to the control; you are putting a handicap on yourself that doesn’t need to be there, especially if you are playing competitively. Combo timings can be adjusted for easily, what can’t be adjusted for are things that are on the edge of your reaction time(blocks, wiff punishes, hit confirms). Even if you can’t directly feel the difference, your consistency is being affected. Suppose depending on your concentration you block an 18 frame overhead on frame 17(block) 33%, 18(block) 33% 19(hit) 33%. If you add an extra frame of lag you are going from a 66% consistency to 33% consistency. Of course those percentages are just for examples sake and would change depending on the situation, but it’s easy to see how this lag can make an appreciable difference.
Just because the majority of players have the most room for improvement in their gameplay, doesn’t mean they should just accept this small handicap when purchasing a stick.
I completely agree with this. I don’t play fighting games that much these days, but I play shmups and other arcade games. Input lag is very important to the experience even when it’s 1P versus the CPU when reaction and timing are critical.
There aren’t likely to be large deviations in the values being collected since Teyah is collecting frame data, 0 frames (tie), +1, +2, -1, etc. In this case, the %RSD is not relevant since he is observing the different animations by hand to assign the frame data for each trial and it is very unlikely that he misinterprets the frame data. The lack of resolution of the data is what will lead to an inaccurate result (essentially by binning the data into discrete frames rather than collecting millisecond data). This is where the notion of having a large sample of results leading to higher accuracy comes from. Since the larger the data set, the closer the sample mean will be to the actual population mean for each comparison. With more data, you get a more accurate curve-fit of the data to estimate the underlying distribution.
This is a plot that illustrates the distribution fitting that is being done. Although this curve fit is done with a different method (maximum likelihood to get the distribution parameters), you can see the results are about the same. The frame data is converted to millisecond bins (0ms=0frames, 16.6ms=1frame, 33.3ms=2frames, etc.). Once the curve fit is done, the mean of the distribution is the value that Teyah reports in his database.
This approach really is valid! Sure, it would be good to get more data, test more games, etc., but let’s not shoot the messenger and cast suspicion on the results. If there are valid questions about the approach taken let’s hear them. Surely if there is a bias or mistake in the approach then it can be demonstrated and reproduced but it is not fair to cast doubt on Teyah’s results without some solid reasons.
I think this kind of information is very valuable. The companies does not give this kind of data to consumers and we don’t have so many people with knowledge and disposition to do that, like Teyah did.
If the consumers don’t know about the existence of any differences between the PCBs in the market, the companies does not have any motivation to improve their products, in the matter of input delay.
Ah this is a damn shame. I think I’ll be sending my HRAP 4 Kai back to Amazon. I didn’t know about the input lag. One frame can be the difference between successfully blocking an attack or getting hit with a combo. I think I’ll wait until I can get a good price on a TE stick nearer SFV’s release.
Considering you’ll be putting hundreds of dollars/hours into your new PS4 stick and you still have the option to return it, if you care about minimizing lag then I’d return it. Provided you have an alternative stick in mind - in this case, the v4 is a very similar substitute that doesn’t lag nearly as much.
The lag results of the new Hori line are spreading around pretty quickly on Japanese sites as well. I hope this puts pressure on Hori to correct the issue on their next line of sticks.
I don’t really believe any of this really affects anyone that much since people with what are apparently terrible PCB’s have won majors. I think people just get used to the lag.
Well… going slightly off topic and playing devil’s advocate, we don’t know that as fact. As far as I remember no one has been asked to open up their arcade stick at a tournament to see how it may have been modified.
Didn’t @Phreakazoid discover a mechanical exploit in SSF4 a while back? IIRC he made a chip that, once installed, would enable the user to “auto-blink”.
I think part of the problem not opening up and inspecting sticks at tournaments is a issue of logistics.
No Tournament Organizer had the time, resources and personnel to spare on this. You also have to make sure all the personnel inspecting sticks to be trained what to look for.
Depending on the nature of the cheat mod, even a seasoned modder or full blown electronics engineer could miss something.
Yeah, I’ve been thinking about the logistics side of things. In a perfect world there would be an officially recognised federation of modders who would test sticks prior to an event, much like how athletes are tested. Or anyone making it to the the top 16 has their sticks tested, but that has its own problems. Added to the fact that ‘cheat chips’ can be made to be nigh on undetectable…
Even tournaments providing their own equipment is barely a solution. Smaller tournaments, perhaps.
There isn’t any point to inspecting sticks, nobody would be able to visually detect your custom auto-magic firmware anyway. Special wiring for s-plinking and such is generally allowed.
That what I was thinking, someone makes this custom firmware for their board and me and you will never notice.
The special wiring for s-plinking and such I though was more of a gray area, like there things you can do and not do.
Good thing about the old cabs, as you eliminate the controller PCB/encoder the input lag from controls disappear.
Any lag then would come from the local machine lag, especially if CRT screens are involved.
The Encoder chip ( a 4021 serial chip) first appeared in the NES controllers only appear as it was cheaper to include the chip than to give every input it’s own dedicated signal line to the console.
Silicone was cheaper than cables. And the logic at the time was the microsecond delay would be not notice by players. I don’t know the lag input the NES game pad operated on, sub 1ms times?
I know with testing software people for sub 1ms times on the SNES controller using the 240p Test Suit rom (the home brew app was for testing video setting and video lag, one test is a manual lag test).
Honestly now that I think about it, @Teyah did you eve consider how the240p test suite might apply to future testing?
Its mostly for older/retro systems so I know its out of your main focus, as you are mainly working on current gen systems.
Too bad we can’t get anything like this for modern consoles. Maybe the PS3 if you jail break it and made a custom port of the Test software. But that more work that I expect out of anyone.
No, however we would be able to see obvious things like any PCB changes or wiring mods (like select button blinking).
I just have this vision in my head of MadCatz using Hori VX:SA pcbs on the sly
For small tournaments then yeah, I’d happily make people play on a cab, and if people start whining saying want to use a pad then fuck 'em. But for something like EVO you’d need to find an awful lot of identical cabs to cover all the pools.
I wonder if the Bring your own controller rules can be applies/ twisted to a Neo Geo Plus/Super Gun Sticks only (just 1 joystick/4 directionals and 6 action buttons, start and select) and the tournament provides their own encoders which are in project boxes, and you plug your stick in using a DB 15 cable. The only thing that should be in your stick is straight wiring and switches, no actual PCBs allowed. You also never have to worry about what console the tourney runs on as everyone forced to the same NEO GEO Plus standard. And as long as that tournament all have the same PCB/encoders for each project box stationed for that console, there no issue if one PCB lags more than another as everyone be forced to use the same PCB.
Some people are very capable of hiding mods. Make it look as if it wasn’t modded at all. It’s all moot anyway since no one does any hardware cheats to warrant any checking.
