Quote:
Originally Posted by
sharky974 
No way I would trust the manual test. You can easily, subconsciously or otherwise, compensate for whatever lag there is and get 0 lag readings on any manual test. Not to mention, human reaction time alone is quite large.
I did literally dozens of runs with the guitar hero manual calibration on several sets, and in the end decided the results were worthless. I could get anywhere from 0 to 100 ms lag on the same set if I did enough runs. I dont get near that variability with the auto test. I fond myself way too easily "anticipating" when to hit the button. And I definitely believe human bias creeps in. I wanted my newly purchased set to perform better than my old one, so I had to constantly guard against bias when testing the new set (hitting the button early to get a lower lag number), since I can pretty much get whatever numbers I want out of the manual test. Then, I often wondered if by being so cautious to not be biased for the new set, I was actually biasing against it...
You can believe what you want, but I think you're believing what you want to...the higher auto calibration figure is likely correct.
If you can find a CRT to test, try that. I bet it will convince you the auto test is working.
All this has me more curious than ever, tonight I will try to do more testing with the RB2 auto. I'll test the 13" CRT some more, since I only did one run on it the first time. And I'll test my PC monitor, a 22" Westinghouse. I'm assuming the lag for it will be something like 20 ms, much lower than the HDTV's, but higher than the CRT...
Quote:
Originally Posted by
rahzel
Well you obviously haven't read through my posts.
My manual calibrations with my A650 are inline with people's results using the stop watch test whereas the auto calibration was consistently 15-20ms later. I've used both the built-in manual calibration and the complete manual calibration and have consistently got between 15-25ms, which is considerably lower than the auto calibration offset of 40-43ms (which again is 15-25ms later than my manual tests).
I've been playing drums on RB2 for years and I play much better with the offset set to 25ms or lower. Not to mention it's been proven that S-IPS panels have lower input lag than PVA panels.
Why would I believe what I want to when it would affect my performance in RB2? I wouldn't set incorrectly just because I
want my newly purchased TV to have low input lag. Ultimately, all I care about is what gives me better performance.
I don't think I will ever get my hands on a CRT monitor. Best thing I can do is get a laptop.
edit: Another thought: it could be possible that HMX purposely set the auto calibration so it makes the offset slightly larger for the more inexperienced players that are more likely to be late on a note than early.
I realize my findings aren't definitive... I just thought I'd share my findings using what I have.
i'm not sure why you guys are pursuing this test method, its likely to be flawed either way and most of all will never be used to produce real results do the simple fact that you can never prove the results
I need photo evidence proof that a TV has a X amount of lag before i will report its lag figure on the first post.
I know you guys got computers because your typing on them as you post and then you also have a basic PC monitor as well, cheaper the monitor the better because it means its a TN panel and its got next to no input lag, laptop screens work as well provided they are TN panels and most are. if your not sure what type of panel you have in you monitor or laptop theres a few easy ways to tell.
if your monitors color gets terrible or even inverts when you look at it from an off angle, most notably vertical off angles its a TN monitor no question about it. if they do not exhibit this problem then they are either VA or IPS. VA are the bad panel's that should be avoided for control monitors.
they can be identified easily by putting up a black screen and tapping on the display.
on an IPS panel you will just see a distorted picture where you finger directly press.
on a TN you will see some flash around your finger but not all that much.
on a VA panel you will see a big light halo for as long as your pressing on the screen.
VA Panels are pretty uncommon in PC monitors so its not a huge concern.
lastly i'm sure you got a digital camera, even if it does not have an adjustable shutter speed or ability to set a very fast shutter speed the results you get will still be within 1 frame of error (16ms) and if you take enough pictures you can determine how fast the TV is absolutely by either averaging all the results or just tossing all but the fastest results for the same round of tests in the same picture mode/resolution because this is what happens when you have a slow shutter speed
with a typical 1/60 shutter speed you capture the TV and monitor almost always inbetween 2 frames because when the shutter opens they are on frame 1 at some point in frame 1's hold time of 16.66ms the chances are very high that the shutter will remain open long enough for frame 2 to come into view by the camera.
what this results in is you see numbers overlapping numbers and ghost numbers from the previous frame(s) due to LCD's poor pixel response time
this is why running multiple timers is useful. only compare timer numbers between the same timer of course but having more than one timer running on screen raises your chances of getting a photo with both display's showing numbers as close together as possible and if you run 2-3 or more timers on screen it will only take about 10 pictures to get 3-4 low lag results out of the photo's even with a slow camera and that is your lowest lag figure.
alternatively if your running vista/win7 x64 I have a copy of a semi free program i paid about 5 dollars for off a European PC monitor review site. they wrote this timer to increase accuracy of the results by using directX rather than flash so the frame rates are sky high (2000+ FPS on my PC) and it runs depending on the resolution your in 6-8 timers down both the left and right edges of the screen. they are synchronized from left to right so the top left timer is the same as the top right timer and each timer below the top one is offset by 1ms
this means you get 6-8 chances per photo taken to get a clean low result
this timer is also configurable for different colors in the even your LCD monitor has a smearing problem with a certian color combination but this is unlikely with most recent displays
here is the website for the program if you wish to purchase a copy for yourself, as i said in my somewhat recently edited first post if you PM me your e-mail i will send you the Vista/win7 x64 version of the program that i have. they allow people to distribute the program for private use for free to friends. if you do want to pay for another version of the program that i do not have i would appreciate it if you contact me so i can get a copy of it to have to distribute to other users wanting to test their TV's
http://smtt.thomasthiemann.com/index_en.html
