Question on input lag and HD TVs

maybe i can't look at a laggy display and know exactly how bad it is just by that, but i know lag when i see it. when i tested the tv via hdmi it lagged by roughly 8ms compared to the laptop used to test it (the laptop clocked in at 2ms of lag compared to the benchmark crt, so if you add that to it it's more like 10ms of delay).

for the record, when i tested it using component (via an rgb to y/pb/cr transcoder, as the laptop only had hdmi and vga out), there was no lag at all.

so, maybe i didn't mentally calculate that it was lagging by exactly 8ms just by seeing it, but i felt the presence of lag (missing visual cues in both sthd and battle fantasia was a bit unsettling) and it proved to indeed be there.

<div class="ubbcode-block"><div class="ubbcode-header">Quote:</div><div class="ubbcode-body">Again, specifically how do you 'test' a difference of 8ms of time using a clock whose hand only moves in discrete 16ms increments . . . </div></div>

you did click the link i posted a few posts back, right? timecode tests are the only way to test for input lag. you'd be surprised how even "virtually lag free" displays can perform quite poorly once you put them to the test (no pun intended).

bottom line: the only way to completely eliminate lag is to use an analog display (crt) using an analog connection (composite, s-video or component). lcd's can come as close as they want, but will never be able to achieve through years of "improvement" what "outdated" technology does by nature.

edit: after re-reading your post, i think i know what you mean now.

there was a visual delay of 8ms compared to the laptop which was off by 2ms, so let's say 10ms of lag. then add in the fact that it's lagging on top of the 16ms the tv takes to rescan and we end up with 26ms, roughly half a frame of lag. not necessarily the hardest to detect, seeing as it can easily screw up your timing for one frame links and such.

 

Super Turbo arcade machines running CPS2 have 2 to 4 frames of lag from the time you press the button until you see it happen on the screen. The PS2 version of Super Turbo (on Capcom Classics Collection vol. 2) has 5 - 8 frames of lag from the time you press a button until the time the move comes out on the screen.

So both have input delay, but compared to the arcade, CCC has additional frames of lag that really throw people off who are used to the arcade.

If you test out a television using VF5, and find out that there is a one or two frame delay, it might be arcade perfect. Does anyone here know how many frames it takes for a button press to show up on screen for VF5 running on Sega's Linderbergh unit? If it is 0 - 1 frames then yes, no console is going to be arcade perfect, in terms of input delay no matter what connection you use.

The point I'm making is that input delay is present at the arcade, so trying to find a TV with instant response to a console is meaningless for some games (like ST). If VF has a frame of delay at the arcade, then arcade perfection is one frame of delay.

All in all, I loathe HDTVs for this problem, in a rush to manufacture these things gamers have been left in the dust at the drawing board for this technology. Consoles have always had input delay on them, but with HDTVs now, the delay is incredibly noticeable on some models.

 

You're using a camera with a time resolution of 125 milliseconds, comparing displays that fully update only every 16 milliseconds, and you're claiming you can measure a difference in lag to an accuracy of 8 milliseconds? I mean, yeah, its kind of fun to take pictures, but at that point its just noise, not science.

Even if you had a fast enough camera, at 8 ms you're looking at the equivalent of comparing whether the electron beam made it 25% of the way down the screen on the "laggy" CRT vs 75% of the way down the screen on the "perfect" crt. . . which due to the way phosphors work makes basically no difference to your eye.

Like instant overhead pointed out, even in the arcade there is always a delay for a game to read an input, update the state of the engine, and create next frame for display. Half a frame of lag (which is 8 ms, not 26ms) isn't going to screw up your timing for anything. It's just as likely that your input happened to come in during the middle of a cycle and wasn't processed until the next one. Even another computer can't synchronize inputs to a game with 100% accuracy, you're going to miss a cycle here and there. I see this all the time with my programmable equipment.

The bad HDTVs are the ones that have like 50-100ms of lag, that IS going to affect your game, and IS noticeable. 8ms is negligible.

 

i agree. it was just the mere fact that using a digital connection added any lag at all was worth bringing up since that's where the conversation segued.

i do appreciate the insight though (from both you and instantoverhead). i guess i never really took into account that there could be more forces at work than i initially thought.

 

I remind that one frame is about 16,7 ms. Lag less than that is indeed ok for play purposes, but I swear that Ive seen a HDTV setup where lag must have been 100-200 ms. That is what I want to avoid.

 

Yup, I agree. Too bad I havent seen a reliable site for distinguishing hdtvs with whatever minimal lag is inherent in digital signal processing from the hdtvs that have multiple frames of lag. I've seen some for computer monitors, but not for hdtvs. . . and places like SRK are full of people saying "d00d my hdtv doesnt lag" without any kind of testing.

 

I was reading this thread and I thikn I somewhat understand what's being talked about? Anyway, I hate on you all that, but I got to say I agree with this post.

It's all relative to the standard right?

Also, aren't most new arcade cabinets and machines using HDTVs anyway? From some videos where I see of VF5, VF5R, they're playing on the cabinet with HDTVs. And SF4 cabs had HDTVs, and so does T6 and T6BR.