Lagless HD Screen?

[Edward_Tz]

For the lulz, that just popped on [H] http://hardforum.com/showthread.php?t=1853884&page=257

0ms monitor huh. That Bodnar tester sure is accurate...


Ed Oscuro this is why there needs to be some kind of control. That monitor will be paraded as the best from now on.

I'm curious how this tester can ever read lower than 8ms as the average. Doesn't it prove the inaccuracy of the Bodnar tester if those are the numbers we get?

[Xyga]

It's a 55" TV ! The guy on [H] is very happy but who's gonna explain to him why and how this is wrong ?...

[Sixfortyfive]

I also wanted to mention the 8.5 (or 8.3333...ms) issue. So yeah, in theory and in reality the Bodnar tester's mid-screen reading should be no less than roughly this number.

I just want to expand on this a little bit. The theoretical minimum for this value doesn't have to be 8.3ms. I believe it's tied to the screen's refresh rate. Here's an example of the LB tester being used on a CRT VGA monitor:



(I had to use an HDMI-VGA adapter to convert the signal and pass it through my capture card because I had no gender-change adapter at the time. I don't think their effect on the output is statistically relevant.)

The LB tester measures the time elapsed between (a) the current frame being sent from the tester to the display, and (b) the time at which the sensor bar that you're measuring is finally drawn to the screen. CRTs and most LCDs draw to the screen in a similar fashion, starting with the top line and ending with the bottom, then going back to the top to start the next frame. At 60 Hz, it takes about 16.67ms to complete a single frame (including the non-visible data like blanking and other things that the more technically-inclined among us could probably explain, if needed). With that in mind, the numbers in the above picture should make sense: the top bar measurement is close to 0 (as there is basically no processing delay in a CRT, so the picture starts being drawn immediately), the bottom bar measurement is close to 16 (as it takes about 16.67ms to complete a single frame), and the middle bar measurement is about halfway in-between.

Compare this to the result seen in a 120 Hz flat-panel:



The top bar is 13ms instead of 1ms. There's more processing involved, so it makes sense that the LCD takes longer to start drawing the picture than the CRT does. But notice that the bottom bar is 20ms.

120 Hz: frame completes in about 7ms (20 - 13)
60 Hz: frame completes in about 14ms (15 - 1)

So I think it's reasonable to say that you can actually get a lower measurement on the middle bar than 8.3ms. You'd just need a display with a faster refresh rate than 60 Hz and close to 0 processing delay.

But more to the point: that just backs up your original argument. I think the numbers provided by the tester and similar methods are more valuable overall than CRT comparisons for this and other reasons. It doesn't make sense to treat a CRT's measurements as a reference for "zero" lag. Every CRT is different (Fudoh sites an 8.5ms value up-thread; my CRT monitor in this post is sub-8.0), and it's theoretically possible for a flat-panel with a high refresh rate to beat a 60 Hz CRT (although I don't expect that to be the case for common displays anytime soon, if ever).

This, however:

For the lulz, that just popped on [H] http://hardforum.com/showthread.php?t=1853884&page=257

is puzzling, because a measurement of literally zero doesn't make much sense.

I'm curious how this tester can ever read lower than 8ms as the average. Doesn't it prove the inaccuracy of the Bodnar tester if those are the numbers we get?

It probably wouldn't hurt to make an effort at understanding how something works before spouting off with a know-nothing know-it-all attitude.

[Jack Burton]

I believe the P-series 4k Vizio sets all have a gaming mode that's around 17ms through a particular input. The sky's the limit on screen size there.

I have the 60" model. Great contrast, very little lag, and huge. I've read that the sony models have better motion resolution, but after a few firmware updates on the vizio, I can't find anything I'm not happy with. I did buy the extended warranty because I read there were reliability problems with the vizios.

[Edward_Tz]

I'm curious how this tester can ever read lower than 8ms as the average. Doesn't it prove the inaccuracy of the Bodnar tester if those are the numbers we get?

It probably wouldn't hurt to make an effort at understanding how something works before spouting off with a know-nothing know-it-all attitude.

I'm not an expert but I do understand not all displays start from top to bottom and also why you want the average. Nothing in your post answers that question but thanks for saying I have a "know-nothing know-it-all attitude" when I was asking a question. You bring up another question though.

So I think it's reasonable to say that you can actually get a lower measurement on the middle bar than 8.3ms. You'd just need a display with a faster refresh rate than 60 Hz and close to 0 processing delay.

I thought the Bodnar worked at 60hz. Which is why it confuses me as to how it could measure lower than 8ms.

Maybe I'm misunderstanding this. Lets say in theory and ignore different display technology. Wouldn't a 60hz display mean the fastest you could ever draw the next frame be about 16ms if timed from the very start of the prior frame? So wouldn't that mean in theory it's impossible to ever have lower than 8ms of lag averaged? Assuming you measure the lag from the top, middle, and bottom of the screen.

Am I misinformed here? I want to understand this.

[Ed Oscuro]

@ Sixfortyfive: Yes, I've simplified things by assuming 60Hz. Leo Bodnar's site says this:

The minimum result possible at the bottom of the screen should be 16ms (t = 1 / f = 1 / (60 Hz) = 16.67 ms).

It is very interesting that you've got a display which is showing a lower figure than this.

The only obvious thing that comes to mind is that the CRT is displaying fewer than 1080p lines, but the lines are still in the same position. I have to wonder if the figure would change with a 720p tester.

What happens when displaying a 60Hz signal on a display that's faster than 120Hz is definitely another unknown. With the example here, there's still no middle bar measurement less than 8.3333...ms; instead it looks like the display "caught up" to the data already buffered / latched. I wouldn't think you can get that to a lower number as the entire half frame has to be transmitted from the Bodnar, which should still take 8.333...ms. Again, the only guess I have is that the CRT results hold the essential clue here.

I think one thing to take away from this is that a more flexible tester design may be needed for supporting resolutions and framerates outside 720p60 / 1080p60. The lag tests for displays faster than 60Hz still seem accurate from those I've seen, but it would take away some sources of confusion if we knew that a 144Hz monitor was getting a 144Hz signal.

@ Edward: In your latest post I think you're right. This is an area where I'd agree the black box nature of the Leo Bodnar makes understanding the theory behind its measurements more difficult. It could well still be accurate but we obviously need to know more.

[Sixfortyfive]

So I think it's reasonable to say that you can actually get a lower measurement on the middle bar than 8.3ms. You'd just need a display with a faster refresh rate than 60 Hz and close to 0 processing delay.

I thought the Bodnar worked at 60hz. Which is why it confuses me as to how it could measure lower than 8ms.

The video signal that the tester sends might be 60fps, but the light sensor that you place over the screen is much faster than that. It has to be faster since it's capable of reporting fluctuations of one-tenth of a millisecond.

What happens when displaying a 60Hz signal on a display that's faster than 120Hz is definitely another unknown.

I'd like someone with a 240 Hz panel to try it out and see what happens; I don't have one to test. The 60 Hz and 120 Hz screens I've tested have been pretty consistent, be they CRT or LCD: the 60 Hz displays have a ~14ms difference from top sensor to bottom sensor and the 120 Hz displays take about half that time.

[Edward_Tz]

The light sensor doesn't change that there's 60 frames a second going to that TV. (simplified)

It does not make sense to me with what I understand for any tester to measure lower than 8ms averaged given the variables we're discussing in this thread.

[Sixfortyfive]

The light sensor doesn't change that there's 60 frames a second going to that TV. (simplified)

It does not make sense to me with what I understand for any tester to measure lower than 8ms averaged given the variables we're discussing in this thread.

You're measuring the lag of the display (how long it takes for the display to show the picture). The framerate of the video source is irrelevant for this specific purpose. You're basically measuring how long it takes for a single frame to be sent from the source and drawn to the screen; the previous and following frames aren't part of the equation.

The "8ms" minimum value for the middle bar assumes a 60 Hz display. 60 Hz = 16.67ms per frame. Middle bar = halfway through the frame = about half the time of a single frame. And those are all approximations. As seen in the CRT monitor example above, the middle bar is a little less than 8.0 milliseconds, likely because there's more to a single frame than just the pixel data that's being drawn to the screen.

Double the display's refresh rate and you cut the time it takes for the display to finish drawing a frame to the screen in half (ignoring how long it takes for the display to start drawing it). Even if the source video is only sending 60 frames per second, the display can still show a single frame faster than 16.67ms and either wait or do other stuff (e.g. motion interpolation) before the next frame comes in.

Pretty sure that the timing of the tester's video output (60fps) and the timing of the photo sensor are unrelated to each other.

EDIT: It would be illuminating to see how the tester actually behaves on a fast 240 Hz panel and if it's consistent with all of the above. I can't say for sure if it's the case.

I wouldn't think you can get that to a lower number as the entire half frame has to be transmitted from the Bodnar, which should still take 8.333...ms.

Does this have to be the case? I wouldn't know. Someone more knowledgeable than me shed some light on this. When a video source sends a 60fps signal over HDMI, does it take a full 16.67ms to send a full frame, or does it take just a fraction of that time and then wait for the next frame? Would this change depending on which HDMI spec the device uses? HDMI 1.4, for example, specifies support for 1080p @ 120Hz. I think it would be reasonable to expect (but admittedly not guaranteed) that an HDMI 1.4 device could send a complete frame in 8.3 ms or less instead of 16.7ms, regardless of whether the frame rate of the video actually is 120, 60, or some other value.

[Edward_Tz]

How does it explain a lower than 8ms reading from the Bodnar tester though?

[Sixfortyfive]

How does it explain a lower than 8ms reading from the Bodnar tester though?

I don't understand why you're fixated on this specific value. Why do 8ms and 16ms have to be hard limits for the middle and bottom bars? Keeping in mind that the CRT example upthread already falls below those numbers.

A device that outputs a 1080p60 signal over HDMI has to be able to encode and send a full frame in 16.67 ms or less. If the HDMI spec used is capable of sending a 1080p120 signal, for example, then it is capable of sending a full frame in 8.3 ms or less. So, what's to say that it can't send a frame in less than 8 ms, then sit around and wait another 8 ms to start sending the next frame?

I don't have the answers to those questions. It'd be interesting to know what the theoretical minimum readings are for the device. Whatever that minimum is, though, I'm pretty comfortable in saying that probably no existing consumer flat-panel display reaches it.

And it's much less of a concern than the troubles involved in a camera test.

[Edward_Tz]

Because about 8ms (not exact alright) is the CRT lag the Bodnar measures. It is essentially lagless. You can't physically have something with less lag that wouldn't be negligible. If someone is going to claim otherwise is outright bs. If the Bodnar measures less than 8ms(again not exact) then it's not measuring lag. It's measuring a faster scan rate.

The photo tests with a CRT shows you exactly how much slower the comparison display is. You're comparing the change. The test has a control and I think it's the better test.
The Bodnar test is just giving you numbers. It's measuring the time it takes to display something.

That's how I see it.

[Sixfortyfive]

It is essentially lagless. You can't physically have something with less lag that wouldn't be negligible.

You have a fundamental misunderstanding of what's actually being measured. I don't think I can spell things out any more explicitly than I already have.

Yes, even the scan rate is part of the total display lag. It's a major reason why the Bodnar tester outputs 3 different sensor bars to 3 different regions of the screen in the first place. The extra 16 ms a screen takes to finish drawing a frame is relevant, especially when there are screens out there that can complete that same operation in half the time or less.

I understand the value in knowing how much lag an LCD has "compared to a CRT," but you can get that value more easily with a Bodnar tester than you can with a camera test: just subtract 8 ms from the Bodnar test result, and that's your value for "lag vs. a CRT."

[Edward_Tz]

It is essentially lagless. You can't physically have something with less lag that wouldn't be negligible.

You have a fundamental misunderstanding of what's actually being measured. I don't think I can spell things out any more explicitly than I already have.

All the Bodnar is measuring is light and time. I know exactly what it's measuring.

A CRT that takes 16ms to scan from the top to the bottom of the screen isn't slower than that TV that the Bodnar tester measured 0ms on. No input is going to be registered and displayed faster on that TV then on the CRT. That TV may finish displaying the frame a bit quicker but the lag would be the exact same.

[Sixfortyfive]

No input is going to be registered and displayed faster on that TV then on the CRT.

A 120fps source displayed on a 120 Hz CRT is entirely possible. Such a combination would be displayed in half the time as a 60fps source on a 60 Hz CRT. Is the Bodnar tester precise enough to accurately measure the former? Maybe not. But this is the reason why you shouldn't consider the 60 Hz CRT to have "zero lag;" it's technically possible to display an image faster than that.

You can ignore that 0.0 ms reading on that LCD posted up-thread. It's pretty obvious that the tester isn't picking up a signal from the display and the poster is flipping out about it for no reason.

[Edward_Tz]

Even in theory it wouldn't. You could have the 120hz or 60hz TV display the whole frame in how ever realistically fast you want it to be displayed in. It's still not faster than the CRT. Could be equal.

When that console outputs that signal with the new frame in it that CRT is displaying it as fast as the circuit will allow. It's laglass.

You can't speed up that next frame only slowdown the current one.

Now if that TV did display the whole frame instantly then you can argue that the TV had a whole frame up longer than the CRT. That's not lag though.

[Galgomite]

I believe the P-series 4k Vizio sets all have a gaming mode that's around 17ms through a particular input. The sky's the limit on screen size there.

I have the 60" model. Great contrast, very little lag, and huge. I've read that the sony models have better motion resolution, but after a few firmware updates on the vizio, I can't find anything I'm not happy with. I did buy the extended warranty because I read there were reliability problems with the vizios.

Glad to hear you say that. I saw one (the 50" I think) in a store and was extremely impressed with black levels. Good purchase, sir! Surprised there isn't more chatter about that TV, sites like DisplayLag don't even include it.