In reviewing displays, one aspect that seems to go almost untouched by reviewers is that of input lag. For people like myself who don’t play many video games, this idea is an afterthought that we don’t even worry about. With computer displays, this is an area that attracts far more attention but also has a lot of work to go into it, making it hard to do accurately or easily. When I saw details earlier this year on how the Input Lag Tester from Leobodnar.com worked, I got on the list to get one as soon as it was available, which it finally is this month. Does this simple device make testing input lag easy for once?
Currently, when I test for input lag for computer displays at AnandTech, it is a complex process. I have to hook up my Sony CRT, which is widescreen and capable of 1920×1200 resolution, to the computer and then mirror the displays between it and the monitor under test. Then I go ahead and run SMTT, a program that lets me measure both input lag and pixel response time for the two devices. Once that is running, I have to take 20-30 pictures with my SLR camera, then review all of those images and enter the numbers into a spreadsheet to calculate these values. In the end, measuring lag takes at least 30-45 minutes per display, and more if there are game modes on the display.
With all of this hassle, I don’t do this testing typically for TVs and projectors, where the devices are located far away from the computer and CRT monitor, and testing is harder to accomplish. While SMTT does do a very good job and is much better than the previous FPS counter methods used, it isn’t fast or easy to do. The new lag tester under review works in a much simpler way. You simply hook up the HDMI cable to the output, hold down a button, and then a test pattern will appear on your screen. A photo sensor on the back of the lag tester reads a flashing white bar on the screen, and the delay from sending the signal to reading the flash is calculated continually and displayed on the screen.
Compared to the SMTT method, this is done in 30 seconds with a number you can quickly write down and with no spreadsheet. Testing game mode simply means turning it on, holding the button, and doing another test. No SLR, no CRT, no display mirroring, just a quick number that you can write down and know the input lag. Here are some of the numbers that I obtained in 5 minutes of testing with the Lag Tester, using different modes on the displays.
|Lag Test Results|
|Samsung PN50B650 Plasma|
|Game Mode Off||90.4 ms|
|Game Mode On||49.5 ms|
|Sony VPL-HW50ES Projector|
|MotionFlow Off||67 ms|
|MotionFlow On||62.9 ms|
|Nixeus VUE27 Monitor|
|HDMI Input||28.4 ms|
The first thing we see is that the game mode on the Samsung makes a very noticeable difference in performance. As people debate how well-calibrated the displays of people creating games are, for many, sacrificing a bit of visual accuracy for the lower lag is certainly worth it. We also see on Sony that MotionFlow creates a small improvement in lag time. This is a surprise, as frame interpolation typically requires a frame to be buffered, which is a minimum of 17ms of lag introduced.
The Sony refreshes at 240 Hz, so with a 1080p60 frame, each frame is repeated 4 times to get that one frame without any interpolation. With the MotionFlow interpolation enabled, it seems that the first flashing frame is interpolated into the last black frame before it, so you get it 1/240 of a second faster, or around 4ms. The real issue here is that there seems to be a full frame buffered on the Sony at all times, even without MotionFlow enabled, but enabling it gives you a slightly faster response time.
The real test was with the Nixeus display, as that has been tested with SMTT before, so I have lag and pixel response times for it. With SMTT, it measured 30.2 ms for combined pixel response and lag. With the Lag Tester, it measured 28.4 ms of lag. Why is there a difference here? With the lag tester it is testing how quickly it goes from a black screen to a measurable white block for the light sensor in the device. On SMTT, we are looking at the full pixel response time, which would be black to peak white, not just measurable white. Because of this difference, the times with the Lag Tester will be lower.
That said, this isn’t really much of an issue. SMTT times are worst-case scenarios, as we do a total sub-pixel change (blue to yellow, for example) and measure only the peak reading, which will take the longest. The eye will see the change far before the peak level is reached, so the Lag Tester results might be more real-world-based than SMTT. We also can’t separate pixel response and input lag measures on the Lag Tester, as we can with SMTT. This removes a bit of granularity from the results but doesn’t change the effective feel of the lag.
There are a couple of other issues with the Lag Tester when compared to using SMTT. You can’t test multiple inputs with the Lag Tester, such as DisplayPort, because it is designed for TVs more than PC displays. Additionally, you can’t use a resolution other than 1080p or 720p, as it is fixed at one resolution and cannot be changed by the user. There is also no way to adjust the colors being used from black and white to something else like SMTT because of how the Lag Tester works. Finally, you might wish to be able to save the results instead of reading them off the screen, but that’s really just me being picky now.
I would have liked to see at least some instructions come with the tester, but I figured it out in two minutes and then was able to start measuring displays instantly. So far, the results seem to line up with what I would expect, and it does it so fast and easily that I can do it on display without having to budget time to do it. It is certainly a special interest item that only reviewers might care about, but for me, it will be a good part of my reviewing toolkit so I can see how other displays stack up, even if I’m not a good enough video gamer for it to matter for me.