How we test Input Latency
Controller to screen times and how we do it!
As input latency is something I cover in my list of expanding game and hardware tests I wanted to detail my methods and how it all works, you can check out the video covering this below with some game tests also shown.
Many sites cover this and some have devised various methods that can capture and allow you to analyse the time it takes. These all fall into a similar method though, capturing the time you press the chosen input and when the onscreen action is updated, with many factors affecting this. From the widely used and known Benheck version that can be used on XboxOne, PC and Xbox360 with a signal board representing input which is used in the industry. Along with other methods which interrupt the output from your TV feed and blank out the sections when a button is pressed which may also add a small amount of input latency itself.
My method is quite simple, I find the traces on the controller board itself that I can solder onto using the via. This is then fed out into a single line feed delivering a few volts when any of the buttons or analogue sticks are pressed. Not a great deal of volts but enough to light up a small LED that I have attached to the screen, so whenever any button is pressed the light shows when the controller received the input and then with a High-speed capture camera the frames between this lighting up and the action on screen are counted, giving us an accurate number that reflects the input latency. Because each action will have a different level of polling latency the average is given across multiple tests from punching, shooting, moving etc and this becomes your final measurement.
Because my method is wired directly from the controller and not from an output feed it means that I can fully test them in wired or wireless mode to demonstrate if this has a significant effect on the performance, across multiple pieces of hardware. I capture the 60Hz screen with a minimum of a 240fps camera and a maximum of 1000 frames per second. Although this is not really required due to the screen refresh rate itself becoming the limiting factor way before the camera speed ever will, which is true of all displays anyone may own.
Does the TV matter?
But as I have covered in detail already a few years ago with my “How to set up your TV for games” your display itself and even more so the relevant settings will have a great impact on your chosen games response times, far greater than you may think, check that out below to learn how to get the best out of your TV. I will also have my new HDR one up soon that will do the same for your 4K screen and output image quality.But for these latency tests I use a calibrated Monitor to get the best setting and lowest input lag possible, the Acer model I use is giving me around 9ms of input latency so under a standard frame refresh rate for a 60fps title, so this is deducted from the count to give a true engine result, this being the polling rate from your input being read and influencing the rendering on screen.
The process then is all very simple, plug in the chosen controller to the LED connector, video the screen and then press buttons to see what happens. Playing back the video, you can simply count the frames from when the light is green and the screen updates from your chosen input, simple. This clean and unobtrusive method is the best way to test, as it does not involve affecting the input from the controller or output to the display in anyway and as such is fully reflective of what you will experience yourself, TV type and settings non-with standing. With that said the input times that should really make the difference and be relevant are the ones that hit the screen, a ~10ms monitor here is faster than pretty much 99.9% of commercial LED or OLED screens you will buy for your house, so with that included for the tests it is still a best case scenario unless you have an expensive higher input screen which are not common, these tests are reflective of the majority of users that will be watching this video and playing these games.
So now you can see how I test the games and hardware and just how it influences the games that you play. As I covered in my long list of titles tested & demonstrating that the XboxOne can sometimes use triple buffering on its titles to help improve performance at the cost of latency, whereas the PS4 runs the more standard double buffered rendering approach. This can also show the latter if/when it occurs along with titles that can perform differently on each machine. I will be back soon with more tests but the one I wanted to demonstrate here was how much having the wrong setting on a TV can increase your control and response from titles. By using a Sony W6 LED 1080P screen which has one of the lowest input lags on a standard display coming in around 14.7ms it is an incredibly fast & responsive screen. The newer W7 beats this by another millisecond but you get the point. If you turn on the image options to improve the picture quality (which do not really, outside of setting it up correctly) then you can nearly triple this figure which will of course include turning game mode off. Giving you an increase in Input latency of nearly 50ms, and many other sets can add even higher gaps with the wrong settings. So, if you do nothing else today, check out the video below and get your TV set-up correctly for games, and check back on the site and channel soon for more in-depth tests.