This is the explanation for how I test for input lag and what setup I use.
The basic setup is that I use an LED light that I have wired into the circuit of the down movement input of my arcade stick. I have also added a watch battery to the circuit so that the light does not eat the arcade stick input and also shines brighter (the watch battery is the secret sauce).
The reason why the down input has been selected is because certain shmups have built in delays into their shots for whatever reason (like Akai Katana), and also because the input delay of movement is the most important thing to test for. Also, I only test samples that occur when NO SLOWDOWN is happening. Input lag scales with slowdown (it would make no sense if it didn’t), so measuring lag during slowdown will only skew results. However, there are some games, like SDOJ, that seem to experience extra delay during the middle of the stages, even when there does not appear to be any slowdown occurring.
In addition to this LED light, I also use a VGA CRT computer monitor at 480p. I do this because a CRT at 480p is the fastest responding display type that I am aware of (though a CRT at any progressive resolution should be the same). Also, all Vsync and post processing effects are disabled when possible. Vsync being on or off is noted in the data. In that same vein, with the input methods I use, I do NOT use any sort of input converters and strive to use the most responsive controllers/arcade sticks that I am aware of.
For the consoles that do not have native analog output (damn them) like the PS4 and Switch, I use this special HDMI to VGA cable called the HD Gamerfury 2. I have tested this cable for lag and, as far as my testing has shown, the cable appears to be lagless. (It better be, considering I paid $100 for it and it uses an active power source).
Lastly, I have captured the footage at 60 fps and then use frame advance in VLC to count the frames between the light turning on and the ship moving. Each test is around 30 samples and results are typically very consistent between the samples. In the past, I ended up having a problem with my previous camera which advertised 60 fps, but only delivered around 30 fps. Since then I have been sure to check the codec information on each and every video with my new camera.
So this is how I am achieving the numbers in these tests. Overall, I am confident in the results because I have minimized any sort of guesswork and am working with the most responsive equipment I am aware of. In fact, one thing to keep in mind about these results is that they are likely LOWER than what you will typically experience because I am eliminating display lag. So, depending on the setup, you may have higher levels of latency, but you shouldn’t be able to achieve lower levels than these tests.
I think this is going to be a very useful resource for the community, plus it’s a personal obsession of mine anyway.