11
votes
How fast is a typical keypress?
I recently learned about the Wooting two HE, a keyboard with analog input. The interesting feature is something they call "rapid trigger", which defines the actuation point as when the key changes direction (up or down), rather than the key reaching a specific height.
I'm wondering how much faster this could let someone press or release a key. I tried searching online but I can't find anyone who has measured the average time for typists to reach the actuation point, or even the average velocity of a key-press.
In case that's not where you found out about it, look into the osu! community's discussion on the keyboard. Since it's a game where very fast key presses matter a lot, it made a big splash when players started using it. You may be able to find some numbers on it on the forum or subreddit.
I searched around, but almost all of the discussion is whether rapid trigger counts as cheating (I don't think it does, it's comparable to tablet vs mouse).
I have a regular mechanical keyboard with Cherry MX Speed (linear) switches. I just took a 60 fps video of me pressing a key a few times at my guess of my typical speed.
It always took about 2 frames, so if we assume an actuation point halfway between the top and bottoming out, the average key press or release takes
2 frames * (1/60 seconds per frame) * (1/2 keyboard actuation distance) = 0.016666 seconds = 16.666 millisecondsThat seems very fast to me, I would have expected something on the scale of 50 to 100 ms. I kind of doubt this test is accurate.
You need to test delay all the way to being registered by the computer. There's also delay for the board to register the key press and send it to the computer but that's probably minimal... but it all ads up.
Linus Tech Tips did a video on AliExpress keyboards a few days ago and way at the end he says that their lab tested one board and got around 20ms latency and thought that was a lot IIRC. Again though, that was not leading key press time... more the microcontroller.
Regarding that new analog board... they're going to have to be doing a little bit of smoothing of analog inputs to make sure they're not getting false inputs. That's going to slow down registering of key presses a small amount. I think the benefit is likely on the reset time since the key doesn't need to come all the way back up to be pressed again. Also, maybe you could maybe set it to register a second key press on the up-stroke?
Regardless, I can't see any of this mattering for most people. IMO reading keys as analog buttons is probably more relevant to more people.
I'm aware there's additional latency beyond the physical key, but I'm assuming that it will be comparable to other keyboards. If my video test was actually accurate, then in the worst case of LTT's 20 ms the rapid trigger feature could nearly halve input latency. This physical latency could be the highest piece of the latency chain by an order of magnitude.
I think rapid trigger and analog input are mutually exclusive options (how do you define an actuation distance for an analog input?). If so, then there is no smoothing to consider for latency.
I skimmed through a bunch of YouTube reviews of the keyboard while looking for physical latency measurements, and it seems CS:GO and Valorant players can feel a significant difference in their ability to strafe.
I'm interested in this because I do play a game primarily focused on keyboard input, TagPro. It's an online game, and I can feel a difference in ping as low as 20-30 ms. If the physical latency of this keyboard is reduced by as much as I think it is with rapid trigger enabled then it would be a huge reduction in latency.
Ya, I think I'm basically just off base on some of my thoughts.
Wooting is claiming sub 1ms latency so that kind of blows any points I made out of the water. If you're coming from a slower keyboard then the the combined improvement would be substantial.
I imagine it will become THE gaming keyboard while I stick to my tactile switches like a scrub.
I'm wondering, wouldn't it start getting actually advantageous when you only do half or any other press below a full press? I press the keys completely in and therefore the speed would be more determined by the springs than anything else. This is probably flawed logic, so don't take it as fact but only as an input for further discussion.
I think a full press would benefit too, because you can "de-actuate" the key by moving it upward by any distance.
The springs would only be a factor if you can lift your finger off the key faster than the spring will push the key off. I don't even know if that's physically possible - even using my whole arm to try to lift my finger faster, it doesn't feel like my finger leaves the key before the key fully extends.
I understand this doesn't answer your question but that description of "rapid-trigger" is quite similar to how the "hair-trigger" setting for triggers in steam input works. Also, a big benefit of such a setting is that you can do rapid inputs not only faster, but also more consistently, because you don't have to worry about whether or not you crossed the actuation point.
Off-topic but a question for you, how is the Wooting Two? I had a Wooting One back when they first launched but found that a lot of games don't support it, what is it like today?
Back then I tried using it in many games and ran into issues where the game would see the analog input as a thumb stick input from a controller. The game would constantly switch between keyboard and Xbox controller inputs (where it would switch between keyboard and Xbox button icons). Are there many games now where you can use the analog button to control your run speed (like with a controller) without having this issue?
I don't have mine yet, I only ordered it yesterday. I don't play any games where analog input is super useful to me, so I haven't looked into support.
The official discord can probably help.
I found this YouTube video with slow motion of key-presses on a Corsair K70.
It's at 240 fps. Each press down takes anywhere from 4 to 10 frames, and the release is ~4-6 frames. That means a press is ~16.667-41.667 ms, and release (by finger, not by spring) is ~16.667-25 ms.