Frame Timing Test

What a frame actually is

A frame is one drawn image, and a game's timing windows are measured in them rather than in seconds. At 60 frames per second each one lasts 16.67 milliseconds. At 30 fps a frame is 33.3 ms, which is why older console games feel more forgiving. At 144 fps it is 6.9 ms.

That unit matters because game logic is discrete. An input either lands on the frame the window is open or it does not, and there is no partial credit. A move with a two frame window at 60 fps gives you roughly 33 ms of real time, total, to press.

This is not a reaction test

A one frame window is 16.7 ms. Human reaction to something unexpected is around 250 ms. Those numbers are not close, and the gap is the entire reason this test works the way it does: nothing here is reactable, so nothing here is hidden.

The marker moves at a constant, visible speed toward a line you can see. You are not being surprised. You are being asked whether you can execute a press at a moment you already know is coming, which is precisely what a frame perfect input in a fighting game or a speedrun demands. Players do not react to a link. They learn its rhythm and commit.

If you want the other skill, the reaction time test measures response to an unpredictable signal, and the two correlate far less than people expect. Being quick does not make you precise.

How to read your result

Hits out of ten tells you whether you cleared the window you chose. Mean error is the number worth tracking over time, because it does not change when you change the difficulty.

Early or late is the useful number

The bias figure is the most actionable thing this test gives you. A mean error of 25 ms that is consistently late is a much easier problem than 25 ms scattered either side: a consistent offset is a single correction, and scatter is a precision problem that takes real practice.

Most people run late on their first few attempts, and some of that is not them. On a 60 Hz screen the marker's drawn position can be up to 16.7 ms behind where it truly is, so you are aiming at slightly stale information. The measurement compares your press against the exact crossing time regardless, so the number is honest, but a small late bias on a low refresh display is expected rather than a personal failing.

Frame windows in real games

One frame windows are the ones that built the reputation. Certain links in older fighting games require an input on exactly the right frame, which is why they get drilled in training mode for hours rather than attempted on instinct. Speedrunning has the same category of trick: a jump or a menu input that only works on a single frame, practised until the rhythm is automatic.

Most real windows are more generous than one frame, and the useful question is not whether you can hit a single frame on demand but how tight your spread is. A player whose mean error is half a frame will land almost every three frame window in a real game, under pressure, which matters far more than one lucky frame perfect input in practice.

How to actually get tighter

Track the marker rather than counting. Counting produces a rhythm that drifts, and the randomised delay before each sweep exists to stop counting working at all.

Commit early rather than deciding late. A press you begin before you are certain lands more accurately than one you deliberate over, because the deliberation itself costs time and adds scatter. This is the same effect that makes second guessing a familiar word wrong on the verbal memory test.

Then use the bias. If you are late by 20 ms consistently, aim 20 ms earlier and check whether the spread narrows. If it does not, the problem was scatter rather than offset, and that only improves with repetition. Raising the frame rate setting does not make you more precise, it just relabels the same milliseconds into more frames, which is worth remembering before reading too much into a frame count.

For a test that mixes precision with target finding instead, the aim trainer is the closest thing here, and the F1 start lights sit halfway between this and a pure reaction test: a known procedure with an unpredictable release.