kbwenLab

Illusion Ruler

Adjust until they look equal, then see how far off you were.

Quick guide

Use cases, answers, and nearby tools

Compact below-tool notes that help first-run users and repeated visitors move faster without changing the main interface.

Chinese search: 視錯覺、錯視、視錯覺測試、錯覺測試、錯視圖形、錯視實驗、繆勒萊爾錯覺、艾賓浩斯錯覺、波根多夫錯覺、垂直水平錯覺、同時對比錯覺、Cornsweet 錯覺、眼睛會被騙、錯覺強度、龐佐錯覺、佐爾納錯覺、懷特錯覺、錯視量尺

How to use

Run a clean first pass

  1. Drag anywhere on the stage until the two things look equal — same length, same size, or same lightness. Commit when you are satisfied.
  2. The context fades out and a red ruler drops in showing how far off you were. Three trials per illusion, and in one of them the inducing context is removed entirely — that is the control.
  3. After three trials the tool subtracts your control error from your context average. That difference is the illusion’s real strength on you. Finish all nine for a full ruler, or stop whenever you like — what you skipped is marked not measured, never counted as zero.

Examples

Real jobs this page helps with

  • Why a no-context control trial?Because everyone has a baseline matching bias. Asked to make two plain lines equal, most people are still a percent or two off — some habitually long, some short. The control measures that baseline. Illusion strength is the context error minus the control error. Every "illusion test" that reports raw error is not measuring you; it is guessing with a percent sign attached.
  • Simultaneous contrast and White's: one grey, two opposite answersIn simultaneous contrast a grey patch on a dark surround looks lighter, which is what you would guess. White's illusion puts the patch ON a black stripe and it still looks lighter — even though it now has more white beside it, which is the opposite of what contrast predicts. Having both is what makes your ruler say something interesting: whether your visual system is led by how bright the surround is, or by which edge it runs along.
  • Poggendorff and Zöllner: two different ways of losing directionPoggendorff interrupts a diagonal with two rails and asks you to put the right segment back on the same line — the one people overshoot most. Zöllner takes two genuinely parallel lines, crosses them with oblique hatches, and they stop looking parallel, because the visual system overestimates acute angles. One tests continuation across an occluder, the other tests the tilt of the line itself.
  • Then drain the illusion and watch it goEach measured illusion unlocks one slider. Pull it down and the fins shorten, the surrounding circles shrink, the rails retract, the two surrounds merge. At zero you are looking at the control itself — not a separate diagram, the same parameter at zero, which is why "the control" and "the slider at the bottom" are literally the same state.

FAQ

What people usually want to know

Why measure a no-context control for every illusion?

Because everyone carries a baseline matching bias — even with no context at all, most people land one or two percent off when making two lines equal. The control measures that baseline, and illusion strength is your context error minus it. Without the correction the number is not a measurement.

Why no Hermann grid, Kanizsa triangle or motion after-effect?

Because they have nothing you can adjust to perceptual equality. The one entry requirement is that an illusion be nullable: there has to be a physical parameter you can set until the two look the same, or there is no error to compute. Those three can only be looked at. Delboeuf is nullable but is a variant of Ebbinghaus on the same mechanism, so including it would be measuring one thing twice. The nine here are nine different mechanisms: terminators, perspective, surrounding size, collinearity, orientation contrast, orientation itself, and three lightness effects with three different causes.

Can I compare my number with other people?

No, and you should not. Screen calibration, viewing distance and a short session all move the absolute value. The tool prints no "typical adult" figure anywhere, because every such number would need individual source-checking and an unchecked one is fabrication. What is honest to compare is your own data: which illusion is strongest on you, and how far apart your two readings were.

Are those two greys in simultaneous contrast really the same?

They are, and they are painted from the same sRGB hex, so the claim holds whatever your screen is doing. Everything you see between them comes from what is behind them. Commit and the two surrounds converge to one grey — the patches become plainly the same colour.

Why is the target value never shown as a number?

If the first reveal told you the answer was 200, trials two and three would stop being a perceptual task — you would just dial back to 200. So each trial’s reference is jittered within ±9% and the interface only ever prints percentages. Varying the standard across trials is standard psychophysics practice, not coyness.

Does a high score mean something is wrong with my eyes?

No. How strongly an illusion works on you reflects how your visual system integrates context, not visual acuity, and certainly not intelligence. There is no personality or IQ reading here. If you are genuinely worried about your vision, see an optometrist.

Do I have to measure all nine?

No. Three trials on one illusion gives you that illusion’s strength immediately, about forty seconds. Stop whenever you like — anything unmeasured is labelled not measured on the ruler, never counted as a zero. All nine takes roughly five minutes.

Related tools

One Shade Off Scale Sense Contour → Terrain Profile