Free Value Perception Drill

A value drill that keeps the sign of the error

An answer of 6.5 for a value of 5 is one and a half steps out. It is also one and a half steps too light, and that second fact is the useful one: a painter who reads every dark a step too dark can correct that habit on the palette tomorrow.

So each answer is scored as your answer minus the true value. At the end of a session the page gives the average size of your errors and, beside it, the average with its direction kept.

The six tasks

Name the value shows a grey patch on mid grey. On a surround puts it on a ground at least two steps darker or lighter. Spot in a picture circles a place in a painting or in your own photo. Value of a colour shows a saturated red, orange, yellow, green, blue or violet. All four are answered with the slider, in half steps.

Rank five shows five greys to press from darkest to lightest; the score is the number of pairs left in the wrong order. Which is lighter shows two greys on different grounds, or two saturated colours of different hues.

The simultaneous contrast lab shows what a ground does to any colour you choose; this drill tests whether you can read through it.

How the true value is measured

Value runs from 1, black, to 10, white. It is read from the colour's lightness in the OKLab colour space, the same measure the value checker uses, and always from the colour the screen is showing. For a picture it is the average inside the circle; the drill looks for a place close to one flat value, and says so where brushwork leaves none.

No patch is darker than value 2. Below that, the steps of this scale are greys that most screens show as the same black, and scoring you on them would be scoring your monitor.

When the drill calls a lean

Answers are grouped: darks are values 1 to 4, mid values 4.5 to 6.5, lights 7 to 10; then a darker and a lighter surround; then each hue family. A group gets a sentence such as “You read darks 1.2 steps too dark” only when it has five answers or more, its average is a quarter step or more, and that average is at least twice its own standard error. Otherwise the page says there is no clear lean, or that it is too early to say.

That rule is this tool's own, and cautious on purpose. One session rarely fills every group, so the first screen pools your last ten.

Your screen is part of the test

A grey is only as true as the screen showing it. Night mode, a warm tint, a dimmed laptop and a steep viewing angle all move the darks, and the drill cannot see your screen to correct for them. Use normal brightness and look straight on.

Then test the habit on paper: the value study maker turns a reference into a few flat values.

What it leaves out

The drill has no levels, lives, badges or leaderboards: the only score is your error. It does not judge a painting, and it puts no clock on an answer. The picture task uses one photo of yours, or three paintings: The Harvesters by Pieter Bruegel the Elder, Young Woman with a Water Pitcher by Johannes Vermeer and Wheat Field with Cypresses by Vincent van Gogh, all in The Metropolitan Museum of Art and marked public domain in its open-access records. History stays in this browser, up to 200 sessions; clearing site data removes it, so export it to keep it.