Skip to content

Fix your proportions

Mark the center of six bars of different lengths. Find out whether your marks tend to miss in the same direction.

Divide by eyeYour turn
TipDraw it, then press Measure it.

Mark the middle of every bar. Short ticks, by eye — no measuring.

0 of 6 bars marked
Free · unlimited · no account

The technique

Divide a distance accurately by eye

4 steps
  1. 01

    Stop looking for the middle

    Rather than guessing a midpoint in isolation, compare the two pieces that your tick would make. Do they look equal? The comparison gives you a concrete check on your first estimate.

  2. 02

    Place a mark, then judge the leftover

    Put the tick roughly in the middle of the bar, then ask which piece looks longer. If one does, use that difference to plan an adjustment.

  3. 03

    Move it and ask again

    Take a second look and place a corrected tick if one half still looks longer. The later tick on that bar is the one used. Compare again before moving on; there is no promised percentage improvement.

  4. 04

    Check the answer sideways

    On paper, turn the page ninety degrees and compare the halves again. A different view can help you notice a mismatch. On screen, compare your upright and horizontal bars; keep a comfortable posture rather than twisting your head.

How it works

What it can name

The faults, drawn

These percentages can mean nothing by themselves. Instead, each problem has its own fix. The examples below are all real strokes that I generated with the same code that I used to test the engine.

  • Leaning late

    The ticks tend to sit past the middle, making the first piece longer. Compare the two pieces before moving each mark.

  • Leaning early

    The ticks tend to sit short of the middle, making the second piece longer. Compare the two pieces before moving each mark.

  • Scatter

    Misses go in both directions. Check the two pieces on each bar rather than shifting all the marks the same way.

  • Uprights read high

    The upright marks sit higher relative to their middles than the flat-bar marks do. Compare top and bottom lengths as carefully as left and right.

  • Lost over distance

    The long bars have larger average errors than the short ones. Imagine dividing the pieces again to compare shorter spans.

  • One careless mark

    One mark sits noticeably farther from its midpoint than the others. Recheck that bar before changing every tick.

How the number is made

What the score actually measures

Accuracy

rms(mark − true) ÷ bar length65% of the score

How far your marks are from the middle of the bar as a fraction of the total length of the bar. This makes long and short bars equivalent.

Worst mark

largest single miss ÷ bar length35% of the score

Your largest miss among the six marks. Accuracy measures the set as a whole; this term highlights the worst individual mark so an average does not hide it. It cannot tell whether you rushed.

Habit

mean of the signed missesreported, not scored

Whether your misses tend to share a direction. A signed average shows that tendency; misses in opposite directions can cancel each other, so the scatter matters too.

Consistency

standard deviation of the missesreported, not scored

How widely your misses scatter around their average. This complements the directional bias: a small average miss can still hide marks scattered on both sides of the midpoint.

Upright vs flat

mean(uprights) − mean(flat bars)reported, not scored

Do standing bars read differently from lying bars? A difference is named only when it is large enough compared with the scatter within the groups. Six marks can suggest a pattern, not prove a lasting bias.

Near vs far

mean|error| long − mean|error| shortreported, not scored

Does your judgment get better as the two ends of the bar get further apart?

Errors are measured as a fraction of each bar’s length. A two-pixel miss means more on a short bar than on a long one. Placing a tick asks for less line control than drawing a whole circle, but screen size and input precision can still affect it. Paper practice gives you another way to check your estimates. Every threshold, and the evidence behind it.

Practising proportion by comparing lengths

Proportion is one reason a careful drawing can still feel off. The lines and shading may work while the size or position of a major form differs from what you intended. It can be hard to spot that in your own work, so comparing a few clear lengths gives you something specific to check. Small differences are not automatically mistakes in a stylized character.

The circle and line drills ask you to control a whole stroke. This exercise asks you to judge a length and place a small tick. That reduces the amount of line control involved, but it does not remove the effect of your hand or device. Look for patterns across attempts rather than treating one miss as a diagnosis.

You can practise the same comparison on paper: mark a midpoint by eye, then measure the two halves. Compare what you notice, without expecting a paper attempt and a screen attempt to receive identical results.

Why six bars, and why they are different lengths

One bar shows where one estimate landed. Several bars let you compare repeated misses, different orientations and different lengths. They offer more context, although they still cannot tell us exactly why you placed a mark there.

Three bars lie flat and three stand upright. Their different lengths also let the drill compare longer and shorter bars. The six marks can therefore show whether orientation or length seems to affect this attempt.

Each orientation has three marks, so the engine can compare the gap between groups with the scatter inside them. It only names a difference when that gap clears its rule threshold. That reduces misleading feedback from scattered marks, but six estimates are still a small sample. Check whether the pattern repeats before treating it as a habit.

Check whether your marks share a direction

If you place a tick above the true midpoint of an upright bar, its top half will be shorter. Notice whether that happens repeatedly in your own attempts rather than assuming it must happen to everyone.

Use the same comparison when planning a drawing: check the height of a head against the body, or the space above and below a feature. Match those relationships to your intended character or reference. One simple midpoint test does not establish where every person’s eyes or waist should sit.

A repeated bias gives you something specific to check. The page shows the size and direction of the misses in this attempt. If the pattern appears again, try comparing the two halves more deliberately. A repeated tendency can be useful information without being a fixed habit.

When you have got it

Try for three attempts above 85 with no named fault on your usual device. Then compare a different setup if you have one. On paper, draw six bars with a ruler, mark their middles by eye, and use the ruler afterwards to check them. The check comes after your estimate.

Questions

Why are my proportions always wrong?

Try comparing the two parts instead of relying only on the first estimate. Place a mark, ask which half looks longer, then adjust if needed. This gives you a repeatable way to check proportion; it does not guarantee a particular improvement.

Does it matter that I am using a mouse and not a pencil?

It can matter. A tick takes less continuous line control than a circle, but placing it precisely can still feel different with a mouse, finger or pen. Compare attempts on the same setup first, and treat paper practice as a separate check.

It says my upright marks sit high. Is that a real thing?

It describes the marks in this attempt. The engine names the upright-versus-flat difference only when it is large compared with the scatter within each group. That is a useful clue, not proof of a lasting perceptual bias. Try again and see whether the direction repeats.

Can I use a ruler or count pixels?

For this exercise, place the ticks by eye. A ruler or pixel count would answer a different question, so the result would not tell you much about your estimate. On paper, measuring afterwards is useful for checking what you saw.

Why does it want the marks in no particular order?

Each tick is matched to the nearest bar, so you can work in any order. Keep the tick close to the intended bar to make that match clear. If a mark is assigned incorrectly, undo it and place it again more clearly.

I marked one bar twice. Which one counts?

The later tick on that bar counts. Changing your mind after comparing the two halves is part of the practice, so a second tick is treated as a correction.

Can I erase or start again?

You can erase a misplaced tick or use Undo, then place it again. The engine measures the marks that remain. Make the new placement by eye and compare the two halves before moving on.

Is my drawing sent to an AI model?

No. The measurements run in JavaScript in your browser using the recorded marks. The same marks and scoring settings give the same result. This exercise does not send the drawing to an AI model.

Next drill

Draw the other half

The missing half of a half drawn vase must be completed from the rim down to the foot of the vase.

Draw draw the other half

Drew something real?

Rate My Sketch

These drills measure geometry. For a sketch of a subject, Luna offers AI feedback on five technique dimensions and marks suggested corrections on your image.

Upload a drawing