How Pixactly scoring works.
A Pixactly result has two different scales: pixel error, where lower is better, and size match, where higher is better. This guide follows the current game's calculations so you can reproduce a result and decide which number to use for practice. All worked values below are illustrative.
Total off: add the two absolute errors
The game first subtracts the target dimension from the drawn dimension
and rounds each difference to the nearest whole pixel. It then adds
the absolute values: round score = |width difference| + |height difference|.
Opposite directions do not cancel out.
For a target of 180 × 120px and a drawing of 194 × 111px, the signed
differences are +14px and −9px. The displayed total off is
14 + 9 = 23px. That score measures the two side lengths;
it does not measure area, overlap, or the rectangle's position on the canvas.
Size match: compare each side as a ratio
For each positive dimension, the game divides the smaller of the drawn
and target sizes by the larger, multiplies by 100, and rounds:
axis match = round(100 × smaller size / larger size).
A zero drawn dimension receives 0%. The round's Size match
is the rounded average of the two already rounded axis percentages.
| Measurement | Calculation | Rounded result |
|---|---|---|
| Width | 100 × 180 / 194 | 93% |
| Height | 100 × 111 / 120 | 93% |
| Size match | (93 + 93) / 2 | 93% |
This is a ratio of lengths, not a percentage of correct guesses or a percentile among players. Doubling a side or halving it both produce 50% on that axis. For a 100px target, drawing 90px gives 90%, while drawing 110px gives 91% after rounding. Both are 10px off: equal pixel errors need not have identical match percentages.
Why the same pixel miss can show a different percentage
| Target | Drawing | Total off | Size match |
|---|---|---|---|
| 40 × 40px | 50 × 40px | 10px | 90% |
| 200 × 200px | 210 × 200px | 10px | 98% |
In the first example, width match is 80% and height match is 100%. In the second, width match rounds to 95% and height match is 100%, making the rounded mean 98%. The extra 10px is a smaller share of the larger target. Use pixel error to describe the distance missed and size match to describe the miss relative to the target.
What the five-round result averages
The final score adds all five round scores. For example, rounds of
7, 9, 13, 13, and 17px produce 59px. Time contributes
nothing to that score. The browser keeps the lowest completed total
as your personal best on that browser profile.
Width accuracy and Height accuracy are the rounded arithmetic means of their five axis match percentages. Each round has equal weight, regardless of target size. They are not calculated by dividing the total error by all target dimensions combined.
Average width off and Average height off average the absolute pixel errors and round to whole pixels. Width errors of 5, 2, 8, 8, and 9px average 6.4px, displayed as 6px. Height errors of 2, 7, 5, 5, and 8px average 5.4px, displayed as 5px. Multiplying those displayed averages by five gives 55px, not the 59px total, because the averages have been rounded.
Record direction before leaving each round
Average error cannot tell you whether to draw wider or narrower. A drawing 10px too wide and one 10px too narrow contribute the same width error. Compare the drawn and target outlines during each round's feedback and note the direction before continuing. The final summary does not retain those outlines or display signed differences.
Use the manual practice log to keep those observations. If directions alternate, a near-zero signed average may hide large misses; keep absolute error alongside it. Targets change between sets, so one lower total is an observation, not proof that a correction works consistently.