Reading-error calculator

Every clicked point has a small error. The calculator shows what that is in reality at a given scale and segment length, and how much of it comes from calibrating on a short dimension.

Free, no sign-in

Result

Source of errorTypicalMaximum
Picking the points of the measured segment0.71% relative
0.0707 m absolute
1% relative
0.1 m absolute
Calibration3.5% relative
0.354 m absolute
5% relative
0.5 m absolute
Total3.6% relative
0.361 m absolute
6% relative
0.6 m absolute

On the sheet: measured segment 100 mm, reference segment 20 mm.

Calibration error dominates: the reference segment is 5× shorter than the measured one. Calibrate on a longer dimension.

The result is an estimate: the arithmetic is exact, but measuring accuracy depends on the drawing, the printout and how you read it.

Calculations run in your browser. The numbers you type are not sent to a server, and no account is needed.

The model assumes independent errors for the points. It ignores errors of the drawing itself (line width, paper distortion, scan distortion), which can be larger than the pointing error.

Reference segment length and calibration error

For a pointing error e = 0.25 mm on the sheet; typical error = e·√2 ÷ segment length. The longer the reference segment, the smaller the relative calibration error.

Reference segment length and calibration error
Reference segment on the sheetRelative calibration error (typical)Error over 10 m (typical)
20 mm1.77%177 mm
50 mm0.71%70.7 mm
100 mm0.35%35.4 mm
200 mm0.18%17.7 mm
400 mm0.09%8.84 mm
800 mm0.04%4.42 mm

Practical conclusion

Calibrate on the longest dimension you can. Calibration error is inversely proportional to the length of the reference segment: a segment 10 times longer gives a 10 times smaller calibration error. For long measurements use the overall building dimension or grid lines, not a door width.

Formulas

  • e = pointing error of one point on the sheet (mm)
  • Error of the distance between two points: typically e · √2, worst case 2e
  • Error in reality = error on the sheet × N (for a 1:N scale)
  • Relative error of the measurement: δm = e · √2 ÷ segment length on the sheet
  • Relative calibration error: δc = e · √2 ÷ reference segment length on the sheet
  • Total: typically √(δm² + δc²), maximum δm + δc (with 2e instead of e · √2)
  • Pixels to sheet: e [mm] = px × 25.4 ÷ (dpi × zoom ÷ 100)

Frequently asked questions

How does the pointing error affect a measurement?
A distance is defined by two points, so its error is typically e·√2 and at worst 2e, where e is the error of one point. Multiplied by the scale denominator it gives the error in reality: 0.5 mm at 1:100 is about 71 mm typically.
Why should I calibrate on a long dimension?
Calibration error carries into every later measurement as a fixed percentage. The same pointing error is a large percentage on a short segment and a small one on a long segment. A reference segment 10 times longer reduces the calibration error 10 times.
How do I convert pixels to millimetres on the sheet?
e [mm] = px × 25.4 ÷ (dpi × zoom ÷ 100). At 96 dpi and 100% one pixel is about 0.26 mm of the sheet; at 400% zoom it is about 0.066 mm.
Does this result guarantee measuring accuracy?
No. It estimates the effect of picking the points only. Actual accuracy also depends on the quality of the drawing, printout or scan and on whether the scale is accurate.

Guides

Other tools

Measure on the whole drawing instead

The calculator handles a single dimension. In the Skalnivo app you open a file (PDF, DWG, DXF, photo), calibrate the scale with one known dimension and measure distances and areas directly on the drawing, without converting by hand.