tutorials

Digitize your first chart

A practical walkthrough of the five-step flow, plus a quick recipe for each of the seven chart types. For the concepts behind each step, see the docs.

Before you start

  • You need an invite code to create an account - ask whoever invited you, or an admin, for one.
  • Have your chart image ready, under 10 MB. Screenshots, scanned figures, and photos of printed charts all work.
  • Nothing you upload leaves your browser. See Data & privacy if you want the details before you start.

1. Upload

Open the Digitizer and drop in your chart image (or click to browse). Plotion decodes it entirely client-side and holds it in IndexedDB, so refreshing the page won't lose your work.

2. Setup

Pick the chart type that matches your image. If the chart type uses axis scales (XY, Bar, Geo), choose the scale for X and Y independently - linear, log, date/time, or category. If either axis is categorical, type the category labels here, in order.

Getting the axis scale right matters more than it looks: a year-over-year chart is a dateaxis even if the labels look like plain numbers (“2020”, “2021”...), and a chart with unevenly spaced gridlines that get closer together toward one end is usually log, not linear.

3. Calibrate

Set the calibration targets for your chosen chart type (see each recipe below for exactly which ones). For axis-based targets, click a point on the image whose value you know from the chart's own axis labels, then type that value in the popover that appears. Two points per axis is enough - pick ones that are far apart on the axis for better precision, not two gridlines right next to each other.

If you calibrate and then realize you picked the wrong axis scale in Setup, going back and changing it resets that axis's calibration (Plotion will confirm before doing this) - your points' pixel positions are kept, values are just recalculated after you recalibrate.

4. Pick

Click each data point on the image. Use the Series panel to add a series before you start on a new line/group, give it a name and color, and pick line (points get connected, in order) or scatter (points stand alone). The panel shows which series is active; new clicks go there.

Points and calibration targets can both be dragged to adjust them after the fact - each drag is one undo step. The points table lists everything you've placed, with its computed value and margin of error, and lets you move a point to a different series or delete it.

5. Export

Choose a file format (CSV, JSON, or XLSX) and, for chart types that support it, long or wide. When in doubt: long is the safer default for further analysis in R/Python/pandas; wide is more convenient for a quick look in a spreadsheet, one column per series.

Recipe: XY line chart

  1. Setup: chart type XY, pick a scale for X and Y (e.g. date for X, linear for Y).
  2. Calibrate: click and type values for X1, X2 (two known X positions), then Y1, Y2.
  3. Pick: add one series per line on the chart, set each to line, and click along each curve left to right.
  4. Export: long format keeps one row per point across all series; wide gives you one Y column per series aligned by X.

Recipe: stacked Bar

  1. Setup: chart type Bar, orientation (vertical/horizontal), grouping mode stacked, and type in the category labels.
  2. Calibrate: the category axis is fixed to those labels; calibrate the value axis with two known values (e.g. 0 and 100).
  3. Pick: add one series per stacked segment (bottom to top = first to last in the panel), then click the top edge of each segment for every category. Clicking the same category+series slot again replaces that value.
  4. Reorder series by dragging in the panel if the stacking order needs to change - this affects how segment values are computed.

Recipe: Pie / donut

  1. Setup: chart type Pie, set Direction to match the chart (clockwise is the common case).
  2. Calibrate: click Center, then click Start on the rim where the first slice begins (usually 12 o'clock).
  3. Pick: one series per slice, one click each on the rim where that slice ends, going around in Direction order.
  4. Check the closure error shown in the export panel - a large value means your Center or slice edges were placed imprecisely.

Recipe: Polar

  1. Setup: chart type Polar, radius scale (linear or log), and Direction.
  2. Calibrate: click Center, then Start (type its radius value), then R2 at a different radius (type its value too).
  3. Pick: click points anywhere on the plot. For a connected curve, set the series to line - it will connect points in angle order, not radius order.

Recipe: Ternary

  1. Setup: chart type Ternary.
  2. Calibrate: click the three triangle corners, vertex A, vertex B, vertex C - no values to type, their positions define 100%/0%/0% automatically.
  3. Pick: click each data point inside the triangle. Axis C (and its share of the composition) is computed for you on export - you never enter it directly.

Recipe: Geo (lat/lon)

  1. Setup: chart type Geo (axes are locked to linear automatically).
  2. Calibrate: click and type decimal-degree values for two longitude references and two latitude references, using gridlines or labeled coordinates on the map.
  3. Pick: click each location marker. Values come out as decimal longitude/latitude, not degrees-minutes-seconds.
  4. Remember this assumes an equirectangular (non-projected) map - fine for a regional or schematic map, less accurate for a whole-world map with a real projection.

Recipe: Scale-bar

  1. Setup: chart type Scale-bar.
  2. Calibrate: click Origin (your (0,0) reference point), then click Ref and type the real-world distance from Origin to Ref (using the scale bar drawn on the image).
  3. Pick: click each feature you want to measure. Distance from Origin is computed for you on export.
  4. Keep your typed distance unit consistent (all cm, or all mm, etc.) - Plotion doesn't validate or convert units.

Tips & troubleshooting

  • Wrong Direction on Pie/Polar: if percentages or angles look inverted or wildly off, double-check Direction (clockwise vs. counterclockwise) in Setup before re-picking anything.
  • Category snapping feels off: category picks snap to the nearest label position; if your bars aren't evenly spaced in the source image, place your two category calibration clicks as far apart as possible to average out the error.
  • Log axis and margin of error: on a log-scaled axis, the same pixel imprecision produces a bigger value-error toward the high end of the axis - a wider margin of error there is expected, not a bug.
  • Lost your work after closing the tab? Your session auto-saves to this browser only. Reopening Plotion in the same browser on the same device restores it; a different browser or device starts fresh.
  • Still stuck? See the docs for how each chart type's calibration model works, or contact us.