For researchers

Pull data out of any published figure

Extract numerical data from journal article charts, technical reports, and conference papers — without redrawing or estimating by eye.

Try it on a paper figure See all use cases →

Today

What you're stuck on

Source data not shared

Supplementary materials rarely include the raw numbers behind a figure. You need the data to extend or validate the work.

Hand-digitizing eats hours

Manually clicking points on a scanned figure takes 20-30 minutes — and the error bars are wider than the points.

Reproducibility blocked

Methods can be replicated, but only if the underlying data is available. Otherwise you're starting from estimates.

Workflow

A typical research workflow

01
Screenshot a figure from a PDF or paper. PNG, JPEG, or SVG — exactly how your paper exports it.
02
Drag axis lines onto the gridlines and enter the published min/max values. Linear, log, and categorical axes all supported.
03
Export the calibrated points as CSV or JSON. Pipe into Python, R, MATLAB, or whatever you analyze in.

Why it fits

Built for the way you actually work

Works with paper figures

PDF screenshots, JPEG plates, even camera photos of printed figures. Whatever the source quality, the calibration math is the same.

Multiple series per chart

Control + experimental conditions, multiple treatments, different sample groups — extract each as a separate labeled series in one pass.

Reproducible by construction

Pixel-precise axis calibration with explicit min/max values. Anyone re-running the extraction from the same image gets the same numbers.

Coverage

Chart types you'll handle

ScatterLineBarColumnStacked barAreaLog-scale axes

Linear and log-scale axes both supported. For error bars and multi-panel figures, extract each panel separately — the workflow is identical.

"I extract data from dozens of paper figures every month for my research. The multi-series support is what won me over — bar, scatter, all in one tool."

PhD Candidate

Materials Science

FAQ

Common questions

Does it handle log-scale axes?

Yes. Set the axis type when calibrating — values are interpolated logarithmically for log axes, linearly for linear axes. Most science figures use one or the other.

Can I extract multiple series from the same figure?

Yes. Add a series for each group (control / treatment, condition A / B / C, etc.), switch active series as you click. Every extracted point keeps its series label in the CSV output.

Will my figure get uploaded anywhere?

Manual extraction runs entirely in your browser — your figure never leaves your machine, important for unpublished data or pre-publication figures. AI auto-extraction is opt-in; only then is the cropped image sent to our AI provider for a single read, and it isn't retained.

Can I cite this tool in methods?

Yes — please do. Reference as: "Data was extracted from published figures using DataFromChart (datafromchart.com)." A formal citation format is in the works.

Other use cases

Also built for

For consultants

Get the chart in your client's deck into a working spreadsheet. Drop the screenshot in, paste straight into Excel.

For financial analysts

Reconstruct any chart from a research report. Quarterly figures, earnings projections, sector trends — all from screenshot to CSV.