Extract data from chart images to CSV

DataFromChart and Engauge Digitizer solve the same problem from opposite directions. DataFromChart is a hosted browser app with AI-assisted calibration and XLSX-with-chart export. Engauge is open-source desktop (GPLv2, Win / macOS / Linux) with deep axis support and a Qt UI. Pick DataFromChart for zero-install with modern exports; pick Engauge for open-source on local disk, polar or date axes, or offline.

At a glance

	DataFromChart	Engauge Digitizer
License	Free (hosted)	Free, open source (GPLv2)
Install	None — browser	Desktop app (Win / macOS / Linux)
Offline use	No	Yes
Axis types	Linear, log, date	Linear, log, polar, date
Manual calibration	Yes	Yes
AI calibration	Yes	No
Color auto-extraction	Yes (live preview)	Yes (segment fill)
CSV / TSV	Yes	Yes
XLSX	Yes (chart embedded)	No
Project files	Yes	Yes (`.dig` files)
Undo / redo	Yes	Yes (real stack)

What each one is best at

DataFromChart wins on onboarding and output. No install, no Qt, no version-pinning. XLSX embeds the chart image and axis labels with units. AI calibration handles endpoints on familiar chart types. PNG to data in under five minutes.

Engauge wins on offline reliability and long sessions. Session-wide undo stack, no network, handles weird axes (polar, date with custom ticks) browser tools skip. Segment-fill works well on smooth curves. Project files (.dig) make a chart resumable across days.

Workflow comparison

Same four-step shape — upload, calibrate, extract, export — but ergonomics differ.

Calibration. Engauge asks for axis type up front, then endpoint clicks; DataFromChart detects type during classification and offers AI endpoint placement. Polar: Engauge. Standard XY linear: DataFromChart skips a step.

Auto-extraction. Engauge does segment-fill — pick a starting pixel, the tool walks the colored region into a polyline. DataFromChart does color-mask — pick a color, set tolerance, snap every matching pixel. Segment-fill wins on smooth curves; color-mask wins on scatter where segments aren’t connected.

Export. Engauge: CSV and .dig. DataFromChart: CSV, TSV, XLSX (chart embedded), JSON.

Undo / redo. Engauge: full session stack. DataFromChart: undo within current layer, not across full history.

When to pick DataFromChart

Zero-install in a browser tab.
XLSX with chart embedded for visual verification.
Linear or log chart, AI-assisted calibration.
Chromebook or locked-down Mac where native installs are friction.
Live color-tolerance preview, not a numeric dial.

When to pick Engauge Digitizer

Network-restricted, needs offline.
Polar axes or date axes with custom tick formats.
Source code in hand (GPLv2).
Prefer segment-fill on smooth lines over color-mask on scatter.
Auditing a methodology that names Engauge specifically.

Install friction

Engauge ships native binaries:

Windows: signed installer; ~30 MB.
macOS: unsigned .dmg; right-click → Open past Gatekeeper.
Linux: source build with Qt5, or community binary by distro. Ubuntu 24.04+ is straightforward; older distros may need Qt back-porting.

DataFromChart has no install — open the extractor, drop a chart, export.

If install is a deal-breaker, DataFromChart is the default. If Engauge is installed with a workflow built around it, the feature delta isn’t worth disrupting.

Accuracy

Same math, same ceiling — within ~0.5% on a clean image with careful calibration. Image and user, not tool.

The practical difference: on a smooth curve where you want every pixel, segment-fill produces denser output than color-mask at the same tolerance. If downstream needs ~100+ points per curve, segment-fill is worth the install.

Cost

Both free. Engauge under GPLv2; DataFromChart hosted with a free tier covering individual extraction.

Try DataFromChart in your browser first. If install friction is what’s stopping you, the extractor gets you to data in a tab. If you end up needing polar axes or offline operation, Engauge is the fallback.