Scanpath Studio¶

See eye-tracking-while-reading data the way the reader saw it — words at their true on-screen positions, with fixations, saccades, heatmaps, and animated replay layered on top. Publication-ready figures included.

Try the live demo Get started Desktop app

The Scanpath Studio app: a scanpath drawn true-to-scale over the stimulus text, with the layer controls on the right

What it does¶

True-to-scale scanpaths

The stimulus text is re-rendered at its exact on-screen geometry, so fixations, saccades, and word boxes line up with what the reader actually saw — or overlay the original stimulus image. See how the rendering works.
Canonical reading measures

FFD, gaze duration, go-past, total fixation duration, skips, and regressions per word — precomputed EyeLink IA values are used when present, computed natively when not.
Corpus analysis

Question-oriented views: profile a text across readers, a reader across trials, or one or two groups — with shared measure pickers, spread bands, effect sizes, and per-view CSV downloads.
Vertical drift correction

The ten Carr et al. (2021) line-assignment algorithms, applied in place or compared side-by-side in a grid, ported natively (no GPL dependency).
Any dataset

Auto-detects EyeLink / Gazepoint / snake-case columns, with a guided upload wizard and manual mapping override. OneStop, PoTeC, and MultiplEYE load out of the box. See the data format.
Everything exports

HTML / PNG / SVG / PDF figures (also as separable layers), GIF / MP4 replays, tidy CSV/Parquet tables, and shareable deep links. The scanpath and replay figures are also scriptable via the Python API and CLI.

Ways to run it¶

In the browser

The hosted demo runs the bundled 3-participant OneStop sample — nothing to install.
From PyPI
```
pip install scanpath-studio
scanpath-studio
```
Python 3.11–3.14 — see Getting started.
As a desktop app

A double-clickable build for Windows / macOS / Linux — no Python needed, and private data never leaves your machine. Download & instructions.

The app in one paragraph¶

Two views, toggled from the header: Scanpath Visualization (one trial — the layered plot plus Annotations, Stimulus & questions, Comparisons, Line assignment, Export, Data Inspection, and Share subtabs) and Corpus Analysis (Per text · Per reader · Groups). Layers — text, fixations, saccades, word boxes, heatmap, stimulus image — toggle independently; Animate replays the reading, Compare overlays a second trial, and the Comparisons subtab ranks same-text scanpaths by similarity. Everything obeys the active trial filters, and the single-trial scanpath and replay figures are reproducible headless through the same pipeline.

Citing Scanpath Studio

If you use Scanpath Studio in your research, please cite it — the citation metadata lives in CITATION.cff (also surfaced in the app's About panel), and the bundled demo data comes from OneStop Eye Movements.