figscribe

A lean desktop GUI for scientific plots: build and fine-tune figures interactively, and get clean, standalone Python code (Matplotlib) as the result instead of a proprietary project file.

Intended for scientists and engineers who find Origin unwieldy but don't want to hand-write every detail of a Matplotlib figure just to fine-tune it.

Status

Very early — planning stage. Nothing usable has been implemented yet; the PyPI entry is a placeholder release that reserves the name.

The full project plan lives in plan/.

Design principle

The declarative plot model is the single source of truth:

Plot model (typed, JSON-serializable)
   ├──> Renderer:       model -> Matplotlib figure (GUI preview AND export)
   ├──> Code generator:  model -> standalone Python script
   └──> Persistence:    model <-> JSON project file

The GUI only ever reads and mutates the model, through commands with undo/redo — it never manipulates Matplotlib objects directly. Preview and export share the same renderer, so what you see is what you get. The generated code reproduces the same figure and runs without figscribe installed.

Scope

Planned for the MVP: data import (CSV, HDF5, NPY/NPZ), line/scatter/errorbar plots, histograms, 2D images and color meshes, curve fitting via scipy.optimize.curve_fit, multi-panel layouts with shared axes, LaTeX labels, undo/redo, export to PDF/SVG/PNG at exact millimeter sizes, and Matplotlib style templates for journal-column widths.

Deliberately out of scope: spreadsheet-style data editing, a statistics suite, a custom scripting language, and reading generated or third-party Python code back in.

Stack

Python >= 3.11, PySide6, Matplotlib, NumPy, SciPy, h5py.

License

MIT, see LICENSE.