Tavotto for Matplotlib

Edit figures visually. Keep scripts untouched.

Tavotto opens the figures matplotlib has already drawn: titles, legends, curves and the page layout are adjusted on the canvas. Each change is saved on its own; the script stays as it was.

A real figure, at the size it will print.

Choose a desktop version

Download for Windows Windows 64-bit · .exe v0.17.0 · 93 MB Download for macOS macOS Apple silicon · .dmg v0.17.0 · 175 MB

Try it in your browser Intel Mac version Other versions and checksums

Free & Open Source · 100% Local Execution

Installing on Windows

Windows may display a SmartScreen warning because our open-source code-signing certificate is pending verification. Click "More info" → "Run anyway" to proceed.

All binaries are distributed via official GitHub Releases with verifiable SHA-256 checksums.

Once installed, Tavotto runs on your own machine. Your figures and scripts are not uploaded.

Code signing policy

Click anything. See it clearly.

Native Matplotlib Hierarchy

Tavotto reads the figure as the objects matplotlib drew, not as pixels. Every title, tick label, axis and curve can be selected on its own.

axes_0.title Title
axes_0.xlabel Axis label
axes_0.ylabel Axis label
axes_0.lines_0 Line
axes_0.lines_1 Line
axes_0.legend Legend
axes_0.xticks Tick set
axes_0.yticks Tick set

Live Visual Tweaks

Adjust font sizes, line weights, stroke styles, or positioning on the fly. Matplotlib re-renders the underlying figure instantly.

Try it on a sample, or your own .py.

axes_0.title Title

fontsize

9.0 → 12.0 pt

Non-Destructive Sidecar Overrides

Every edit is recorded beside the figure; the Python file is not modified. Re-run the script in Tavotto with new data and the edits are applied again.

restoring 1 edit identical

fig1_kinetics.py

"""示例 1：两条动力学曲线 + 图例（最典型的可参数化面板）。
在 Tavotto 里双击这张图，可以直接点中标题、坐标轴标签、图例、任意一条曲线，
改字号/颜色/线宽或拖动位置——改动存成 override，本文件一个字都不会被动。
"""
import numpy as np
import matplotlib.pyplot as plt
from paper_style import COL_1, PALETTE, save
def main():
t = np.linspace(0, 60, 200)
fast = 1 - np.exp(-t / 8)
slow = 1 - np.exp(-t / 24)
fig, ax = plt.subplots(figsize=(COL_1, COL_1 * 0.72))
ax.plot(t, fast, color=PALETTE[0], lw=1.2, label="Catalyst (k = 0.125 $\\mathrm{min^{-1}}$)")
ax.plot(t, slow, color=PALETTE[1], lw=1.2, ls="--", label="Blank (k = 0.042 $\\mathrm{min^{-1}}$)")
ax.set_xlabel("Reaction time (min)")
ax.set_ylabel("Conversion")
ax.set_title("Reaction kinetics")
ax.set_xlim(0, 60)
ax.set_ylim(0, 1.05)
ax.legend(loc="lower right")
save(fig, "Fig1_kinetics")
if __name__ == "__main__":
main()

fig1-overrides.json

                {  "index": 0,  "patch": {    "gid": "axes_0.title",    "prop": "fontsize",    "value": 12.0  },  "hash_after": "sha256:56591382a70626624a8df4ae9a5a17a9358e0844f65a030f2bd0d36ee6905bb9",  "warnings": [],  "moved_gids": [    "axes_0.title"  ],  "bbox_after": {    "axes_0.title": [      0.32596875,      0.03180555555555542,      0.37306249999999996,      0.06730855305989585    ]  }}

Millimeter-Accurate Multi-Panel Layout

Panels (a), (b) and (c) are laid out on one page measured in millimetres; this one is 150 × 112.5 mm. Scale a panel and its type scales with it.

The real composed page, 150 by 112.5 millimetres: panel (a) is the reaction kinetics figure, panel (b) a scatter plot with a linear fit, panel (c) a bar chart of yields with error bars. — 150 × 112.5 mm

150 × 112.5 mm

(a) Fig1_kinetics.pdf 80 × 57.6 mm 0.9163
(b) Fig2_correlation.pdf 59.0894 × 51.573 mm 0.9646
(c) Fig2_yield.pdf 58.4941 × 47.279 mm 0.9745

Automated Publication Pre-Flight Checks

Type sizes, line weights and layout are checked before you export.

lab-publication-v1 1.0.0

error cleared (a) (b) (c) Figure text at an effective 7.72pt is not above the absolute floor of 8pt
suggestion cleared (a) (b) The profile expects legend_text to be bold (currently normal)
warn cleared (a) (b) (c) The effective line width 1.20pt is not one of the profile's steps (0.5/0.75/1/1.5pt) The effective line width 0.53pt is not one of the profile's steps (0.5/0.75/1/1.5pt)
suggestion cleared (a) (b) The axis title “Conversion” is not in the profile's “Title (unit)” form
warn cleared (a) (b) The legend's effective font size 7.72pt is outside the profile's 8.5–9pt
suggestion cleared (a) All 2 lines are drawn without markers, which is hard to tell apart in black-and-white print or for colorblind readers — consider markers or distinct line styles
suggestion cleared (b) There is a fitted curve but no confidence band — submissions usually require the uncertainty of the fit

error 1 · warn 2 · not_verifiable 0 · suggestion 4 · blocking true error 0 · warn 1 · not_verifiable 0 · suggestion 4 · blocking false

1 raised 12 → 9 pt blocking true
2 raised 22 → 9.5 / 10 pt blocking false

Single-Source PDF & PNG Output

The PDF keeps real, selectable text, formulas included. The PNG is rasterised from that same PDF, so the two formats show the same figure.

Fig1_page.pdf 150 × 112.5 mm · 425.2 × 318.9 pt 291 characters of real text · 96.0 KB
Fig1_page.png 3544 × 2658 px · 600 dpi 306.2 KB
Fig1_page_proof.json lab-publication-v1 1.0.0 error 1 · warn 2 · not_verifiable 0 · suggestion 4 · blocking true

The launch film 33 s · With music