Common academic drawing mistakes—and how to avoid them

Published: 2026-05-01 Category: Academic figures Reading time: ~10 min Tags: academic figures, drawfig, LaTeX figures, TikZ, paper figures, figure guidelines

---

Introduction

Reviewers may spend 30 extra seconds on your figures—and that glance can shape their impression of the whole paper. Blurry exports, colour schemes that fail for colour-blind readers, captions that ignore journal style—these rarely trigger automatic rejection, but they signal sloppiness. This article covers six common mistakes and concrete fixes, whether you use TikZ, matplotlib, Visio, or DrawFig.

---

Mistake 1 — Low resolution that fails in print

Root cause

Figures built in slide tools and exported as default 72 DPI PNG look fine on screen but blur when the journal scales them. RGB screens also differ from CMYK print.

Fix

Format	Recommended	Use case
PDF (vector)	Infinite scale	Submission, print
SVG (vector)	Infinite scale	Web, online journals
PNG (raster)	≥ 300 DPI	Transparency needed
TIFF (raster)	≥ 600 DPI	Strict print houses

Best practice: Archive and submit vectors (PDF/SVG). Rasterise from vector at 600 DPI if required—never re-compress the same bitmap. In DrawFig, choose PDF or SVG export—free with no sign-in—for true vector output.

---

Mistake 2 — Fonts that do not match the paper

Root cause

Visio defaults to Arial; your LaTeX body uses Times or Latin Modern. Worse: text rasterised into pixels—unsearchable in PDF.

Fix

Rule: Figure typography should match body text in family and size.

• LaTeX + TikZ: Inherit document fonts—avoid hard-coded \fontspec{Arial}.

% Wrong
\tikzset{every node/.style={font=\fontspec{Arial}}}

% Right
\tikzset{every node/.style={font=\small}}

• DrawFig → TikZ: Compiled nodes use LaTeX defaults. For PNG/SVG slides, pick a matching font in DrawFig’s style panel.
• matplotlib:

import matplotlib.pyplot as plt
plt.rcParams["font.family"] = "serif"
plt.rcParams["font.serif"] = ["Times New Roman", "Times"]

---

Mistake 3 — Arbitrary colour that fails in print or for colour-blind readers

Root cause

Neon red/green pairs look fine on screen; ~8% of male readers cannot distinguish them; print may desaturate fluorescent hues.

Fix

Use colour-blind-safe palettes (Okabe–Ito is a proven set):

Role	Colour	Hex
Primary 1	Blue	#0072B2
Primary 2	Orange	#E69F00
Primary 3	Sky blue	#56B4E9
Primary 4	Vermillion	#D55E00
Accent	Pink	#CC79A7
Neutral	Grey	#999999

In DrawFig, enter hex values or apply preset academic colour themes. Also: Encode series with line style (solid/dash/dot) and marker shape—not colour alone—so greyscale print still works.

---

Mistake 4 — Caption format that ignores the journal

Root cause

Captions like “Figure 1: results” may violate “Fig. 1. Experimental results.” Captions must often be standalone—readable without the surrounding paragraph.

Fix

% IEEE: no trailing period
\caption{System architecture overview}

% ACM: with period
\caption{System architecture overview.}

% Springer-style label
\caption{\textbf{Fig. 1.} System architecture overview.}

Caption checklist: 1. State what the figure shows—not “as shown above” 2. Include key conditions (temperature, learning rate, …) 3. Expand abbreviations on first use 4. For statistics: sample size and significance markers

---

Mistake 5 — Wrong figure size, forced scaling in LaTeX

Root cause

Random canvas size → \resizebox squashes text and lines.

Fix

Plan width before drawing. IEEE-style two-column templates:

• Single column: ~88 mm (3.46 in)
• Full width: ~183 mm (7.2 in)

Set DrawFig canvas to 88 mm or 183 mm when possible so \input{fig.tex} needs no scaling.

% Avoid if the figure was not sized correctly
\resizebox{\columnwidth}{!}{\input{fig.tex}}

% Prefer
\input{fig.tex}

If you must scale geometry only:

\tikzset{every node/.style={transform shape=false}}

---

Mistake 6 — Ignoring accessibility

Root cause

Flat PNGs and figures without text layers are opaque to screen readers. ACM and IEEE CS increasingly expect alt text.

Fix

1. Prefer PDF/SVG with live text layers.
2. Write concise alt text (≤ ~150 characters) per figure.
3. In LaTeX, consider accsupp where supported.
4. Self-test: Print in greyscale; cover the figure—can caption + body text still convey the message?

---

Pre-submission checklist

• [ ] Vector PDF/SVG (or high-DPI raster from vector)
• [ ] Figure fonts match or harmonise with body text
• [ ] Colour-blind-safe palette; redundant encodings (line/marker)
• [ ] Caption matches target journal (periods, casing, labels)
• [ ] Width matches column; minimal forced \resizebox
• [ ] Alt text prepared

Tick all six for genuinely publication-ready figures.

---

Get started

👉 Open DrawFig editor (free canvas) Vector PDF/SVG export, TikZ export, academic palettes—no install required. Canvas drag-and-drop and SVG/PNG/PDF export are free with no sign-in. TikZ export (3 credits/use) and AI canvas (5 credits/use) require sign-in. 30 credits daily (accumulated). See credit rules.