What is the difference between a default figure from a plotting library and a beautiful data visualisation? The answer lies in the number of plotting parameters tuned. This is especially important when preparing figures for research papers: we need to make sure that our figure is readable, the colours are intuitive and minimalistic, and the layout is decluttered.
However, when you open a code editor and see hundreds of parameters in a plotting library, the task of optimising them becomes overwhelming. I’ve had this idea for a while: to show the whole visualisation process and break it down into simple, manageable steps. What if we create a figure, say in Julia or Python, using all default plotting parameters, and then improve those parameters one by one? It turns out that we can learn a lot about figures through this iterative process. I did just that in my recent tutorial, where I talk in depth about the visualisation process behind creating beautiful figures:
In this video, I share my code in Python, showing how multiple plotting parameters can be optimised. After coding, I post-process the figures using vector graphics software (e.g., CorelDRAW) to make final adjustments and improvements. And finally, I test how the figures look when placed in LaTeX and Word manuscripts.
I hope this tutorial will inspire you to tune more parameters when creating figures for research papers.
You can find all coding examples in my repository: https://github.com/AndreyChurkin/BeautifulFigures/
When creating the figures, I was following these design principles:
- The overall quality → Use vector graphics (e.g., SVG, PDF) for pixel-free, scalable, publication-ready figures
- Readability → Use clear fonts of proper sizes and optimise the layout of figures
- Simplify and declutter → Remove unnecessary elements, make your figure simple and effective
- Colours → Use fewer colours, apply colours strategically and select harmonious colour schemes
- Message and story → Think about the message or story of your figure. What do you want to tell the reader?
- Consistent style → Maintain consistent fonts, colours and formatting across all figures
- To avoid → Avoid pie charts and 3D plots (they often mislead the reader and make it difficult to analyse data)
- Time → Allocate enough time to make a good figure — iterate, refine, and polish
I elaborate more on these principles in the previous tutorial:
I plan to make more videos about this topic, and maybe even a full professional course on data visualisation for research. I would love to hear from you. What challenges do you face when creating figures? Are there specific problems you would like me to cover in future videos? Please drop me an email if you are interested in such a course. I would appreciate your feedback. Thank you!
Andrey Churkin (Андрей Чуркин) 2025
