scientific-visualization

Apply a publication style preset, set figure size to the target journal's column width, use colorblind-friendly colors, label axes with units, and remove top and right spines. Export as PDF or EPS at 300-1000 DPI using a save helper that handles format-specific settings.

The Okabe-Ito palette is the most widely recommended colorblind-safe choice, distinguishable across all forms of color vision deficiency. For continuous data use perceptually uniform colormaps like viridis or cividis, and avoid jet, rainbow, and red-green diverging maps.

Seaborn provides high-level statistical plotting with automatic confidence intervals and faceting, ideal for box plots, violin plots, and heatmaps. Matplotlib offers finer control for complex multi-panel layouts, and seaborn figures accept all matplotlib customizations since it builds on matplotlib.

Most journals require 300 DPI minimum for photographs, 600 DPI for combination figures, and 1000 DPI for line art. Vector formats PDF and EPS are preferred for graphs, while TIFF or PNG are used for images; JPEG should never be used for scientific data.

Use matplotlib GridSpec to define the layout, add subplots for each panel, then place bold letter labels (A, B, C or lowercase for Nature) via ax.text with transform=ax.transAxes. Keep styling consistent across panels and verify text remains readable at final print size.

Blurriness comes from exporting raster images below 300 DPI or scaling up small figures. Text shrinkage happens when figures are designed larger than the journal column width; design at the exact final size with fonts of at least 6-7 pt.