Graphical tools

Fra Robin

(Forskjeller mellom versjoner)
Gå til: navigasjon, søk
(Tips on Exporting Figures)
(Tips on Exporting Figures)
(11 mellomrevisjoner ikke vist.)
Linje 22: Linje 22:
** Matplotlib
** Matplotlib
*** see [https://github.com/jbmouret/matplotlib_for_papers Mouret's tutorial] for publication quality plots
*** see [https://github.com/jbmouret/matplotlib_for_papers Mouret's tutorial] for publication quality plots
-
*** check the cheat sheets [https://raw.githubusercontent.com/matplotlib/cheatsheets/master/cheatsheets-1.png 1] & [https://raw.githubusercontent.com/matplotlib/cheatsheets/master/cheatsheets-2.png 2]
+
*** check the [https://github.com/matplotlib/cheatsheets cheat sheets]
** Seaborn (statistical data visualization; uses Matplotlib internally. See for example this [https://elitedatascience.com/python-seaborn-tutorial tutorial])
** Seaborn (statistical data visualization; uses Matplotlib internally. See for example this [https://elitedatascience.com/python-seaborn-tutorial tutorial])
* R
* R
Linje 44: Linje 44:
** Preferably use vector graphics such as svg, pdf or eps (can be edited with vector graphics editors)
** Preferably use vector graphics such as svg, pdf or eps (can be edited with vector graphics editors)
** Check your colour maps! (see https://www.nature.com/articles/s41467-020-19160-7 or https://matplotlib.org/stable/tutorials/colors/colormaps.html)
** Check your colour maps! (see https://www.nature.com/articles/s41467-020-19160-7 or https://matplotlib.org/stable/tutorials/colors/colormaps.html)
-
** Check font size, font type, line widths, marker sizes, proportions, aspect ratios and resolution!
+
** Check font size, font type, line width, marker size, proportion, aspect ratio and resolution! Tuning those parameters makes a significant impact on how your figures are perceived!
** Example plots with mostly default (left) and adapted (right) plotting parameters.
** Example plots with mostly default (left) and adapted (right) plotting parameters.
-
** [[Fil:TestT.png|400px|middle]][[Fil:Test.png|250px|middle]]  
+
** [[Fil:TestT .png|400px|middle]][[Fil:Test.png|250px|middle]]  
 +
** Make sure that your figures are still readable when printing them in grayscale
* Sometimes it is neccessary to use png format (e.g. in case of render graphics or plots with an essential transparency effect)
* Sometimes it is neccessary to use png format (e.g. in case of render graphics or plots with an essential transparency effect)
** In that case, one should pick a proper resolution for the export file while accounting for its file size
** In that case, one should pick a proper resolution for the export file while accounting for its file size
Linje 59: Linje 60:
*** R: https://github.com/daqana/tikzDevice
*** R: https://github.com/daqana/tikzDevice
*** Matlab/Octave: https://github.com/matlab2tikz/matlab2tikz
*** Matlab/Octave: https://github.com/matlab2tikz/matlab2tikz
-
** It is possible to either input the code directly through <code>\input{}</code> or to compile the figure first [https://blog.modelworks.ch/producing-stand-alone-figures-with-tikz-in-latex/ through] the ''standalone'' class
+
*** It is possible to either input the code directly through <code>\input{}</code> or to compile the figure first [https://blog.modelworks.ch/producing-stand-alone-figures-with-tikz-in-latex/ through] the ''standalone'' class
 +
* To see how plotting parameters can be tuned given concrete examples, see [https://github.com/juuf/IN5490/blob/5a4a6bb0098402166ceb623bcfd5c3402c5b4f72/IN5490.ipynb here]. Based on an example data set, first distributional plots are shown. Then the default plotting parameters are tuned.

Versjonen fra 3. nov 2021 kl. 13:14

Innhold

Figures/Illustrations

  • Vector Graphics Editors
    • Inkscape (free and open-source)
    • Adobe Illustrator (for students only through UiO programkiosk)
    • Directly in LaTeX (PGF/TikZ: https://github.com/pgf-tikz/pgf; some examples)
    • LibreOffice Draw (free and open-source)
    • Through presentation software such as Powerpoint, Keynote, …
    • Web-based options: Google draw (in Google disk), Draw.io, Figma, Lucid charts, …
  • Raster Graphics Editors
    • Gimp (free and open-source)
    • Adobe Photoshop (for students only through UiO programkiosk)
    • Photopea (online editor)
  • 3D Computer Graphics Editors
    • Blender (free and open-source)
    • Autodesk Maya
    • see also here

Plotting Apps

  • Python
    • Matplotlib
    • Seaborn (statistical data visualization; uses Matplotlib internally. See for example this tutorial)
  • R
    • ggplot2 (with Rstudio)
  • Matlab (available at UiO)
  • Octave (free alternative to matlab)
  • Gnuplot
  • Desmos.com (web based)
  • javaFX
  • excel (UiO)

Digitalize Figures

To qualitatively compare one's own data with other published data, it is sometimes needed to obtain the concrete data of the respective publication. In that case there are multiple ways to do that:

Tips on Exporting Figures

  • Before saving the graphics in the respective programs
    • Avoid bitmap graphics if possible and especially jpg!
    • Preferably use vector graphics such as svg, pdf or eps (can be edited with vector graphics editors)
    • Check your colour maps! (see https://www.nature.com/articles/s41467-020-19160-7 or https://matplotlib.org/stable/tutorials/colors/colormaps.html)
    • Check font size, font type, line width, marker size, proportion, aspect ratio and resolution! Tuning those parameters makes a significant impact on how your figures are perceived!
    • Example plots with mostly default (left) and adapted (right) plotting parameters.
    • Make sure that your figures are still readable when printing them in grayscale
  • Sometimes it is neccessary to use png format (e.g. in case of render graphics or plots with an essential transparency effect)
    • In that case, one should pick a proper resolution for the export file while accounting for its file size
    • Possible ways to compress the png file is to use for example pngquant, AdvanceCOMP or ImageMagick
  • Exporting to LaTeX
  • To see how plotting parameters can be tuned given concrete examples, see here. Based on an example data set, first distributional plots are shown. Then the default plotting parameters are tuned.
Personlige verktøy