In graphic design, color is one of the most important considerations. For scientific figures, color is an important part of the data visualization to provide aid in visual clarity and effective delivery of the presented data. Also, the right use of color might be important depending on a certain data type in the paper. While the aesthetic quality of scientific figures is not a primary focus of scientific research and paper, figures with poorly selected colors can substantially harm the reader’s experience and clarity of the presented data. In this tip post, I wish to cover the basics of color in graphic design and provide several tips and resources for further in-depth learning on the use of color for data visualization.

RGB vs. CMYK
In graphic design, there are two major color models: RGB and CMYK models.

RGB stands for Red, Green, and Blue and it is majorly used for screen-based (or web-based) images. The RGB model is an additive color model based on the mixing of visual light (meaning the color becomes brighter with more mixing). Since the RGB model is based on the color space for visual light, digital images from measurement devices (e.g., camera, microscope, scanner) are also generated in the RGB color mode as the image sensor reads the intensity of red, green, and blue signals from the input.

CMYK stands for Cyan, Magenta, Yellow, and Key (black) and it is majorly used for printed images. CMYK model is a subtractive color model based on the painting of color (meaning the color becomes darker with more mixing). Since the CMYK model is based on printed/painted colors, it is the standard color model for printing-based publication including academic journals and magazines (except online-only journals).

While it is not common to distinguish RGB and CMYK color models in scientific research and figure preparation, it is often an important consideration for a certain type of image data. For example, a fluorescent microscope takes all original images in the RGB color models. However, because of the difference in RGB and CMYK color spaces, the same image becomes slightly different in color when it is being saved in the CMYK color model (which is the standard for most printed paper figures) as shown in the above image (confocal microscope images of rat tissue with green and blue fluorescent channels). Hence, it would need the care to keep the original color model for image data in analysis and use the consistent color model in figures to avoid visual confusion and/or misrepresentation of data.

Tip 1 – Use resources for in-depth learning
The use of color in data visualization is a very important technique in various research fields where the graphical representation of a large amount of data is required. Given the depth of the topic, I wish to introduce several very useful resources instead of diving too much in this tip post:

• Ten Simple Rules for Better Figures by Rougier et al.
• Your Friendly Guide to Colors in Data Visualization by Lisa Charlotte Muth
• Simple Tools for Mastering Color in Scientific Figures by Patrícia Chrzanová Pečnerová

Tip 2 – Make a color code for each paper
Inconsistent use of colors in figures is one of the most common sources of bad data visualization and reader experience. Hence, it is very useful to build a color code for each paper before preparing figures based on which you can ensure consistent use of colors in all figures in the paper. There is no strict rule for color codes but there are several points to consider. I also provide various example color codes with a downloadable Illustrator file below.

• Do not overuse different colors.
• Light to dark gradient based on few color choices (or theme colors) can give the feeling of consistency and orderliness.
• Avoid the use of saturated colors.
• Use complementary colors wisely.

Various color codes (AI file)

​As a more practical example, I provide the color code (with a downloadable Illustrator file) that I used for this paper.

Example color code (AI file)

Tip 3 – Use color wisely
Colors in scientific figures can be very effective tools to aid understanding and delivery of information to the readers. For example, different colors can be used to highlight an important step or part in your experiment/data as shown in the below examples. With right and smart use, colors can be one of the most effective tools in our scientific communications.

Example 1 – Highlights in chemical schemes

​Tip 4 – Be mindful of color-vision deficiency & color-blindness
The main purpose of scientific figures and colors in them is to aid communications of our scientific findings. Hence, it is important to be mindful of the inclusive use of colors for our colleagues with color-vision deficiency and color-blindness. Thankfully, this topic has been well studied with scientifically derived guidelines and suggestions to learn such as this paper (this is a must-read article!).

​​Disclaimer. The contents are my personal opinion and do not represent the view of any institution or company I am affiliated/employed. If you find any incorrect information, please feel free to let me know via my email.

​The capability to use tools has given humankind enormous power to enable many things. It’s not a very different story for scientific figure preparation! While there is simple software like Microsoft Painter (my first digital drawing software) and Microsoft PowerPoint (which is surprisingly popular among researchers to “draw” things not just for presentation slides), more dedicated professional graphic design software can definitely be a great help. In this tip post, I wish to introduce several common and popular design software for 2D and 3D graphics.

2D Graphics
In scientific figures, 2D graphical elements (schematic illustrations, plots) are the most common type of graphics. 2D graphic design software can be divided into two categories: vector and raster design software (for more information on vector vs raster formats, please check this tip post).

Let’s start with 2D vector design software. Despite being commercial with a relatively high price (although cheaper with academic discount), Adobe Illustrator has become a dominant software for 2D vector design in scientific figure preparation due to its ready compatibility with other commonly used Adobe software packages (Photoshop, Acrobat, InDesign, etc) as well as broad adoption in the publishing industry. I also use Adobe Illustrator for most of my 2D graphic design for scientific figures, so the upcoming tip posts will be mostly based on Adobe Illustrator.

However, there are open-source alternatives that are freely available with comparable functionality. Among them, Inkscape is one of the most popular and powerful alternatives to Adobe Illustrator. Note that many standard vector graphic formats (.eps, .svg, .pdf) are compatible across different software, so the choice of software can be made based on the availability (often through university/institute software volume licensing) and personal preference.

Next is 2D raster design software. Unlike non-scientific purposes, raster graphic design has very limited use in scientific figure preparation. The main reason is that inappropriate manipulation of raw image data (commonly in raster formats) is a major concern regarding research ethics and integrity. Hence, it is strongly recommended to familiarize with the guideline for scientific digital image manipulation before using these 2D raster design software (some good reading materials like this and this).

Adobe Photoshop is arguably the most well-known 2D raster graphic design software due to its long history and powerful functions (even photoshopping is in the dictionary). However, Adobe Photoshop is a relatively expensive commercial software package (unless provided by university/institute through volume licensing) and its powerful functionalities are often not really relevant to scientific figure preparation as scientific digital image data mostly need only minimal edits (like cropping, etc). Hence, very simple built-in functions of Microsoft PowerPoint (or Keynote if you use Mac) can be enough for many occasions. If you may need open-source alternatives to Adobe Photoshop, I find GIMP (GNU Image Manipulation Program) can be a great option as a free alternative with comparable functionality.

3D Graphics
While 2D graphics are the most common type in scientific figures, 3D graphics are often useful to describe complicated structures or ideas. 3D graphics are typically prepared by computer-aided design (CAD) software. While the division is not very clear, I roughly divide commonly used 3D CAD software into two categories: engineering CAD and general CAD.

Engineering CAD software is a dedicated program for engineering design & validation that are often incorporated in the curriculum of undergrad/grad programs. While engineering CAD software is not mainly used to make scientific figures, their capability to build 3D models of complicated structures with specific dimensions and design parameters together with photo-realistic rendering function fits perfectly for 3D graphics for scientific figures. There is many engineering CAD software, but I am most familiar with Solidworks and Autodesk Inventor (probably because I am a mechanical engineer). While the upcoming tip posts will focus more on 2D graphics, I will use Solidworks for 3D graphics although the general approach should be similar in different CAD software.

Apart from engineering CAD software, there is a range of general CAD software for more general design purposes including animation, games, etc. One key difference between engineering and general CAD software is the extensive design validation functionality. General CAD software focuses mostly on building 3D models and their photo-realistic rendered images. Among many general CAD software, SketchUp (popular in architecture design) and Autodesk 3DS Max are pretty commonly used for various purposes. Note that these engineering and general CAD software are commercial and very expensive (unless provided by university/institute through volume licensing). Blender is a great open-source free alternative for commercial CAD software for 3D model building and rendering.

​Disclaimer. The contents are my personal opinion and do not represent the view of any institution or company I am affiliated/employed. If you find any incorrect information, please feel free to let me know via my email.

In the era of digital publishing, scientific figures are prepared and published in digital graphics. Hence, it is useful to know the basics on digital graphics relevant to scientific figure preparations. In this tip post, I wish to introduce the most basic concepts about digital graphics including 1) vector vs. raster images, 2) image resolution, and 3) digital image formats. I also find that this guide document from Nature also provides good information regarding the basics of digital graphics.

Vector vs. Raster Images
Although there are tons of different file formats for digital images, there are two big categories of digital images: vector and raster images.

Vector images are digital images whose geometry is defined by points on a Cartesian plane (2D) connected by lines and curves to form polygons (the name vector comes from the fact that it is made of many vector lines to form polygons). The key benefit/feature of vector images is their capability to being magnified without losing detail/resolution – because vector image is defined by the points on a Cartesian plane that can be easily transformed in the same plane without losing information. It’s linear algebra in action! Because of this capability to keep high resolution regardless of image manipulations (magnification, etc) and easiness in edits without losing details, vector images are the most preferred type for scientific figure preparation for schematic illustrations, plots, and other graphical elements other than raster image-based data.

Raster or bitmap images are digital images based on color information on a predefined grid of dots or pixels (the name raster comes from the fact that it is made of dots in a raster or grid). Raster images are the most common type of digital images because image sensors in measurement devices (camera, microscope, CCD, or photodiode), as well as screen capture, mostly generate raster images due to their straightforwardness (just converting pixel information in sensor/screen into raster image). The key disadvantage of raster images is that their resolution is predefined and degrade by magnifications contrary to vector images (see the above image for example).

Image Resolution
Since the major purpose of scientific figures is the delivery of information, high enough resolution in digital images to clearly convey the needed information is a critical requirement for scientific figures. Vector images have an obvious advantage in terms of image resolution as their resolution is independent of magnification. While some journals adopt vector graphics in their PDF version of papers, still the majority of paper figures are published in raster images (in such cases, original vector images are rasterized). Furthermore, many scientific image data are only collected in raster images. Hence, image resolution is an important consideration in scientific figure preparation.

The most common unit of image resolution in publishing is dpi (dots per inch). Dpi is a pretty straightforward unit – higher dots per inch, higher resolution. The above image shows an example of the difference between different dpi for the same image. 300 dpi is the most common resolution for printing (and therefore, for publishing-quality paper figures), giving high resolution where individual dots are not visible in the human eye. 72 dpi is a common choice of resolution for web-based content, but you can see a bit grainy texture due to the low resolution.

Digital Image Formats
There are a lot of digital image formats, but there are a handful of commonly used formats in scientific publishing. For vector images, EPS and PDF formats are commonly used. AI and postscript (SVG) formats are also used widely. For raster images, TIFF format is the most commonly used one for publishing while JPEG, PNG, and GIF are widely used too. Note that vector formats have a lot higher editability as they conserve individual editable elements instead of making the whole figure into a flattened raster image. Therefore, I recommend using vector format until the figure reaches the final version before rasterizing (if needed). Some journals require the submission of vector format figures to allow appropriate copyediting.