Lecture 5: Image Formats

Brian J. Smith

2026-01-28

My first big first-author paper

Smith et al. (2016) Biological Invasions


I assumed that after the paper was accepted, the production team would let me know if they thought the size or the aspect ratios of my figures were awkward.


They didn’t.

Digital Image Formats

Digital Image Formats

Raster

  • By far, the most common digital image format
  • Composed of picture elements (“pixels”)
  • Discrete grid of pixels contains color or intensity information
  • Fixed resolution, determined by pixel size
  • AKA bitmap, referring to a spatial arrangement of bits

 

Vector

  • Far less common format
  • Composed of mathematical geometry
  • Continuous space
  • Scales well to any resolution thanks to mathematical definitions unrelated to pixels

Digital Image Formats

Raster

  • JPEG
  • GIF
  • PNG
  • TIFF
  • etc.

 

Vector

  • SVG
  • PDF
  • EPS
  • etc.

Image Resolution


A raster image’s resolution describes how much detail it has.


We often express resolution using pixel count, expressed either as a single value (e.g., 2.1 megapixels) or as a pair of numbers showing width and height (e.g., 1920 x 1080).

Image Resolution


For example a “full HD” image has dimensions 1920 x 1080.

If you are familiar with megapixels, that’s:

\[1920 \times 1080 = 2,073,600 \approx 2.1\,\mathrm{Megapixels}\]

Image Resolution

While pixels ultimately describe the resolution of an image, whether or not the image appears to be high-quality depends on how large it is displayed. If you display an image at a certain physical size, you can refer to the resolution in:

  • DPI (dots per inch; a term related to dots of ink in a printed image)
  • PPI (pixels per inch; technically more correct on screen).

You will often see DPI/PPI used interchangeably.

Image Resolution

Some common DPI values:

DPI Use
72 Standard Mac display
96 Standard Windows display
300-400 Modern Apple Retina HD display
300-720 Inkjet printer
600-2400 Laser printer

File Size

Vector image file sizes are determined by the number of vectors.


Uncompressed, raster image file sizes are related to the image’s resolution and color depth.


Color depth is the number of bits used to indicate the color of a single pixel and is measured in bits per pixel (bpp).

File Size

If that image had 24-bit color (web standard), its file size would be:

\[1920 \times 1080 \times 24 = 49,766,400\,\mathrm{bits} \times \frac{1\,\mathrm{bit}}{8\,\mathrm{bytes}} = 6,220,800\,\mathrm{bytes}\approx5.9\,\mathrm{MiB}\]

Color Depth

The number of possible colors for \(n\)-bit color depth is \(2^n\).

Bit Depth Bits per pixel Examples
1-bit \(2^1=2\) Usually black-and-white
2-bit \(2^2=4\) Many early computer systems; original Game Boy
3-bit \(2^3=8\) RGB; early home computers with TV displays
8-bit \(2^8=256\) 1980s PCs; modern smartwatches; Mars Exploration Rovers1
15-bit \(2^{15}=32,768\) Super Nintendo, Game Boy Color
24-bit \(2^{24}\approx16M\) “True color”2; modern computer and phone displays

Image Compression


Data compression is the process of encoding information using fewer bits that the original representation.


Image compression is data compression optimized for image data by taking advantage of visual perception or statistical properties of the data.

If you’re interested in image compression, the Wikipedia page Image compression has a lot of detail.

Image Compression


Image compression can be lossless or lossy.


Lossy compression usually (but not always) results in smaller file sizes, but it can introduce noticeable artifacts.

Image Compression

Photo taken by Brian on a Nikon ZIII Mirrorless Camera

2048 x 1365 pixels (2.8 Megapixels)

(Note most web browsers (except Safari) don’t display TIFFs.)

Compression Image File Size
JPEG, 100 Quality JPEG, full quality 0.99 MiB
JPEG, 10 Quality JPEG, high compression 88.2 KiB
TIFF, No compression TIFF, no compression 8.03 MiB
TIFF, LZW compression TIFF, LZW compression 2.88 MiB

Scientific Publishing


Most journals have explicit figure guidelines in their “Instructions for Authors” material.


For example, Science has an entire figure guideline PDF.


They require images to be at least 300 PPI, they prefer vector over raster graphics, and they give exact measurements for one-column and two-column figures.

R Graphical Devices

R Graphical Devices


You can read about graphical devices in R on the help page.

?Devices

# OR

help("Devices")


You can also find the help page rendered on the web here.

R Graphical Devices

You will notice that the help file indicates that many of the devices are platform-specific. You can check what devices are available on your current system 1:

capabilities()
       jpeg         png        tiff       tcltk         X11        aqua 
       TRUE        TRUE        TRUE        TRUE       FALSE       FALSE 
   http/ftp     sockets      libxml        fifo      cledit       iconv 
       TRUE        TRUE       FALSE        TRUE       FALSE        TRUE 
        NLS       Rprof     profmem       cairo         ICU long.double 
       TRUE        TRUE        TRUE        TRUE        TRUE        TRUE 
    libcurl 
       TRUE

R Graphical Devices

Additional graphical devices can be accessed through packages. I often use the AGG library through the package ragg (https://ragg.r-lib.org/).


ragg claims to be faster (I haven’t tried) and higher quality (I can confirm) than base R graphics (grDevices).


It is also system-independent, so you should get the same output from Windows, Mac, or Linux.

Code Demo

Questions?



BCB5200 Home