The Archive of Player Experience is a collection of visualizations of different players’ playthroughs of particular games. The Archive utilizes ImagePlot, open-source software developed by the Software Studies Initiative, to extract images from playthrough videos and plot them according to characteristics of the images. In its initial stage, the imageplots included are basic plots of images according to time, effectively providing a distance visualization of a given player’s experience with the game.

This project is meant to reveal where, how, and how much players’ experiences of a game differ. In particular, the imageplots allow for an analysis of game narrative that accounts for variation, interactivity, and difference, enabling the assessment of how linear or multilinear a given game’s narrative structure is. This method can also account for individual players’ circumstances and contexts, allowing scholars to consider differences in players’ personal experiences of the games.

While these are the initial questions of the project, the sort of distant visualization of games that ImagePlot allows for could lead to many other questions, such as comparisons across genre, time period, or other measure for comparison.

What Is ImagePlot?

ImagePlot is open-source software that takes a collection of images (even collections in the thousands and tens of thousands) and plots them on a basic x-axis, y-axis graph according to qualities of the images. It does so using the images themselves, and a .csv file containing numerical measures of things such as hue, saturation, or date. These measures can be generated using the related ImageMeasure macro or similar macros. You can even plot images according to time, as with the imageplots of game playthroughs included in this project.

ImagePlot and ImageMeasure are macros (basically, scripts/strings of code used by another piece of software) for the image software ImageJ, maintained by NIH and often used to analyze large collections of images (such as x-rays or brain images) in medical settings and the sciences.

So, basically, ImagePlot uses ImageJ to read a .csv file and organize images according to the measures that file contains.

How To Read ImagePlots

When interpreting and analyzing images using ImagePlot, it’s critical that you know which measures are being plotted in the imageplot, meaning which measures correspond with the x-axis and y-axis. Without knowing which measures are plotted to which axes, it will be difficult to assess what the imageplot tells you.

For example, the imageplots currently included in the Archive of Player Experience are basic plots of time on the x-axis, and a randomly generated y-axis. This means that each imageplot of a game playthrough will have the earliest time (the beginning of the video) on the left, and the latest time (the end of the video) at the right. ImagePlot plots time according to the filename of the images, with the first image of the video being file 1, and each image thereafter increasing by one (essentially, first image is file 1, second is file 2, etc.). However simply plotting time–associated with filename–on the x-axis would result in all of the images being stacked on top of each other in a single line, making them difficult to read. To address this issue, these imageplots utilize a random y-axis: basically, each file is also assigned a random y-value between 1 and 50, so when the images are plotted they won’t be stacked on top of each other.

The result is an imageplot that can be read left to right, visualizing the entire playthrough of a game. The beginning of the game is on the left, the end of it is on the right, and any images at the same x-axis location on the graph are from the same part of the game.


ImagePlot provides a method for visualizing entire playthroughs of games, and for making comparisons between those playthroughs much easier thanks to allowing a kind of distance-reading or bird’s-eye view of the playthroughs. That said, there are several important limitations to this method:

  1. The random y-axis sometimes leaves gaps in the imageplot where no image was assigned a particular y-value. This becomes less common with larger quantities of images (which game playthroughs often have thanks to their length) where it is more likely that an image will be assigned every y-value.
  2. It is possible to zoom in and see individual images, including what was happening at particular moments, but this is entirely dependent on the quality of the images. A site like this one uses JPEGS, which are not the highest quality, and limit what one can see with zooming in.
  3. While the imageplots themselves can reveal a lot about the structure and variance present in a game, one needs to reference the original videos and collect more data from them in order to make meaningful comparisons. For example, it’s helpful to go back to the original videos and get specific times and durations in order to help compare across videos and imageplots with more than just eyeballing-it precision.