Visualizing Binary Data

Original visualization and Author’s comments isn’t so much a statement as an exploration of what files on computers “look like”. In essence, it takes a binary file containing arbitrary data and generates an interactive visualization which allows users to inspect regions of the file for values such as byteclass (useful for distinguishing text from other data, for example), local entropy, and byte-level details. Admittedly, the target audience of this presentation is likely technically-minded, but the key ideas apply to a lot of areas. To keep it short, I’ll go over just two of them.


The first is the representation of parametrized data in 2D. Many visualizations try to present multidimensional data in 2D, often using 3D computer graphics and animation. This visualization tackles the opposite problem: presenting a 1D sequence of points in 2D. The author exploits spatial locality by clustering contiguous regions of similar type, but also allows simple sequential display of the data on the grid. He also explains the techniques he used right in the help menu.

The second feature is that information isn’t withheld. For credibility, a presentation should be verifiable by giving sources or making the raw data available. This visualization takes this idea a step further by showing the raw data directly alongside the visualization, albeit in readable chunks. An even cooler feature is that the data which the user is focusing on is highlighted in each of the components (see the screenshot above). This creates a visual mapping between the raw data and the presentation, a powerful technique for showing connections across levels of abstraction and inviting users to ask their own questions.