Visualizing VR Data

Post by Wells Johnston

Virtual reality has the potential to revolutionize the content industry, and with it, we have uncovered a whole new way to analyze human behavior as audiences consume media. Consumer media analytics have always been about understanding demographics, calculating ad revenue, and gauging performance. The new VR medium introduces an exciting opportunity to explore and understand how people engage with content.

Unlike traditional video, where the viewport is a flat surface, virtual reality allows us to track where people look. This generates an astonishing amount of data, representing everyone’s experience in a piece of VR content. At Littlstar, we have built a VR analytics platform to harness this data, and we are excited about exposing it in innovative ways. Our goal is to make our data as accessible and meaningful as possible, and provide unprecedented access to understanding and even monetizing user engagement.

One way we have exposed our data is through heatmap visualizations of where people looked in VR. Using color gradients overlaid on the original 360 video (or VR scene), we are able to represent the density of viewers looking at each area throughout the video. The result is a beautiful, fluid heatmap visualization that demonstrates how people engaged with the scene.

World of Tanks heatmap

“The technology, and more importantly, the data it unearths, could be a significant catalyst for entertainment studios, production firms and brands.”
- Tim Peterson, Marketing Land

To further analyze this data, we use a statistical method called cluster analysis to find what we call “hotspots”, which are essentially areas of interest throughout any immersive experience. Performing these calculations and finding hotspots involves so much data and processing that it needs to be distributed over a number of machines in the cloud, and the computation takes hours. After cluster analysis, we further analyze the type of movement people exhibit at these hotspots, and can make conclusions about the audience’s cognitive response. The opportunities to explore this type of data are endless, and we are just getting started.

The above beach scene lacks focus, therefore no clear hotspot is detected.
ABC recently partnered with Lexus to film a scene in VR for it’s hit primetime series “Quantico.” 
As you can see, the first scene (above) lacks focus, so people look everywhere and no hotspot is detected. In the second image, shot by ABC, we see people converge on the Lexus logo, which is determined to be a “hotspot.”

One industry that will be greatly affected by VR analytics is marketing and advertising. Since we know where people are looking in virtual reality, we can determine what people see. This information enables a platform for identifying objects in VR, what people looked at them, and their behavior afterwards. On a large scale, this data becomes meaningful. Say a brand wants to know what percentage of people saw an ad placement, or how different genders responded to a scene; our platform makes this kind of analysis possible.

Another scene from ABC’s Quantico

“…here is a really exciting way to be able to showcase exactly what the engagement, what the viewability is of the sponsor’s assets.”
- Jeffrey Weinstock, VP and Creative Director for ABC’s Integrated Marketing department

Creating an immersive narrative in VR is a big challenge, especially since viewers can divert their attention from the story so easily. As the heatmap above demonstrates, slight movement to the right of the scene triggers almost half of the viewers to follow in that direction. Similarly, when the screen turns on, you can see everyone quickly shift their focus. Behavior in VR can be studied this way, and we hope our platform will help guide the storytelling process for future content.

If you are a virtual reality content creator, you now have access to detailed analytics for all of your content on Littlstar.

Wells Johnston is the Chief Data Scientist at Littlstar. He leads research and development of Littlstar’s VR analytics platform.