We use different techniques to visualize the world around us at multiple speeds, like slow motion and time-lapse photography.
Image credit: arXiv: 2208.11885 [cs.CV]
A recent article on arXiv.org recommends Video Time Pyramid, an approach to visualizing changes at different time intervals and as a visual tool for navigating and exploring pyramid.
The pyramid consists of a collection of new videos, each of which distills the changes that occurred at a specific time period. It allows for a smooth viewing experience with no aliasing or flickering effects. Video Spectrogram plots time versus time scale, showing clear patterns for dramatic cyclical changes such as day/night. Users can scan the overview first and then drill down to a lower time period to see more details.
The researchers validated this method using multiple long-lived webcam datasets of outdoor scenes as diverse as construction sites.
What can we learn about a scene from watching it for months or years? A video shot over a long period of time will depict interesting phenomena in many time ranges, but identifying and viewing them is a challenge. Videos are too long to be viewed in full, and some situations are too slow to experience in real time, such as tape retracting. Timelapse filming is a popular approach to summarizing long videos and visualizing slow intervals. However, the elapsed time is limited to a single selected time frequency, and flickering is often present due to aliasing and temporal discontinuities between frames. In this paper, we propose Video Time Pyramid, a technique that addresses these limitations and expands the ability to visualize the passage of time. Inspired by spatial image pyramids from computer vision, we have developed an algorithm that builds video pyramids in the time domain. Each level of the Video Time Pyramid shows a different time period; for example, videos from a monthly timeframe are generally suitable for visualizing seasonal changes, while videos from a one-minute timeframe are best for visualizing sunrise or the movement of clouds in the sky . To help explore the different levels of the pyramid, we also recommend a Video Spectrum to visualize the amount of activity across the entire pyramid, providing an overview of scene dynamics and searchability. understand and explore phenomena across time and time scales. To demonstrate our approach, we built a Time Pyramid with video from ten outdoor scenes, each containing month or year data. We compare video temporal pyramid layers with naive timelapse and find that our pyramids allow viewing of permanent changes without aliasing. We also demonstrate that Video Spectroscopy facilitates the discovery and discovery of phenomena on pyramidal levels, by allowing both overview and detail-focused views.
Research articles: Swift, ME, Ayers, W., Pallanck, S. and Wehrwein, S., “Visualizing the Passage of Time Using a Temporary Pyramid Video”, 2022. Link: https://arxiv.org/abs/2208.11885
