Here to show off a project that took 18 months, 7KLOC, and more importantly, 194 kilometres of walking! In a nutshell, I wanted to make the famous silhouette animation “Bad Apple” with Strava, which is an exercise tracking app that visualizes where you walk/run on a map. To use the earth as my canvas, I had to do a lot of coding in the prep work to plan where I would walk for each frame of the animation. As I explain in the video, my problem solving/coding steps were something like this:
- split video into frames, determine a good framerate - detect contours with OpenCV for each frame. Since frames are very diverse, I had to start with some heuristics for choosing contours and ultimately make a GUI to manually curate all 438 of them (e.g. some frames had many small contours while others had a few important ones, and the video is constantly inverting colours). Output of this work was an intermediate json list of x,y coordinates representing the path along contours - convert these cartesian coordinates into the Universal Transverse Mercator coordinate system, which represents the earth in a cartesian system (as opposed to GPS which doesn’t lend to adding relative distances in degrees very well) - convert UTM to GPS using a package, then to the GPX file format which is universally accepted - Import GPX files into an app that shows where I am and where I need to walk (I used Gaia GPS). Then I turned on Strava in the background and walked for ~18 months until this was done.
I also had to make a bunch of custom software to visualize these walks and make montages of them, like a script to pull polylines from the Strava API, a custom polyline renderer, a script to censor my strava activities a bit, and much more. I’m so proud of this work, it feels like it stands out in a sea of code-only bad apple projects that have become easier to automate with AI, and I hope it can go down as one of the greats in the bad apple colossus of work!
Video: https://youtu.be/HxdQawDIlnU GitHub: https://github.com/kevinlinxc/badapple-strava
Happy to answer any questions!