From Veo to Motion Transfer: Building Demo Videos for a BJJ Training App
TL;DR: My BJJ coach told me daily HIIT was the fastest way to build match stamina, so I built BJJ FLOW — a solo-training app for Brazilian Jiu-Jitsu with BJJ-specific drills, a mat-side timer, demo videos for every move, and fully offline voice guidance. The plan was to generate all the demo videos with AI. It worked for push-ups. It completely failed for the hip escape — the most basic movement…
My BJJ coach advised me that daily HIIT sessions would best build stamina for match conditions. Consequently, I developed BJJ FLOW, an app tailored to Brazilian Jiu-Jitsu training. The app features BJJ-specific drills, a mat-side timer, demo videos for each move, and offline voice guidance. Initially, I intended to generate all the demo videos using AI, but this approach failed for the hip escape, a fundamental BJJ movement. The AI model had never witnessed this motion and could not replicate it.
To overcome this challenge, I decided to film myself executing every movement and then use Wan 2.2 to transfer the motion to a virtual athlete. This required capturing multiple source videos of my hip escapes, sit-throughs, breakfalls, and stretches from various angles at the dojo and at home. Once the videos were filmed, I employed motion transfer techniques using Wan 2.2 to generate the demo videos for each move in the app.
The process involved stabilizing the footage, extracting a person mask, and using OpenCV and ffmpeg for post-processing to create the final animated videos.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.