How to generate videos

The actual pipeline (verified)

BioReactor-video does not produce an MP4 by itself — it dumps raw binary frames (grid + VOF field per timestep) to run_dir/frames/. render_videos.py is the separate step that actually renders and encodes them:

make build-video
build/BioReactor-video runs/my_run/params.json
uv run python scripts/render_videos.py runs/my_run

Needs ffmpeg on PATH (module load ffmpeg on OSCAR) and Basilisk's own gl headless-rendering libs to have linked when build-video was compiled (libgl1-mesa-dev libglu1-mesa-dev on Debian/Ubuntu — see Setup).

Produces, in runs/my_run/:

File	Content
volume_fraction.mp4	VOF interface, body frame (rotates with the bag)
volume_fraction_lab.mp4	Same field, lab frame (fixed camera)

render_videos.py cleans up frames/ once it's done. See Your first simulation for a real example, including the actual rendered output.

When this runs automatically

chain.py automates the whole build→run→render sequence for every segment when a chain config sets videos: true — it switches to config/slurm_video_template.sh, which calls render_videos.py itself at the end of the SLURM job. For a one-off local run (as above), you run the two steps yourself.

The other video pathway (not verified here)

scripts/submit_video_run.py + config/slurm_video_template.sh exist as a second, standalone way to submit a single video run via SLURM directly. It calls the exact same render_videos.py step, so it should produce the same volume_fraction*.mp4 files — but that specific script hasn't been run end-to-end while writing this page, unlike the direct-invocation path above.