The One Punch Man Anime Dataset on Kaggle aligns 14,629 detected shots with 259,129 timestamped comments from live chat, YouTube comments and MyAnimeList episode threads across 36 episodes. Anime analysis usually studies the animation or what the audience said about it, almost never both on the same clock.
14,629 shots on one side, 259,129 scored comments on the other
The dataset pairs OpenCV shot detection with live chat replay logs, cleaned and timestamped.
Vision side: shot boundaries and per-shot motion intensity, so the visual pace of an episode can be measured across a season.
Language side: VADER sentiment and NRC emotion labels already computed, so nobody has to rerun inference.
Personally identifiable information is hashed before anything ships, so the dataset does not republish who said what.
Every One Punch Man piece on this site reads from it
Aligned, the two sides show what viewers reacted to second by second, and what the animation was doing at that moment.
The immersion analysis, the character race and the animation quality piece all use this dataset.
Publishing the data next to those conclusions means someone can check them.
