Hacker News new | comments | show | ask | jobs | submit login

Hmmm... I wonder if there's potential to use this concept for video. Not this implementation, naturally, but the concept.

I'm a little too sleep deprived to think through this entirely, but just off the top of my head, it seems like over the course of a few frames that an edge in motion would wind up as a series of triangles where one of the points remains static while the other two shift away from it -- in other words, the Trigrad approach would yield motion blur as an artifact. And then when the next key frame comes along, the remaining point gets reselected, probably further along the motion path... So much like a normal differential approach you wouldn't need to store all of the point locations each frame, just the ones that change and which triangle they belong to.

It might be hard to make it stream-friendly though, since obviously the compression efficiency depends heavily on the storage structure (see pjtr's comments)...




There's been some work on encoding video with triangles. See "Video Compression Using Data-Dependent Triangulations" (http://www-home.htwg-konstanz.de/~umlauf/Papers/cgv08.pdf) for example.




Applications are open for YC Winter 2018

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact

Search: