Skip to content

Instantly share code, notes, and snippets.

@LeFreq
Created March 24, 2021 01:47
Show Gist options
  • Save LeFreq/f12b398b656de8b47a4adc26612d78b8 to your computer and use it in GitHub Desktop.
Save LeFreq/f12b398b656de8b47a4adc26612d78b8 to your computer and use it in GitHub Desktop.
The deltas of successive frames should give decent cmopression (~10%), but you can conceivably get log2 n compression with categorization of your video data.
To acheieve these high-rates of compression you need to categorize the seen as sets within sets. A "car" is in the set "street" is in the set "city". Each element of a set can have a further refiinement to a set of "qualities". The element "car" can have the qualities "suburu", "deep green", "open", "in motion", etc.)
Highest bit of categorization (say 32 bits for 4 billion different object in the world) is natural vs. civilization. Consider the "chirality" of the data another 2-bifurcation. For exaple man-make objects made with natural objects or man-made objects in the context of nature.
Nature -> (landscape, aquatic, air&space, ) -> landscape (trees, rivers, mountains, seascapes) -\
This is a start. Someone can take this and use it as they wish, CC-A.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment