Generative video prediction models that directly predict pixel values produce blurry outputs because they cannot enumerate the ~10^15 million possible next frames the way language models enumerate discrete tokens, forcing them to average pixel values across uncertain outcomes.
causalpending
Evidence Quote
“For full HD video in the most general case, each pixel can take on 256 discrete values... meaning there are something like 10 to the power of 15 million possible next video frames”
Created: 8/12/2026, 6:35:53 PM
Your notes
Loading notes...