I'm curious what exactly they mean when they say Live Photos are 'not a movie, it's a photo!'
So what's the file format? How does one encode sound to a moving photo and not call it a movie?
- It's the full photo data (12 MP) for each frame
- It captures 1.5 seconds before and after you press the capture button
- (Speculation based on a dot-point in the keynote) It's stored using some frame-to-frame compression thing, probably similar to a movie but with the option to recover any individual frame as a full size, full quality image
I reckon much of their photo technology comes from their acquisition of the SnappyCam app and that developer's custom JPEG encoding.
This, but they missed the opportunity to call it a single I-Frame with only P-frames in each direction, if they actually did some effort in terms of compression efficiency.
Another thing is that any 4k capable codec is also probably magnitudes more space efficient than baseline jpeg for stills, with or without added temporal information.
Sound as metadata, multiple frames in a single file pointed to by metadata? First "snapshot" frame the first, so works with normal software? Lotsa ways to play games.