It seems like an oversimplification to imply all processing exists on a gradient with no significant lines we can draw between one type and another. E.g. conten

It seems like an oversimplification to imply all processing exists on a gradient with no significant lines we can draw between one type and another. E.g. content identification has a pretty clear qualitative difference from whatever "dumb" mappings do to get from RAW to values in some color space. One isn't just "more processing" and the other "less processing", colloquially.