There is no capacity limit. You can learn unlimited, hundreds, of words per day months on end and retain more than 90% accuracy with multiple choice questions. The only limit is time.
Steve Kaufmann looks at movies only as “rewards” since he is looking for an efficient use of his time, and movies don’t constitute a dense uninterrupted flow of speech, and he is right about that. That being said, stuff like vlogs would be great and a very efficient use of time. The flow is much faster than reading would ever be, you can hear the pronunciation, get visual ques, and listen and read at the same time (it is the same as reading if you have subtitles JUST MORE). Not to mention, that media can be regarded as a natural graded content since vocab needed for that is around 7,000 while vocab for reading literature can span from 9,000 for an easy book and 12,000 to 15,000 for a more richer book.
I find media to be a great transition medium to gradually go from there to bilingual books with audio.
Initial exposure to a lot of audio would also greatly allow you to visualize in your mind the correct pronunciation of the words you read, so it is just a much logical step to start with and transition from.
Starting with novels is not a great idea in my opinion as that is a very tall order, but once you’re advance enough that would be the only way to go, as exposure to such vocab through media would be extremely limited and hence much much more slower to make any additional progress.
Example of dense vlog type content: