0

Could one make some general assumptions about size and perceived quality with regards to I, B and P frames.

My first presumption would be something like:

Quality(I) > Quality(B) > Quality(P)

and

Size(I) > Size(B) > Size(P)

By Size I mean the frame size in bytes.

Let's keep it simple and consider one slice per frame.

holzkohlengrill
  • 1,044
  • 19
  • 29

1 Answers1

1

Generally speaking I is larger than P is larger than B. This does not need to be true, but generally is.

Now, you can not compare "quality" between the types. A P frame has only stores the differences between itself and it reference(s). A decoded P frame may improve the fadility of its reference, or it may reduce it depending on the information delta, and how may bits were used to encode that information.

szatmary
  • 29,969
  • 8
  • 44
  • 57
  • Thanks for this great answer. If I understand correctly the B-frame was primary invented aiming at a file size reduction while maintaining the same quality rather than improving overall quality? – holzkohlengrill Jul 14 '16 at 08:37
  • @holzkohlengrill improving quality at a given bitrate, or reducing bitrate for a given quality are the same thing. Every single feature/option added to a codec it there to support this goal. Or to reducing encoding cost. – szatmary Jul 14 '16 at 14:55