[MEncoder-users] Videos with fps 1000.000

[Rich Felker]

On Tue, Mar 20, 2007 at 02:46:32AM -0600, Mike Hodson wrote:
> >My point is that you're not going to get significant increases in
> >image quality by lowering the framerate, because there will be much
> >larger differences to encode, and thus more bits needed for each
> >frame.
> Yes, but sometimes you need to slow way down to keep something
> readable. From 30fps motion of someone talking, to maybe 2-5fps of
> some computerised text, thats acceptable imho.

If you're talking about computerized text appearing a character at a
time or such, you will not save any bits by omitting frames unless the
codec royally sucks, The amount of information to be encoded is the
same.

On the other hand if you're talking about fast motion (panning,
perspective change, etc.) while text is visible and you need to ensure
readability of the text, you may be at least partly correct. I still
would never watch 2-5 fps content...it's just too painful.

> >This is possible with almost any container, including even AVI, and
> >done in practice quite often.
> Perhaps I am just unaware of how to properly do this, as Ive been
> wanting to properly encode some anime dvds of mine without the
> blurring or choppiness of 24/dropping CG or 30/having the 5th frame
> blur)

-vf pullup will output the right frame sequence, but mencoder does not
have good vfr output support, so the output file won't be quite what
you'd want... :( Maybe eventually we'll get it fixed.

> >No one using a 1/120 time base is encoding 120 frames per second. If
> >you see "120 fps", it's just an application misinterpreting a time
> >base as a framerate.
> Again, my ignorance to the specifics of file formats and whatnot is
> probably showing here. I was always under the impression there would
> be some sort of image data stored for every frame in the file, and
> that given enough of this extra data it would greatly increase
> filesize (megabytes worth) if encoded that way.

It's true that something is stored for each _frame_. The error in your
reasoning is that it's not 120 frames per second. It's 120 timer ticks
per second. AVI does have some degree of wasted overhead per timer
tick too. Not horrible, but not irrelevant either. Matroska and NUT of
course do not have any such waste.

Rich