[FFmpeg-user] Changing DAR ?

Ronald F. Guilmette rfg at tristatelogic.com
Tue Oct 8 21:11:41 CEST 2013


In message <CAPCGgdS_JFurFUCyyUJ8vX8bxZvSUiqa8BxrRFrQy7gGkBbC-g at mail.gmail.com>
Anatol <anatol2002 at gmail.com> wrote:

>Check:
>http://ffmpeg.org/ffmpeg-filters.html#setdar_002c-setsar

Thank you.  The information I found there has proven to be somewhat
helpful, however...

1)  It's relly too bad that a link to the page you just pointed me to
isn't found under the description of the -vf option on this page:

     http://www.ffmpeg.org/ffmpeg.html#Video-Options

or even here:

      http://www.ffmpeg.org/ffmpeg.html#filter_005foption

2)  The descriptions of both setdar= and setsar= sub-options at the location
you just pointed me to are extraordinarily unclear.  More specifically,
what, exactly is that equation near the beginning supposed to represent?

The verbage at the top of this section seems to be saying that setdar=
will cause the *SAR* to be changed!

      "The setdar filter sets the Display Aspect Ratio ...

      This is done by changing the specified Sample (aka Pixel) Aspect
      Ratio..."

Huh?  Obviously, what is said here is not what was actually meant.  And
also, of course, the true meaning and significance of the equation shown
is nowhere explained.

Then there is this:

   The filters accept the following options:

   'r, ratio, dar (setdar only), sar (setsar only)'

Huh??  Is that a literal 'r'?  Should that be written with or without the
surrounding single quotes?

My impression is that 'r' in this context is supposed to represent an
actual literal (floating point?) number, such as "1.7777", but that sure
is not clear.  Likewise, I guess that "ratio" in this context really means
something like 16/9, however even that is quite confusing.  Earlier, I had
tried using setdar=16:9 because that (colon) notation seems to be supported
by other parts of ffmpeg.  However I got neither any error nor even any
warning, as far as I could see, and yet there was no effect produced when
I tried using -vf setdar=16:9".  So I guess that ffmpeg demands different
notations for the same ratio in different context, yes?

Nothing like a bit of inconsistancy to confuse the users!

I also have no idea what context, if any, the 'max' sub-parameter might be
useful in.  Do people ever really try something foolish like this?

      -vf setdar=10000/1

And if they do, and things go haywire, don't they get what they deserve?

Last but not least, we have these two examples in the following section:

    setdar=dar=16/9
    setdar=ratio=16/9:max=1000

Now I am totally flumoxed!  What would be the difference between these
two video filters?

     setdar=dar=16/9
     setdar=ratio=16/9

and also, what would happen if I just said:

	-vf setdar=16/9
?

(I believe this last thing above is actually what I tried to use in my
earlier experiments.  I don't remember seeing any errors or warning when
trying to use that.  Is it in fact erroneous?)

Obviously, "-vf" is an option for ffmpeg.  To my way of thinking, that
makes "setdar=" what I would call a "sub-option".  But now it appears
that that sub-option has its own specific, special, and peculiar set of
sub-sub-options (e.g.  "dar", "ratio", "max").

To say that this is all made somewhat less than clear by the documentation
would be an understatement, I think.


More information about the ffmpeg-user mailing list